Fix relative index URLs - #230

This is for the index I host together with my custom Pyodide deployment. The index lists all packages that were built just for Pyodide.

Each version of the lab has its own index, so e.g. there is

• https://lab.climet.eu/main/pypa/simple/ for the main branch version
• https://lab.climet.eu/v0.3.0/pypa/simple/ for v0.3.0

When navigating the links in the web browser, everything works, so the relative links seem to be correctly set up. From my understanding, the simple HTML index supports relative URLs with the same meaning as when using a web browser.

I could of course switch to absolute URLs but don't really want to bake in the full URL since it would directly tie everything to where the lab is currently being hosted.

I see. Thanks.

So, if the relative URL starts with / (anaconda case), then it should be concatenated with the base URL, otherwise if, it should be concatenated to the full URL.

It looks like urllib.parse.urljoin handles this correctly. Let me see if I can fix the logic with it.

>>> urljoin("https://my.index.com/simple/hello/", "../mywhl.whl")
'https://my.index.com/simple/mywhl.whl'
>>> urljoin("https://my.index.com/simple/hello/", "/mywhl.whl")
'https://my.index.com/mywhl.whl'

Oh, okay, looks like it is already fixed in #174. I think we can merge this then. Thanks!