From: Oleg Broytman <phd@phdru.name>
Date: Sun, 8 Aug 2010 19:52:35 +0000 (+0000)
Subject: Fixed parsing in case of unknown entity.
X-Git-Tag: v4.5.3~133
X-Git-Url: https://git.phdru.name/?a=commitdiff_plain;h=83ac2470649ec5932119ee66dd1ce6829278977c;p=bookmarks_db.git

Fixed parsing in case of unknown entity.


git-svn-id: file:///home/phd/archive/SVN/bookmarks_db/trunk@243 fdd5c36f-1aea-0310-aeeb-c58d7e2b6c23
---

diff --git a/Robots/parse_html.py b/Robots/parse_html.py
index 80c7aa8..fd8dc92 100755
--- a/Robots/parse_html.py
+++ b/Robots/parse_html.py
@@ -34,8 +34,12 @@ def recode_entities(title, charset):
    for part in entity_re.split(title):
       if part not in ("&amp;", "&lt;", "&gt;", "&quot;") and \
             entity_re.match(part):
-         part = unichr(name2codepoint.get(part[1:-1], part)).encode(charset)
-      output.append(part)
+         _part = name2codepoint.get(part[1:-1], None)
+         if _part is None:
+             _part = part
+         else:
+             _part = unichr(_part).encode(charset)
+      output.append(_part)
    title = ''.join(output)
 
    output = []