Provision refs resolver

[longhotsummer]

Replace the old internal refs resolver with a more robust provision refs resolver that can also resolve provisions in external works.

The bulk of the tests were copied over from the old resolver, especially the edge cases. We now can resolve a lot more of these than before.

A single resolver supports multiple languages, currently English and Afrikaans. Adding new languages should hopefully be relatively simple.

https://www.loom.com/share/730117beb7c048e1b857c1914ee1952e

https://www.loom.com/share/12ac35fd26ab4c07aeb850252f6f1620

Outstanding

• 'subsection (1)(a) of section 27'

[longhotsummer]

'paragraph' vs subsection mark-up (some missed)

Do you think it's safe to make something like 'paragraph (16)' look for paragraph, subsection and subparagraph elements? And probably equivalently for subsection and subparagraph.

[longhotsummer]

'sub-section', 'section one' passed over

I fixed sub-section but not section one.

[goose-life]

'paragraph' vs subsection mark-up (some missed)

Do you think it's safe to make something like 'paragraph (16)' look for paragraph, subsection and subparagraph elements? And probably equivalently for subsection and subparagraph.

I want to say yes? Especially since you seem to have solved the other issues around where to look first, and finding the 'of section 2(3)' matches before their preceding 'paragraph (a)'.

I suppose it does open us up to potentially more false matches, but if it's a question of just adding more elements to look for I think we can try it out and see? (rather than a big rewrite)

Here are my suggestions from my test drive:

• 'paragraph' – look for paragraph, subparagraph, subsection (in that order of preference, if that's a thing)
• 'subparagraph' – look for subparagraph, paragraph
• 'regulation' – look for section
• 'subregulation' – look for subsection, paragraph, subparagraph

[longhotsummer]

Support for synonyms, regulations and sub-regulations added.

[goose-life]

okay we're getting fancy!

[longhotsummer]

[longhotsummer]

The bulk of the raised issues are fixed. I had to disable the Afrikaans translation of 'or'. I don't have a holistic fix currently for indicating to the parser which language it should be working in.

[goose-life]

two minor suggestions, happy for you to leave them out (especially tables) if you think it best

[longhotsummer]

@goose-life I've added crossHeading to the list of elements to ignore, too.