From a272ddcc7869ec2e700ab6d397fc1854760f7e64 Mon Sep 17 00:00:00 2001
From: epriestley <git@epriestley.com>
Date: Sun, 4 May 2014 10:48:49 -0700
Subject: [PATCH] Fix two parsing issues for diffs with damaged whitespace

Summary:
Fixes T4941. If a diff has had trailing whitespace stripped, we will fail to handle empty lines correctly (previously, these lines had a leading space when the original tool emitted them).

(This probably stopped working around the time we began retaining newlines.)

Test Plan: The diff in T4941 now parses and renders correctly.

Reviewers: asherkin, btrahan

Reviewed By: btrahan

Subscribers: epriestley

Maniphest Tasks: T4941

Differential Revision: https://secure.phabricator.com/D8968
---
 .../parser/DifferentialHunkParser.php         | 35 ++++++++++++++++---
 1 file changed, 30 insertions(+), 5 deletions(-)

diff --git a/src/applications/differential/parser/DifferentialHunkParser.php b/src/applications/differential/parser/DifferentialHunkParser.php
index ef4b696a0e..4964e109c3 100644
--- a/src/applications/differential/parser/DifferentialHunkParser.php
+++ b/src/applications/differential/parser/DifferentialHunkParser.php
@@ -423,16 +423,41 @@ final class DifferentialHunkParser {
       $lines = phutil_split_lines($lines);
 
       $line_type_map = array();
+      $line_text = array();
       foreach ($lines as $line_index => $line) {
         if (isset($line[0])) {
           $char = $line[0];
-          if ($char == ' ') {
-            $line_type_map[$line_index] = null;
-          } else {
-            $line_type_map[$line_index] = $char;
+          switch ($char) {
+            case " ":
+              $line_type_map[$line_index] = null;
+              $line_text[$line_index] = substr($line, 1);
+              break;
+            case "\r":
+            case "\n":
+              // NOTE: Normally, the first character is a space, plus, minus or
+              // backslash, but it may be a newline if it used to be a space and
+              // trailing whitespace has been stripped via email transmission or
+              // some similar mechanism. In these cases, we essentially pretend
+              // the missing space is still there.
+              $line_type_map[$line_index] = null;
+              $line_text[$line_index] = $line;
+              break;
+            case "+":
+            case "-":
+            case "\\":
+              $line_type_map[$line_index] = $char;
+              $line_text[$line_index] = substr($line, 1);
+              break;
+            default:
+              throw new Exception(
+                pht(
+                  'Unexpected leading character "%s" at line index %s!',
+                  $char,
+                  $line_index));
           }
         } else {
           $line_type_map[$line_index] = null;
+          $line_text[$line_index] = '';
         }
       }
 
@@ -444,7 +469,7 @@ final class DifferentialHunkParser {
         $type = $line_type_map[$cursor];
         $data = array(
           'type'  => $type,
-          'text'  => (string)substr($lines[$cursor], 1),
+          'text'  => $line_text[$cursor],
           'line'  => $new_line,
         );
         if ($type == '\\') {