fix(qoder-cn): 解析新版 Qoder CN（com.qodercn.app.stable）的 ~/.qoder-cn/projects JSONL 会话

[Hu9956]

背景

PR #549 修复了国际版新版 Qoder（com.qoder.app.stable）的 JSONL 会话追踪，并为 CN 预留了完整的接线（sync 块、qoderCnNew 游标、status 展示），但当时假设 CN 与国际版共享 ~/.qoder/projects，resolveQoderCnProjectsDir 的默认值也指向了国际目录。

实测新版 Qoder CN（com.qodercn.app.stable，app 0.1.8 / agent 1.1.44）把会话写在兄弟目录 ~/.qoder-cn/projects，于是 sync.js / status.js 中「CN 目录与国际目录分离才解析」的防重复守卫恒为假——新版 CN 的 usage 从未被解析（而旧 CN 库 QoderCN/SharedClientCache/cache/db/local.db 已于 2026-08 停写，CN 计数就此断流）。

改动

• resolveQoderCnProjectsDir 默认改 ~/.qoder-cn/projects（QODER_CN_PROJECTS_DIR 仍可覆盖；Windows WSL 探测 .qoder-cn）；QODER_HOME / QODER_PROJECTS_DIR（国际版专属）不会影响 CN 解析
• 剥离 CN BYOK 模型 id 内嵌的安装级 provider UUID（qoder-custom-<uuid>/glm-5.3-flash → glm-5.3-flash），桶键跨重装稳定、不按用户碎片化；官方 id（qmodel_*）原样保留
• sync.js / status.js 逻辑零改动，仅更新过时注释（目录修正后守卫自动生效）
• 测试 +3：CN 目录默认/覆盖回归（含「CN ≠ 国际」不变量）、CN 真实形态端到端（Anthropic 语义精确 token、流式重复 message id 去重、subagent transcript、无 usage 的 auto 行跳过、二次同步幂等）

验证

• npm test：2643 项 / 2635 过；6 个失败为 upstream/main 基线同集合的环境性失败（stash 对比确认，零回归）
• 真机只读端到端（macOS，CN 真实数据）：41 个会话文件 → 1237 条 usage → 22 个桶；与独立脚本复算逐位一致（input 327,343,057 / cache_read 322,214,912 / cache_creation 0 / output 842,875 / 总 650,400,844）；二次同步 0 桶
• 旧版回退安全：迁移导入的 24 个旧会话 JSONL 均无 usage 行，不会与旧库已入库数据重复计数

Summary by CodeRabbit

• Bug Fixes
  • Qoder CN projects are now read from the separate ~/.qoder-cn/projects directory, preventing duplicate reporting with international projects.
  • CN usage reports now consistently recognize BYOK model identifiers and group usage under the underlying model name.
  • Improved handling of cached-token usage, streamed duplicate messages, and automatic model rows in CN session data.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Team

Run ID: 355f9e60-653d-4644-8730-8692ecec384a

📥 Commits

Reviewing files that changed from the base of the PR and between 271bbd8 and d334b0d.

📒 Files selected for processing (4)

• src/commands/status.js
• src/commands/sync.js
• src/lib/rollout.js
• test/qoder-new-parser.test.js

Included review availability: Your plan provides up to 4 included reviews per hour; 3 remain after this review.

---

📝 Walkthrough

Walkthrough

Qoder CN project resolution now uses ~/.qoder-cn/projects. BYOK model identifiers are normalized without provider UUID prefixes. Tests cover path selection, token aggregation, duplicate messages, and repeated synchronization.

Changes

Qoder CN session handling

Layer / File(s)	Summary
Separate CN resolution and model normalization src/lib/rollout.js, src/commands/status.js, src/commands/sync.js	CN sessions use ~/.qoder-cn/projects, with directory divergence checks documented. BYOK UUID prefixes are removed before model normalization.
Resolver and incremental parser coverage test/qoder-new-parser.test.js	Tests cover CN path precedence, Anthropic cache-token aggregation, duplicate message suppression, BYOK bucket keys, cursor handling, and no-op reprocessing.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Merge Risk: ⚪ Minimal · up to d334b

此变更将新版 Qoder CN 会话发现切换到独立目录，并稳定化 BYOK 模型统计键。现有覆盖验证了目录隔离、解析聚合、去重与重复同步无新增数据，当前可合并。

Suggested reviewers: kewen-dev

Sequence Diagram(s)

sequenceDiagram
  participant CNSessionFiles
  participant parseQoderNewIncremental
  participant qoderNewModelFromRecord
  participant BucketAndCursorState
  CNSessionFiles->>parseQoderNewIncremental: Provide CN transcript and subagent JSONL rows
  parseQoderNewIncremental->>qoderNewModelFromRecord: Normalize model identifier
  qoderNewModelFromRecord->>BucketAndCursorState: Use bare model id and preserve cursor identifier
  parseQoderNewIncremental->>BucketAndCursorState: Aggregate usage and update cursor

Loading

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 2 functions across 3 files. (1 skipped: 1 …	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly describes the main change: fixing parsing for the new Qoder CN JSONL sessions stored in ~/.qoder-cn/projects.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

Full details: Docstring Coverage

Explanation

Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 2 functions across 3 files. (1 skipped: 1 too large.)

• Fix all pre-merge checks with AI

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[xiufengsun]

已通过 #589 合并，保留了你的原始提交。已验证 CN 独立目录、BYOK 模型归一化、去重及重复同步幂等性，完整回归和跨平台 CI 均通过。无需再修改；本次尚未发布新的 npm/桌面版本，谢谢！