Add values to "rel" instead of overwriting existing ones

[buhtz]

Hello,

let me start with an example.

<a href="https://fosstodon.org/@backintime" rel="me">foobar</a>

This input is transformed into this, with an overwritten rel-attribute:

<a href="https://fosstodon.org/@backintime" rel="nofollow">foobar</a>

But what I want is

<a href="https://fosstodon.org/@backintime" rel="me nofollow">foobar</a>

readme_renderer/readme_renderer/clean.py

Lines 79 to 85 in 71f7cba

	cleaned = nh3.clean(
	html,
	tags=ALLOWED_TAGS,
	attributes=ALLOWED_ATTRIBUTES,
	link_rel="nofollow",
	url_schemes={"http", "https", "mailto"},
	)

Background

I am aware that you use hn3 package for this. I asked there and got the answer that there is not option on the side of hn3 to modify that behavior. messense/nh3#71

The broader picture here is that I need verify a mailman3 list on Mastodon. To make this happen the list landing page in mailman3 need to contain a rel=me link like you see above.

Approach

I can imagine a regex based approach/workaround and would provide a PR if you accept the concept of the solution. What do you think?

The approach described in short steps:

1. Before calling hn3.clean() record (via regex) all a-tags (and if you want link-tags contain in head-tags) containing a "rel" attribute.
2. Call hn3.clean()
3. Iterate over the recorded cases
   1. Use their hn3.clean() version to find them in the sanitized HTML.
   2. Add the old rel-values to the rel-attribute.

What do you think?