OoO En cette soirée bien amorcée du mardi 29 janvier 2008, vers 22:23, Eric Evans <[EMAIL PROTECTED]> disait:
> There are some parsing functions in urllib and urlparse but none of them > seemed to provide for a very clean solution. > The attached patch should correctly parse urls with or without a > username/password pair as well as setting up the appropriate header for > proxy basic auth. > Again, I don't have a ready means of testing this. I've made sure that > it's parsing the url and setting the instance attributes correctly, and > that it doesn't effect normal usage of boto. If you could apply the > patch and let me know how it works for you, it would be appreciated. Hi Eric ! I have tested your patch: - without http_proxy: OK - with empty http_proxy: OK - without password: OK - with password: OK > + pattern = re.compile( > + '(?:http://)?' \ > + '(?:(?P<user>\w+):(?P<pass>\w+)@)?' \ > + '(?P<host>[\w\-\.]+)' \ > + '(?::(?P<port>\d+))?' I think that \w is too restrictive for password. I can have special characters in password. I would suggest [EMAIL PROTECTED] instead of \w+. It will catch: http://user:[EMAIL PROTECTED]:3128 Well, in fact, you are better with .+ because you can catch: http://user:[EMAIL PROTECTED]@localhost:3128 As long that there is '@' in the string, we have a user/password scheme. And '@' cannot appear in the hostname or in the port, so we are safe. Since password can be empty, I would suggest .* : pattern = re.compile( '(?:http://)?' \ '(?:(?P<user>\w+):(?P<pass>.*)@)?' \ '(?P<host>[\w\-\.]+)' \ '(?::(?P<port>\d+))?' Thanks for your patch ! -- BOFH excuse #231: We had to turn off that service to comply with the CDA Bill.
