"""
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
# 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):
try:
content_type = headers["Content-Type"]
try:
- content_type, charset = content_type.split(';')
+ # 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
self.log(" no charset in Content-Type header")
- if content_type == "text/html":
+ if content_type in ("text/html", "application/xhtml+xml"):
parser = parse_html(fname, charset, self.log)
bookmark.real_title = parser.title
if parser.refresh: