5

This shell command succeeds

$ curl -A "Mozilla/5.0 (X11; Linux x86_64; rv:18.0) Gecko/20100101 Firefox/18.0 (compatible;)" http://fifa-infinity.com/robots.txt

and prints robots.txt. Omitting the user-agent option results in a 403 error from the server. Inspecting the robots.txt file shows that content under http://www.fifa-infinity.com/board is allowed for crawling. However, the following fails (python code):

import logging
import mechanize
from mechanize import Browser

ua = 'Mozilla/5.0 (X11; Linux x86_64; rv:18.0) Gecko/20100101 Firefox/18.0 (compatible;)'
br = Browser()
br.addheaders = [('User-Agent', ua)]
br.set_debug_http(True)
br.set_debug_responses(True)
logging.getLogger('mechanize').setLevel(logging.DEBUG)
br.open('http://www.fifa-infinity.com/robots.txt')

And the output on my console is:

No handlers could be found for logger "mechanize.cookies"
send: 'GET /robots.txt HTTP/1.1\r\nAccept-Encoding: identity\r\nHost: www.fifa-infinity.com\r\nConnection: close\r\nUser-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:18.0) Gecko/20100101 Firefox/18.0 (compatible;)\r\n\r\n'
reply: 'HTTP/1.1 403 Bad Behavior\r\n'
header: Date: Wed, 13 Feb 2013 15:37:16 GMT
header: Server: Apache
header: X-Powered-By: PHP/5.2.17
header: Vary: User-Agent,Accept-Encoding
header: Connection: close
header: Transfer-Encoding: chunked
header: Content-Type: text/html
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/moshev/Projects/forumscrawler/lib/python2.7/site-packages/mechanize/_mechanize.py", line 203, in open
    return self._mech_open(url, data, timeout=timeout)
  File "/home/moshev/Projects/forumscrawler/lib/python2.7/site-packages/mechanize/_mechanize.py", line 255, in _mech_open
    raise response
mechanize._response.httperror_seek_wrapper: HTTP Error 403: Bad Behavior

Strangely, using curl without setting the user-agent results in "403: Forbidden" rather than "403: Bad Behavior".

Am I somehow doing something wrong, or is this a bug in mechanize/urllib2? I don't see how simply getting robots.txt can be "bad behaviour"?

Moshev
  • 556
  • 6
  • 14
  • And another example of header sniffing gone bad. The server on the other side is looking at more that the UA agent, check what headers `curl` is sending, compare them to what `mechanize` is using, adjust, rince, repeat. This is *not* a python problem. – Martijn Pieters Feb 13 '13 at 15:48
  • This problem look awefully similar to [urllib2.HTTPError: HTTP Error 403: Forbidden](https://stackoverflow.com/questions/13303449/urllib2-httperror-http-error-403-forbidden/46213623#46213623) – Supreet Sethi Nov 06 '17 at 16:31

1 Answers1

9

As verified by experiment, you need add an Accept header to specify acceptable content types(any type will do, as long as "Accept" header exists). For example, it will work after changing:

br.addheaders = [('User-Agent', ua)]

to:

br.addheaders = [('User-Agent', ua), ('Accept', '*/*')]
Hui Zheng
  • 10,084
  • 2
  • 35
  • 40