Support hash redirects / heading redirects

[reteps]

Resolves #69

• First hash is used if no top-level one found
• Injects into existing pages if needed

I won't claim this is the cleanest implementation, but it works! I tested it on my own site with these rules:

      redirect_maps:
        index.md#foo: 'concepts/index.md#bam'
        foo.md#bar: 'concepts/index.md#bam'
        foo.md: 'getting-started/index.md'

where foo.md doesn't exist, and index.md does.

[reteps]

@pawamoy Love to get your input!

[pawamoy]

Hi @reteps, thanks a lot for the PR! I'll try to review this tomorrow 🙂

[reteps]

Thank you! I think there may be an implementation bug, as the tests don't pass.

[pawamoy]

There are a few things that make this PR a bit hard to review:

• there seem to be formatting changes, could you revert them? (maintain previous line length)
• you have renamed the "redirect maps" variables to "redirect entries", is there a practical reason? if not, could you revert those too?

I am currently unable to see where is the impactful change 😅

[reteps]

Hi @pawamoy, could you point out what portion is formatting changes? I genuinely have added about 130 LoC. As for the redirect entries, there is a helper function explaining its difference from the redirect maps. I added this so that I can keep track of overall and hash based redirects, for each base page.

[pawamoy]

Ah, my bad, I was probably looking at a single commit instead of the full diff 🙇

[pawamoy]

Made an initial review. The code looks good, though I didn't try looking at the semantics, whether the change makes sense or if there are better ways, etc. To be honest I don't have a lot of mental bandwidth to invest in MkDocs maintenance, so I'm not sure when I'll be able to provide a more meaningful review. Please bear with me 🙏 Others are welcome to review this PR too 🙂

[pawamoy]

thought: I wonder if we shouldn't escape the variables properly. Something like ... = {old_anchor!r}, which I believe would work in Javascript code as well?

[reteps]

Can you explain a scenario where this would be useful?

[pawamoy]

Can anchors contain "? Or would such characters be encoded in the percent format? If yes and not encoded, such anchors would break this Javascript code.

[reteps]

My understanding is that pretty much all headings in mkdocs go through a slugification process that removes special characters (mkdocs-material docs on this). I have personally never seen a link with this type of character in it, but I can add the escaping if you would like.

[yarikoptic]

FWIW, this functionality would be immediately useful for us too, so thank you @reteps and @pawamoy for working this out! Let me know if you need any assistance

[reteps]

Hi @yarikoptic I can give you push access to my fork if you would like to immediately work on this; I won't have some time for another two weeks.

[reteps]

I think the code-style nits are now resolved, but there is still a bug because the tests are still failing. If anyone want to help pick this up, I can provide push access to this branch.

[reteps]

Some of the existing test cases redirect away from an existing page -- and with this addition, this causes those tests to fail.

For example,

existing_pages = [
    "README.md",
]

And the test case checks that a file index.html is written out containing a redirect to new.md for test_page_dest_path_directory_urls.

    ("README.md", "index.html", "index.html"),

This doesn't seem like desirable behavior -- should I change the test suite?

[reteps]

@pawamoy any thoughts?