More Pythonic?

[Jeremy Hylton]

The way I dealt with this problem in my urllib replacement (see
http://the-tech.mit.edu/~jeremy/python/urllib2.py) was to raise an
exception where the exception instance is actually a valid HTTP
response object.  

A typical usage would be something like this:

try:
    resp = urllib2.urlopen('http://frob.nitz/')
except HTTPError:
    print resp.code, resp.msg

The following excerpt shows just the exception class, which extends
the addinfourl object from the original urllib module.

class HTTPError(URLError, addinfourl):
    """Raised when HTTP error occurs, but also acts like non-error return"""

    def __init__(self, url, code, msg, hdrs, fp):
	addinfourl.__init__(self, fp, hdrs, url)
	self.code = code
	self.msg = msg
	self.hdrs = hdrs
	self.fp = fp
	# XXX
	self.filename = url
	
    def __str__(self):
	return 'HTTP Error %s: %s' % (self.code, self.msg)

    def __del__(self):
	# XXX is this safe? what if user catches exception, then
	# extracts fp and discards exception?
	self.fp.close()

Incidentally, have you taken a look at the urllib2 code before?  If it 
isn't sufficient for your needs, I would be interested to get some
comments or criticism.

Jeremy