This might be related to bug #110692, since it seems to be related to proxies.

A program I have that works under older versions of Python now fails with this traceback:


Traceback (most recent call last):
  File "IGNRatings.py", line 289, in ?
    main(sys.argv[1:])
  File "IGNRatings.py", line 133, in main
    inStream = urllib.urlopen(url)
  File "E:\Python20\lib\urllib.py", line 61, in urlopen
    return _urlopener.open(url)
  File "E:\Python20\lib\urllib.py", line 163, in open
    return getattr(self, name)(url)
  File "E:\Python20\lib\urllib.py", line 259, in open_http
    h = httplib.HTTP(host)
  File "E:\Python20\lib\httplib.py", line 624, in __init__
    self._conn = self._connection_class(host, port)
  File "E:\Python20\lib\httplib.py", line 324, in __init__
    self._set_hostport(host, port)
  File "E:\Python20\lib\httplib.py", line 330, in _set_hostport
    port = int(host[i+1:])
ValueError: invalid literal for int():


My URL is nothing special:

    'http://dreamcast.ign.com/review_lists/a.html'


I have tracked the problem down a bit:

In urllib.py, line 147, the statement

    proxy = self.proxies[type]

assigns the string

    'http://http://proxy:8080'

to "proxy". My guess is that's not right. Anyway, the following line assigns

    type = 'http'
    proxy = '//http://proxy:8080'

and the next line assigns

    host = 'http:'
    selector = '//proxy:8080'

'http:' as a host name then goes on to cause trouble when an HTTPConnection is constructed.

Hope this info helps you track it down.

Bob Alexander
[email protected]

Fixed in 2.0b2.

The earlier code assumed "protocol=host;protocol=host;..." or "host", but Windows may also use "protocol=host" (just one entry), as well as "protocol://host".  This code needs some more work, so I'll leave the bug open for now.

Fredrik, can you fix the code right or assign it to someone else who can? Or decide to close the bug report after all.

Okay, I've failed to find case for which the new code
doesn't work, so let's close this one.