From: Oleg Broytman <phd@phdru.name>
Date: Fri, 13 Aug 2010 13:06:20 +0000 (+0000)
Subject: Lookup title in html if not found in head.
X-Git-Tag: v4.5.3~99
X-Git-Url: https://git.phdru.name/?p=bookmarks_db.git;a=commitdiff_plain;h=331fbf809d7bd3cc14faeb7b3a72b25bc54f6dfb

Lookup title in html if not found in head.


git-svn-id: file:///home/phd/archive/SVN/bookmarks_db/trunk@277 fdd5c36f-1aea-0310-aeeb-c58d7e2b6c23
---

diff --git a/Robots/parse_html_html5.py b/Robots/parse_html_html5.py
index 2302051..43e8d74 100644
--- a/Robots/parse_html_html5.py
+++ b/Robots/parse_html_html5.py
@@ -83,4 +83,16 @@ def parse_html(filename, charset=None, log=None):
                 icon = node.attributes['href']
                 break
 
+    else:
+        for node in html.childNodes:
+            if node.name == 'title':
+                if node.childNodes:
+                    title = node.childNodes[0].value
+                    break
+                else:
+                    title = ''
+
+        if title is None:
+            return None
+
     return HTMLParser(charset, meta_charset, title, refresh, icon)