From 62954ac1573082249dbe1a748d3b1982703edbbd Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Sybren=20A=2E=20St=C3=BCvel?= <sybren@stuvel.eu>
Date: Fri, 15 Sep 2017 16:47:40 +0200
Subject: [PATCH] Latest assets/comments: using Mongo aggregation instead of
 Python code

---
 pillar/api/latest.py | 118 +++++++++++++++++++++----------------------
 1 file changed, 59 insertions(+), 59 deletions(-)

diff --git a/pillar/api/latest.py b/pillar/api/latest.py
index c834e713..8cffbb78 100644
--- a/pillar/api/latest.py
+++ b/pillar/api/latest.py
@@ -1,5 +1,8 @@
+import functools
 import itertools
+import typing
 
+import bson
 import pymongo
 from flask import Blueprint, current_app
 
@@ -12,99 +15,95 @@ def keep_fetching(collection, db_filter, projection, sort, py_filter,
                   batch_size=12):
     """Yields results for which py_filter returns True"""
 
-    projection['_deleted'] = 1
+    db_filter['_deleted'] = {'$ne': True}
     curs = collection.find(db_filter, projection).sort(sort)
     curs.batch_size(batch_size)
 
     for doc in curs:
-        if doc.get('_deleted'):
-            continue
-        doc.pop('_deleted', None)
         if py_filter(doc):
             yield doc
 
 
-def latest_nodes(db_filter, projection, py_filter, limit):
-    nodes = current_app.data.driver.db['nodes']
+def _public_project_ids() -> typing.List[bson.ObjectId]:
+    """Returns a list of ObjectIDs of public projects.
+
+    Memoized in setup_app().
+    """
+
+    proj_coll = current_app.db('projects')
+    result = proj_coll.find({'is_private': False}, {'_id': 1})
+    return [p['_id'] for p in result]
+
+
+def latest_nodes(db_filter, projection, limit):
+    """Returns the latest nodes, of a certain type, of public projects.
+
+    Also includes information about the project and the user of each node.
+    """
 
     proj = {
         '_created': 1,
         '_updated': 1,
+        'user.full_name': 1,
+        'project._id': 1,
+        'project.url': 1,
+        'project.name': 1,
+        'name': 1,
+        'node_type': 1,
+        'parent': 1,
+        **projection,
     }
-    proj.update(projection)
 
-    latest = keep_fetching(nodes, db_filter, proj,
-                           [('_created', pymongo.DESCENDING)],
-                           py_filter, limit)
+    nodes_coll = current_app.db('nodes')
+    pipeline = [
+        {'$match': {'_deleted': {'$ne': True}}},
+        {'$match': db_filter},
+        {'$match': {'project': {'$in': _public_project_ids()}}},
+        {'$sort': {'_created': pymongo.DESCENDING}},
+        {'$limit': limit},
+        {'$lookup': {"from": "users",
+                     "localField": "user",
+                     "foreignField": "_id",
+                     "as": "user"}},
+        {'$unwind': {'path': "$user"}},
+        {'$lookup': {"from": "projects",
+                     "localField": "project",
+                     "foreignField": "_id",
+                     "as": "project"}},
+        {'$unwind': {'path': "$project"}},
+        {'$project': proj},
+    ]
 
-    result = list(itertools.islice(latest, limit))
-    return result
-
-
-def has_public_project(node_doc):
-    """Returns True iff the project the node belongs to is public."""
-
-    project_id = node_doc.get('project')
-    return is_project_public(project_id)
-
-
-# TODO: cache result, for a limited amt. of time, or for this HTTP request.
-def is_project_public(project_id):
-    """Returns True iff the project is public."""
-
-    project = current_app.data.driver.db['projects'].find_one(project_id)
-    if not project:
-        return False
-
-    return not project.get('is_private')
+    print('QUERY: db.nodes.aggregate(%r)' % pipeline)
+    latest = nodes_coll.aggregate(pipeline)
+    return list(latest)
 
 
 @blueprint.route('/assets')
 def latest_assets():
     latest = latest_nodes({'node_type': 'asset',
                            'properties.status': 'published'},
-                          {'name': 1, 'project': 1, 'user': 1, 'node_type': 1,
+                          {'name': 1, 'node_type': 1,
                            'parent': 1, 'picture': 1, 'properties.status': 1,
                            'properties.content_type': 1,
                            'permissions.world': 1},
-                          has_public_project, 12)
-
-    embed_user(latest)
-    embed_project(latest)
+                          12)
 
     return jsonify({'_items': latest})
 
 
-def embed_user(latest):
-    users = current_app.data.driver.db['users']
-
-    for comment in latest:
-        user_id = comment['user']
-        comment['user'] = users.find_one(user_id, {
-            'auth': 0, 'groups': 0, 'roles': 0, 'settings': 0, 'email': 0,
-            '_created': 0, '_updated': 0, '_etag': 0})
-
-
-def embed_project(latest):
-    projects = current_app.data.driver.db['projects']
-
-    for comment in latest:
-        project_id = comment['project']
-        comment['project'] = projects.find_one(project_id, {'_id': 1, 'name': 1,
-                                                            'url': 1})
-
-
 @blueprint.route('/comments')
 def latest_comments():
     latest = latest_nodes({'node_type': 'comment',
                            'properties.status': 'published'},
-                          {'project': 1, 'parent': 1, 'user': 1,
+                          {'parent': 1,
                            'properties.content': 1, 'node_type': 1,
                            'properties.status': 1,
                            'properties.is_reply': 1},
-                          has_public_project, 10)
+                          10)
 
     # Embed the comments' parents.
+    # TODO: move to aggregation pipeline.
     nodes = current_app.data.driver.db['nodes']
     parents = {}
     for comment in latest:
@@ -118,11 +117,12 @@ def latest_comments():
         parents[parent_id] = parent
         comment['parent'] = parent
 
-    embed_project(latest)
-    embed_user(latest)
-
     return jsonify({'_items': latest})
 
 
 def setup_app(app, url_prefix):
+    global _public_project_ids
+
     app.register_api_blueprint(blueprint, url_prefix=url_prefix)
+    cached = app.cache.cached(timeout=3600)
+    _public_project_ids = cached(_public_project_ids)