From: Oleg Broytman <phd@phdru.name>
Date: Tue, 4 Mar 2008 10:55:43 +0000 (+0000)
Subject: Ignore case for DOCTYPE.
X-Git-Tag: v4.5.3~176
X-Git-Url: https://git.phdru.name/?a=commitdiff_plain;h=05955edd8eb929042d179dcbbd81333af2714128;p=bookmarks_db.git

Ignore case for DOCTYPE.


git-svn-id: file:///home/phd/archive/SVN/bookmarks_db/trunk@200 fdd5c36f-1aea-0310-aeeb-c58d7e2b6c23
---

diff --git a/Robots/parse_html_beautifulsoup.py b/Robots/parse_html_beautifulsoup.py
index c2a4032..ca8ee2e 100644
--- a/Robots/parse_html_beautifulsoup.py
+++ b/Robots/parse_html_beautifulsoup.py
@@ -39,7 +39,7 @@ class BadDeclParser(BeautifulSoup):
              except SGMLParseError:
                  # Could not parse the DOCTYPE declaration
                  # Try to just skip the actual declaration
-                 match = re.search(r'<!DOCTYPE([^>]*?)>', self.rawdata[i:], re.MULTILINE)
+                 match = re.search(r'<!DOCTYPE([^>]*?)>', self.rawdata[i:], re.MULTILINE|re.I)
                  if match:
                      toHandle = self.rawdata[i:match.end()]
                  else: