Skip to content

fix(stt): normalize language code to ISO 639-1 for OpenAI Whisper#1939

Merged
piorpua merged 1 commit intomainfrom
fix/sentry-ELECTRON-G3
Mar 30, 2026
Merged

fix(stt): normalize language code to ISO 639-1 for OpenAI Whisper#1939
piorpua merged 1 commit intomainfrom
fix/sentry-ELECTRON-G3

Conversation

@kaizhou-lab
Copy link
Copy Markdown
Collaborator

@kaizhou-lab kaizhou-lab commented Mar 30, 2026

Summary

  • Sentry Issue: ELECTRON-G3 (3 events, 3 users across Chile/Colombia/Hong Kong)
  • OpenAI Whisper API requires ISO 639-1 language codes (e.g. en) but the languageHint and config values may arrive in BCP 47 format (e.g. en-us), causing 422 Unprocessable Entity responses
  • Fix: extract the primary language subtag (language.split('-')[0].toLowerCase()) before appending to the form data

Changes

  • src/process/bridge/services/SpeechToTextService.ts: normalize language code before formData.append('language', ...)
  • tests/unit/SpeechToTextService.test.ts: new test file covering BCP 47 → ISO 639-1 conversion, plain ISO 639-1 passthrough, no-language omission, and error response handling

Verification

  • Unit tests: 5/5 pass
  • Type check: clean
  • Lint: 0 errors
  • Process: main (unit tests sufficient)

Test plan

  • Unit tests cover BCP 47 language hint conversion (e.g. en-us → en)
  • Unit tests cover config language conversion (e.g. zh-CN → zh)
  • Unit tests verify plain ISO 639-1 codes pass through unchanged
  • Unit tests verify language field omitted when no hint/config
  • Unit tests verify error handling on non-ok response

@kaizhou-lab
fix(stt): normalize language code to ISO 639-1 before OpenAI Whisper …
269b332 
…request (ELECTRON-G3)

OpenAI Whisper API requires ISO 639-1 language codes (e.g. "en") but
the languageHint and config values may arrive in BCP 47 format
(e.g. "en-us"), causing 422 Unprocessable Entity responses. Extract
the primary subtag before appending to the form data.
@kaizhou-lab kaizhou-lab marked this pull request as ready for review March 30, 2026 09:58
@piorpua piorpua added the bot:reviewing Review in progress (mutex) label Mar 30, 2026
@piorpua
Copy link
Copy Markdown
Contributor

piorpua commented Mar 30, 2026

Code Review：fix(stt): normalize language code to ISO 639-1 for OpenAI Whisper (#1939)

变更概述

本 PR 修复 Sentry 问题 ELECTRON-G3：SpeechToTextService 向 OpenAI Whisper API 发送语言参数时，直接传递了 BCP 47 格式（如 en-us），而 Whisper 仅接受 ISO 639-1 格式（如 en），导致 3 名用户遭遇 422 错误。改动仅涉及 SpeechToTextService.ts 一行逻辑及新增的 5 个单元测试。

方案评估

结论：✅ 方案合理

用 language.split('-')[0].toLowerCase() 提取主语言标签，精准修复 BCP 47 → ISO 639-1 的转换问题，改动最小且与 Whisper API 文档要求一致。现有 if (language) 守卫覆盖了空值情况，无需额外处理。测试覆盖了 BCP 47 转换、原生 ISO 639-1 直通、无语言字段省略和错误响应四个核心场景。

问题清单

✅ 未发现明显问题，代码质量良好，建议批准合并。

注：bunx oxlint 报告的 no-extraneous-class 警告（SpeechToTextService 类，第 143 行）为已存在的历史问题，与本次改动无关，不作为本 PR 阻塞点。

汇总

无问题。

结论

✅ 批准合并 — 修复精准，测试完整，无阻塞性问题。

本报告由本地 pr-review skill 生成，包含完整项目上下文，无截断限制。

CONCLUSION: APPROVED
IS_CRITICAL_PATH: false
PR_NUMBER: 1939

@piorpua
Copy link
Copy Markdown
Contributor

piorpua commented Mar 30, 2026

✅ 已自动 review，无阻塞性问题，正在触发自动合并。

@piorpua piorpua merged commit 868bdaf into main Mar 30, 2026
17 checks passed
@piorpua piorpua deleted the fix/sentry-ELECTRON-G3 branch March 30, 2026 11:37
@piorpua piorpua added bot:done Auto-merged by bot and removed bot:reviewing Review in progress (mutex) labels Mar 30, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

bot:done Auto-merged by bot

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@kaizhou-lab @piorpua