Tweak commit snapshot production about html-build

[annevk]

Idea: use sed to perform a series of replacements on source. Then generate a new single-page copy from that. I think that should work for Commit Snapshots and Review Drafts. And it's pretty easy.

I think it should work for @tobie's whatpr work too.

[annevk]

FWIW, it seems reasonably doable in bash, but the newline situation is rather painful (not finished):

HTML_FAKE_SHA="alalaladjfkasd234j33242"

INPUT="source"
OUTPUT="new-source"

sed -e 's/<h2 id="living-standard" class="no-num no-toc">Living Standard/<h2 id="living-standard" class="no-num no-toc">Commit Snapshot/' \
    -e "s/^  <title w-nodev>HTML Standard/  <title w-nodev>HTML Standard (Commit Snapshot $HTML_FAKE_SHA)/" \
    -e 's/^  <h2 class="no-num no-toc" id="contents" w-nosplit w-nodev>Full table of contents<\/h2>/  <div data-fill-with="warning">\'$'\n''   <details class="annoying-warning" open=""><summary>This Is a Commit Snapshot of the Standard<\/summary>\'$'\n'"    <p>This document contains the contents of the standard as of the <a href=\"https:\/\/github.com\/whatwg\/html\/commit\/$HTML_FAKE_SHA\">$HTML_FAKE_SHA commit<\/a>, and should only be used as a historical reference. This commit may not even have been merged into master.<\/p><p> Do not attempt to implement this version of the specification. Do not reference this version as authoritative in any way. Instead, see <a href=\"https:\/\/url.spec.whatwg.org\/\">https:\/\/url.spec.whatwg.org\/<\/a> for the living standard.<\/p><\/details><\/div><h2 class=\"no-num no-toc\" id=\"contents\" w-nosplit w-nodev>Full table of contents<\/h2>/" \
    < "$INPUT" > "$OUTPUT"
diff -up "$INPUT" "$OUTPUT"

[annevk]

The reason I somewhat favor this over instructions to Wattsi is because the latter is build around the no-html/no-dev notion, which seem mutually exclusive? And Pascal is somewhat trickier to deal with than some replacements.

[domenic]

I don't think they're mutually exclusive; we have w-nosplit and w-nohtml, and the multipage versions are both "html" and "split" at the same time. We'd want to introduce a new type of thing, maybe w-noliving...

I think that might be technically nicer in a few ways, but in the end it's up to the person doing the work, and you've already made good progress.

[annevk]

Wattsi generates all the variants by default so I guess we'll have to modify it either way. 😟

A snapshot could indeed be generated alongside dev/ and multipage/. A Review Draft would be trickier. We'd basically need a new way to invoke Wattsi to make it only produce that for a given input.

In html/ I guess we'd have a similar review.sh script that generates review-drafts/05-2018.wattsi or some such? And whenever we process that we'd invoke wattsi in such a way to only generate a singlepage.

[annevk]

My various WIP PRs against html/html-build/wattsi address 1-4 from OP. 5-6 are a little tricky due to HTML's unusual heading at the top. I'd prefer we address those in a follow-up.

[annevk]

So am I correct that addressing 5-6 would be adding some links to the HTML source and letting wattsi take care of putting the right data in them? And given that we already pass on the SHA wattsi wouldn't necessarily need API changes so this shouldn't affect whatpr, right?

[domenic]

That sounds about right, yeah. Although I'm not sure we pass the SHA to Wattsi vs. doing it at some other step?

[tobie]

LMK if there's any changes I need to make on the PR PReview side.