Test the language chooser -> Bloom data mapping

[andrew-polk]

Problem

Picking a language in the language chooser hands Bloom five things that cross from TypeScript into C#: the language tag, the name to show, whether the user renamed it, whether the script reads right-to-left, and the country. Neither side of that handoff had a single automated test.

It has not been quiet code. Three shipped bugs were defects in the TypeScript half alone — BL-15190 (choosing a script didn't change the name Bloom stored), BL-13982 (the right-to-left setting stopped surviving a restart) and BL-14426 (the C# side couldn't read the name field at all). A fourth defect of the same shape would have reached users the same way.

What the PR does

The TypeScript half. Moves getLanguageData() and ILanguageData, unchanged, out of LanguageChooserDialog.tsx into their own languageData.ts, so tests can reach the function without loading the dialog's MUI and bloomApi chain. The function body is byte-for-byte what it was. Adds 12 tests over the five fields — including the difference between "the script says left-to-right" and "the script didn't say", which is what BL-13982 turned on — plus 4 more pinning the corners where this function deliberately differs from EthnoLib's defaultDisplayName(language, script), so folding the two together has to be a decision rather than an accident.

The C# half. Bloom decides whether a name is the language's own or one the user typed by comparing DesiredName with DefaultName — written out inline, four times over, in the handlers for Language 1, 2, 3 and the sign language. That could not be tested as written; a test could only have re-implemented the comparison and asserted on itself. It now has a name, LanguageChangeEventArgs.IsCustomName, and all four sites call it. Behavior is unchanged, the duplication is gone, and 9 NUnit tests cover it.

PublishApi has its own, different rule for the same question, and this PR does not touch it — that divergence is described in the remarks on IsCustomName and pinned by two tests, so it cannot be unified by accident. Fixing it is BL-16760, in the PR stacked on this one.

Also records why the sign-language handler, unlike Language 1-3, does not apply args.IsRtl: a sign language has no text direction.

Every production change here is behavior-preserving.

Devin review

---

This change is 

[andrew-polk]

(Claude Opus 5 (1M context)) Consulted Devin on 2026-08-26 04:59 UTC up to commit cca9bc19496896891d426c4fb78348b9aa999ec2.

Clean: no bugs, no investigate flags, no informational items.

[greptile-apps]

Greptile Summary

The PR extracts the language-chooser data mapping into a directly testable module and centralizes the existing C# custom-name comparison without changing its behavior.

• Adds TypeScript coverage for language tags, names, text direction, and country mapping.
• Adds NUnit coverage for custom-name classification and its intentional difference from PublishApi.
• Rewires both chooser entry points to the extracted mapping module.

Important Files Changed

Filename	Overview
src/BloomBrowserUI/collection/languageData.ts	Extracts the existing language-selection mapping unchanged into a standalone module.
src/BloomBrowserUI/collection/languageData.test.ts	Adds focused Vitest coverage for all serialized language-data fields and documented edge cases.
src/BloomExe/web/controllers/LanguageChangeEventArgs.cs	Introduces an IsCustomName property equivalent to the previously repeated string comparison.
src/BloomExe/Collection/CollectionSettingsDialog.cs	Replaces four equivalent inline custom-name comparisons with the shared property.
src/BloomTests/web/controllers/LanguageChangeEventArgsTests.cs	Adds NUnit coverage for custom-name classification and the intentionally distinct PublishApi rule.

_{Reviews (2): Last reviewed commit: "Test the language chooser -> Bloom data ..." | Re-trigger Greptile}

[andrew-polk]

(Claude Opus 5 (1M context)) Consulted Devin on 2026-08-26 up to commit 0b4f608eb14443b02b78740930b2cdbacd6cd1a5 (adds the C# custom-name coverage).

No bugs, no investigate flags. Two informational items, both confirming rather than questioning the change: that the C# IsCustomName extraction is behavior-preserving, and that getLanguageData moved without change. Neither mirrored as a review thread (informational items are low signal).

Note: pr-automation and Greptile did not register for this push — GitHub was having problems at the time — so Devin was triggered manually via its review page.