plain.html is not consistent with deployed index.html

[Josh-Cena]

Summary

The output plain.html is not consistent with index.html, in particular the anchors are different.

URL

https://developer.mozilla.org/en-US/docs/Web/CSS/Value_definition_syntax#question_mark

Reproduction steps

1. Run yarn build files/en-us/web/css/value_definition_syntax/index.md --nohtml
2. Check build/en-us/docs/web/css/value_definition_syntax/plain.html; search for "question_mark"
3. See <h3 id="question_mark_">

Expected behavior

The anchor is question_mark, because that's what the deployed page has (and what yarn start displays)

Actual behavior

The anchor is question_mark_

Device

Desktop

Browser

Chrome

Browser version

Stable

Operating system

Mac OS

Screenshot

No response

Anything else?

Apologies in advance if this is something that's changed very recently. My Yari version is the latest.

Validations

• I have read the Community Participation Guidelines.
• I have verified that there isn't already an issue that reports the same bug to avoid creating a duplicate.
• I have checked that this is a concrete bug. For Q&A open a GitHub Discussion.

[caugner]

We don't make any guarantees about the contents of those plain.html files, as these are only used internally for our AI Help index. We really shouldn't write them as part of a regular yarn build run.

[Josh-Cena]

Since generating index.html is deprecated, what do we use to get the output HTML, then?

[caugner]

Rendering index.html as part of the build is deprecated, because it has been split to a separate yarn render:html command. Let me know if that helps.

[Josh-Cena]

Could you add that to content's scripts? I'm not sure if this is distributed in the Yari CLI.

Also, the index.html is much bigger than plain.html and hard to parse. Having a stable plain.html is still much appreciated, but it's fine if you want to treat that as an implementation detail.

[LeoMcA]

I'm intrigued that the plain.html files don't have the same ids as index.html, we should probably fix that in its own right for AI Help.

Otherwise, @Josh-Cena can you help me understand your workflow here? Why are you building and manually parsing the HTML files, rather than running yarn start and viewing the output live-built in your browser? What's the sequence of commands + actions you're (ideally) running?

[Josh-Cena]

@LeoMcA: I'm parsing plain.html as part of my link checker that detects broken anchor links. I run it on build output so it can handle links generated by macros.

This bug doesn't block me now because I have since switched to parse index.json instead, but I think this is still nice-to-have for Yari.