From 918694a0964656840d3bfb771d74544b61199631 Mon Sep 17 00:00:00 2001
From: Damien Elmes <git@ichi2.net>
Date: Wed, 22 May 2013 09:45:58 +0900
Subject: [PATCH] more media regexp fixes

- split quoted and unquoted image case into separate regexps, since we can't
  include a group reference inside a character set
- disallow spaces in the non-quoted case
- this should fix matching on images with other attributes again
---
 anki/media.py       | 16 +++++++++++-----
 tests/test_media.py |  6 +++++-
 2 files changed, 16 insertions(+), 6 deletions(-)

diff --git a/anki/media.py b/anki/media.py
index d7e51cbe4..5ee7bc260 100644
--- a/anki/media.py
+++ b/anki/media.py
@@ -12,10 +12,14 @@ from anki.latex import mungeQA
 
 class MediaManager(object):
 
-    # other code depends on this order, so don't reorder
-    regexps = ("(?i)(\[sound:(?P<fname>[^]]+)\])",
-               "(?i)(<img[^>]+src=(?P<str>[\"']?)"+
-                "(?P<fname>[^>]+)(?P=str)[^>]*>)")
+    soundRegexps = ["(?i)(\[sound:(?P<fname>[^]]+)\])"]
+    imgRegexps = [
+        # src element quoted case
+        "(?i)(<img[^>]+src=(?P<str>[\"'])(?P<fname>[^>]+?)(?P=str)[^>]*>)",
+        # unquoted case
+        "(?i)(<img[^>]+src=(?!['\"])(?P<fname>[^ >]+)[^>]*?>)",
+    ]
+    regexps = soundRegexps + imgRegexps
 
     def __init__(self, col, server):
         self.col = col
@@ -175,7 +179,9 @@ If the same name exists, compare checksums."""
                 return tag
             return tag.replace(
                 fname, urllib.quote(fname.encode("utf-8")))
-        return re.sub(self.regexps[1], repl, string)
+        for reg in self.imgRegexps:
+            string = re.sub(reg, repl, string)
+        return string
 
     # Rebuilding DB
     ##########################################################################
diff --git a/tests/test_media.py b/tests/test_media.py
index 308ddd297..49c349be9 100644
--- a/tests/test_media.py
+++ b/tests/test_media.py
@@ -23,7 +23,11 @@ def test_strings():
     mid = d.models.models.keys()[0]
     assert mf(mid, "aoeu") == []
     assert mf(mid, "aoeu<img src='foo.jpg'>ao") == ["foo.jpg"]
-    assert mf(mid, "aoeu<img src=foo bar.jpg>ao") == ["foo bar.jpg"]
+    assert mf(mid, "aoeu<img src='foo.jpg' style='test'>ao") == ["foo.jpg"]
+    assert mf(mid, "aoeu<img src='foo.jpg'><img src=\"bar.jpg\">ao") == [
+            "foo.jpg", "bar.jpg"]
+    assert mf(mid, "aoeu<img src=foo.jpg style=bar>ao") == ["foo.jpg"]
+    assert mf(mid, "<img src=one><img src=two>") == ["one", "two"]
     assert mf(mid, "aoeu<img src=\"foo.jpg\">ao") == ["foo.jpg"]
     assert mf(mid, "aoeu<img src=\"foo.jpg\"><img class=yo src=fo>ao") == [
             "foo.jpg", "fo"]