add BCP-47 validation cases ported from libpalaso

[hahn-kev]

I was checking to see if using ICU4X Locale could improve our isValid check (it didn't), but as part of that I imported the BCP-47 validation tests from libpalaso, so I figured I'd open a PR to contribute those tests. Feel free to dump it if it doesn't seem useful.

---

AI summary

Broadens test coverage for isValidBcp47Tag in @ethnolib/find-language with well-formedness cases ported from libpalaso's IetfLanguageTag tests (SIL.WritingSystems.Tests/IetfLanguageTagTests.cs). No behavior change to the validator itself — the implementation still uses the existing BCP-47 grammar regex.

Why: We evaluated swapping isValidBcp47Tag to an ICU4X-backed parser to see if it was more complete than the regex. It wasn't: ICU4X follows UTS-35, a strict subset of BCP-47, so it rejects well-formed tags the library itself emits (extlang tags like sgn-ads-GH, grandfathered tags like i-ami, 4–8-char primary languages, and language-less x- tags). Wrapping it with a regex fallback made it behaviorally identical to the plain regex — pure cost, no gain — so the ICU swap was dropped. The libpalaso test corpus surfaced during that comparison was worth keeping.

Changes:

• languageTagUtils.spec.ts: +6 test cases covering
  • well-formed tags with each subtag position filled (tpi-AR, tpi-Lepc-BR-fonipa-x-blah, qed-Lepc-x-rubbish, qed)
  • extlang subtags up to the 3-max, and rejection beyond it
  • Unicode/transform extensions (en-US-u-ca-gregory, en-t-jp, de-DE-u-co-phonebk)
  • private-use subtag length limit (8 chars ok, 16 rejected)
  • empty-subtag and illegal-character rejections (-, --, en--US, qed-?~, trailing x-)
• languageTagUtils.ts: doc comment clarifying that isValidBcp47Tag checks syntactic validity only, not registry membership — unlike libpalaso's registry-backed IetfLanguageTag.IsValid.

Note on scope: libpalaso's IsValid additionally validates subtags against the IANA/ISO registries (so e.g. zzz is invalid there). That semantic level is deliberately out of scope here; these tests only assert well-formedness, matching the existing isValidBcp47Tag contract.

Test plan

• npm run testonce in components/language-chooser/common/find-language — 88 tests pass (was 82).
• npx tsc --noEmit — clean.

---

This change is 

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
font-chooser-react-mui	Ready	Preview	Sep 2, 2026 2:29am UTC

[andrew-polk]

components/language-chooser/common/find-language/languageTagUtils.spec.ts line 822 at r1 (raw file):

  it("should return false for more than three extlang subtags", () => {
    expect(isValidBcp47Tag("abcdefgh-abc-def-ghi-jkl")).toBeFalsy();

Devin says

---

components/language-chooser/common/find-language/languageTagUtils.spec.ts:R822

Extlang-limit test does not test the extlang limit

abcdefgh-abc-def-ghi-jkl is rejected because an 8-letter primary language subtag cannot carry any extlang, not because it exceeds three extlangs. The assertion would pass even if the extlang-count check were broken, so it tests nothing. Only en-abc-def-ghi-jkl exercises the limit.

[andrew-polk]

@andrew-polk reviewed 2 files and all commit messages.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on hahn-kev).

[andrew-polk]

@andrew-polk reviewed 1 file and all commit messages, and resolved 1 discussion.
Reviewable status: complete! all files reviewed, all discussions resolved (waiting on hahn-kev).