fix(log): preserve quota in usage statistics

[jimmyleocn]

Agent

• Tool: Codex
• Tool version: 0.150.1
• Model (full id): gpt-5.6-sol
• Host (CLI / IDE / GitHub coding agent / other): IDE
• Date (UTC): 2026-08-31
• AI assistance disclosure: 本变更由 AI 辅助定位、实现和验证，提交者已审阅代码及本 PR 描述，并对其准确性负责。

Links

• Closes bug: 使用日志顶部用量显示 ¥0，但消费明细 quota 非零 #7106
• Related: perf: SumUsedQuota dual serial aggregation doubles table scan on /api/log/stat #6106, perf: merge dual serial aggregation in SumUsedQuota into single conditional query #6107

User request

修复使用日志统计中 quota 始终为 0、但消费日志明细和 RPM/TPM 正常的问题，并确认问题存在于未修改的上游代码。

Out of scope — refuse

• Matched: no
• If yes, what was told to the user (stop here; do not open a PR): not applicable

本变更不涉及 Coding Plan、逆向渠道、第三方 API wrapper、Codex 协议、透传模式或第三方托管服务。

Kind

• Bug fix
• New feature
• Performance / refactor
• Docs
• Other:

Issue facts

• Actual behavior: /api/log/stat 返回的 quota 为 0，但 RPM/TPM、消费日志明细以及数据库中的 logs.quota 均为非零。
• Impact: 管理员和用户会看到错误的用量汇总，可能误判为没有产生消费。底层消费日志和实际扣费数据没有丢失。
• Frequency: 在当前 GORM v1.25.12 下，只要 quota 聚合结果非零并继续执行 RPM/TPM 扫描，该问题即可稳定复现。
• Evidence that the problem is in new-api rather than the client or upstream:
  • 未修改的 upstream main 仍将两个不同投影的查询结果扫描到同一个 Stat。
  • 数据库直接汇总 logs.quota 为非零，但统计接口返回的 data.quota 为 0。
  • 独立 GORM 最小复现证明第二次 Scan 会将第一次扫描得到的 Quota 清零。
  • 该统计流程不涉及上游模型请求。
• Applicable types and their fields:
  • Relay: not applicable。
  • Billing: 消费日志和扣费记录正常，错误仅发生在只读统计结果。
  • Frontend: 前端直接显示统计接口返回的错误 quota；本 PR 不修改前端。
  • Deployment: 问题已在 PostgreSQL 部署中复现；回归测试使用 SQLite。

Change

SumUsedQuota 原先依次将 quota 查询和 RPM/TPM 查询扫描到同一个 Stat：

tx.Scan(&stat)
rpmTpmQuery.Scan(&stat)

第一个查询只返回 quota，第二个查询只返回 rpm 和 tpm。GORM v1.25.12 初始化第二次结构体扫描时会清零目标值，因此第一次查询写入的 Quota 被重置为 0。

本 PR 将 RPM/TPM 查询扫描到独立的局部结构体，然后只把 Rpm 和 Tpm 赋回最终结果：

var rateStat struct {
    Rpm int
    Tpm int
}
rpmTpmQuery.Scan(&rateStat)
stat.Rpm = rateStat.Rpm
stat.Tpm = rateStat.Tpm

这样可以保留第一次聚合得到的 quota，同时维持原有 RPM/TPM 查询及最近 60 秒统计规则。

新增回归测试，验证一次统计请求能够同时返回正确的 quota、RPM 和 TPM。

Research

Duplicate / prior art

• Search queries (issues, PRs):
  • SumUsedQuota
  • rpmTpmQuery
  • usage quota 0
  • log stat quota
  • quota RPM TPM
• What already existed and why this is not a duplicate:
  • bug: 使用日志顶部用量显示 ¥0，但消费明细 quota 非零 #7106 是本 PR 对应的 Bug Issue。
  • perf: SumUsedQuota dual serial aggregation doubles table scan on /api/log/stat #6106 和 perf: merge dual serial aggregation in SumUsedQuota into single conditional query #6107 讨论将两次聚合合并为一次查询，属于性能优化。
  • 本 PR 修复的是 GORM 二次扫描覆盖 quota 的正确性问题，不合并查询，也不改变现有统计语义。
  • 截至 2026-08-31，未发现修复该正确性问题的现有 PR。

Docs and code

• https://docs.newapi.ai/ : 未发现将使用日志 quota 汇总配置为 0 的选项或预期行为。
• https://deepwiki.com/QuantumNous/new-api : 日志统计由后端接口提供 quota、RPM 和 TPM，没有要求存在消费记录时 quota 返回 0。
• README / repo docs: 项目将用量统计和成本核算列为支持场景，没有与该现象对应的配置说明。
• Code paths and what they imply for this change:
  • controller/log.go：GetLogsStat 和 GetLogsSelfStat 调用 model.SumUsedQuota。
  • model/log.go：quota 查询和 RPM/TPM 查询原先扫描到同一个 Stat。
  • web/src/features/usage-logs/api.ts：使用日志统计请求 /api/log/stat 或 /api/log/self/stat。
  • 前端直接使用接口返回的 quota，因此根因位于后端统计结果映射。

Alternatives considered

• Option A: 使用独立的 RPM/TPM 结果结构体，再将两个字段赋回最终结果。
• Option B: 使用条件聚合，将 quota 和最近 60 秒 RPM/TPM 合并为一条 SQL。
• Why this approach: Option A 是最小、直接且跨数据库的正确性修复，不改变过滤条件和 RPM/TPM 最近 60 秒的业务语义。单查询性能优化应在独立 PR 中处理。

Files

Path	Why
model/log.go	防止 RPM/TPM 查询覆盖已经聚合得到的 quota
model/log_stat_test.go	验证 quota、RPM 和 TPM 可以在同一次统计中正确返回

Behavior

• Before: quota 查询得到非零值，但第二次 Scan(&stat) 将其清零，最终接口返回 quota: 0。
• After: RPM/TPM 使用独立结构体接收，最终结果同时保留正确的 quota、RPM 和 TPM。
• Explicit non-goals / leftover work:
  • 不修改消费日志写入、扣费或用户余额。
  • 不修改 RPM/TPM 最近 60 秒的统计规则。
  • 不合并两次数据库查询。
  • 不修改前端。
  • 不处理 perf: SumUsedQuota dual serial aggregation doubles table scan on /api/log/stat #6106/perf: merge dual serial aggregation in SumUsedQuota into single conditional query #6107 中的性能优化。

Verification

Only what was actually run.

• Commands and results:
  • go test ./model -run TestSumUsedQuotaPreservesQuotaWhenLoadingRateStats -count=1 -p 1 -parallel 1
    • PASS
  • go test ./model -count=1 -p 1 -parallel 1
    • PASS
  • go test ./controller -count=1 -p 1 -parallel 1
    • PASS
  • git diff --cached --check
    • PASS
  • GitNexus impact analysis:
    • Risk: LOW
    • Direct callers: GetLogsStat, GetLogsSelfStat
    • No affected execution flows detected
• Manual steps and observed result:
  • 使用相同扫描结构分别测试 GORM v1.25.2 和 v1.25.12。
  • v1.25.2 第二次扫描后保留 Quota: 12345。
  • v1.25.12 第二次扫描后将其重置为 Quota: 0。
  • 改用独立 RPM/TPM 结构体后，回归测试得到 Quota: 12345、Rpm: 1、Tpm: 30。
• UI: 本 PR 不包含前端变更；问题及根因证据已记录在 bug: 使用日志顶部用量显示 ¥0，但消费明细 quota 非零 #7106。
• Tests added or updated, or why none:
  • 新增 TestSumUsedQuotaPreservesQuotaWhenLoadingRateStats。
• Databases / providers / platforms exercised:
  • SQLite：自动化回归测试。
  • PostgreSQL：Issue 中的实际问题复现环境。
  • Provider: not applicable。
  • Platform: Windows amd64 开发环境。

Risks

• Failure modes: 若未正确赋回 Rpm 或 Tpm，速率统计可能返回 0；新增回归测试同时覆盖 quota、RPM 和 TPM。
• Billing / quota / auth impact: 不修改计费、额度扣减、认证或日志写入，仅修正只读统计接口的返回结果。
• Follow-ups: 如需减少两次表扫描，可在 perf: SumUsedQuota dual serial aggregation doubles table scan on /api/log/stat #6106/perf: merge dual serial aggregation in SumUsedQuota into single conditional query #6107 的范围内单独进行性能优化。

Scope check

• Single focused change: yes
• Secrets included: no
• Out of scope (Coding Plan / reverse-engineered channel / third-party wrapper / Codex): no

Summary by CodeRabbit

• Bug Fixes

  • Fixed quota reporting so logged quota values are preserved while request and token usage metrics are calculated.
  • Improved accuracy of aggregate usage statistics for matching consumption records.

• Tests

  • Added coverage to verify quota, request, and token metrics are reported together correctly.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: d534e23b-9887-4be0-8576-4297e4111c09

📥 Commits

Reviewing files that changed from the base of the PR and between 2b6f1df and 96daf7a.

📒 Files selected for processing (2)

• model/log.go
• model/log_stat_test.go

Included review availability: Your plan provides up to 10 included reviews per hour; 9 remain after this review.

---

Walkthrough

SumUsedQuota now loads RPM and TPM into a separate structure, preserving the quota from the earlier query. A regression test verifies quota, RPM, and TPM values for matching consumption logs.

Changes

Quota statistics

Layer / File(s)	Summary
Preserve quota during rate aggregation model/log.go, model/log_stat_test.go	SumUsedQuota copies RPM and TPM from a local structure without overwriting quota. The test verifies all three returned values.

Estimated code review effort: 2 (Simple) | ~10 minutes

Merge Risk: ⚪ Minimal · up to 96daf

This change corrects usage statistics so quota is preserved while RPM and TPM remain available, without changing billing, authorization, logging, API shape, or deployment behavior. No actionable merge-blocking risk remains beyond normal checks and review.

Suggested reviewers: calcium-ion

Poem

A rabbit checked the quota stack
RPM and TPM came hopping back
The saved total stayed in sight
A test kept every number right
“No more zero,” whispered the night

🚥 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 2 files.	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 primary change: preserving quota in usage statistics.
Linked Issues check	✅ Passed	The changes address issue #7106 by preventing the RPM/TPM query from overwriting the aggregated quota. The regression test verifies that quota, RPM, and TPM remain available together.
Out of Scope Changes check	✅ Passed	The changes are limited to the usage-statistics aggregation fix and its regression test. No unrelated code changes are present.

• 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.}