X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=Robots%2Fbkmk_rsimple.py;h=afeeb254e2b519a33e145612e1b57de3116ec649;hb=3bc7824ef023e60e84ba4402e658836304c248e3;hp=0dee51e8837ae3db9922704d78bd3e08b7325df7;hpb=39c6f1c85c93c021fca528bb1a5cca91f53c196c;p=bookmarks_db.git diff --git a/Robots/bkmk_rsimple.py b/Robots/bkmk_rsimple.py index 0dee51e..afeeb25 100644 --- a/Robots/bkmk_rsimple.py +++ b/Robots/bkmk_rsimple.py @@ -1,7 +1,7 @@ """ Simple, strightforward robot - Written by Oleg BroytMann. Copyright (C) 2000-2007 PhiloSoft Design. + Written by Oleg Broytman. Copyright (C) 2000-2008 PhiloSoft Design. """ import sys, os @@ -40,6 +40,8 @@ class MyURLopener(urllib.URLopener): # Error 301 -- also relocated (permanently) http_error_301 = http_error_302 + # Error 307 -- also relocated (temporary) + http_error_307 = http_error_302 # Error 401 -- authentication required def http_error_401(self, url, fp, errcode, errmsg, headers, data=None): @@ -157,9 +159,10 @@ class robot_simple(Robot): try: content_type = headers["Content-Type"] try: + # extract charset from "text/html; foo; charset=UTF-8, bar; baz;" content_type, charset = content_type.split(';', 1) content_type = content_type.strip() - charset = charset.split('=')[1].strip() + charset = charset.split('=')[1].strip().split(',')[0] self.log(" HTTP charset : %s" % charset) except (ValueError, IndexError): charset = None