From: Christian Heller Date: Sun, 17 Jan 2016 19:38:57 +0000 (+0100) Subject: Add content-type check to URL interpreter. X-Git-Url: https://plomlompom.com/repos/%7B%7Bprefix%7D%7D/process?a=commitdiff_plain;h=d074cd51f4378884e977b26490af1c6d1dcc81c8;p=plomlombot-irc.git Add content-type check to URL interpreter. --- diff --git a/plomlombot.py b/plomlombot.py index 179bc2a..7d915dc 100644 --- a/plomlombot.py +++ b/plomlombot.py @@ -103,6 +103,10 @@ while 1: for i in range(len(matches)): url = matches[i] webpage = urllib.request.urlopen(url) + content_type = webpage.info().get_content_type() + if not content_type in ('text/html', 'text/xml', + 'application/xhtml+xml'): + continue charset = webpage.info().get_content_charset() content = webpage.read().decode(charset) title = str(content).split('')[1].split('')[0]