This repository has been archived on 2023-10-09. You can view files and clone it, but cannot push or open issues or pull requests.
Files
blender-archive/source/blender/editors/space_file/filelist.c

2687 lines
77 KiB
C
Raw Normal View History

/*
* ***** BEGIN GPL LICENSE BLOCK *****
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version 2
* of the License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software Foundation,
2010-02-12 13:34:04 +00:00
* Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
*
* The Original Code is Copyright (C) 2007 Blender Foundation.
* All rights reserved.
*
* The Original Code is: all of this file.
*
* Contributor(s): none yet.
*
* ***** END GPL LICENSE BLOCK *****
*/
2011-02-27 20:29:51 +00:00
/** \file blender/editors/space_file/filelist.c
* \ingroup spfile
*/
/* global includes */
#include <stdlib.h>
#include <math.h>
#include <string.h>
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
#include <sys/stat.h>
#include <time.h>
#ifndef WIN32
# include <unistd.h>
#else
# include <io.h>
# include <direct.h>
#endif
#include "MEM_guardedalloc.h"
#include "BLI_blenlib.h"
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
#include "BLI_fileops.h"
#include "BLI_fileops_types.h"
#include "BLI_fnmatch.h"
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
#include "BLI_ghash.h"
#include "BLI_hash_md5.h"
#include "BLI_linklist.h"
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
#include "BLI_math.h"
#include "BLI_stack.h"
#include "BLI_task.h"
#include "BLI_threads.h"
#include "BLI_utildefines.h"
#ifdef WIN32
# include "BLI_winstuff.h"
#endif
#include "BKE_context.h"
#include "BKE_global.h"
#include "BKE_library.h"
#include "BKE_icons.h"
#include "BKE_idcode.h"
#include "BKE_main.h"
#include "BLO_readfile.h"
#include "DNA_space_types.h"
#include "ED_datafiles.h"
#include "ED_fileselect.h"
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
#include "ED_screen.h"
#include "IMB_imbuf.h"
#include "IMB_imbuf_types.h"
#include "IMB_thumbs.h"
#include "PIL_time.h"
#include "WM_api.h"
#include "WM_types.h"
#include "UI_resources.h"
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
#include "UI_interface_icons.h"
#include "atomic_ops.h"
#include "filelist.h"
/* ----------------- FOLDERLIST (previous/next) -------------- */
typedef struct FolderList {
struct FolderList *next, *prev;
char *foldername;
} FolderList;
ListBase *folderlist_new(void)
{
ListBase *p = MEM_callocN(sizeof(*p), __func__);
return p;
}
void folderlist_popdir(struct ListBase *folderlist, char *dir)
{
const char *prev_dir;
struct FolderList *folder;
folder = folderlist->last;
if (folder) {
/* remove the current directory */
MEM_freeN(folder->foldername);
BLI_freelinkN(folderlist, folder);
folder = folderlist->last;
if (folder) {
prev_dir = folder->foldername;
BLI_strncpy(dir, prev_dir, FILE_MAXDIR);
}
}
/* delete the folder next or use setdir directly before PREVIOUS OP */
}
void folderlist_pushdir(ListBase *folderlist, const char *dir)
{
struct FolderList *folder, *previous_folder;
previous_folder = folderlist->last;
/* check if already exists */
if (previous_folder && previous_folder->foldername) {
if (BLI_path_cmp(previous_folder->foldername, dir) == 0) {
return;
}
}
/* create next folder element */
folder = MEM_mallocN(sizeof(*folder), __func__);
folder->foldername = BLI_strdup(dir);
/* add it to the end of the list */
BLI_addtail(folderlist, folder);
}
const char *folderlist_peeklastdir(ListBase *folderlist)
{
struct FolderList *folder;
if (!folderlist->last)
return NULL;
folder = folderlist->last;
return folder->foldername;
}
int folderlist_clear_next(struct SpaceFile *sfile)
{
struct FolderList *folder;
/* if there is no folder_next there is nothing we can clear */
if (!sfile->folders_next)
return 0;
/* if previous_folder, next_folder or refresh_folder operators are executed it doesn't clear folder_next */
folder = sfile->folders_prev->last;
if ((!folder) || (BLI_path_cmp(folder->foldername, sfile->params->dir) == 0))
return 0;
/* eventually clear flist->folders_next */
return 1;
}
/* not listbase itself */
void folderlist_free(ListBase *folderlist)
{
if (folderlist) {
FolderList *folder;
for (folder = folderlist->first; folder; folder = folder->next)
MEM_freeN(folder->foldername);
BLI_freelistN(folderlist);
}
}
ListBase *folderlist_duplicate(ListBase *folderlist)
{
if (folderlist) {
ListBase *folderlistn = MEM_callocN(sizeof(*folderlistn), __func__);
FolderList *folder;
BLI_duplicatelist(folderlistn, folderlist);
for (folder = folderlistn->first; folder; folder = folder->next) {
folder->foldername = MEM_dupallocN(folder->foldername);
}
return folderlistn;
}
return NULL;
}
/* ------------------FILELIST------------------------ */
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
typedef struct FileListInternEntry {
struct FileListInternEntry *next, *prev;
char uuid[16]; /* ASSET_UUID_LENGTH */
int typeflag; /* eFileSel_File_Types */
int blentype; /* ID type, in case typeflag has FILE_TYPE_BLENDERLIB set. */
char *relpath;
char *name; /* not striclty needed, but used during sorting, avoids to have to recompute it there... */
BLI_stat_t st;
} FileListInternEntry;
typedef struct FileListIntern {
ListBase entries; /* FileListInternEntry items. */
FileListInternEntry **filtered;
char curr_uuid[16]; /* Used to generate uuid during internal listing. */
} FileListIntern;
#define FILELIST_ENTRYCACHESIZE_DEFAULT 1024 /* Keep it a power of two! */
typedef struct FileListEntryCache {
size_t size; /* The size of the cache... */
int flags;
/* Block cache: all entries between start and end index. used for part of the list on diplay. */
FileDirEntry **block_entries;
int block_start_index, block_end_index, block_center_index, block_cursor;
/* Misc cache: random indices, FIFO behavior.
* Note: Not 100% sure we actually need that, time will say. */
int misc_cursor;
int *misc_entries_indices;
GHash *misc_entries;
/* Allows to quickly get a cached entry from its UUID. */
GHash *uuids;
/* Previews handling. */
TaskPool *previews_pool;
ThreadQueue *previews_todo;
ThreadQueue *previews_done;
double previews_timestamp;
int previews_pending;
} FileListEntryCache;
/* FileListCache.flags */
enum {
FLC_IS_INIT = 1 << 0,
FLC_PREVIEWS_ACTIVE = 1 << 1,
};
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
typedef struct FileListEntryPreview {
char path[FILE_MAX];
unsigned int flags;
int index;
ImBuf *img;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
} FileListEntryPreview;
typedef struct FileListFilter {
unsigned int filter;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
unsigned int filter_id;
char filter_glob[64];
char filter_search[66]; /* + 2 for heading/trailing implicit '*' wildcards. */
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
short flags;
} FileListFilter;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
/* FileListFilter.flags */
enum {
FLF_HIDE_DOT = 1 << 0,
FLF_HIDE_PARENT = 1 << 1,
FLF_HIDE_LIB_DIR = 1 << 2,
};
2012-06-06 22:38:39 +00:00
typedef struct FileList {
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
FileDirEntryArr filelist;
short prv_w;
short prv_h;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
short flags;
short sort;
FileListFilter filter_data;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
struct FileListIntern filelist_intern;
struct FileListEntryCache filelist_cache;
/* We need to keep those info outside of actual filelist items, because those are no more persistent
* (only generated on demand, and freed as soon as possible).
* Persistent part (mere list of paths + stat info) is kept as small as possible, and filebrowser-agnostic.
*/
GHash *selection_state;
short max_recursion;
short recursion_level;
struct BlendHandle *libfiledata;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
/* Set given path as root directory, may change given string in place to a valid value. */
void (*checkdirf)(struct FileList *, char *);
/* Fill filelist (to be called by read job). */
void (*read_jobf)(struct FileList *, const char *, short *, short *, float *, ThreadMutex *);
/* Filter an entry of current filelist. */
bool (*filterf)(struct FileListInternEntry *, const char *, FileListFilter *);
} FileList;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
/* FileList.flags */
enum {
FL_FORCE_RESET = 1 << 0,
FL_IS_READY = 1 << 1,
FL_IS_PENDING = 1 << 2,
FL_NEED_SORTING = 1 << 3,
FL_NEED_FILTERING = 1 << 4,
};
#define SPECIAL_IMG_SIZE 48
#define SPECIAL_IMG_ROWS 4
#define SPECIAL_IMG_COLS 4
enum {
SPECIAL_IMG_FOLDER = 0,
SPECIAL_IMG_PARENT = 1,
SPECIAL_IMG_REFRESH = 2,
SPECIAL_IMG_BLENDFILE = 3,
SPECIAL_IMG_SOUNDFILE = 4,
SPECIAL_IMG_MOVIEFILE = 5,
SPECIAL_IMG_PYTHONFILE = 6,
SPECIAL_IMG_TEXTFILE = 7,
SPECIAL_IMG_FONTFILE = 8,
SPECIAL_IMG_UNKNOWNFILE = 9,
SPECIAL_IMG_LOADING = 10,
SPECIAL_IMG_BACKUP = 11,
SPECIAL_IMG_MAX
};
2012-06-18 13:01:24 +00:00
static ImBuf *gSpecialFileImages[SPECIAL_IMG_MAX];
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static void filelist_readjob_main(struct FileList *, const char *, short *, short *, float *, ThreadMutex *);
static void filelist_readjob_lib(struct FileList *, const char *, short *, short *, float *, ThreadMutex *);
static void filelist_readjob_dir(struct FileList *, const char *, short *, short *, float *, ThreadMutex *);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
/* helper, could probably go in BKE actually? */
static int groupname_to_code(const char *group);
static unsigned int groupname_to_filter_id(const char *group);
static void filelist_filter_clear(FileList *filelist);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static void filelist_cache_clear(FileListEntryCache *cache, size_t new_size);
/* ********** Sort helpers ********** */
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static int compare_direntry_generic(const FileListInternEntry *entry1, const FileListInternEntry *entry2)
{
2010-04-25 15:24:18 +00:00
/* type is equal to stat.st_mode */
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (entry1->typeflag & FILE_TYPE_DIR) {
2015-08-20 08:51:35 +10:00
if (entry2->typeflag & FILE_TYPE_DIR) {
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
/* If both entries are tagged as dirs, we make a 'sub filter' that shows first the real dirs,
* then libs (.blend files), then categories in libs. */
if (entry1->typeflag & FILE_TYPE_BLENDERLIB) {
if (!(entry2->typeflag & FILE_TYPE_BLENDERLIB)) {
return 1;
}
}
else if (entry2->typeflag & FILE_TYPE_BLENDERLIB) {
return -1;
}
else if (entry1->typeflag & (FILE_TYPE_BLENDER | FILE_TYPE_BLENDER_BACKUP)) {
if (!(entry2->typeflag & (FILE_TYPE_BLENDER | FILE_TYPE_BLENDER_BACKUP))) {
return 1;
}
}
else if (entry2->typeflag & (FILE_TYPE_BLENDER | FILE_TYPE_BLENDER_BACKUP)) {
return -1;
}
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
else {
return -1;
}
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
else if (entry2->typeflag & FILE_TYPE_DIR) {
return 1;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
/* make sure "." and ".." are always first */
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (FILENAME_IS_CURRENT(entry1->relpath)) return -1;
if (FILENAME_IS_CURRENT(entry2->relpath)) return 1;
if (FILENAME_IS_PARENT(entry1->relpath)) return -1;
if (FILENAME_IS_PARENT(entry2->relpath)) return 1;
2015-01-07 11:40:44 +11:00
return 0;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static int compare_name(void *UNUSED(user_data), const void *a1, const void *a2)
2015-01-07 11:40:44 +11:00
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
const FileListInternEntry *entry1 = a1;
const FileListInternEntry *entry2 = a2;
char *name1, *name2;
2015-01-07 11:40:44 +11:00
int ret;
if ((ret = compare_direntry_generic(entry1, entry2))) {
return ret;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
name1 = entry1->name;
name2 = entry2->name;
return BLI_natstrcmp(name1, name2);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static int compare_date(void *UNUSED(user_data), const void *a1, const void *a2)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
const FileListInternEntry *entry1 = a1;
const FileListInternEntry *entry2 = a2;
char *name1, *name2;
int64_t time1, time2;
2015-01-07 11:40:44 +11:00
int ret;
2015-01-07 11:40:44 +11:00
if ((ret = compare_direntry_generic(entry1, entry2))) {
return ret;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
time1 = (int64_t)entry1->st.st_mtime;
time2 = (int64_t)entry2->st.st_mtime;
if (time1 < time2) return 1;
if (time1 > time2) return -1;
name1 = entry1->name;
name2 = entry2->name;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
return BLI_natstrcmp(name1, name2);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static int compare_size(void *UNUSED(user_data), const void *a1, const void *a2)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
const FileListInternEntry *entry1 = a1;
const FileListInternEntry *entry2 = a2;
char *name1, *name2;
uint64_t size1, size2;
2015-01-07 11:40:44 +11:00
int ret;
2015-01-07 11:40:44 +11:00
if ((ret = compare_direntry_generic(entry1, entry2))) {
return ret;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
size1 = entry1->st.st_size;
size2 = entry2->st.st_size;
if (size1 < size2) return 1;
if (size1 > size2) return -1;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
name1 = entry1->name;
name2 = entry2->name;
return BLI_natstrcmp(name1, name2);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static int compare_extension(void *UNUSED(user_data), const void *a1, const void *a2)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
const FileListInternEntry *entry1 = a1;
const FileListInternEntry *entry2 = a2;
char *name1, *name2;
2015-01-07 11:40:44 +11:00
int ret;
if ((ret = compare_direntry_generic(entry1, entry2))) {
return ret;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if ((entry1->typeflag & FILE_TYPE_BLENDERLIB) && !(entry2->typeflag & FILE_TYPE_BLENDERLIB)) return -1;
if (!(entry1->typeflag & FILE_TYPE_BLENDERLIB) && (entry2->typeflag & FILE_TYPE_BLENDERLIB)) return 1;
if ((entry1->typeflag & FILE_TYPE_BLENDERLIB) && (entry2->typeflag & FILE_TYPE_BLENDERLIB)) {
if ((entry1->typeflag & FILE_TYPE_DIR) && !(entry2->typeflag & FILE_TYPE_DIR)) return 1;
if (!(entry1->typeflag & FILE_TYPE_DIR) && (entry2->typeflag & FILE_TYPE_DIR)) return -1;
if (entry1->blentype < entry2->blentype) return -1;
if (entry1->blentype > entry2->blentype) return 1;
}
else {
const char *sufix1, *sufix2;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (!(sufix1 = strstr(entry1->relpath, ".blend.gz")))
sufix1 = strrchr(entry1->relpath, '.');
if (!(sufix2 = strstr(entry2->relpath, ".blend.gz")))
sufix2 = strrchr(entry2->relpath, '.');
if (!sufix1) sufix1 = "";
if (!sufix2) sufix2 = "";
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if ((ret = BLI_strcasecmp(sufix1, sufix2))) {
return ret;
}
}
name1 = entry1->name;
name2 = entry2->name;
return BLI_natstrcmp(name1, name2);
}
void filelist_sort(struct FileList *filelist)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if ((filelist->flags & FL_NEED_SORTING) && (filelist->sort != FILE_SORT_NONE)) {
switch (filelist->sort) {
case FILE_SORT_ALPHA:
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
BLI_listbase_sort_r(&filelist->filelist_intern.entries, compare_name, NULL);
break;
case FILE_SORT_TIME:
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
BLI_listbase_sort_r(&filelist->filelist_intern.entries, compare_date, NULL);
break;
case FILE_SORT_SIZE:
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
BLI_listbase_sort_r(&filelist->filelist_intern.entries, compare_size, NULL);
break;
case FILE_SORT_EXTENSION:
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
BLI_listbase_sort_r(&filelist->filelist_intern.entries, compare_extension, NULL);
break;
case FILE_SORT_NONE: /* Should never reach this point! */
default:
BLI_assert(0);
}
filelist_filter_clear(filelist);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
filelist->flags &= ~FL_NEED_SORTING;
}
}
void filelist_setsorting(struct FileList *filelist, const short sort)
{
if (filelist->sort != sort) {
filelist->sort = sort;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
filelist->flags |= FL_NEED_SORTING;
}
}
/* ********** Filter helpers ********** */
static bool is_hidden_file(const char *filename, FileListFilter *filter)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
char *sep = (char *)BLI_last_slash(filename);
bool is_hidden = false;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (filter->flags & FLF_HIDE_DOT) {
if (filename[0] == '.' && filename[1] != '.' && filename[1] != '\0') {
2014-12-04 11:18:27 +01:00
is_hidden = true; /* ignore .file */
}
else {
2012-06-18 13:01:24 +00:00
int len = strlen(filename);
if ((len > 0) && (filename[len - 1] == '~')) {
2014-12-04 11:18:27 +01:00
is_hidden = true; /* ignore file~ */
}
}
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (!is_hidden && (filter->flags & FLF_HIDE_PARENT)) {
if (filename[0] == '.' && filename[1] == '.' && filename[2] == '\0') {
is_hidden = true; /* ignore .. */
}
}
if (!is_hidden && ((filename[0] == '.') && (filename[1] == '\0'))) {
is_hidden = true; /* ignore . */
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
/* filename might actually be a piece of path, in which case we have to check all its parts. */
if (!is_hidden && sep) {
char tmp_filename[FILE_MAX_LIBEXTRA];
BLI_strncpy(tmp_filename, filename, sizeof(tmp_filename));
sep = tmp_filename + (sep - filename);
while (sep) {
BLI_assert(sep[1] != '\0');
if (is_hidden_file(sep + 1, filter)) {
is_hidden = true;
break;
}
*sep = '\0';
sep = (char *)BLI_last_slash(tmp_filename);
}
}
return is_hidden;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static bool is_filtered_file(FileListInternEntry *file, const char *UNUSED(root), FileListFilter *filter)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
bool is_filtered = !is_hidden_file(file->relpath, filter);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (is_filtered && filter->filter && !FILENAME_IS_CURRPAR(file->relpath)) {
if (file->typeflag & FILE_TYPE_DIR) {
if (file->typeflag & (FILE_TYPE_BLENDERLIB | FILE_TYPE_BLENDER | FILE_TYPE_BLENDER_BACKUP)) {
if (!(filter->filter & (FILE_TYPE_BLENDER | FILE_TYPE_BLENDER_BACKUP))) {
is_filtered = false;
}
}
else {
if (!(filter->filter & FILE_TYPE_FOLDER)) {
is_filtered = false;
}
}
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
else {
if (!(file->typeflag & filter->filter)) {
is_filtered = false;
}
}
if (is_filtered && (filter->filter_search[0] != '\0')) {
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (fnmatch(filter->filter_search, file->relpath, FNM_CASEFOLD) != 0) {
is_filtered = false;
}
}
}
return is_filtered;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static bool is_filtered_lib(FileListInternEntry *file, const char *root, FileListFilter *filter)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
bool is_filtered;
char path[FILE_MAX_LIBEXTRA], dir[FILE_MAXDIR], *group, *name;
BLI_join_dirfile(path, sizeof(path), root, file->relpath);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (BLO_library_path_explode(path, dir, &group, &name)) {
is_filtered = !is_hidden_file(file->relpath, filter);
if (is_filtered && filter->filter && !FILENAME_IS_CURRPAR(file->relpath)) {
if (file->typeflag & FILE_TYPE_DIR) {
if (file->typeflag & (FILE_TYPE_BLENDERLIB | FILE_TYPE_BLENDER | FILE_TYPE_BLENDER_BACKUP)) {
if (!(filter->filter & (FILE_TYPE_BLENDER | FILE_TYPE_BLENDER_BACKUP))) {
is_filtered = false;
}
}
else {
if (!(filter->filter & FILE_TYPE_FOLDER)) {
is_filtered = false;
}
}
}
if (is_filtered && group) {
if (!name && (filter->flags & FLF_HIDE_LIB_DIR)) {
is_filtered = false;
}
else {
unsigned int filter_id = groupname_to_filter_id(group);
if (!(filter_id & filter->filter_id)) {
is_filtered = false;
}
}
}
if (is_filtered && (filter->filter_search[0] != '\0')) {
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (fnmatch(filter->filter_search, file->relpath, FNM_CASEFOLD) != 0) {
is_filtered = false;
}
}
}
}
else {
is_filtered = is_filtered_file(file, root, filter);
}
return is_filtered;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static bool is_filtered_main(FileListInternEntry *file, const char *UNUSED(dir), FileListFilter *filter)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
return !is_hidden_file(file->relpath, filter);
}
static void filelist_filter_clear(FileList *filelist)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
filelist->flags |= FL_NEED_FILTERING;
}
2012-06-18 13:01:24 +00:00
void filelist_filter(FileList *filelist)
{
int num_filtered = 0;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
const int num_files = filelist->filelist.nbr_entries;
FileListInternEntry **filtered_tmp, *file;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (filelist->filelist.nbr_entries == 0) {
return;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (!(filelist->flags & FL_NEED_FILTERING)) {
/* Assume it has already been filtered, nothing else to do! */
return;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
filelist->filter_data.flags &= ~FLF_HIDE_LIB_DIR;
if (filelist->max_recursion) {
/* Never show lib ID 'categories' directories when we are in 'flat' mode, unless
* root path is a blend file. */
char dir[FILE_MAXDIR];
if (!filelist_islibrary(filelist, dir, NULL)) {
filelist->filter_data.flags |= FLF_HIDE_LIB_DIR;
}
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
filtered_tmp = MEM_mallocN(sizeof(*filtered_tmp) * (size_t)num_files, __func__);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
/* Filter remap & count how many files are left after filter in a single loop. */
for (file = filelist->filelist_intern.entries.first; file; file = file->next) {
if (filelist->filterf(file, filelist->filelist.root, &filelist->filter_data)) {
filtered_tmp[num_filtered++] = file;
}
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (filelist->filelist_intern.filtered) {
MEM_freeN(filelist->filelist_intern.filtered);
}
filelist->filelist_intern.filtered = MEM_mallocN(sizeof(*filelist->filelist_intern.filtered) * (size_t)num_filtered,
__func__);
memcpy(filelist->filelist_intern.filtered, filtered_tmp,
sizeof(*filelist->filelist_intern.filtered) * (size_t)num_filtered);
filelist->filelist.nbr_entries_filtered = num_filtered;
// printf("Filetered: %d over %d entries\n", num_filtered, filelist->filelist.nbr_entries);
filelist_cache_clear(&filelist->filelist_cache, filelist->filelist_cache.size);
filelist->flags &= ~FL_NEED_FILTERING;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
MEM_freeN(filtered_tmp);
}
void filelist_setfilter_options(FileList *filelist, const bool hide_dot, const bool hide_parent,
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
const unsigned int filter, const unsigned int filter_id,
const char *filter_glob, const char *filter_search)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
bool update = false;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (((filelist->filter_data.flags & FLF_HIDE_DOT) != 0) != (hide_dot != 0)) {
filelist->filter_data.flags ^= FLF_HIDE_DOT;
update = true;
}
if (((filelist->filter_data.flags & FLF_HIDE_PARENT) != 0) != (hide_parent != 0)) {
filelist->filter_data.flags ^= FLF_HIDE_PARENT;
update = true;
}
if ((filelist->filter_data.filter != filter) || (filelist->filter_data.filter_id != filter_id)) {
filelist->filter_data.filter = filter;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
filelist->filter_data.filter_id = filter_id;
update = true;
}
if (!STREQ(filelist->filter_data.filter_glob, filter_glob)) {
BLI_strncpy(filelist->filter_data.filter_glob, filter_glob, sizeof(filelist->filter_data.filter_glob));
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
update = true;
}
if ((BLI_strcmp_ignore_pad(filelist->filter_data.filter_search, filter_search, '*') != 0)) {
BLI_strncpy_ensure_pad(filelist->filter_data.filter_search, filter_search, '*',
sizeof(filelist->filter_data.filter_search));
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
update = true;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (update) {
/* And now, free filtered data so that we know we have to filter again. */
filelist_filter_clear(filelist);
}
}
/* ********** Icon/image helpers ********** */
void filelist_init_icons(void)
{
short x, y, k;
ImBuf *bbuf;
ImBuf *ibuf;
BLI_assert(G.background == false);
#ifdef WITH_HEADLESS
bbuf = NULL;
#else
Color Management, Stage 2: Switch color pipeline to use OpenColorIO Replace old color pipeline which was supporting linear/sRGB color spaces only with OpenColorIO-based pipeline. This introduces two configurable color spaces: - Input color space for images and movie clips. This space is used to convert images/movies from color space in which file is saved to Blender's linear space (for float images, byte images are not internally converted, only input space is stored for such images and used later). This setting could be found in image/clip data block settings. - Display color space which defines space in which particular display is working. This settings could be found in scene's Color Management panel. When render result is being displayed on the screen, apart from converting image to display space, some additional conversions could happen. This conversions are: - View, which defines tone curve applying before display transformation. These are different ways to view the image on the same display device. For example it could be used to emulate film view on sRGB display. - Exposure affects on image exposure before tone map is applied. - Gamma is post-display gamma correction, could be used to match particular display gamma. - RGB curves are user-defined curves which are applying before display transformation, could be used for different purposes. All this settings by default are only applying on render result and does not affect on other images. If some particular image needs to be affected by this transformation, "View as Render" setting of image data block should be set to truth. Movie clips are always affected by all display transformations. This commit also introduces configurable color space in which sequencer is working. This setting could be found in scene's Color Management panel and it should be used if such stuff as grading needs to be done in color space different from sRGB (i.e. when Film view on sRGB display is use, using VD16 space as sequencer's internal space would make grading working in space which is close to the space using for display). Some technical notes: - Image buffer's float buffer is now always in linear space, even if it was created from 16bit byte images. - Space of byte buffer is stored in image buffer's rect_colorspace property. - Profile of image buffer was removed since it's not longer meaningful. - OpenGL and GLSL is supposed to always work in sRGB space. It is possible to support other spaces, but it's quite large project which isn't so much important. - Legacy Color Management option disabled is emulated by using None display. It could have some regressions, but there's no clear way to avoid them. - If OpenColorIO is disabled on build time, it should make blender behaving in the same way as previous release with color management enabled. More details could be found at this page (more details would be added soon): http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Color_Management -- Thanks to Xavier Thomas, Lukas Toene for initial work on OpenColorIO integration and to Brecht van Lommel for some further development and code/ usecase review!
2012-09-15 10:05:07 +00:00
bbuf = IMB_ibImageFromMemory((unsigned char *)datatoc_prvicons_png, datatoc_prvicons_png_size, IB_rect, NULL, "<splash>");
#endif
if (bbuf) {
2012-06-18 13:01:24 +00:00
for (y = 0; y < SPECIAL_IMG_ROWS; y++) {
for (x = 0; x < SPECIAL_IMG_COLS; x++) {
int tile = SPECIAL_IMG_COLS * y + x;
if (tile < SPECIAL_IMG_MAX) {
ibuf = IMB_allocImBuf(SPECIAL_IMG_SIZE, SPECIAL_IMG_SIZE, 32, IB_rect);
2012-06-18 13:01:24 +00:00
for (k = 0; k < SPECIAL_IMG_SIZE; k++) {
memcpy(&ibuf->rect[k * SPECIAL_IMG_SIZE], &bbuf->rect[(k + y * SPECIAL_IMG_SIZE) * SPECIAL_IMG_SIZE * SPECIAL_IMG_COLS + x * SPECIAL_IMG_SIZE], SPECIAL_IMG_SIZE * sizeof(int));
}
gSpecialFileImages[tile] = ibuf;
}
}
}
IMB_freeImBuf(bbuf);
}
}
void filelist_free_icons(void)
{
int i;
BLI_assert(G.background == false);
2012-06-18 13:01:24 +00:00
for (i = 0; i < SPECIAL_IMG_MAX; ++i) {
IMB_freeImBuf(gSpecialFileImages[i]);
gSpecialFileImages[i] = NULL;
}
}
void filelist_imgsize(struct FileList *filelist, short w, short h)
{
filelist->prv_w = w;
filelist->prv_h = h;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static FileDirEntry *filelist_geticon_get_file(struct FileList *filelist, const int index)
{
BLI_assert(G.background == false);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
return filelist_file(filelist, index);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
ImBuf *filelist_getimage(struct FileList *filelist, const int index)
{
FileDirEntry *file = filelist_geticon_get_file(filelist, index);
return file->image;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static ImBuf *filelist_geticon_image_ex(const unsigned int typeflag, const char *relpath)
{
ImBuf *ibuf = NULL;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (typeflag & FILE_TYPE_DIR) {
if (FILENAME_IS_PARENT(relpath)) {
ibuf = gSpecialFileImages[SPECIAL_IMG_PARENT];
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
else if (FILENAME_IS_CURRENT(relpath)) {
ibuf = gSpecialFileImages[SPECIAL_IMG_REFRESH];
}
else {
ibuf = gSpecialFileImages[SPECIAL_IMG_FOLDER];
Drag and drop 2.5 integration! Finally, slashdot regulars can use Blender too now! :) ** Drag works as follows: - drag-able items are defined by the standard interface ui toolkit - each button can get this feature, via uiButSetDragXXX(but, ...). There are calls to define drag-able images, ID blocks, RNA paths, file paths, and so on. By default you drag an icon, exceptionally an ImBuf - Drag items are registered centrally in the WM, it allows more drag items simultaneous too, but not implemented ** Drop works as follows: - On mouse release, and if drag items exist in the WM, it converts the mouse event to an EVT_DROP type. This event then gets the full drag info as customdata - drop regions are defined with WM_dropbox_add(), similar to keymaps you can make a "drop map" this way, which become 'drop map handlers' in the queues. - next to that the UI kit handles some common button types (like accepting ID or names) to be catching a drop event too. - Every "drop box" has two callbacks: - poll() = check if the event drag data is relevant for this box - copy() = fill in custom properties in the dropbox to initialize an operator - The dropbox handler then calls its standard Operator with its dropbox properties. ** Currently implemented Drag items: - ID icons in browse buttons - ID icons in context menu of properties region - ID icons in outliner and rna viewer - FileBrowser icons - FileBrowser preview images Drag-able icons are subtly visualized by making them brighter a bit on mouse-over. In case the icon is a button or UI element too (most cases), the drag-able feature will make the item react to mouse-release instead of mouse-press. Drop options: - UI buttons: ID and text buttons (paste name) - View3d: Object ID drop copies object - View3d: Material ID drop assigns to object under cursor - View3d: Image ID drop assigns to object UV texture under cursor - Sequencer: Path drop will add either Image or Movie strip - Image window: Path drop will open image ** Drag and drop Notes: - Dropping into another Blender window (from same application) works too. I've added code that passes on mousemoves and clicks to other windows, without activating them though. This does make using multi-window Blender a bit friendler. - Dropping a file path to an image, is not the same as dropping an Image ID... keep this in mind. Sequencer for example wants paths to be dropped, textures in 3d window wants an Image ID. - Although drop boxes could be defined via Python, I suggest they're part of the UI and editor design (= how we want an editor to work), and not default offered configurable like keymaps. - At the moment only one item can be dragged at a time. This is for several reasons.... For one, Blender doesn't have a well defined uniform way to define "what is selected" (files, outliner items, etc). Secondly there's potential conflicts on what todo when you drop mixed drag sets on spots. All undefined stuff... nice for later. - Example to bypass the above: a collection of images that form a strip, should be represented in filewindow as a single sequence anyway. This then will fit well and gets handled neatly by design. - Another option to check is to allow multiple options per drop... it could show the operator as a sort of menu, allowing arrow or scrollwheel to choose. For time being I'd prefer to try to design a singular drop though, just offer only one drop action per data type on given spots. - What does work already, but a tad slow, is to use a function that detects an object (type) under cursor, so a drag item's option can be further refined (like drop object on object = parent). (disabled) ** More notes - Added saving for Region layouts (like split points for toolbar) - Label buttons now handle mouse over - File list: added full path entry for drop feature. - Filesel bugfix: wm_operator_exec() got called there and fully handled, while WM event code tried same. Added new OPERATOR_HANDLED flag for this. Maybe python needs it too? - Cocoa: added window move event, so multi-win setups work OK (didnt save). - Interface_handlers.c: removed win->active - Severe area copy bug: area handlers were not set to NULL - Filesel bugfix: next/prev folder list was not copied on area copies ** Leftover todos - Cocoa windows seem to hang on cases still... needs check - Cocoa 'draw overlap' swap doesn't work - Cocoa window loses focus permanently on using Spotlight (for these reasons, makefile building has Carbon as default atm) - ListView templates in UI cannot become dragged yet, needs review... it consists of two overlapping UI elements, preventing handling icon clicks. - There's already Ghost library code to handle dropping from OS into Blender window. I've noticed this code is unfinished for Macs, but seems to be complete for Windows. Needs test... currently, an external drop event will print in console when succesfully delivered to Blender's WM.
2010-01-26 18:18:21 +00:00
}
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
else if (typeflag & FILE_TYPE_BLENDER) {
ibuf = gSpecialFileImages[SPECIAL_IMG_BLENDFILE];
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
else if (typeflag & FILE_TYPE_BLENDERLIB) {
ibuf = gSpecialFileImages[SPECIAL_IMG_UNKNOWNFILE];
}
else if (typeflag & (FILE_TYPE_MOVIE)) {
ibuf = gSpecialFileImages[SPECIAL_IMG_MOVIEFILE];
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
else if (typeflag & FILE_TYPE_SOUND) {
ibuf = gSpecialFileImages[SPECIAL_IMG_SOUNDFILE];
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
else if (typeflag & FILE_TYPE_PYSCRIPT) {
ibuf = gSpecialFileImages[SPECIAL_IMG_PYTHONFILE];
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
else if (typeflag & FILE_TYPE_FTFONT) {
ibuf = gSpecialFileImages[SPECIAL_IMG_FONTFILE];
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
else if (typeflag & FILE_TYPE_TEXT) {
ibuf = gSpecialFileImages[SPECIAL_IMG_TEXTFILE];
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
else if (typeflag & FILE_TYPE_IMAGE) {
ibuf = gSpecialFileImages[SPECIAL_IMG_LOADING];
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
else if (typeflag & FILE_TYPE_BLENDER_BACKUP) {
ibuf = gSpecialFileImages[SPECIAL_IMG_BACKUP];
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
else {
ibuf = gSpecialFileImages[SPECIAL_IMG_UNKNOWNFILE];
}
Drag and drop 2.5 integration! Finally, slashdot regulars can use Blender too now! :) ** Drag works as follows: - drag-able items are defined by the standard interface ui toolkit - each button can get this feature, via uiButSetDragXXX(but, ...). There are calls to define drag-able images, ID blocks, RNA paths, file paths, and so on. By default you drag an icon, exceptionally an ImBuf - Drag items are registered centrally in the WM, it allows more drag items simultaneous too, but not implemented ** Drop works as follows: - On mouse release, and if drag items exist in the WM, it converts the mouse event to an EVT_DROP type. This event then gets the full drag info as customdata - drop regions are defined with WM_dropbox_add(), similar to keymaps you can make a "drop map" this way, which become 'drop map handlers' in the queues. - next to that the UI kit handles some common button types (like accepting ID or names) to be catching a drop event too. - Every "drop box" has two callbacks: - poll() = check if the event drag data is relevant for this box - copy() = fill in custom properties in the dropbox to initialize an operator - The dropbox handler then calls its standard Operator with its dropbox properties. ** Currently implemented Drag items: - ID icons in browse buttons - ID icons in context menu of properties region - ID icons in outliner and rna viewer - FileBrowser icons - FileBrowser preview images Drag-able icons are subtly visualized by making them brighter a bit on mouse-over. In case the icon is a button or UI element too (most cases), the drag-able feature will make the item react to mouse-release instead of mouse-press. Drop options: - UI buttons: ID and text buttons (paste name) - View3d: Object ID drop copies object - View3d: Material ID drop assigns to object under cursor - View3d: Image ID drop assigns to object UV texture under cursor - Sequencer: Path drop will add either Image or Movie strip - Image window: Path drop will open image ** Drag and drop Notes: - Dropping into another Blender window (from same application) works too. I've added code that passes on mousemoves and clicks to other windows, without activating them though. This does make using multi-window Blender a bit friendler. - Dropping a file path to an image, is not the same as dropping an Image ID... keep this in mind. Sequencer for example wants paths to be dropped, textures in 3d window wants an Image ID. - Although drop boxes could be defined via Python, I suggest they're part of the UI and editor design (= how we want an editor to work), and not default offered configurable like keymaps. - At the moment only one item can be dragged at a time. This is for several reasons.... For one, Blender doesn't have a well defined uniform way to define "what is selected" (files, outliner items, etc). Secondly there's potential conflicts on what todo when you drop mixed drag sets on spots. All undefined stuff... nice for later. - Example to bypass the above: a collection of images that form a strip, should be represented in filewindow as a single sequence anyway. This then will fit well and gets handled neatly by design. - Another option to check is to allow multiple options per drop... it could show the operator as a sort of menu, allowing arrow or scrollwheel to choose. For time being I'd prefer to try to design a singular drop though, just offer only one drop action per data type on given spots. - What does work already, but a tad slow, is to use a function that detects an object (type) under cursor, so a drag item's option can be further refined (like drop object on object = parent). (disabled) ** More notes - Added saving for Region layouts (like split points for toolbar) - Label buttons now handle mouse over - File list: added full path entry for drop feature. - Filesel bugfix: wm_operator_exec() got called there and fully handled, while WM event code tried same. Added new OPERATOR_HANDLED flag for this. Maybe python needs it too? - Cocoa: added window move event, so multi-win setups work OK (didnt save). - Interface_handlers.c: removed win->active - Severe area copy bug: area handlers were not set to NULL - Filesel bugfix: next/prev folder list was not copied on area copies ** Leftover todos - Cocoa windows seem to hang on cases still... needs check - Cocoa 'draw overlap' swap doesn't work - Cocoa window loses focus permanently on using Spotlight (for these reasons, makefile building has Carbon as default atm) - ListView templates in UI cannot become dragged yet, needs review... it consists of two overlapping UI elements, preventing handling icon clicks. - There's already Ghost library code to handle dropping from OS into Blender window. I've noticed this code is unfinished for Macs, but seems to be complete for Windows. Needs test... currently, an external drop event will print in console when succesfully delivered to Blender's WM.
2010-01-26 18:18:21 +00:00
return ibuf;
}
Drag and drop 2.5 integration! Finally, slashdot regulars can use Blender too now! :) ** Drag works as follows: - drag-able items are defined by the standard interface ui toolkit - each button can get this feature, via uiButSetDragXXX(but, ...). There are calls to define drag-able images, ID blocks, RNA paths, file paths, and so on. By default you drag an icon, exceptionally an ImBuf - Drag items are registered centrally in the WM, it allows more drag items simultaneous too, but not implemented ** Drop works as follows: - On mouse release, and if drag items exist in the WM, it converts the mouse event to an EVT_DROP type. This event then gets the full drag info as customdata - drop regions are defined with WM_dropbox_add(), similar to keymaps you can make a "drop map" this way, which become 'drop map handlers' in the queues. - next to that the UI kit handles some common button types (like accepting ID or names) to be catching a drop event too. - Every "drop box" has two callbacks: - poll() = check if the event drag data is relevant for this box - copy() = fill in custom properties in the dropbox to initialize an operator - The dropbox handler then calls its standard Operator with its dropbox properties. ** Currently implemented Drag items: - ID icons in browse buttons - ID icons in context menu of properties region - ID icons in outliner and rna viewer - FileBrowser icons - FileBrowser preview images Drag-able icons are subtly visualized by making them brighter a bit on mouse-over. In case the icon is a button or UI element too (most cases), the drag-able feature will make the item react to mouse-release instead of mouse-press. Drop options: - UI buttons: ID and text buttons (paste name) - View3d: Object ID drop copies object - View3d: Material ID drop assigns to object under cursor - View3d: Image ID drop assigns to object UV texture under cursor - Sequencer: Path drop will add either Image or Movie strip - Image window: Path drop will open image ** Drag and drop Notes: - Dropping into another Blender window (from same application) works too. I've added code that passes on mousemoves and clicks to other windows, without activating them though. This does make using multi-window Blender a bit friendler. - Dropping a file path to an image, is not the same as dropping an Image ID... keep this in mind. Sequencer for example wants paths to be dropped, textures in 3d window wants an Image ID. - Although drop boxes could be defined via Python, I suggest they're part of the UI and editor design (= how we want an editor to work), and not default offered configurable like keymaps. - At the moment only one item can be dragged at a time. This is for several reasons.... For one, Blender doesn't have a well defined uniform way to define "what is selected" (files, outliner items, etc). Secondly there's potential conflicts on what todo when you drop mixed drag sets on spots. All undefined stuff... nice for later. - Example to bypass the above: a collection of images that form a strip, should be represented in filewindow as a single sequence anyway. This then will fit well and gets handled neatly by design. - Another option to check is to allow multiple options per drop... it could show the operator as a sort of menu, allowing arrow or scrollwheel to choose. For time being I'd prefer to try to design a singular drop though, just offer only one drop action per data type on given spots. - What does work already, but a tad slow, is to use a function that detects an object (type) under cursor, so a drag item's option can be further refined (like drop object on object = parent). (disabled) ** More notes - Added saving for Region layouts (like split points for toolbar) - Label buttons now handle mouse over - File list: added full path entry for drop feature. - Filesel bugfix: wm_operator_exec() got called there and fully handled, while WM event code tried same. Added new OPERATOR_HANDLED flag for this. Maybe python needs it too? - Cocoa: added window move event, so multi-win setups work OK (didnt save). - Interface_handlers.c: removed win->active - Severe area copy bug: area handlers were not set to NULL - Filesel bugfix: next/prev folder list was not copied on area copies ** Leftover todos - Cocoa windows seem to hang on cases still... needs check - Cocoa 'draw overlap' swap doesn't work - Cocoa window loses focus permanently on using Spotlight (for these reasons, makefile building has Carbon as default atm) - ListView templates in UI cannot become dragged yet, needs review... it consists of two overlapping UI elements, preventing handling icon clicks. - There's already Ghost library code to handle dropping from OS into Blender window. I've noticed this code is unfinished for Macs, but seems to be complete for Windows. Needs test... currently, an external drop event will print in console when succesfully delivered to Blender's WM.
2010-01-26 18:18:21 +00:00
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
ImBuf *filelist_geticon_image(struct FileList *filelist, const int index)
{
FileDirEntry *file = filelist_geticon_get_file(filelist, index);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
return filelist_geticon_image_ex(file->typeflag, file->relpath);
}
static int filelist_geticon_ex(
const int typeflag, const int blentype, const char *relpath, const bool is_main, const bool ignore_libdir)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if ((typeflag & FILE_TYPE_DIR) && !(ignore_libdir && (typeflag & (FILE_TYPE_BLENDERLIB | FILE_TYPE_BLENDER)))) {
if (FILENAME_IS_PARENT(relpath)) {
return is_main ? ICON_FILE_PARENT : ICON_NONE;
}
else if (typeflag & FILE_TYPE_APPLICATIONBUNDLE) {
return ICON_UGLYPACKAGE;
}
else if (typeflag & FILE_TYPE_BLENDER) {
return ICON_FILE_BLEND;
}
else if (is_main) {
/* Do not return icon for folders if icons are not 'main' draw type (e.g. when used over previews). */
return ICON_FILE_FOLDER;
}
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (typeflag & FILE_TYPE_BLENDER)
return ICON_FILE_BLEND;
else if (typeflag & FILE_TYPE_BLENDER_BACKUP)
return ICON_FILE_BACKUP;
else if (typeflag & FILE_TYPE_IMAGE)
return ICON_FILE_IMAGE;
else if (typeflag & FILE_TYPE_MOVIE)
return ICON_FILE_MOVIE;
else if (typeflag & FILE_TYPE_PYSCRIPT)
return ICON_FILE_SCRIPT;
else if (typeflag & FILE_TYPE_SOUND)
return ICON_FILE_SOUND;
else if (typeflag & FILE_TYPE_FTFONT)
return ICON_FILE_FONT;
else if (typeflag & FILE_TYPE_BTX)
return ICON_FILE_BLANK;
else if (typeflag & FILE_TYPE_COLLADA)
return ICON_FILE_BLANK;
else if (typeflag & FILE_TYPE_TEXT)
return ICON_FILE_TEXT;
else if (typeflag & FILE_TYPE_BLENDERLIB) {
const int ret = UI_idcode_icon_get(blentype);
if (ret != ICON_NONE) {
return ret;
}
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
return is_main ? ICON_FILE_BLANK : ICON_NONE;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
int filelist_geticon(struct FileList *filelist, const int index, const bool is_main)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
FileDirEntry *file = filelist_geticon_get_file(filelist, index);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
return filelist_geticon_ex(file->typeflag, file->blentype, file->relpath, is_main, false);
}
/* ********** Main ********** */
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static void filelist_checkdir_dir(struct FileList *UNUSED(filelist), char *r_dir)
{
BLI_make_exist(r_dir);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static void filelist_checkdir_lib(struct FileList *UNUSED(filelist), char *r_dir)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
char dir[FILE_MAXDIR];
if (!BLO_library_path_explode(r_dir, dir, NULL, NULL)) {
/* if not a valid library, we need it to be a valid directory! */
BLI_make_exist(r_dir);
}
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static void filelist_checkdir_main(struct FileList *filelist, char *r_dir)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
/* TODO */
filelist_checkdir_lib(filelist, r_dir);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static void filelist_entry_clear(FileDirEntry *entry)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (entry->name) {
MEM_freeN(entry->name);
}
if (entry->description) {
MEM_freeN(entry->description);
}
if (entry->relpath) {
MEM_freeN(entry->relpath);
}
if (entry->image) {
IMB_freeImBuf(entry->image);
}
/* For now, consider FileDirEntryRevision::poin as not owned here, so no need to do anything about it */
if (!BLI_listbase_is_empty(&entry->variants)) {
FileDirEntryVariant *var;
for (var = entry->variants.first; var; var = var->next) {
if (var->name) {
MEM_freeN(var->name);
}
if (var->description) {
MEM_freeN(var->description);
}
if (!BLI_listbase_is_empty(&var->revisions)) {
FileDirEntryRevision *rev;
for (rev = var->revisions.first; rev; rev = rev->next) {
if (rev->comment) {
MEM_freeN(rev->comment);
}
}
BLI_freelistN(&var->revisions);
}
}
/* TODO: tags! */
BLI_freelistN(&entry->variants);
}
2015-08-20 08:51:35 +10:00
else if (entry->entry) {
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
MEM_freeN(entry->entry);
}
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static void filelist_entry_free(FileDirEntry *entry)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
filelist_entry_clear(entry);
MEM_freeN(entry);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static void filelist_direntryarr_free(FileDirEntryArr *array)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
FileDirEntry *entry, *entry_next;
for (entry = array->entries.first; entry; entry = entry_next) {
entry_next = entry->next;
filelist_entry_free(entry);
}
BLI_listbase_clear(&array->entries);
array->nbr_entries = 0;
array->nbr_entries_filtered = -1;
array->entry_idx_start = -1;
array->entry_idx_end = -1;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static void filelist_intern_entry_free(FileListInternEntry *entry)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (entry->relpath) {
MEM_freeN(entry->relpath);
}
if (entry->name) {
MEM_freeN(entry->name);
}
MEM_freeN(entry);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static void filelist_intern_free(FileListIntern *filelist_intern)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
FileListInternEntry *entry, *entry_next;
for (entry = filelist_intern->entries.first; entry; entry = entry_next) {
entry_next = entry->next;
filelist_intern_entry_free(entry);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
BLI_listbase_clear(&filelist_intern->entries);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
MEM_SAFE_FREE(filelist_intern->filtered);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static void filelist_cache_previewf(TaskPool *pool, void *taskdata, int UNUSED(threadid))
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
FileListEntryCache *cache = taskdata;
FileListEntryPreview *preview;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
// printf("%s: Start (%d)...\n", __func__, threadid);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
/* Note we wait on queue here. */
while (!BLI_task_pool_canceled(pool) && (preview = BLI_thread_queue_pop(cache->previews_todo))) {
ThumbSource source = 0;
// printf("%s: %d - %s - %p\n", __func__, preview->index, preview->path, preview->img);
BLI_assert(preview->flags & (FILE_TYPE_IMAGE | FILE_TYPE_MOVIE | FILE_TYPE_FTFONT |
FILE_TYPE_BLENDER | FILE_TYPE_BLENDER_BACKUP | FILE_TYPE_BLENDERLIB));
if (preview->flags & FILE_TYPE_IMAGE) {
source = THB_SOURCE_IMAGE;
}
else if (preview->flags & (FILE_TYPE_BLENDER | FILE_TYPE_BLENDER_BACKUP | FILE_TYPE_BLENDERLIB)) {
source = THB_SOURCE_BLEND;
}
else if (preview->flags & FILE_TYPE_MOVIE) {
source = THB_SOURCE_MOVIE;
}
else if (preview->flags & FILE_TYPE_FTFONT) {
source = THB_SOURCE_FONT;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
IMB_thumb_path_lock(preview->path);
preview->img = IMB_thumb_manage(preview->path, THB_LARGE, source);
IMB_thumb_path_unlock(preview->path);
BLI_thread_queue_push(cache->previews_done, preview);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
// printf("%s: End (%d)...\n", __func__, threadid);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static void filelist_cache_preview_ensure_running(FileListEntryCache *cache)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (!cache->previews_pool) {
TaskScheduler *scheduler = BLI_task_scheduler_get();
TaskPool *pool;
int num_tasks = max_ii(2, BLI_system_thread_count() / 2);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
pool = cache->previews_pool = BLI_task_pool_create(scheduler, NULL);
cache->previews_todo = BLI_thread_queue_init();
cache->previews_done = BLI_thread_queue_init();
while (num_tasks--) {
BLI_task_pool_push(pool, filelist_cache_previewf, cache, false, TASK_PRIORITY_HIGH);
}
IMB_thumb_locks_acquire();
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
cache->previews_timestamp = 0.0;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static void filelist_cache_previews_clear(FileListEntryCache *cache)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
FileListEntryPreview *preview;
if (cache->previews_pool) {
while ((preview = BLI_thread_queue_pop_timeout(cache->previews_todo, 0))) {
// printf("%s: TODO %d - %s - %p\n", __func__, preview->index, preview->path, preview->img);
MEM_freeN(preview);
cache->previews_pending--;
}
while ((preview = BLI_thread_queue_pop_timeout(cache->previews_done, 0))) {
// printf("%s: DONE %d - %s - %p\n", __func__, preview->index, preview->path, preview->img);
if (preview->img) {
IMB_freeImBuf(preview->img);
}
MEM_freeN(preview);
cache->previews_pending--;
}
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
// printf("%s: remaining pending previews: %d\n", __func__, cache->previews_pending);
}
static void filelist_cache_previews_free(FileListEntryCache *cache, const bool set_inactive)
{
if (cache->previews_pool) {
BLI_thread_queue_nowait(cache->previews_todo);
BLI_thread_queue_nowait(cache->previews_done);
BLI_task_pool_cancel(cache->previews_pool);
filelist_cache_previews_clear(cache);
BLI_thread_queue_free(cache->previews_done);
BLI_thread_queue_free(cache->previews_todo);
BLI_task_pool_free(cache->previews_pool);
cache->previews_pool = NULL;
cache->previews_todo = NULL;
cache->previews_done = NULL;
IMB_thumb_locks_release();
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (set_inactive) {
cache->flags &= ~FLC_PREVIEWS_ACTIVE;
}
BLI_assert(cache->previews_pending == 0);
cache->previews_timestamp = 0.0;
}
static void filelist_cache_previews_push(FileList *filelist, FileDirEntry *entry, const int index)
{
FileListEntryCache *cache = &filelist->filelist_cache;
BLI_assert(cache->flags & FLC_PREVIEWS_ACTIVE);
if (!entry->image &&
!(entry->flags & FILE_ENTRY_INVALID_PREVIEW) &&
2015-08-20 08:51:35 +10:00
(entry->typeflag & (FILE_TYPE_IMAGE | FILE_TYPE_MOVIE | FILE_TYPE_FTFONT |
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
FILE_TYPE_BLENDER | FILE_TYPE_BLENDER_BACKUP | FILE_TYPE_BLENDERLIB)))
{
FileListEntryPreview *preview = MEM_mallocN(sizeof(*preview), __func__);
BLI_join_dirfile(preview->path, sizeof(preview->path), filelist->filelist.root, entry->relpath);
preview->index = index;
preview->flags = entry->typeflag;
preview->img = NULL;
// printf("%s: %d - %s - %p\n", __func__, preview->index, preview->path, preview->img);
filelist_cache_preview_ensure_running(cache);
BLI_thread_queue_push(cache->previews_todo, preview);
cache->previews_pending++;
}
}
static void filelist_cache_init(FileListEntryCache *cache, size_t cache_size)
{
cache->block_cursor = cache->block_start_index = cache->block_center_index = cache->block_end_index = 0;
cache->block_entries = MEM_mallocN(sizeof(*cache->block_entries) * cache_size, __func__);
cache->misc_entries = BLI_ghash_ptr_new_ex(__func__, cache_size);
cache->misc_entries_indices = MEM_mallocN(sizeof(*cache->misc_entries_indices) * cache_size, __func__);
copy_vn_i(cache->misc_entries_indices, cache_size, -1);
cache->misc_cursor = 0;
/* XXX This assumes uint is 32 bits and uuid is 128 bits (char[16]), be careful! */
cache->uuids = BLI_ghash_new_ex(
BLI_ghashutil_uinthash_v4_p, BLI_ghashutil_uinthash_v4_cmp, __func__, cache_size * 2);
cache->size = cache_size;
cache->flags = FLC_IS_INIT;
}
static void filelist_cache_free(FileListEntryCache *cache)
{
if (!(cache->flags & FLC_IS_INIT)) {
return;
}
filelist_cache_previews_free(cache, true);
/* Note we nearly have nothing to do here, entries are just 'borrowed', not owned by cache... */
MEM_freeN(cache->block_entries);
BLI_ghash_free(cache->misc_entries, NULL, NULL);
MEM_freeN(cache->misc_entries_indices);
BLI_ghash_free(cache->uuids, NULL, NULL);
}
static void filelist_cache_clear(FileListEntryCache *cache, size_t new_size)
{
if (!(cache->flags & FLC_IS_INIT)) {
return;
}
filelist_cache_previews_clear(cache);
/* Note we nearly have nothing to do here, entries are just 'borrowed', not owned by cache... */
cache->block_cursor = cache->block_start_index = cache->block_center_index = cache->block_end_index = 0;
if (new_size != cache->size) {
cache->block_entries = MEM_reallocN(cache->block_entries, sizeof(*cache->block_entries) * new_size);
}
BLI_ghash_clear_ex(cache->misc_entries, NULL, NULL, new_size);
if (new_size != cache->size) {
cache->misc_entries_indices = MEM_reallocN(cache->misc_entries_indices,
2015-08-20 08:51:35 +10:00
sizeof(*cache->misc_entries_indices) * new_size);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
}
copy_vn_i(cache->misc_entries_indices, new_size, -1);
BLI_ghash_clear_ex(cache->uuids, NULL, NULL, new_size * 2);
cache->size = new_size;
}
FileList *filelist_new(short type)
{
FileList *p = MEM_callocN(sizeof(*p), __func__);
filelist_cache_init(&p->filelist_cache, FILELIST_ENTRYCACHESIZE_DEFAULT);
p->selection_state = BLI_ghash_new(BLI_ghashutil_uinthash_v4_p, BLI_ghashutil_uinthash_v4_cmp, __func__);
switch (type) {
case FILE_MAIN:
p->checkdirf = filelist_checkdir_main;
p->read_jobf = filelist_readjob_main;
p->filterf = is_filtered_main;
break;
case FILE_LOADLIB:
p->checkdirf = filelist_checkdir_lib;
p->read_jobf = filelist_readjob_lib;
p->filterf = is_filtered_lib;
break;
default:
p->checkdirf = filelist_checkdir_dir;
p->read_jobf = filelist_readjob_dir;
p->filterf = is_filtered_file;
break;
}
return p;
}
void filelist_clear_ex(struct FileList *filelist, const bool do_cache, const bool do_selection)
{
if (!filelist) {
return;
}
filelist_filter_clear(filelist);
if (do_cache) {
filelist_cache_clear(&filelist->filelist_cache, filelist->filelist_cache.size);
}
filelist_intern_free(&filelist->filelist_intern);
filelist_direntryarr_free(&filelist->filelist);
if (do_selection && filelist->selection_state) {
BLI_ghash_clear(filelist->selection_state, MEM_freeN, NULL);
}
}
void filelist_clear(struct FileList *filelist)
{
filelist_clear_ex(filelist, true, true);
}
void filelist_free(struct FileList *filelist)
{
if (!filelist) {
printf("Attempting to delete empty filelist.\n");
return;
}
filelist_clear_ex(filelist, false, false); /* No need to clear cache & selection_state, we free them anyway. */
filelist_cache_free(&filelist->filelist_cache);
if (filelist->selection_state) {
BLI_ghash_free(filelist->selection_state, MEM_freeN, NULL);
filelist->selection_state = NULL;
}
memset(&filelist->filter_data, 0, sizeof(filelist->filter_data));
filelist->flags &= ~(FL_NEED_SORTING | FL_NEED_FILTERING);
filelist->sort = FILE_SORT_NONE;
}
void filelist_freelib(struct FileList *filelist)
{
if (filelist->libfiledata)
BLO_blendhandle_close(filelist->libfiledata);
filelist->libfiledata = NULL;
}
BlendHandle *filelist_lib(struct FileList *filelist)
{
return filelist->libfiledata;
}
static const char *fileentry_uiname(const char *root, const char *relpath, const int typeflag, char *buff)
{
char *name = NULL;
if (typeflag & FILE_TYPE_BLENDERLIB) {
char abspath[FILE_MAX_LIBEXTRA];
char *group;
BLI_join_dirfile(abspath, sizeof(abspath), root, relpath);
BLO_library_path_explode(abspath, buff, &group, &name);
if (!name) {
name = group;
}
}
/* Depending on platforms, 'my_file.blend/..' might be viewed as dir or not... */
if (!name) {
if (typeflag & FILE_TYPE_DIR) {
name = (char *)relpath;
}
else {
name = (char *)BLI_path_basename(relpath);
}
}
BLI_assert(name);
return name;
}
const char *filelist_dir(struct FileList *filelist)
{
return filelist->filelist.root;
}
/**
* May modify in place given r_dir, which is expected to be FILE_MAX_LIBEXTRA length.
*/
void filelist_setdir(struct FileList *filelist, char *r_dir)
{
BLI_assert(strlen(r_dir) < FILE_MAX_LIBEXTRA);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
BLI_cleanup_dir(G.main->name, r_dir);
filelist->checkdirf(filelist, r_dir);
if (!STREQ(filelist->filelist.root, r_dir)) {
BLI_strncpy(filelist->filelist.root, r_dir, sizeof(filelist->filelist.root));
filelist->flags |= FL_FORCE_RESET;
}
}
void filelist_setrecursion(struct FileList *filelist, const int recursion_level)
{
if (filelist->max_recursion != recursion_level) {
filelist->max_recursion = recursion_level;
filelist->flags |= FL_FORCE_RESET;
}
}
bool filelist_force_reset(struct FileList *filelist)
{
return (filelist->flags & FL_FORCE_RESET) != 0;
}
bool filelist_is_ready(struct FileList *filelist)
{
return (filelist->flags & FL_IS_READY) != 0;
}
bool filelist_pending(struct FileList *filelist)
{
return (filelist->flags & FL_IS_PENDING) != 0;
}
/**
* Limited version of full update done by space_file's file_refresh(), to be used by operators and such.
* Ensures given filelist is ready to be used (i.e. it is filtered and sorted), unless it is tagged for a full refresh.
*/
int filelist_files_ensure(FileList *filelist)
{
if (!filelist_force_reset(filelist) || !filelist_empty(filelist)) {
filelist_sort(filelist);
filelist_filter(filelist);
}
return filelist->filelist.nbr_entries_filtered;;
}
static FileDirEntry *filelist_file_create_entry(FileList *filelist, const int index)
{
FileListInternEntry *entry = filelist->filelist_intern.filtered[index];
FileDirEntry *ret;
FileDirEntryRevision *rev;
ret = MEM_callocN(sizeof(*ret), __func__);
rev = MEM_callocN(sizeof(*rev), __func__);
rev->size = (uint64_t)entry->st.st_size;
rev->time = (int64_t)entry->st.st_mtime;
ret->entry = rev;
ret->relpath = BLI_strdup(entry->relpath);
ret->name = BLI_strdup(entry->name);
ret->description = BLI_strdupcat(filelist->filelist.root, entry->relpath);
memcpy(ret->uuid, entry->uuid, sizeof(ret->uuid));
ret->blentype = entry->blentype;
ret->typeflag = entry->typeflag;
BLI_addtail(&filelist->filelist.entries, ret);
return ret;
}
static void filelist_file_release_entry(FileList *filelist, FileDirEntry *entry)
{
BLI_remlink(&filelist->filelist.entries, entry);
filelist_entry_free(entry);
}
static FileDirEntry *filelist_file_ex(struct FileList *filelist, const int index, const bool use_request)
{
FileDirEntry *ret = NULL, *old;
FileListEntryCache *cache = &filelist->filelist_cache;
const size_t cache_size = cache->size;
int old_index;
if ((index < 0) || (index >= filelist->filelist.nbr_entries_filtered)) {
return ret;
}
if (index >= cache->block_start_index && index < cache->block_end_index) {
const int idx = (index - cache->block_start_index + cache->block_cursor) % cache_size;
return cache->block_entries[idx];
}
if ((ret = BLI_ghash_lookup(cache->misc_entries, SET_INT_IN_POINTER(index)))) {
return ret;
}
if (!use_request) {
return NULL;
}
// printf("requesting file %d (not yet cached)\n", index);
/* Else, we have to add new entry to 'misc' cache - and possibly make room for it first! */
ret = filelist_file_create_entry(filelist, index);
old_index = cache->misc_entries_indices[cache->misc_cursor];
if ((old = BLI_ghash_popkey(cache->misc_entries, SET_INT_IN_POINTER(old_index), NULL))) {
BLI_ghash_remove(cache->uuids, old->uuid, NULL, NULL);
filelist_file_release_entry(filelist, old);
}
BLI_ghash_insert(cache->misc_entries, SET_INT_IN_POINTER(index), ret);
BLI_ghash_insert(cache->uuids, ret->uuid, ret);
cache->misc_entries_indices[cache->misc_cursor] = index;
cache->misc_cursor = (cache->misc_cursor + 1) % cache_size;
#if 0 /* Actually no, only block cached entries should have preview imho. */
if (cache->previews_pool) {
filelist_cache_previews_push(filelist, ret, index);
}
#endif
return ret;
}
FileDirEntry *filelist_file(struct FileList *filelist, int index)
{
return filelist_file_ex(filelist, index, true);
}
int filelist_file_findpath(struct FileList *filelist, const char *filename)
{
int fidx = -1;
if (filelist->filelist.nbr_entries_filtered < 0) {
return fidx;
}
/* XXX TODO Cache could probably use a ghash on paths too? Not really urgent though.
2015-09-02 18:39:19 +10:00
* This is only used to find again renamed entry, annoying but looks hairy to get rid of it currently. */
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
for (fidx = 0; fidx < filelist->filelist.nbr_entries_filtered; fidx++) {
FileListInternEntry *entry = filelist->filelist_intern.filtered[fidx];
if (STREQ(entry->relpath, filename)) {
return fidx;
}
}
return -1;
}
FileDirEntry *filelist_entry_find_uuid(struct FileList *filelist, const int uuid[4])
{
if (filelist->filelist.nbr_entries_filtered < 0) {
return NULL;
}
if (filelist->filelist_cache.uuids) {
FileDirEntry *entry = BLI_ghash_lookup(filelist->filelist_cache.uuids, uuid);
if (entry) {
return entry;
}
}
{
int fidx;
for (fidx = 0; fidx < filelist->filelist.nbr_entries_filtered; fidx++) {
FileListInternEntry *entry = filelist->filelist_intern.filtered[fidx];
if (memcmp(entry->uuid, uuid, sizeof(entry->uuid)) == 0) {
return filelist_file(filelist, fidx);
}
}
}
return NULL;
}
void filelist_file_cache_slidingwindow_set(FileList *filelist, size_t window_size)
{
/* Always keep it power of 2, in [256, 8192] range for now, cache being app. twice bigger than requested window. */
size_t size = 256;
window_size *= 2;
while (size < window_size && size < 8192) {
size *= 2;
}
if (size != filelist->filelist_cache.size) {
filelist_cache_clear(&filelist->filelist_cache, size);
}
}
/* Helpers, low-level, they assume cursor + size <= cache_size */
static bool filelist_file_cache_block_create(FileList *filelist, const int start_index, const int size, int cursor)
{
FileListEntryCache *cache = &filelist->filelist_cache;
{
int i, idx;
for (i = 0, idx = start_index; i < size; i++, idx++, cursor++) {
FileDirEntry *entry;
/* That entry might have already been requested and stored in misc cache... */
if ((entry = BLI_ghash_popkey(cache->misc_entries, SET_INT_IN_POINTER(idx), NULL)) == NULL) {
entry = filelist_file_create_entry(filelist, idx);
BLI_ghash_insert(cache->uuids, entry->uuid, entry);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
cache->block_entries[cursor] = entry;
}
return true;
}
return false;
}
static void filelist_file_cache_block_release(struct FileList *filelist, const int size, int cursor)
{
FileListEntryCache *cache = &filelist->filelist_cache;
{
int i;
for (i = 0; i < size; i++, cursor++) {
FileDirEntry *entry = cache->block_entries[cursor];
// printf("%s: release cacheidx %d (%%p %%s)\n", __func__, cursor/*, cache->block_entries[cursor], cache->block_entries[cursor]->relpath*/);
BLI_ghash_remove(cache->uuids, entry->uuid, NULL, NULL);
filelist_file_release_entry(filelist, entry);
#ifndef NDEBUG
cache->block_entries[cursor] = NULL;
#endif
}
}
}
/* Load in cache all entries "around" given index (as much as block cache may hold). */
bool filelist_file_cache_block(struct FileList *filelist, const int index)
{
FileListEntryCache *cache = &filelist->filelist_cache;
const size_t cache_size = cache->size;
const int nbr_entries = filelist->filelist.nbr_entries_filtered;
int start_index = max_ii(0, index - (cache_size / 2));
int end_index = min_ii(nbr_entries, index + (cache_size / 2));
int i;
if ((index < 0) || (index >= nbr_entries)) {
// printf("Wrong index %d ([%d:%d])", index, 0, nbr_entries);
return false;
}
/* Maximize cached range! */
if ((end_index - start_index) < cache_size) {
if (start_index == 0) {
end_index = min_ii(nbr_entries, start_index + cache_size);
}
else if (end_index == nbr_entries) {
start_index = max_ii(0, end_index - cache_size);
}
}
BLI_assert((end_index - start_index) <= cache_size) ;
// printf("%s: [%d:%d] around index %d (current cache: [%d:%d])\n", __func__,
// start_index, end_index, index, cache->block_start_index, cache->block_end_index);
/* If we have something to (re)cache... */
if ((start_index != cache->block_start_index) || (end_index != cache->block_end_index)) {
if ((start_index >= cache->block_end_index) || (end_index <= cache->block_start_index)) {
int size1 = cache->block_end_index - cache->block_start_index;
int size2 = 0;
int idx1 = cache->block_cursor, idx2 = 0;
// printf("Full Recaching!\n");
if (cache->flags & FLC_PREVIEWS_ACTIVE) {
filelist_cache_previews_clear(cache);
}
if (idx1 + size1 > cache_size) {
size2 = idx1 + size1 - cache_size;
size1 -= size2;
filelist_file_cache_block_release(filelist, size2, idx2);
}
filelist_file_cache_block_release(filelist, size1, idx1);
cache->block_start_index = cache->block_end_index = cache->block_cursor = 0;
/* New cached block does not overlap existing one, simple. */
if (!filelist_file_cache_block_create(filelist, start_index, end_index - start_index, 0)) {
return false;
}
cache->block_start_index = start_index;
cache->block_end_index = end_index;
}
else {
// printf("Partial Recaching!\n");
/* At this point, we know we keep part of currently cached entries, so update previews if needed,
* and remove everything from working queue - we'll add all newly needed entries at the end. */
if (cache->flags & FLC_PREVIEWS_ACTIVE) {
filelist_cache_previews_update(filelist);
filelist_cache_previews_clear(cache);
}
// printf("\tpreview cleaned up...\n");
if (start_index > cache->block_start_index) {
int size1 = start_index - cache->block_start_index;
int size2 = 0;
int idx1 = cache->block_cursor, idx2 = 0;
// printf("\tcache releasing: [%d:%d] (%d, %d)\n", cache->block_start_index, cache->block_start_index + size1, cache->block_cursor, size1);
if (idx1 + size1 > cache_size) {
size2 = idx1 + size1 - cache_size;
size1 -= size2;
filelist_file_cache_block_release(filelist, size2, idx2);
}
filelist_file_cache_block_release(filelist, size1, idx1);
cache->block_cursor = (idx1 + size1 + size2) % cache_size;
cache->block_start_index = start_index;
}
if (end_index < cache->block_end_index) {
int size1 = cache->block_end_index - end_index;
int size2 = 0;
int idx1, idx2 = 0;
// printf("\tcache releasing: [%d:%d] (%d)\n", cache->block_end_index - size1, cache->block_end_index, cache->block_cursor);
idx1 = (cache->block_cursor + end_index - cache->block_start_index) % cache_size;
if (idx1 + size1 > cache_size) {
size2 = idx1 + size1 - cache_size;
size1 -= size2;
filelist_file_cache_block_release(filelist, size2, idx2);
}
filelist_file_cache_block_release(filelist, size1, idx1);
cache->block_end_index = end_index;
}
// printf("\tcache cleaned up...\n");
if (start_index < cache->block_start_index) {
/* Add (request) needed entries before already cached ones. */
/* Note: We need some index black magic to wrap around (cycle) inside our cache_size array... */
int size1 = cache->block_start_index - start_index;
int size2 = 0;
int idx1, idx2;
if (size1 > cache->block_cursor) {
size2 = size1;
size1 -= cache->block_cursor;
size2 -= size1;
idx2 = 0;
idx1 = cache_size - size1;
}
else {
idx1 = cache->block_cursor - size1;
}
if (size2) {
if (!filelist_file_cache_block_create(filelist, start_index + size1, size2, idx2)) {
return false;
}
}
if (!filelist_file_cache_block_create(filelist, start_index, size1, idx1)) {
return false;
}
cache->block_cursor = idx1;
cache->block_start_index = start_index;
}
// printf("\tstart-extended...\n");
if (end_index > cache->block_end_index) {
/* Add (request) needed entries after already cached ones. */
/* Note: We need some index black magic to wrap around (cycle) inside our cache_size array... */
int size1 = end_index - cache->block_end_index;
int size2 = 0;
int idx1, idx2;
idx1 = (cache->block_cursor + end_index - cache->block_start_index - size1) % cache_size;
if ((idx1 + size1) > cache_size) {
size2 = size1;
size1 = cache_size - idx1;
size2 -= size1;
idx2 = 0;
}
if (size2) {
if (!filelist_file_cache_block_create(filelist, end_index - size2, size2, idx2)) {
return false;
}
}
if (!filelist_file_cache_block_create(filelist, end_index - size1 - size2, size1, idx1)) {
return false;
}
cache->block_end_index = end_index;
}
// printf("\tend-extended...\n");
}
}
else if ((cache->block_center_index != index) && (cache->flags & FLC_PREVIEWS_ACTIVE)) {
/* We try to always preview visible entries first, so 'restart' preview background task. */
filelist_cache_previews_update(filelist);
filelist_cache_previews_clear(cache);
}
// printf("Re-queueing previews...\n");
/* Note we try to preview first images around given index - i.e. assumed visible ones. */
if (cache->flags & FLC_PREVIEWS_ACTIVE) {
for (i = 0; ((index + i) < end_index) || ((index - i) >= start_index); i++) {
if ((index - i) >= start_index) {
const int idx = (cache->block_cursor + (index - start_index) - i) % cache_size;
filelist_cache_previews_push(filelist, cache->block_entries[idx], index - i);
}
if ((index + i) < end_index) {
const int idx = (cache->block_cursor + (index - start_index) + i) % cache_size;
filelist_cache_previews_push(filelist, cache->block_entries[idx], index + i);
}
}
}
cache->block_center_index = index;
// printf("%s Finished!\n", __func__);
return true;
}
void filelist_cache_previews_set(FileList *filelist, const bool use_previews)
{
FileListEntryCache *cache = &filelist->filelist_cache;
if (use_previews == ((cache->flags & FLC_PREVIEWS_ACTIVE) != 0)) {
return;
}
/* Do not start preview work while listing, gives nasty flickering! */
else if (use_previews && (filelist->flags & FL_IS_READY)) {
cache->flags |= FLC_PREVIEWS_ACTIVE;
BLI_assert((cache->previews_pool == NULL) && (cache->previews_todo == NULL) && (cache->previews_done == NULL));
// printf("%s: Init Previews...\n", __func__);
/* No need to populate preview queue here, filelist_file_cache_block() handles this. */
}
else {
// printf("%s: Clear Previews...\n", __func__);
filelist_cache_previews_free(cache, true);
}
}
bool filelist_cache_previews_update(FileList *filelist)
{
FileListEntryCache *cache = &filelist->filelist_cache;
TaskPool *pool = cache->previews_pool;
bool changed = false;
if (!pool) {
return changed;
}
// printf("%s: Update Previews...\n", __func__);
while (!BLI_thread_queue_is_empty(cache->previews_done)) {
FileListEntryPreview *preview = BLI_thread_queue_pop(cache->previews_done);
FileDirEntry *entry;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
/* Paranoid (should never happen currently since we consume this queue from a single thread), but... */
if (!preview) {
continue;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
/* entry might have been removed from cache in the mean while, we do not want to cache it again here. */
entry = filelist_file_ex(filelist, preview->index, false);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
// printf("%s: %d - %s - %p\n", __func__, preview->index, preview->path, preview->img);
if (preview->img) {
/* Due to asynchronous process, a preview for a given image may be generated several times, i.e.
* entry->image may already be set at this point. */
if (entry && !entry->image) {
entry->image = preview->img;
changed = true;
}
else {
IMB_freeImBuf(preview->img);
}
}
else if (entry) {
/* We want to avoid re-processing this entry continuously!
* Note that, since entries only live in cache, preview will be retried quite often anyway. */
entry->flags |= FILE_ENTRY_INVALID_PREVIEW;
}
MEM_freeN(preview);
cache->previews_pending--;
BLI_assert(cache->previews_pending >= 0);
}
// printf("%s: Previews pending: %d\n", __func__, cache->previews_pending);
if (cache->previews_pending == 0) {
if (cache->previews_timestamp == 0.0) {
cache->previews_timestamp = PIL_check_seconds_timer();
}
else if (PIL_check_seconds_timer() - cache->previews_timestamp > 1.0) {
/* Preview task is IDLE since more than (approximatively) 1000ms,
* kill workers & task for now - they will be automatically restarted when needed. */
// printf("%s: Inactive preview task, sleeping (%f vs %f)!\n", __func__, PIL_check_seconds_timer(), cache->previews_timestamp);
filelist_cache_previews_free(cache, false);
}
}
return changed;
}
bool filelist_cache_previews_running(FileList *filelist)
{
FileListEntryCache *cache = &filelist->filelist_cache;
return (cache->previews_pool != NULL);
}
/* would recognize .blend as well */
static bool file_is_blend_backup(const char *str)
{
const size_t a = strlen(str);
size_t b = 7;
bool retval = 0;
if (a == 0 || b >= a) {
/* pass */
}
else {
const char *loc;
if (a > b + 1)
b++;
/* allow .blend1 .blend2 .blend32 */
loc = BLI_strcasestr(str + a - b, ".blend");
if (loc)
retval = 1;
}
return (retval);
}
static int path_extension_type(const char *path)
{
if (BLO_has_bfile_extension(path)) {
return FILE_TYPE_BLENDER;
}
else if (file_is_blend_backup(path)) {
return FILE_TYPE_BLENDER_BACKUP;
}
else if (BLI_testextensie(path, ".app")) {
return FILE_TYPE_APPLICATIONBUNDLE;
}
else if (BLI_testextensie(path, ".py")) {
return FILE_TYPE_PYSCRIPT;
}
else if (BLI_testextensie_n(path, ".txt", ".glsl", ".osl", ".data", NULL)) {
return FILE_TYPE_TEXT;
}
else if (BLI_testextensie_n(path, ".ttf", ".ttc", ".pfb", ".otf", ".otc", NULL)) {
return FILE_TYPE_FTFONT;
}
else if (BLI_testextensie(path, ".btx")) {
return FILE_TYPE_BTX;
}
else if (BLI_testextensie(path, ".dae")) {
return FILE_TYPE_COLLADA;
}
else if (BLI_testextensie_array(path, imb_ext_image) ||
(G.have_quicktime && BLI_testextensie_array(path, imb_ext_image_qt)))
{
return FILE_TYPE_IMAGE;
}
else if (BLI_testextensie(path, ".ogg")) {
if (IMB_isanim(path)) {
return FILE_TYPE_MOVIE;
2013-10-31 23:52:44 +00:00
}
else {
return FILE_TYPE_SOUND;
}
}
else if (BLI_testextensie_array(path, imb_ext_movie)) {
return FILE_TYPE_MOVIE;
}
else if (BLI_testextensie_array(path, imb_ext_audio)) {
return FILE_TYPE_SOUND;
}
return 0;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static int file_extension_type(const char *dir, const char *relpath)
{
char path[FILE_MAX];
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
BLI_join_dirfile(path, sizeof(path), dir, relpath);
return path_extension_type(path);
}
int ED_file_extension_icon(const char *path)
{
int type = path_extension_type(path);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
switch (type) {
case FILE_TYPE_BLENDER:
return ICON_FILE_BLEND;
case FILE_TYPE_BLENDER_BACKUP:
return ICON_FILE_BACKUP;
case FILE_TYPE_IMAGE:
return ICON_FILE_IMAGE;
case FILE_TYPE_MOVIE:
return ICON_FILE_MOVIE;
case FILE_TYPE_PYSCRIPT:
return ICON_FILE_SCRIPT;
case FILE_TYPE_SOUND:
return ICON_FILE_SOUND;
case FILE_TYPE_FTFONT:
return ICON_FILE_FONT;
case FILE_TYPE_BTX:
return ICON_FILE_BLANK;
case FILE_TYPE_COLLADA:
return ICON_FILE_BLANK;
case FILE_TYPE_TEXT:
return ICON_FILE_TEXT;
default:
return ICON_FILE_BLANK;
}
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
int filelist_empty(struct FileList *filelist)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
return (filelist->filelist.nbr_entries == 0);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
unsigned int filelist_entry_select_set(
const FileList *filelist, const FileDirEntry *entry, FileSelType select, unsigned int flag, FileCheckType check)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
/* Default NULL pointer if not found is fine here! */
void **es_p = BLI_ghash_lookup_p(filelist->selection_state, entry->uuid);
unsigned int entry_flag = es_p ? GET_UINT_FROM_POINTER(*es_p) : 0;
const unsigned int org_entry_flag = entry_flag;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
BLI_assert(entry);
BLI_assert(ELEM(check, CHECK_DIRS, CHECK_FILES, CHECK_ALL));
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (((check == CHECK_ALL)) ||
((check == CHECK_DIRS) && (entry->typeflag & FILE_TYPE_DIR)) ||
((check == CHECK_FILES) && !(entry->typeflag & FILE_TYPE_DIR)))
{
switch (select) {
case FILE_SEL_REMOVE:
entry_flag &= ~flag;
break;
case FILE_SEL_ADD:
entry_flag |= flag;
break;
case FILE_SEL_TOGGLE:
entry_flag ^= flag;
break;
}
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (entry_flag != org_entry_flag) {
if (es_p) {
if (entry_flag) {
*es_p = SET_UINT_IN_POINTER(entry_flag);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
else {
BLI_ghash_remove(filelist->selection_state, entry->uuid, MEM_freeN, NULL);
}
}
else if (entry_flag) {
void *key = MEM_mallocN(sizeof(entry->uuid), __func__);
memcpy(key, entry->uuid, sizeof(entry->uuid));
BLI_ghash_insert(filelist->selection_state, key, SET_UINT_IN_POINTER(entry_flag));
}
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
return entry_flag;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
void filelist_entry_select_index_set(FileList *filelist, const int index, FileSelType select, unsigned int flag, FileCheckType check)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
FileDirEntry *entry = filelist_file(filelist, index);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (entry) {
filelist_entry_select_set(filelist, entry, select, flag, check);
}
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
void filelist_entries_select_index_range_set(
FileList *filelist, FileSelection *sel, FileSelType select, unsigned int flag, FileCheckType check)
{
/* select all valid files between first and last indicated */
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if ((sel->first >= 0) && (sel->first < filelist->filelist.nbr_entries_filtered) &&
(sel->last >= 0) && (sel->last < filelist->filelist.nbr_entries_filtered))
{
int current_file;
for (current_file = sel->first; current_file <= sel->last; current_file++) {
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
filelist_entry_select_index_set(filelist, current_file, select, flag, check);
}
}
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
unsigned int filelist_entry_select_get(FileList *filelist, FileDirEntry *entry, FileCheckType check)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
BLI_assert(entry);
BLI_assert(ELEM(check, CHECK_DIRS, CHECK_FILES, CHECK_ALL));
if (((check == CHECK_ALL)) ||
((check == CHECK_DIRS) && (entry->typeflag & FILE_TYPE_DIR)) ||
((check == CHECK_FILES) && !(entry->typeflag & FILE_TYPE_DIR)))
{
/* Default NULL pointer if not found is fine here! */
return GET_UINT_FROM_POINTER(BLI_ghash_lookup(filelist->selection_state, entry->uuid));
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
return 0;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
unsigned int filelist_entry_select_index_get(FileList *filelist, const int index, FileCheckType check)
{
FileDirEntry *entry = filelist_file(filelist, index);
if (entry) {
return filelist_entry_select_get(filelist, entry, check);
}
return 0;
}
bool filelist_islibrary(struct FileList *filelist, char *dir, char **group)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
return BLO_library_path_explode(filelist->filelist.root, dir, group, NULL);
}
static int groupname_to_code(const char *group)
{
2014-02-17 20:15:33 +11:00
char buf[BLO_GROUP_MAX];
char *lslash;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
BLI_assert(group);
BLI_strncpy(buf, group, sizeof(buf));
lslash = (char *)BLI_last_slash(buf);
if (lslash)
2012-06-18 13:01:24 +00:00
lslash[0] = '\0';
2014-02-17 20:15:33 +11:00
return buf[0] ? BKE_idcode_from_name(buf) : 0;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static unsigned int groupname_to_filter_id(const char *group)
{
int id_code = groupname_to_code(group);
return BKE_idcode_to_idfilter(id_code);
}
/*
* From here, we are in 'Job Context', i.e. have to be careful about sharing stuff between bacground working thread
* and main one (used by UI among other things).
*/
typedef struct TodoDir {
int level;
char *dir;
} TodoDir;
static int filelist_readjob_list_dir(
const char *root, ListBase *entries, const char *filter_glob,
const bool do_lib, const char *main_name, const bool skip_currpar)
{
struct direntry *files;
int nbr_files, nbr_entries = 0;
nbr_files = BLI_filelist_dir_contents(root, &files);
if (files) {
int i = nbr_files;
while (i--) {
FileListInternEntry *entry;
if (skip_currpar && FILENAME_IS_CURRPAR(files[i].relname)) {
continue;
}
entry = MEM_callocN(sizeof(*entry), __func__);
entry->relpath = MEM_dupallocN(files[i].relname);
if (S_ISDIR(files[i].s.st_mode)) {
entry->typeflag |= FILE_TYPE_DIR;
}
entry->st = files[i].s;
/* Set file type. */
/* If we are considering .blend files as libs, promote them to directory status! */
if (do_lib && BLO_has_bfile_extension(entry->relpath)) {
char name[FILE_MAX];
entry->typeflag = FILE_TYPE_BLENDER;
BLI_join_dirfile(name, sizeof(name), root, entry->relpath);
/* prevent current file being used as acceptable dir */
if (BLI_path_cmp(main_name, name) != 0) {
entry->typeflag |= FILE_TYPE_DIR;
}
}
/* Otherwise, do not check extensions for directories! */
else if (!(entry->typeflag & FILE_TYPE_DIR)) {
if (filter_glob[0] && BLI_testextensie_glob(entry->relpath, filter_glob)) {
entry->typeflag = FILE_TYPE_OPERATOR;
}
else {
entry->typeflag = file_extension_type(root, entry->relpath);
}
}
BLI_addtail(entries, entry);
nbr_entries++;
}
BLI_filelist_free(files, nbr_files);
}
return nbr_entries;
}
static int filelist_readjob_list_lib(const char *root, ListBase *entries, const bool skip_currpar)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
FileListInternEntry *entry;
LinkNode *ln, *names;
int i, nnames, idcode = 0, nbr_entries = 0;
char dir[FILE_MAX], *group;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
bool ok;
struct BlendHandle *libfiledata = NULL;
/* name test */
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
ok = BLO_library_path_explode(root, dir, &group, NULL);
if (!ok) {
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
return nbr_entries;
}
/* there we go */
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
libfiledata = BLO_blendhandle_from_file(dir, NULL);
if (libfiledata == NULL) {
return nbr_entries;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
/* memory for strings is passed into filelist[i].entry->relpath and freed in filelist_entry_free. */
if (group) {
idcode = groupname_to_code(group);
names = BLO_blendhandle_get_datablock_names(libfiledata, idcode, &nnames);
}
else {
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
names = BLO_blendhandle_get_linkable_groups(libfiledata);
nnames = BLI_linklist_count(names);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
BLO_blendhandle_close(libfiledata);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (!skip_currpar) {
entry = MEM_callocN(sizeof(*entry), __func__);
entry->relpath = BLI_strdup(FILENAME_PARENT);
entry->typeflag |= (FILE_TYPE_BLENDERLIB | FILE_TYPE_DIR);
BLI_addtail(entries, entry);
nbr_entries++;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
for (i = 0, ln = names; i < nnames; i++, ln = ln->next) {
const char *blockname = ln->link;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
entry = MEM_callocN(sizeof(*entry), __func__);
entry->relpath = BLI_strdup(blockname);
entry->typeflag |= FILE_TYPE_BLENDERLIB;
if (!(group && idcode)) {
entry->typeflag |= FILE_TYPE_DIR;
entry->blentype = groupname_to_code(blockname);
}
else {
entry->blentype = idcode;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
BLI_addtail(entries, entry);
nbr_entries++;
}
BLI_linklist_free(names, free);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
return nbr_entries;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
#if 0
/* Kept for reference here, in case we want to add back that feature later. We do not need it currently. */
/* Code ***NOT*** updated for job stuff! */
static void filelist_readjob_main_rec(struct FileList *filelist)
{
ID *id;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
FileDirEntry *files, *firstlib = NULL;
ListBase *lb;
int a, fake, idcode, ok, totlib, totbl;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
// filelist->type = FILE_MAIN; // XXX TODO: add modes to filebrowser
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
BLI_assert(filelist->filelist.entries == NULL);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (filelist->filelist.root[0] == '/') filelist->filelist.root[0] = '\0';
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (filelist->filelist.root[0]) {
idcode = groupname_to_code(filelist->filelist.root);
if (idcode == 0) filelist->filelist.root[0] = '\0';
}
2012-06-18 13:01:24 +00:00
if (filelist->dir[0] == 0) {
/* make directories */
#ifdef WITH_FREESTYLE
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
filelist->filelist.nbr_entries = 24;
#else
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
filelist->filelist.nbr_entries = 23;
#endif
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
filelist_resize(filelist, filelist->filelist.nbr_entries);
for (a = 0; a < filelist->filelist.nbr_entries; a++) {
filelist->filelist.entries[a].typeflag |= FILE_TYPE_DIR;
}
filelist->filelist.entries[0].entry->relpath = BLI_strdup(FILENAME_PARENT);
filelist->filelist.entries[1].entry->relpath = BLI_strdup("Scene");
filelist->filelist.entries[2].entry->relpath = BLI_strdup("Object");
filelist->filelist.entries[3].entry->relpath = BLI_strdup("Mesh");
filelist->filelist.entries[4].entry->relpath = BLI_strdup("Curve");
filelist->filelist.entries[5].entry->relpath = BLI_strdup("Metaball");
filelist->filelist.entries[6].entry->relpath = BLI_strdup("Material");
filelist->filelist.entries[7].entry->relpath = BLI_strdup("Texture");
filelist->filelist.entries[8].entry->relpath = BLI_strdup("Image");
filelist->filelist.entries[9].entry->relpath = BLI_strdup("Ika");
filelist->filelist.entries[10].entry->relpath = BLI_strdup("Wave");
filelist->filelist.entries[11].entry->relpath = BLI_strdup("Lattice");
filelist->filelist.entries[12].entry->relpath = BLI_strdup("Lamp");
filelist->filelist.entries[13].entry->relpath = BLI_strdup("Camera");
filelist->filelist.entries[14].entry->relpath = BLI_strdup("Ipo");
filelist->filelist.entries[15].entry->relpath = BLI_strdup("World");
filelist->filelist.entries[16].entry->relpath = BLI_strdup("Screen");
filelist->filelist.entries[17].entry->relpath = BLI_strdup("VFont");
filelist->filelist.entries[18].entry->relpath = BLI_strdup("Text");
filelist->filelist.entries[19].entry->relpath = BLI_strdup("Armature");
filelist->filelist.entries[20].entry->relpath = BLI_strdup("Action");
filelist->filelist.entries[21].entry->relpath = BLI_strdup("NodeTree");
filelist->filelist.entries[22].entry->relpath = BLI_strdup("Speaker");
#ifdef WITH_FREESTYLE
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
filelist->filelist.entries[23].entry->relpath = BLI_strdup("FreestyleLineStyle");
#endif
}
else {
/* make files */
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
idcode = groupname_to_code(filelist->filelist.root);
2012-06-18 13:01:24 +00:00
lb = which_libbase(G.main, idcode);
if (lb == NULL) return;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
filelist->filelist.nbr_entries = 0;
for (id = lb->first; id; id = id->next) {
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (!(filelist->filter_data.flags & FLF_HIDE_DOT) || id->name[2] != '.') {
filelist->filelist.nbr_entries++;
}
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
/* XXX TODO: if databrowse F4 or append/link filelist->flags & FLF_HIDE_PARENT has to be set */
if (!(filelist->filter_data.flags & FLF_HIDE_PARENT))
filelist->filelist.nbr_entries++;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (filelist->filelist.nbr_entries > 0) {
filelist_resize(filelist, filelist->filelist.nbr_entries);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
files = filelist->filelist.entries;
if (!(filelist->filter_data.flags & FLF_HIDE_PARENT)) {
files->entry->relpath = BLI_strdup(FILENAME_PARENT);
files->typeflag |= FILE_TYPE_DIR;
files++;
}
2012-06-18 13:01:24 +00:00
totlib = totbl = 0;
for (id = lb->first; id; id = id->next) {
ok = 1;
if (ok) {
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (!(filelist->filter_data.flags & FLF_HIDE_DOT) || id->name[2] != '.') {
if (id->lib == NULL) {
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
files->entry->relpath = BLI_strdup(id->name + 2);
}
else {
2015-08-20 08:51:35 +10:00
char relname[FILE_MAX + (MAX_ID_NAME - 2) + 3];
2015-07-12 05:50:07 +10:00
BLI_snprintf(relname, sizeof(relname), "%s | %s", id->lib->name, id->name + 2);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
files->entry->relpath = BLI_strdup(relname);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
// files->type |= S_IFREG;
#if 0 /* XXX TODO show the selection status of the objects */
if (!filelist->has_func) { /* F4 DATA BROWSE */
2012-06-18 13:01:24 +00:00
if (idcode == ID_OB) {
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if ( ((Object *)id)->flag & SELECT) files->entry->selflag |= FILE_SEL_SELECTED;
}
2012-06-18 13:01:24 +00:00
else if (idcode == ID_SCE) {
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if ( ((Scene *)id)->r.scemode & R_BG_RENDER) files->entry->selflag |= FILE_SEL_SELECTED;
}
}
#endif
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
// files->entry->nr = totbl + 1;
files->entry->poin = id;
2012-06-18 13:01:24 +00:00
fake = id->flag & LIB_FAKEUSER;
if (idcode == ID_MA || idcode == ID_TE || idcode == ID_LA || idcode == ID_WO || idcode == ID_IM) {
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
files->typeflag |= FILE_TYPE_IMAGE;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
// if (id->lib && fake) BLI_snprintf(files->extra, sizeof(files->entry->extra), "LF %d", id->us);
// else if (id->lib) BLI_snprintf(files->extra, sizeof(files->entry->extra), "L %d", id->us);
// else if (fake) BLI_snprintf(files->extra, sizeof(files->entry->extra), "F %d", id->us);
// else BLI_snprintf(files->extra, sizeof(files->entry->extra), " %d", id->us);
if (id->lib) {
2012-06-18 13:01:24 +00:00
if (totlib == 0) firstlib = files;
totlib++;
}
files++;
}
totbl++;
}
}
/* only qsort of library blocks */
2012-06-18 13:01:24 +00:00
if (totlib > 1) {
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
qsort(firstlib, totlib, sizeof(*files), compare_name);
}
}
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
#endif
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static void filelist_readjob_do(
const bool do_lib,
FileList *filelist, const char *main_name, short *stop, short *do_update, float *progress, ThreadMutex *lock)
{
ListBase entries = {0};
BLI_Stack *todo_dirs;
TodoDir *td_dir;
char dir[FILE_MAX_LIBEXTRA];
char filter_glob[64]; /* TODO should be define! */
const char *root = filelist->filelist.root;
const int max_recursion = filelist->max_recursion;
int nbr_done_dirs = 0, nbr_todo_dirs = 1;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
// BLI_assert(filelist->filtered == NULL);
BLI_assert(BLI_listbase_is_empty(&filelist->filelist.entries) && (filelist->filelist.nbr_entries == 0));
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
todo_dirs = BLI_stack_new(sizeof(*td_dir), __func__);
td_dir = BLI_stack_push_r(todo_dirs);
td_dir->level = 1;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
BLI_strncpy(dir, filelist->filelist.root, sizeof(dir));
BLI_strncpy(filter_glob, filelist->filter_data.filter_glob, sizeof(filter_glob));
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
BLI_cleanup_dir(main_name, dir);
td_dir->dir = BLI_strdup(dir);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
while (!BLI_stack_is_empty(todo_dirs) && !(*stop)) {
FileListInternEntry *entry;
int nbr_entries = 0;
bool is_lib = do_lib;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
char *subdir;
int recursion_level;
bool skip_currpar;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
td_dir = BLI_stack_peek(todo_dirs);
subdir = td_dir->dir;
recursion_level = td_dir->level;
skip_currpar = (recursion_level > 1);
BLI_stack_discard(todo_dirs);
if (do_lib) {
nbr_entries = filelist_readjob_list_lib(subdir, &entries, skip_currpar);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (!nbr_entries) {
is_lib = false;
nbr_entries = filelist_readjob_list_dir(subdir, &entries, filter_glob, do_lib, main_name, skip_currpar);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
for (entry = entries.first; entry; entry = entry->next) {
BLI_join_dirfile(dir, sizeof(dir), subdir, entry->relpath);
BLI_cleanup_file(root, dir);
2015-08-19 23:54:49 +02:00
/* Generate our entry uuid. Abusing uuid as an uint32, shall be more than enough here,
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
* things would crash way before we overflow that counter!
* Using an atomic operation to avoid having to lock thread...
* Note that we do not really need this here currently, since there is a single listing thread, but better
2015-09-02 18:39:19 +10:00
* remain consistent about threading! */
2015-08-19 23:54:49 +02:00
*((uint32_t *)entry->uuid) = atomic_add_uint32((uint32_t *)filelist->filelist_intern.curr_uuid, 1);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
BLI_path_rel(dir, root);
/* Only thing we change in direntry here, so we need to free it first. */
MEM_freeN(entry->relpath);
entry->relpath = BLI_strdup(dir + 2); /* + 2 to remove '//' added by BLI_path_rel */
entry->name = BLI_strdup(fileentry_uiname(root, entry->relpath, entry->typeflag, dir));
/* Here we decide whether current filedirentry is to be listed too, or not. */
if (max_recursion && (is_lib || (recursion_level <= max_recursion))) {
if (((entry->typeflag & FILE_TYPE_DIR) == 0) || FILENAME_IS_CURRPAR(entry->relpath)) {
/* Skip... */
}
else if (!is_lib && (recursion_level >= max_recursion) &&
2015-08-20 08:51:35 +10:00
((entry->typeflag & (FILE_TYPE_BLENDER | FILE_TYPE_BLENDER_BACKUP)) == 0))
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
{
/* Do not recurse in real directories in this case, only in .blend libs. */
}
else {
/* We have a directory we want to list, add it to todo list! */
BLI_join_dirfile(dir, sizeof(dir), root, entry->relpath);
BLI_cleanup_dir(main_name, dir);
td_dir = BLI_stack_push_r(todo_dirs);
td_dir->level = recursion_level + 1;
td_dir->dir = BLI_strdup(dir);
nbr_todo_dirs++;
}
}
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (nbr_entries) {
BLI_mutex_lock(lock);
*do_update = true;
BLI_movelisttolist(&filelist->filelist.entries, &entries);
filelist->filelist.nbr_entries += nbr_entries;
BLI_mutex_unlock(lock);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
nbr_done_dirs++;
*progress = (float)nbr_done_dirs / (float)nbr_todo_dirs;
MEM_freeN(subdir);
}
/* If we were interrupted by stop, stack may not be empty and we need to free pending dir paths. */
while (!BLI_stack_is_empty(todo_dirs)) {
td_dir = BLI_stack_peek(todo_dirs);
MEM_freeN(td_dir->dir);
BLI_stack_discard(todo_dirs);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
BLI_stack_free(todo_dirs);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static void filelist_readjob_dir(
FileList *filelist, const char *main_name, short *stop, short *do_update, float *progress, ThreadMutex *lock)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
filelist_readjob_do(false, filelist, main_name, stop, do_update, progress, lock);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static void filelist_readjob_lib(
FileList *filelist, const char *main_name, short *stop, short *do_update, float *progress, ThreadMutex *lock)
{
filelist_readjob_do(true, filelist, main_name, stop, do_update, progress, lock);
}
static void filelist_readjob_main(
FileList *filelist, const char *main_name, short *stop, short *do_update, float *progress, ThreadMutex *lock)
{
/* TODO! */
filelist_readjob_dir(filelist, main_name, stop, do_update, progress, lock);
}
typedef struct FileListReadJob {
ThreadMutex lock;
char main_name[FILE_MAX];
struct FileList *filelist;
struct FileList *tmp_filelist; /* XXX We may use a simpler struct here... just a linked list and root path? */
} FileListReadJob;
static void filelist_readjob_startjob(void *flrjv, short *stop, short *do_update, float *progress)
{
FileListReadJob *flrj = flrjv;
// printf("START filelist reading (%d files, main thread: %d)\n",
// flrj->filelist->filelist.nbr_entries, BLI_thread_is_main());
BLI_mutex_lock(&flrj->lock);
BLI_assert((flrj->tmp_filelist == NULL) && flrj->filelist);
flrj->tmp_filelist = MEM_dupallocN(flrj->filelist);
BLI_listbase_clear(&flrj->tmp_filelist->filelist.entries);
flrj->tmp_filelist->filelist.nbr_entries = 0;
flrj->tmp_filelist->filelist_intern.filtered = NULL;
BLI_listbase_clear(&flrj->tmp_filelist->filelist_intern.entries);
memset(flrj->tmp_filelist->filelist_intern.curr_uuid, 0, sizeof(flrj->tmp_filelist->filelist_intern.curr_uuid));
flrj->tmp_filelist->libfiledata = NULL;
memset(&flrj->tmp_filelist->filelist_cache, 0, sizeof(flrj->tmp_filelist->filelist_cache));
flrj->tmp_filelist->selection_state = NULL;
BLI_mutex_unlock(&flrj->lock);
flrj->tmp_filelist->read_jobf(flrj->tmp_filelist, flrj->main_name, stop, do_update, progress, &flrj->lock);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static void filelist_readjob_update(void *flrjv)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
FileListReadJob *flrj = flrjv;
FileListIntern *fl_intern = &flrj->filelist->filelist_intern;
ListBase new_entries = {NULL};
int nbr_entries, new_nbr_entries = 0;
BLI_movelisttolist(&new_entries, &fl_intern->entries);
nbr_entries = flrj->filelist->filelist.nbr_entries;
BLI_mutex_lock(&flrj->lock);
if (flrj->tmp_filelist->filelist.nbr_entries) {
/* We just move everything out of 'thread context' into final list. */
new_nbr_entries = flrj->tmp_filelist->filelist.nbr_entries;
BLI_movelisttolist(&new_entries, &flrj->tmp_filelist->filelist.entries);
flrj->tmp_filelist->filelist.nbr_entries = 0;
}
BLI_mutex_unlock(&flrj->lock);
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
if (new_nbr_entries) {
/* Do not clear selection cache, we can assume already 'selected' uuids are still valid! */
filelist_clear_ex(flrj->filelist, true, false);
flrj->filelist->flags |= (FL_NEED_SORTING | FL_NEED_FILTERING);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
/* if no new_nbr_entries, this is NOP */
BLI_movelisttolist(&fl_intern->entries, &new_entries);
flrj->filelist->filelist.nbr_entries = nbr_entries + new_nbr_entries;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static void filelist_readjob_endjob(void *flrjv)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
FileListReadJob *flrj = flrjv;
/* In case there would be some dangling update... */
filelist_readjob_update(flrjv);
flrj->filelist->flags &= ~FL_IS_PENDING;
flrj->filelist->flags |= FL_IS_READY;
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
static void filelist_readjob_free(void *flrjv)
{
FileListReadJob *flrj = flrjv;
// printf("END filelist reading (%d files)\n", flrj->filelist->filelist.nbr_entries);
if (flrj->tmp_filelist) {
/* tmp_filelist shall never ever be filtered! */
BLI_assert(flrj->tmp_filelist->filelist.nbr_entries == 0);
BLI_assert(BLI_listbase_is_empty(&flrj->tmp_filelist->filelist.entries));
filelist_freelib(flrj->tmp_filelist);
filelist_free(flrj->tmp_filelist);
MEM_freeN(flrj->tmp_filelist);
}
BLI_mutex_end(&flrj->lock);
MEM_freeN(flrj);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
void filelist_readjob_start(FileList *filelist, const bContext *C)
{
wmJob *wm_job;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
FileListReadJob *flrj;
/* prepare job data */
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
flrj = MEM_callocN(sizeof(*flrj), __func__);
flrj->filelist = filelist;
BLI_strncpy(flrj->main_name, G.main->name, sizeof(flrj->main_name));
filelist->flags &= ~(FL_FORCE_RESET | FL_IS_READY);
filelist->flags |= FL_IS_PENDING;
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
BLI_mutex_init(&flrj->lock);
/* setup job */
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
wm_job = WM_jobs_get(CTX_wm_manager(C), CTX_wm_window(C), CTX_wm_area(C), "Listing Dirs...",
WM_JOB_PROGRESS, WM_JOB_TYPE_FILESEL_READDIR);
WM_jobs_customdata_set(wm_job, flrj, filelist_readjob_free);
WM_jobs_timer(wm_job, 0.01, NC_SPACE | ND_SPACE_FILE_LIST, NC_SPACE | ND_SPACE_FILE_LIST);
WM_jobs_callbacks(wm_job, filelist_readjob_startjob, NULL, filelist_readjob_update, filelist_readjob_endjob);
/* start the job */
WM_jobs_start(CTX_wm_manager(C), wm_job);
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
void filelist_readjob_stop(wmWindowManager *wm, ScrArea *sa)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
WM_jobs_kill_type(wm, sa, WM_JOB_TYPE_FILESEL_READDIR);
2010-03-30 11:38:06 +00:00
}
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
int filelist_readjob_running(wmWindowManager *wm, ScrArea *sa)
{
Final 'FileBrowser First Stage' merge. It basically rewrites most of filelist.c, with some more limited changes in other areas of filebrowser. From user perspective, it: * Removes some info in 'long' drawing mode (owner, permissions) - OS-specific data that do not really matter in Blender! * Makes short/long display 'fixed' size (among four choices, like thumbnails mode). * Allows to list several layers of dirtree at once, in a flat way (inside .blend files and/or real directories). * Consequently, adds datablocks types filtering. * Uses way less RAM when listing big directories, especially in thumbnail mode (we are talking of several hundred of MiB spared). * Generates thumbnails way faster. From code perspective, it: * Is ready for asset engine needs (on data structure level in filebrowser's listing). * Simplifies and makes 'generic' file listing much lighter. * Separates file listing in three different aspects: ** 'generic' filelisting (in BLI), which becomes a shallow wrapper around stat struct. ** 'filebrowser drawing' filelisting, which only contains current visible subset of the whole list (sliding window), with extra drawing data (strings for size, date/time, preview, etc.). ** 'asset-ready' filelisting, which is used for operations common to 'basic' filehandling and future asset-related one. * Uses uuid's to handle file selection/state in the browser, instead of using flags in filelisting items. * Uses much lighter BLI_task handling for previews, instead of heavy 'job' system (using the new 'notifier' timer to handle UI refresh, in similar way to jobs). * Moves .blend datablocks preview handling to IMB_thumbnail (necessary to avoid storing all datablock previews at once, and gives better consistency and performances too). Revision: https://developer.blender.org/D1316 Thanks to Campbell & Sergey for the reviews. :)
2015-08-19 22:41:39 +02:00
return WM_jobs_test(wm, sa, WM_JOB_TYPE_FILESEL_READDIR);
}