Validate language tags, throwing errors if they aren't valid

[dlongley]

No description provided.

[dlongley]

Use the language tag regex from RDF (not the more strict one from BCP47).

[dlongley]

Turtle uses this

[144s] LANGTAG                        ::= "@" [a-zA-Z]+ ( "-" [a-zA-Z0-9]+ )*

[davidlehn]

Note a PR exists for this on the jsonld.js side:
digitalbazaar/jsonld.js#284

It uses this:

const LANG_RE = /^[a-zA-Z]+(-[a-zA-Z0-9]+)*$/;

I still am concerned about merging that until we have performance hit benchmarks, probably a flag to disable all that checking (when you know you don't need it), and maybe an API for handling failures vs just dropping invalid data silently. Same issues if the check is in this lib.

[dlongley]

We should use BCP47 validation instead, not RDF, RDF will be getting errata/bug report.

[davidlehn]

Such regex insanity:
https://stackoverflow.com/questions/7035825/regular-expression-for-a-language-tag-as-defined-by-bcp47

[dlongley]

Those regexes are for capturing the various components of a language tag; we don't need any of that, we just need to know if it's valid. It should be a much simpler regex.