Contextualization

[JayPanoz]

This PRs address contextualisation of utterances, per spec – that is private but will eventually be published in docs.

Draft because the spec is still being laid out for some verbosity levels.

[danielweck]

Well done :)
A couple of things come to mind:

• in the tests for TextFragments etc. I would add more whitespace in the fixtures, as indentations and other insignificant spaces and tabs are commonly present in the source document markup, interspersed inside hierarchies of (non)sectioning element contents.
• I'm sure you've considered using the "official" TextFragment generator from the Google polyfill, so I wonder why you decided to write your own? (theirs covers more edge cases in HTML markup, including different locales, handling of Unicode surrogate pairs, etc.) https://github.com/GoogleChromeLabs/text-fragments-polyfill/blob/main/src/fragment-generation-utils.js One option would be to include their unit tests in this repository, in order to verify that the simpler naive generator implemented in this PR passes their baseline requirements? (baring in mind, their implementation isn't actively maintained so...to use with caution anyway) https://github.com/GoogleChromeLabs/text-fragments-polyfill/blob/main/test/unit/fragment-generation-utils-test.js
• EDIT: they also have a bunch of tests for matching/highlighting/marking which are less relevant here in this PR, but things like text normalisation could be useful: https://github.com/GoogleChromeLabs/text-fragments-polyfill/blob/abc6ed408b3f20e91d9cbda9977748459f5e3877/test/unit/text-fragment-utils-test.js#L246

[JayPanoz]

Thanks!

Fixtures are definitely something we want to grow with such cases. Thanks for pointing this one out.

As for the polyfill, that was me trying to gauge what I had to deal with instead of defaulting to the polyfill, so that I had a deeper understanding of the complexity at play. This should definitely migrate to it before the PR goes out of draft indeed!