From 98181607fade83338c1e633e18713f7544425ff3 Mon Sep 17 00:00:00 2001 From: Oleg Broytman Date: Wed, 11 Aug 2010 16:59:40 +0000 Subject: [PATCH] Fixed a bug in case there are more than one Content-Type headers. git-svn-id: file:///home/phd/archive/SVN/bookmarks_db/trunk@256 fdd5c36f-1aea-0310-aeeb-c58d7e2b6c23 --- Robots/bkmk_rsimple.py | 13 ++++++++++--- 1 file changed, 10 insertions(+), 3 deletions(-) diff --git a/Robots/bkmk_rsimple.py b/Robots/bkmk_rsimple.py index 2288996..42975d2 100644 --- a/Robots/bkmk_rsimple.py +++ b/Robots/bkmk_rsimple.py @@ -158,6 +158,7 @@ class robot_simple(Robot): if headers: try: content_type = headers["Content-Type"] + self.log(" Content-Type: %s" % content_type) try: # extract charset from "text/html; foo; charset=UTF-8, bar; baz;" content_type, charset = content_type.split(';', 1) @@ -167,7 +168,13 @@ class robot_simple(Robot): except (ValueError, IndexError): charset = None self.log(" no charset in Content-Type header") - if content_type in ("text/html", "application/xhtml+xml"): + for ctype in ("text/html", "application/xhtml+xml"): + if content_type.startswith(ctype): + html = True + break + else: + html = False + if html: parser = parse_html(fname, charset, self.log) bookmark.real_title = parser.title if parser.refresh: @@ -233,8 +240,8 @@ class robot_simple(Robot): icons[icon] = None os.remove(fname) - except KeyError: - pass + except KeyError, key: + self.log(" no header: %s" % key) except IOError, msg: if (msg[0] == "http error") and (msg[1] == -1): -- 2.39.5