oadmin/sub2api
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity
2,718 Commits 2 Branches 0 Tags
b9b52e74c63da4e1c0cfbe4944c7a13d159fa5c8
Commit Graph

1366 Commits

Author SHA1 Message Date
QTom
1f6a73f0db fix(openai): treat 401 {"detail":"Unauthorized"} as permanent auth failure 
- ratelimit_service: detect non-standard OpenAI 401 format and permanently disable account
- account_test_service: mark account error on 401 during connection test

Made-with: Cursor
 2026-04-02 20:44:05 +08:00
QTom
f2e596f6ec fix(oauth): 每次刷新都通过 backend-api 获取最新 plan_type 
账号订阅类型可能每月变化，id_token 中的 plan_type 是签发时的快照，
不一定反映当前状态。移除 plan_type == "" 前置条件，确保每次刷新都
调用 ChatGPT backend-api 获取实时订阅类型并覆盖旧值。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-04-02 20:43:56 +08:00
Yanzhe Lee
77ba9e728d Merge branch 'Wei-Shaw:main' into fix/openai-gateway-content-session-hash-fallback 2026-04-02 01:56:18 +08:00
YanzheL
cf9efefd96 fix(lint): satisfy errcheck for strings.Builder.WriteString calls 2026-04-02 01:03:22 +08:00
YanzheL
4fb1603001 test(gateway): add tests for content-based session hash fallback 
- 20 unit tests for deriveOpenAIContentSessionSeed covering:
  - Empty/nil inputs, model-only, stable across turns
  - Different model/system/first-user produce different seeds
  - Tools, functions, developer role, structured content
  - Responses API: input string, input array, instructions, input_text typed items
  - JSON canonicalization (whitespace/key-order insensitive)
  - Prefix presence, empty tools ignored, messages preferred over input
- 3 integration tests for GenerateSessionHash content fallback:
  - Content fallback produces stable hash
  - Explicit signals override content fallback
  - Empty body still returns empty hash
 2026-04-02 00:11:17 +08:00
YanzheL
c5aac1251d fix(gateway): add content-based session hash fallback for non-Codex clients 
When no explicit session signals (session_id, conversation_id, prompt_cache_key)
are provided, derive a stable session seed from the request body content
(model + tools + system prompt + first user message) to enable sticky routing
and prompt caching for non-Codex clients using the Chat Completions API.

This mirrors the content-based fallback already present in GatewayService.
GenerateSessionHash, adapted for the OpenAI gateway's request formats (both
Chat Completions messages and Responses API input).

JSON fragments are canonicalized via normalizeCompatSeedJSON to ensure
semantically identical requests produce the same seed regardless of
whitespace or key ordering.

Closes #1421
 2026-04-02 00:11:06 +08:00
QTom
b155bc564b fix(antigravity): 修复批量刷新令牌不设置隐私模式的问题 
- refreshSingleAccount ProjectIDMissing 提前返回前补上 EnsureAntigravityPrivacy 调用
- EnsureAntigravityPrivacy 跳过条件从"有任何值"改为"仅 privacy_set 成功时跳过"，
  privacy_set_failed 允许重试，对齐 OpenAI shouldSkipOpenAIPrivacyEnsure 的行为
- 后台 TokenRefreshService.ensureAntigravityPrivacy 同步修改
- ExchangeCode/ValidateRefreshToken 获得令牌后立即调用 setAntigravityPrivacy，
  不依赖后续账号创建流程

Made-with: Cursor
 2026-04-01 12:24:52 +08:00
Wesley Liddick
055c48ab33 Merge pull request #1262 from InCerryGit/main 
fix(openai): preserve bare gpt-5.3-codex-spark across forwarding paths
 2026-04-01 08:31:12 +08:00
YanzheL
4514f3fc11 fix(gemini): resolve customtools alias in mapping lookup 2026-04-01 02:19:42 +08:00
YanzheL
f00351c106 fix(openai): sanitize empty base64 input images 2026-04-01 00:46:38 +08:00
YanzheL
dd5978f222 fix(gemini): normalize ai studio google search tools 2026-04-01 00:45:56 +08:00
YanzheL
0ebe0ce585 fix(gemini): preserve google search in Claude compat tools 2026-04-01 00:33:39 +08:00
Wesley Liddick
83a16dec19 Merge pull request #1407 from DaydreamCoding/feat/cache-driven-rpm-buffer 
feat(gateway): Cache-Driven RPM Buffer
 2026-03-31 14:01:23 +08:00
Wesley Liddick
820c531814 Merge pull request #1406 from DaydreamCoding/feat/group-account-filter 
feat(group-filter): 分组账号过滤控制 — require_oauth_only + require_privacy_set
 2026-03-31 14:01:05 +08:00
QTom
72e5876c64 feat(gateway): Cache-Driven RPM Buffer 
- buffer 公式从 baseRPM/5 改为 concurrency + maxSessions
  保留 baseRPM/5 作为 floor 向后兼容
- 粘性路径 fallback 新增 [StickyCacheMiss] 结构化日志
  reason: rpm_red / gate_check / session_limit / wait_queue_full / account_cleared
- session_limit 路径跳过 wait queue 重试（RegisterSession 拒绝无副作用）
- 典型配置 buffer 从 3 提升至 13，大幅减少高峰期 Prompt Cache Miss

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-03-31 13:24:22 +08:00
QTom
aeed2eb9ad feat(group-filter): 分组账号过滤控制 — require_oauth_only + require_privacy_set 
为 OpenAI/Antigravity/Anthropic/Gemini 分组新增两个布尔控制字段：
- require_oauth_only: 创建/更新账号绑定分组时拒绝 apikey 类型加入
- require_privacy_set: 调度选号时跳过 privacy 未成功设置的账号并标记 error

后端：Ent schema 新增字段 + 迁移、Group CRUD 全链路透传、
      gateway_service 与 openai_account_scheduler 两套调度路径过滤
前端：创建/编辑表单 toggle 开关（OpenAI/Antigravity/Anthropic/Gemini 平台可见）

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-03-31 13:04:55 +08:00
QTom
46bc5ca73b feat(antigravity): 令牌刷新失败及创建账号时也设置隐私 
- token_refresh: 不可重试错误和重试耗尽两条路径添加 ensureAntigravityPrivacy
- admin_service: CreateAccount 为 Antigravity OAuth 账号异步设置隐私

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-03-31 11:42:23 +08:00
InCerry
0b3feb9d4c fix(openai): resolve Anthropic compat mapping from normalized model 
Anthropic compat requests normalize reasoning suffixes before forwarding, but the account mapping step was still using the raw request model. Resolve billing and upstream models from the normalized compat model so explicit account mappings win over fallback defaults.

Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent)

Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
 2026-03-31 10:33:28 +08:00
InCerry
ca8692c747 Merge remote-tracking branch 'upstream/main' 
# Conflicts:
#	backend/internal/service/openai_gateway_messages.go
 2026-03-31 09:38:40 +08:00
qingyuzhang
6b646b6127 fix(openai): fail over passthrough 429 and 529 2026-03-30 22:29:26 +08:00
haruka
49e99e9d51 fix: resolve errcheck lint for sync.Map type assertion 
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-03-30 16:44:15 +08:00
haruka
ad2cd97618 fix: resolve refresh token race condition causing false invalid_grant errors 
When multiple goroutines/workers concurrently refresh the same OAuth token,
the first succeeds but invalidates the old refresh_token (rotation). Subsequent
attempts using the stale token get invalid_grant, which was incorrectly treated
as non-retryable, permanently marking the account as ERROR.

Three complementary fixes:
1. Race-aware recovery: after invalid_grant, re-read DB to check if another
   worker already refreshed (refresh_token changed) — return success instead
   of error
2. In-process mutex (sync.Map of per-account locks): prevents concurrent
   refreshes within the same process, complementing the Redis distributed lock
3. Increase default lock TTL from 30s to 60s to reduce TTL-expiry races

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-03-30 16:23:38 +08:00
Wesley Liddick
6a2cf09ee0 Merge pull request #1349 from touwaeriol/feat/antigravity-internal500-penalty 
feat(antigravity): progressive penalty for consecutive INTERNAL 500 errors
 2026-03-30 15:54:04 +08:00
Wesley Liddick
c6fd88116b Merge pull request #1354 from wucm667/fix/billing-use-requested-model 
fix(billing): 计费始终使用用户请求的原始模型，而非映射后的上游模型
 2026-03-30 15:52:31 +08:00
Wesley Liddick
9a92fa4a60 Merge pull request #1370 from YanzheL/fix/1320-openai-messages-gpt54-xhigh 
fix(gateway): normalize gpt-5.4-xhigh for /v1/messages
 2026-03-30 15:44:34 +08:00
Wesley Liddick
b5642bd068 Merge pull request #1377 from DaydreamCoding/fix/lifecycle-stop-duplicate-close 
fix(lifecycle): TokenRefreshService Stop() 防重复 close
 2026-03-30 15:38:39 +08:00
Wesley Liddick
128f322252 Merge pull request #1376 from weak-fox/fix/privacy-without-refresh-token 
修复缺少 refresh_token 时被临时停调度
 2026-03-30 15:38:27 +08:00
Wesley Liddick
17d7e57a2e Merge pull request #1375 from weak-fox/fix/batch-reset-temp-unsched 
修复重置状态时未清理临时停调度
 2026-03-30 15:37:58 +08:00
shaw
50288e6b01 fix: 修复模型定价文件更新url 2026-03-30 15:36:53 +08:00
shaw
ab3e44e4bd fix: 适配X-Claude-Code-Session-Id头 2026-03-30 11:43:07 +08:00
QTom
61607990c8 fix(lifecycle): TokenRefreshService Stop() 防重复 close 
使用 sync.Once 包裹 close(stopCh)，避免多次调用 Stop() 时
触发 panic: close of closed channel。
 2026-03-30 10:33:06 +08:00
shaw
b65275235f feat: Anthropic oauth/setup-token账号支持自定义转发URL 2026-03-30 09:10:57 +08:00
weak-fox
e298a71834 fix: clear temp unsched when resetting account status 2026-03-30 00:22:02 +08:00
weak-fox
3f6fa1e3db fix: avoid temp unsched when refresh token is missing 2026-03-30 00:21:51 +08:00
YanzheL
f2c2abe628 fix(openai): keep xhigh normalization scoped to messages 2026-03-29 21:09:19 +08:00
YanzheL
8c10941142 fix(openai): normalize gpt-5.4-xhigh compat mapping 2026-03-29 20:52:29 +08:00
wucm667
f5764d8dc6 fix(billing): 计费始终使用用户请求的原始模型，而非映射后的上游模型 
当账号配置了模型映射（如 claude-sonnet-4-6 → glm-5.0）时，系统错误地
使用映射后的上游模型名计算费用。由于上游模型（如 glm-5.0）在定价系统中
没有价格配置，导致计费失败后被静默置为 0，用户不被扣费。

修改 forwardResultBillingModel 优先返回请求模型名，并移除 OpenAI 路径
中 BillingModel 字段对计费模型的覆盖逻辑。
 2026-03-28 16:22:06 +08:00
erio
9abdaed20c style: gofmt antigravity_internal500_penalty.go 2026-03-27 20:18:07 +08:00
erio
eb94342f78 chore: adjust internal500 penalty durations to 30m / 2h 2026-03-27 20:11:24 +08:00
erio
d563eb2336 test: add unit tests for INTERNAL 500 progressive penalty 
Cover isAntigravityInternalServerError body matching,
applyInternal500Penalty tier escalation, handleInternal500RetryExhausted
nil-safety and error handling, and resetInternal500Counter paths.
 2026-03-27 20:11:24 +08:00
erio
3ee6f085db refactor: extract internal500 penalty logic to dedicated file 
Move constants, detection, and penalty functions from
antigravity_gateway_service.go to antigravity_internal500_penalty.go.
Fix gofmt alignment and replace hardcoded duration strings with
constant references.
 2026-03-27 20:11:24 +08:00
erio
7cca69a136 fix: move internal500 counter reset to cover all success paths 
Move the reset logic after urlFallbackLoop so it covers both direct
success and smart retry (429/503) success paths.
 2026-03-27 20:11:24 +08:00
erio
093a5a260e feat(antigravity): progressive penalty for consecutive INTERNAL 500 errors 
When an antigravity account returns 500 "Internal error encountered."
on all 3 retry attempts, increment a Redis counter and apply escalating
penalties:
- 1st round: temp unschedulable 10 minutes
- 2nd round: temp unschedulable 10 hours
- 3rd round: permanently mark as error

Counter resets on any successful response (< 400).
 2026-03-27 20:11:24 +08:00
InCerryGit
b6d46fd52f Merge branch 'Wei-Shaw:main' into main 2026-03-27 17:35:47 +08:00
QTom
c729ee425f fix(gateway): 修复 OpenAI→Anthropic 转换路径 system prompt 被静默丢弃的 bug 
injectClaudeCodePrompt 和 systemIncludesClaudeCodePrompt 的 type switch
无法匹配 json.RawMessage 类型（Go typed nil 陷阱），导致 ForwardAsResponses
和 ForwardAsChatCompletions 路径中用户 system prompt 被替换为仅 Claude Code
banner。新增 normalizeSystemParam 将 json.RawMessage 转为标准 Go 类型。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-03-27 14:51:36 +08:00
QTom
47a544230a fix(privacy): 刷新令牌失败时也尝试设置 OpenAI 隐私模式 
刷新失败不代表 access_token 无效，在后台定时刷新（不可重试错误 +
重试耗尽）和前端批量/单次刷新的失败路径中，均利用可能仍有效的
access_token 调用隐私设置。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-03-27 14:51:36 +08:00
QTom
c13c81f09d feat(privacy): 为 OpenAI OAuth 账号添加前端手动设置隐私按钮 
复用已有的 set-privacy API 端点，Handler 通过 platform 分发到
ForceOpenAIPrivacy / ForceAntigravityPrivacy，前端 AccountActionMenu
扩展隐私按钮支持 OpenAI OAuth 账号。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
 2026-03-27 14:51:36 +08:00
Wesley Liddick
b688ebeefa Merge pull request #1215 from weak-fox/fix/privacy-retry-failed-mode 
fix: 允许 OpenAI privacy_mode修改失败后能在 token 刷新时重试
 2026-03-27 14:46:38 +08:00
shaw
1854050df3 feat(tls-fingerprint): 新增 TLS 指纹 Profile 数据库管理及代码质量优化 
新增功能：
- 新增 TLS 指纹 Profile CRUD 管理（Ent schema + 迁移 + Admin API + 前端管理界面）
- 支持账号绑定数据库中的自定义 TLS Profile，或随机选择（profile_id=-1）
- HTTPUpstream.DoWithTLS 接口从 bool 改为 *tlsfingerprint.Profile，支持按账号指定 Profile
- AccountUsageService 注入 TLSFingerprintProfileService，统一 usage 场景与网关的 Profile 解析逻辑

代码优化：
- 删除已被 TLSFingerprintProfileService 完全取代的 registry.go 死代码（418 行）
- 提取 3 个 dialer 的重复 TLS 握手逻辑为 performTLSHandshake() 共用函数
- 修复 GetTLSFingerprintProfileID 缺少 json.Number 处理的 bug
- gateway_service.Forward 中 ResolveTLSProfile 从重试循环内重复调用改为预解析局部变量
- 删除冗余的 buildClientHelloSpec() 单行 wrapper 和 int64(e.ID) 无效转换
- tls_fingerprint_profile_cache.go 日志从 log.Printf 改为 slog 结构化日志
- dialer_capture_test.go 添加 //go:build integration 标签，防止 CI 失败
- 去重 TestProfileExpectation 类型至共享 test_types_test.go
- 修复 9 个测试文件缺少 tlsfingerprint import 的编译错误
- 修复 error_policy_integration_test.go 中 handleError 回调签名被错误替换的问题
 2026-03-27 14:33:05 +08:00
shaw
d571f300e5 feat(rectifier): 请求整流器增加 API Key 账号签名整流支持 
新增独立开关控制 API Key 账号的签名整流功能，支持配置自定义
匹配关键词以捕获不同格式的上游错误响应。

- 新增 apikey_signature_enabled 开关（默认关闭）
- 新增 apikey_signature_patterns 自定义关键词配置
- 内置签名检测规则对 API Key 账号同样生效
- 自定义关键词对完整响应体做不区分大小写匹配
- 重试二阶段检测仅做模式匹配，不重复校验开关
- Handler 层校验关键词数量（≤50）和长度（≤500）
- API 响应 nil patterns 统一序列化为空数组
- OAuth/SetupToken/Upstream/Bedrock 账号行为不变
 2026-03-26 16:43:38 +08:00