From: Oleg Broytman <phd@phdru.name>
Date: Fri, 13 Aug 2010 14:53:58 +0000 (+0000)
Subject: Test for completely broken HTML.
X-Git-Tag: v4.5.3~94
X-Git-Url: https://git.phdru.name/?a=commitdiff_plain;h=1a001aafb9fefdbc003ee5ce49ca19d4f4f7e3ec;p=bookmarks_db.git

Test for completely broken HTML.


git-svn-id: file:///home/phd/archive/SVN/bookmarks_db/trunk@282 fdd5c36f-1aea-0310-aeeb-c58d7e2b6c23
---

diff --git a/Robots/parse_html_lxml.py b/Robots/parse_html_lxml.py
index 7aba09c..638af4e 100644
--- a/Robots/parse_html_lxml.py
+++ b/Robots/parse_html_lxml.py
@@ -11,6 +11,9 @@ from parse_html_util import HTMLParser
 def parse_html(filename, charset=None, log=None):
     html_tree = parse(filename)
 
+    if t.getroot() is None:
+        return None
+
     title = html_tree.findtext('head/title')
     if title is None:
         title = html_tree.findtext('title')