fix: Restore Chinese language selection for Brand Markets | LLMO-7309

[igor-grubic]

1. Abstract

This restores Chinese as a selectable language for Brand Markets add-market flows by tolerating Semrush catalog names that qualify Chinese with a parenthetical script or variant.

2. Reasoning

Chinese markets are valid in the product, but the Hong Kong Add market dialog no longer offered Chinese when filtering languages. The option source compared local language names to the Semrush catalog by exact English name, which dropped the Chinese option when Semrush names Chinese as a qualified variant such as Chinese (Simplified).

3. High-level overview of the changes

Before this change, both the UI picker and the API language-id resolver required exact catalog-name matches, so an unqualified local Chinese option could be filtered out or fail to resolve. After this change:

• exact catalog names still take precedence;
• catalog names are also matched by their base name after stripping a trailing parenthetical qualifier;
• Chinese can match a qualified Semrush Chinese catalog row while existing Chinese markets continue to render from their stored language label.

4. Required information

• Jira / issue: https://jira.corp.adobe.com/browse/LLMO-7309|LLMO-7309
• Other: <https://cq-dev.slack.com/archives/C091CRHN961/p1788351535374919|Slack report>

6. Affected / used mysticat-workspace projects

• spacecat-api-service: changed API-side Serenity market language resolution so selected Chinese markets can resolve against qualified Semrush catalog names.
• project-elmo-ui: changed the Brand Markets language option source to include local language options whose base English name is present in the Semrush catalog.

8. Test plan

• Verify in dev or stage: open Brand Markets for a Hong Kong brand, add a market, set Country/Market to Hong Kong, filter the Language field by Ch, and confirm Chinese is selectable.
• Verify that an existing Chinese market still renders with the expected language label in the Markets tab.
• Verify creation/publish of a Hong Kong Chinese market against a live Semrush-backed environment, paying attention to whether Simplified vs Traditional Chinese needs follow-up product modeling.

9. Deployment & merge order

• https://github.com/adobe/project-elmo-ui/pull/3061: related picker-side change that consumes this API behavior.
• Ordered sequence: merge/deploy spacecat-api-service first, then merge/deploy project-elmo-ui so the UI picker and backend resolver agree.

🤖 Generated with review-kit

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[MysticatBot]

Hey @igor-grubic,

Verdict: Approve - clean, well-scoped bug fix with good test coverage.
Complexity: TRIVIAL - small targeted fix in one handler, two files.
Changes: Adds base-name aliasing to the Semrush language cache so unqualified language codes (e.g. zh) resolve against qualified catalog rows (e.g. Chinese (Simplified)) (2 files).

Non-blocking (3): minor issues and suggestions

• nit: Orphaned comment fragment starting mid-sentence ("tags. The handler guards...") above the 400s when the language tag is not a real language test. The new tests were inserted between the first line and the rest of a three-line comment block, leaving a dangling fragment. Restore the full comment or rewrite it as a single coherent block. - test/support/serenity/handlers/markets.test.js:671
• suggestion: Log at log.debug when a base-name alias is registered (e.g. "chinese" aliased from "chinese (simplified)") - would help diagnose future catalog changes without adding runtime noise, since the cache rebuilds once per TTL window. - src/support/serenity/handlers/markets.js:120
• suggestion: Pin the locale on localeCompare (e.g. localeCompare(b, 'en')) or use a plain codepoint sort to make the determinism guarantee independent of the Lambda runtime's ambient locale. Practical risk is negligible for ASCII names, but pinning makes the intent explicit. - src/support/serenity/handlers/markets.js:112

---

_{Skill: pr-review | Model: us.anthropic.claude-opus-4-6-v1[1m] | Duration: 2m 37s | Cost: $4.48 | Commit: 593b83365583f2b08215627d9ef19e2f27d30f82
If this code review was useful, please react with 👍. Otherwise, react with 👎.}

[github-actions]

This PR will trigger a minor release when merged.

[MysticatBot]

Hey @igor-grubic,

⚠ Degraded review - no spec document was found for this change (searched the PR links, the touched repos' docs, the architecture/guidelines docs, and linked Jira). This review covers code-level quality but could not validate the change against an agreed design, so confidence is reduced. Add a spec link (PR template section 4) and re-request review for a full-confidence pass.

Verdict: Approve - well-scoped bug fix with solid normalization approach and thorough test coverage.
Complexity: HIGH - medium diff; API surface signal.
Changes: Adds word-order-independent language name normalization and script-variant-aware BCP-47 resolution so Chinese markets resolve against qualified Semrush catalog rows (2 files).

Non-blocking (4): minor issues and suggestions

• nit: Orphaned comment fragment starting mid-sentence ("tags. The handler guards...") above the 400s when the language tag is not a real language test. The new tests were inserted between lines of a multi-line comment block, leaving a dangling fragment. Restore the full sentence or rewrite as a standalone comment. - test/support/serenity/handlers/markets.test.js
• suggestion: Pin the locale on localeCompare (e.g. .localeCompare(b, 'en')) so the sort order is deterministic regardless of the Lambda runtime's ambient LANG/LC_ALL. Practical risk is negligible for ASCII names, but pinning makes the intent explicit. - src/support/serenity/handlers/markets.js:155
• suggestion: Add a queryTokens.length < tokens.size guard on the token-subset fallback so it only fires when the query looks like a base form (fewer tokens than the candidate). This prevents same-token-count false positives if the catalog evolves with entries that happen to share tokens. - src/support/serenity/handlers/markets.js:196
• suggestion: Document the ASCII-English assumption on normalizeLanguageName - the /[^a-z0-9]+/ regex strips non-ASCII characters silently. A @param note like "must be an ASCII-English name" prevents a future contributor from assuming it handles accented or CJK names. - src/support/serenity/handlers/markets.js:90

Previously flagged, now resolved

• Base-name alias logging suggestion is moot - baseLanguageName replaced entirely by the normalized approach.

---

_{Skill: pr-review | Model: us.anthropic.claude-opus-4-6-v1[1m] | Duration: 5m 35s | Cost: $5.87 | Commit: 915daf8c209e4d4e0c8cceba2d03d6bff490a86b
If this code review was useful, please react with 👍. Otherwise, react with 👎.}