From: Oleg Broytman <phd@phdru.name>
Date: Thu, 27 Dec 2007 19:33:24 +0000 (+0000)
Subject: Strip every line in title.
X-Git-Tag: v4.5.3~241
X-Git-Url: https://git.phdru.name/?a=commitdiff_plain;h=5bf527cabbb158190e9b1ac17e19684135d9a6d2;p=bookmarks_db.git

Strip every line in title.


git-svn-id: file:///home/phd/archive/SVN/bookmarks_db/trunk@135 fdd5c36f-1aea-0310-aeeb-c58d7e2b6c23
---

diff --git a/Robots/parse_html.py b/Robots/parse_html.py
index 666c707..d3870fa 100755
--- a/Robots/parse_html.py
+++ b/Robots/parse_html.py
@@ -87,7 +87,8 @@ def parse_html(filename, charset=None, log=None):
          if log: log("   unknown charset: `%s' or `%s'" % (parser.charset, current_charset))
 
    title = recode_entities(title, current_charset)
-   title = title.replace('\r', '').replace('\n', ' ').strip()
+   parts = [s.strip() for s in title.replace('\r', '').split('\n')]
+   title = ' '.join([s for s in parts if s])
    if log: log("   final title    : %s" % title)
    parser.title = title
    return parser