Bug 834582 has an interesting set of URL.s like 'http://git.fedorahosted.org/git/?p=java-deptools.git;a=snapshot;h=2a6a72d74c2fa96f5ed246459b4a259e3e7fa35c;sf=tbz2;foo=/java-deptools-0.tar.bz2'
The problem is how to get the filename of an URL like this. For now, os.path.basename gives the correct answer, but only because the filename happens to be the last parameter. Basically, os.path.basename is undefined on this kind of input. We need a more robust way to handle this; after all, it's fedorahosted's urls we are talking about.
Other kind of odd urls related to this is things like https://www.disk.dsl.o2online.de/FclyPlh/RPMS/speed-dreams/speed-dreams-2.0.0-0.1.rc1_r4420.12.fc16.src.rpm?a=UAl5-rFrWFI, from bugzilla #785371.
Here the parameters should be skipped....
A mess. At the point of the download, we have (worst case) no more info then a bug number, so how determine what to look for?
See also 1e4f4344ec71
A misconception. This is about the source URL, not the normail url:s. Closing.