[Discussion] Sourceforge download patterns

[jloehel]

@tdruez Thank you very much for the improvements of the Sourceforge download patterns. I still have some issues with the following URLs:

https://sourceforge.net/projects/keepass/files/KeePass%202.x/2.55/KeePass-2.55-Source.zip/download
https://sourceforge.net/projects/codeblocks/files/Sources/20.03/codeblocks-20.03.tar.xz/download
https://sourceforge.net/projects/npppluginmgr/files/v1.3.3/PluginManager_1.3.3_UNI.zip/download
https://sourceforge.net/projects/openofficeorg.mirror/files/4.1.12/source/apache-openoffice-4.1.12-60e0a1c111-src.zip/download
https://sourceforge.net/projects/qbittorrent/files/qbittorrent/qbittorrent-4.6.2/qbittorrent-4.6.2.tar.xz/download
https://sourceforge.net/projects/scrcpy.mirror/files/v2.3.1/scrcpy-win64-v2.3.1.zip/download
https://sourceforge.net/projects/zedgraph/files/zedgraph%20source/5.0.9%20%2B%204.5.9/zedgraph_source_v509_459.zip/download

What I have seen so far it's not guaranteed that the project name matches with the files. Some Projects name the files just Source . Like Codeblocks is doing it. Furthermore the name in the archive does not need to match with the project name. zedgraph is really an edge case because the archive has two versions. I am not sure if it's necessary to consider version ranges there. zedgraph is the only case I know.

What do you think about this regex:

^https?://sourceforge.net/projects/(?P<namespace>[^/]+)/files(/(.+/)?)((?P<version_prefix>(?!\d)[a-zA-Z\-]+)?(?P<version>\d[\w\.\-\+\ ]+))(/(.+/)?)(?P<filename>(?P<name>.+)[-_\ ](?P=version_prefix)?(?P=version)(.*))/download$

[tdruez]

What do you think about this regex

@jloehel You should add the URL listing in the url2purl.json with their expected PURLs.
You can then run the tests and adapt the regex as needed until those pass.
Once you have something stable, you can submit a PR and I'll take care of the merge.
Cheers ;)