From 3bbe3e07c526ef31e1e0901c05e077ad406c9300 Mon Sep 17 00:00:00 2001 From: Oleg Broytman Date: Wed, 13 Feb 2008 18:17:53 +0000 Subject: [PATCH] Replace BeautifulSoup's guessed cp1252 with DEFAULT_CHARSET. git-svn-id: file:///home/phd/archive/SVN/bookmarks_db/trunk@176 fdd5c36f-1aea-0310-aeeb-c58d7e2b6c23 --- Robots/parse_html_beautifulsoup.py | 3 +++ 1 file changed, 3 insertions(+) diff --git a/Robots/parse_html_beautifulsoup.py b/Robots/parse_html_beautifulsoup.py index 9b0fade..c7263fe 100644 --- a/Robots/parse_html_beautifulsoup.py +++ b/Robots/parse_html_beautifulsoup.py @@ -67,6 +67,9 @@ def parse_html(filename, charset=None): head = root.html # Some sites put TITLE in HTML without HEAD _charset = root.originalEncoding + if _charset == "windows-1252": # Replace default + _charset = DEFAULT_CHARSET + try: title = head.title.string.encode(_charset) except AttributeError: -- 2.39.2