Commit Graph
22 Commits
Author SHA1 Message Date
Randark 2c250bfd73 fix(monitor-ui): localize the "quota" placeholder model across monitor views
纯配额模式的主模型存储值是占位符 "quota"(数据源是关联账号而非模型),
此前在管理列表主模型列、用户端监控卡片、用户端详情弹窗、运行结果弹窗
四处被当作真实模型名直接渲染,读起来像假模型。

新增 useChannelMonitorFormat.formatMonitorModel:占位符替换为本地化
「配额」标签(复用既有 monitorCommon.checkMode.quota,零新增 i18n 键),
quota_probe/probe 的真实模型名不受影响;纯匹配逻辑(如 15d/30d 可用率
按模型名回查 detail)仍用原始值。

问题定位备注:另一反馈「设置页看不到 channel_monitor_show_quota 开关」
经查为部署镜像过旧(开关随 #5761 于 08-18 合入,反馈环境容器构建于
08-15),上游代码中开关完整存在,更新镜像即可,无需改码。
2026-08-18 10:35:57 +00:00
Randark c9effc4560 fix(frontend): monitor form check-mode restore, account unbinding and mode badge
审计 P2-7 三处前端收尾:

- 切离 antigravity 时对称还原 check_mode:antigravity 强制 quota,但反向
  切走没有还原——编辑存量 antigravity 监控换平台后停留在 quota,目标平台
  未必支持。现切走时还原为 probe 并清掉占位模型 'quota'(回填值),否则
  probe 模式拿 'quota' 当探活模型(与离开 grok 清 DEFAULT_GROK_MODEL
  同一先例)。
- update 切回 probe 显式发 account_id=0:后端语义 0=解绑、null=不动,
  此前 probe 模式 payload 带 null,存量关联永远清不掉。仅 update——
  create 发 0 会落 &0 触发外键违约(API 类型注释同步写明)。
- 管理列表 provider 徽章旁并列 check_mode 徽章(checkModeLabel 此前
  定义了却无处使用):quota 系=蓝、probe=中性灰,三种模式一眼可分;
  i18n 复用既有 monitorCommon.checkMode.*,零新增键。
2026-08-18 10:28:28 +00:00
Randark e2dfb3b8cb refactor(monitor): pass loaded account through quota sources (single load per fetch)
配额 fetcher 缓存未命中时账号被加载两次:fetchUncached GetByID 一次,
下游数据源(GetUsage / CN QueryUsage / CN QueryBalance)按 ID 再各自
GetByID 一次——每次 GetByID 含 proxies/groups 联查,纯属重复劳动。

统一改为「路由前加载一次、指针直传」:
- fetcher 三个数据源接口签名改收 *Account:
  GetUsageForAccount / QueryUsageForAccount / QueryBalanceForAccount
- AccountUsageService 暴露 GetUsageForAccount 直通(getUsageForAccount
  既有逻辑零改动)
- CN 两侧服务抽 validateCodingPlanAccount / validatePayGAccount
  (加载后校验原样),新 ForAccount 入口 = NOT_CONFIGURED 守卫 → 校验 →
  singleflight → 探测;ID 入口 = load + 委托,语义不变
- singleflight key 保持 "cn_quota:<id>" / "cn_balance:<id>":ForAccount
  与 admin handler 的 ID 入口并发探测仍按账号合并
- cn_provider_balance_check_service.checkOne 刻意留在 ID 入口:调度器
  路径按 ID 取的是最新凭据,不复用可能已过期的已加载账号

校验移到 flight 之外且 ForAccount 复用同一套校验,直传无法绕过
平台/模式检查;专项测试固化(无效账号零出站请求、单次加载指针直传
require.Same)。
2026-08-18 10:28:28 +00:00
Randark c41ae19e52 fix(monitor): reject unusable quota data sources and invalid mode combos at write time
P2-2/P2-4/P2-5 from review:

- validateLinkedAccount/revalidateLinkedAccount now check
  monitorAccountQuotaCapability after the platform match, blocking combos
  that would permanently error at runtime: deepseek coding and
  custom-domain kimi coding (no quota endpoint), zhipu payg (no balance
  endpoint), anthropic/openai API-Key accounts (usage query requires
  oauth; anthropic setup-token still allowed via local estimation).
  gemini/grok/antigravity stay permissive. Quota mode surfaces
  CHANNEL_MONITOR_ACCOUNT_NOT_SUPPORTABLE on edit; probe mode silently
  unbinds (same as platform mismatch).
- applyMonitorUpdate re-runs validateCheckMode on the effective
  provider+check_mode whenever either field is patched, so a
  provider-only update can no longer persist antigravity+probe (the
  check is conditioned on provider/check_mode presence so legacy illegal
  rows can still be renamed/disabled).
- normalizeMonitorPrimaryModel only substitutes the "quota" placeholder
  for pure quota mode; quota_probe with an empty model now fails with
  MissingPrimaryModel instead of probing model="quota". The quota branch
  also moves ahead of the grok default, so grok+quota now gets "quota"
  rather than "grok-4.5" (review L-item).

Tests: capability matrix (15 cases incl. anti-over-blocking rows),
linked-account wiring, revalidate quota/probe split, provider-only
update bypass + legacy-row rename, quota_probe missing-model create,
normalize matrix.
2026-08-18 10:28:28 +00:00
Randark 1128df2592 fix(monitor): align quota-fetcher credential/balance semantics with scheduler
P2-1/P2-3 from review:

- fetchCNQuota: credential-invalid now judged by StatusCode 401/403
  (aligned with fetchCNBalance) instead of `!Success && !CredentialValid` —
  CN quota service only sets CredentialValid=true on the success path, so
  500/429/zhipu business errors were all misclassified as failed instead
  of error.
- fetchCNBalance: snapshot carries new BalanceLow flag computed with the
  scheduler's exact criterion (`!Available || allCNBalancesBelowThreshold`)
  against Gateway.CNProviders.BalanceThreshold (ctor now takes cfg; wire
  regenerated). quotaDegradedHint reports "balance low" instead of the old
  `<=0` check, so an account already paused by the scheduler (balance 5 /
  threshold 10) no longer shows green in the monitor.
- threshold helper falls back to viper default 0.5 for nil/<=0 config to
  avoid a zero-threshold regression where balance=0 stops alerting.

Tests: CN quota status-code matrix (rewrites the test that cemented the
old behavior), balance-low matrix (below-threshold / unavailable /
multi-currency healthy), threshold-from-config; PayG stubs now set
Available explicitly (zero-value trap).
2026-08-18 10:28:28 +00:00
Randark 5cbd0c96a2 fix(monitor-ui): 关联账号选择器改服务端搜索+回填,OpenAI 配额模式加消耗提示
- 选择器改为 /admin/accounts?platform=&search= 服务端搜索(50 条/页,
  300ms 防抖由 Select 承担),AbortController + 序列守卫防乱序覆盖
- 已绑定账号不在结果页时用 getById 回填为固定选项,删除「不在列表即
  静默清空绑定」;绑定仅在显式切换 provider 或账号确实失效时清空(带提示)
- OpenAI 平台配额模式显示 Codex 探测额度消耗提示;checkModeQuotaHint
  修正为「不发送探活请求」不再宣称「不产生 LLM 请求」
- grok spec 补齐缺失的 accounts mock;新增账号选择器 7 用例
  (加载/防抖搜索/回填保留/失效清空/提示门控/固定选项/切平台清空)
2026-08-18 09:14:44 +00:00
Randark 03c3f3b6fc feat(ui): Select 组件支持可选远程搜索(remote/loading props + search 事件)
- remote 模式始终显示搜索框且不做本地过滤,输入 300ms 防抖后 emit('search')
- 关闭/卸载时取消排队中的防抖定时器,避免尾随空 query
- loading 时空态显示加载文案;未设 remote 时行为不变(回归用例保障)
2026-08-18 09:14:31 +00:00
Randark 22df600d09 fix(channel-monitor): 配额快照识别值通道失败并加 60s 负缓存与 singleflight
- fetchUsage 显式识别 UsageInfo.Error/ErrorCode/NeedsReauth/IsBanned/IsForbidden
  (antigravity/grok 等平台失败不走 Go error 通道,此前会被误判为 operational)
- 凭据失效语义(401/403)标记 CredentialInvalid → failed 状态;
  grok quota_unknown 已知未知态豁免不判失败
- 失败快照进 60s 负缓存,避免故障/凭据失效期间以最小 15s 间隔打上游
- 同账号并发抓取由 singleflight 合并(脱离调用方 ctx,45s 总超时兜底)
2026-08-18 09:14:20 +00:00
Randark ebcae03afd fix(lint): 补齐 setting_public.go 与配额 fetcher 测试的 gofmt 对齐 2026-08-18 06:29:43 +00:00
Randark 7376f4a48c fix(lint): gofmt 对齐 + 移除未使用的 isSupportedProvider
- 三个文件补 gofmt 规范对齐(结构体尾部注释、const 块字段对齐)
- isSupportedProvider 在 validateProvider 改用能力集合后无引用,删除
2026-08-18 05:57:41 +00:00
Randark 302a10b886 test(monitor-ui): 配额视图渲染与开关门控用例
- 新增 MonitorQuotaView.spec:tier window/label 的 i18n 映射与未知
  token 透出、阈值配色(≥90 红/≥75 黄/其余绿)、进度条宽度钳制、
  多币种余额与单余额回退、plan_level 徽章、错误截断与失败态不渲染 tiers
- 新增 MonitorCard.quota.spec:latest_quota 受系统开关门控,
  无快照或开关关闭时不渲染
- 既有工厂/断言补 check_mode、account_id;provider 网格断言 4→8
2026-08-18 05:10:10 +00:00
Randark a20e1c00cc feat(monitor-ui): 配额模式表单、用量快照视图与 8 平台支持
管理端:
- MonitorFormDialog:新增检查方式三态分段控件(探活/配额/探活+配额),
  provider 网格扩展至 8 平台(+antigravity/kimi/zhipu/deepseek),
  国产平台默认 endpoint 预填、antigravity 强制 quota-only;配额模式
  显示关联账号选择器(复用 adminAPI.accounts.list 按平台过滤),
  隐藏 endpoint/api_key/模型等探活专属字段
- 新建共用 MonitorQuotaView:用量窗口进度条(阈值/样式对齐账号页
  CNProviderQuotaCell)、多币种余额、plan_level 徽章与错误态;
  window/label 机器 token 走 i18n,未知 token 原样透出(前向兼容)
- 列表主模型单元格内联 latest_quota;运行结果弹窗逐行展示配额;
  筛选栏/模板管理/格式化 composable 同步 8 平台(配色对齐
  utils/platformColors.ts);ProviderIcon 搬运 4 家官方 SVG

用户端:
- MonitorCard 支持 latest_quota,渲染受 channel_monitor_show_quota
  系统开关(isChannelMonitorQuotaVisible)控制,服务端剥离为第一道防线

设置与 i18n:
- 管理设置页新增「向用户展示渠道用量/余额」开关(默认关闭);
  zh/en 全量补充 providers/checkMode/quota 文案
2026-08-18 05:09:58 +00:00
Randark c51fd7d0b3 test(channel-monitor): adapt checker body test to 3-arg normalizeMonitorPrimaryModel 2026-08-18 04:51:06 +00:00
Randark 7ab6d3db66 test(channel-monitor): quota mode unit/integration/migration coverage
- fetcher:海外/CN coding/CN payg 分派、账号缺失、凭据失效标记、
  TTL 命中与失败不缓存、UsageInfo→tiers 全窗口归一、状态推导矩阵
- quota mode:RunCheck 三分派(quota 单行 / quota_probe 仅挂主行 /
  probe 不变)、配额失败不翻探活状态、校验矩阵、关联账号复核
  (quota 报错 probe 自动解绑)、quota→probe 切换强制重填 key、
  Duplicate 空明文重加密
- settings:channel_monitor_show_quota 缺省关闭 + 仅 "true" 开启
- 迁移内容断言(仿 176 grok 迁移测试)
- repo 集成测试:check_mode/account_id 往返、quota JSONB 回读、
  探活旧行 NULL 兼容(testcontainers PG 实跑迁移 226)
- 修复 Fetch 在 nil receiver 上的 panic(缓存查询先于原 nil 守卫)
2026-08-18 04:51:02 +00:00
Randark 41344c20ff feat(monitor): wire quota fetcher & expose check_mode in handlers
- handler DTO: create/update 接收 check_mode/account_id,provider oneof 扩至 8 家,
  endpoint/api_key 改为 omitempty(条件必填下沉 service 校验);
  monitor/checkResult/historyItem 响应透传 check_mode/account_id/quota
- 用户端 latest_quota 由 channel_monitor_show_quota 控制,关闭时服务端剥离
- wire: NewChannelMonitorQuotaFetcher 以具体服务类型收参(窄接口包内保留供
  stub),ProvideChannelMonitorRunner 注入后 SetQuotaFetcher
2026-08-18 04:14:15 +00:00
Randark 6a6fd304f6 feat(settings): channel_monitor_show_quota public setting (default off)
- 新增公开设置 key(迁移 226 已插入默认 false),控制用户端监控页
  是否展示配额/余额;管理端不受影响
- 解析 fail-closed:仅字面 "true" 视为开启(对齐 available_channels_enabled
  语义,而非 hide_throughput 的 fail-open)
- 全链路贯通:domain key 常量 → setting_public(公开读取 + ChannelMonitorRuntime
  .ShowQuota + 注入 payload)→ settings_view/parse/update → handler DTO
  (admin/user 响应 + admin 更新请求)→ api_contract_test 两个 wantJSON 块
2026-08-18 04:06:18 +00:00
Randark c44711ac98 feat(channel-monitor): quota mode service layer (fetcher + dispatch + repo)
- 常量:MonitorProvider{Antigravity,Kimi,Zhipu,Deepseek}、MonitorCheckMode 三态、
  配额缓存/告警阈值参数;新增 check_mode 相关业务错误
- 校验:monitorProviders(8)/probeCapableProviders(7) 集合替代 adapter 表判定,
  validateCheckMode 矩阵(antigravity 仅 quota);quota 模式 primary_model
  默认占位 "quota",endpoint/api_key 条件必填
- checker:kimi/deepseek 复用 /v1/chat/completions、zhipu 走
  /api/paas/v4/chat/completions;merge 黑名单与 replace 校验同步扩容
- ChannelMonitorQuotaFetcher:窄接口聚合账号侧三个用量服务 + 账号加载,
  归一为 domain.MonitorQuotaSnapshot;成功快照 5min TTL 缓存防配额端点风暴;
  Fetch 永不返回 error(失败降级 Success=false 快照);状态推导
  operational/degraded(≥90% 或余额耗尽/账号未关联)/failed(401/403)/error
- RunCheck 按 check_mode 分派:quota 单行结果、quota_probe 探活+配额挂主行、
  probe 不变;Update 组合校验 + 关联账号复核(probe 自动解绑失效账号);
  Duplicate 拷贝新字段且 quota 模式空明文重加密(防 runner 误停摆)
- repo:Create/Update/entToServiceMonitor 透传 check_mode/account_id;
  InsertHistoryBatch 写 quota;ListLatestForMonitorIDs 裸 SQL 加 quota 列 +
  scanMonitorQuota JSONB 解包(NULL 安全);ListHistory 透传 quota
2026-08-18 04:03:13 +00:00
Randark 615e6901ea feat(channel-monitor): add quota mode schema (migration 226 + ent)
- 迁移 226:provider CHECK 扩到 8 平台(monitors + request_templates)、
  channel_monitors 加 check_mode/account_id(FK ON DELETE SET NULL)、
  channel_monitor_histories 加 quota JSONB、插入公开设置
  channel_monitor_show_quota(默认 false)
- ent schema 同步:provider 枚举 +4、endpoint 去 NotEmpty(quota 模式存空串)、
  新增 check_mode/account_id 字段、history 加 quota JSON 快照
- domain 新增归一化配额快照类型 MonitorQuotaSnapshot/Tier/Balance
  (放 domain 是因为 ent schema 需要引用,service 会造成 import cycle)
2026-08-18 03:50:01 +00:00
Randark 7e45634df9 chore: remove leftover Sora references after platform removal
PR #1463 removed the Sora platform, but some references survived:

- README/README_CN/README_JA kept the 'Sora status (temporarily
  unavailable)' sections and gateway.sora_* docs that the removal PR
  never touched.
- deploy/config.example.yaml still documented ~130 lines of sora_*
  gateway keys, the top-level sora: direct-client/storage block, and
  token_refresh.sync_linked_sora_accounts - none of which map to any
  field in the config structs anymore.
- The OIDC login PR (02a66a01c, branched off pre-removal main and
  merged 4 days after #1463) re-added the dead
  PublicSettings.SoraClientEnabled field, which no code ever sets.
- A release sync (748a84d87) re-introduced sora i18n keys that the
  later i18n split (d9e514f98) faithfully carried into
  locales/{zh,en}/admin/{overview,settings}.ts. No component references
  any of these keys.

This drops all of the above. Pure deletions, no behavior change.
2026-08-18 00:51:30 +00:00
Randark e728545382 style: gofmt 迁移测试注释(CI golangci-lint gofmt) 2026-08-17 04:26:22 +00:00
Randark 4b667ccd45 fix(review): 处理 PR #5666 四个阻断项(B1-B4)
- B1: 移除根目录 docker-compose.yml(个人镜像测试产物混入)
- B2: 新增迁移 224 放宽 user_platform_quotas CHECK 至 8 平台,补
  BulkInsertInitial 国产平台集成测试与迁移内容断言测试
- B3: CC/Responses×anthropic 四个上游读循环接入间隔超时泵
  (gateway.stream_data_interval_timeout,默认 180s):上游挂住 SSE
  不发数据也不断连时结束排水/组装、关闭 resp.Body 归还连接池位,
  排水超时仍按已累计 usage 返回(与 messages 主路径同语义)
- B4: 配额/余额探测发起前过 cnValidateProbeURL(与网关转发同一套
  security.url_allowlist 策略),被拒时零出站、API key 不离开本机
2026-08-17 04:07:06 +00:00
Randark 901a0439f1 feat: 国产供应商一等支持(Kimi/Zhipu/DeepSeek 多协议 + 配额/余额监控)
后端:
- 协议凭证维度 credentials[api_protocol] ∈ chat_completions(默认)/anthropic/responses(deepseek)
- /v1/messages 零转换直通原生 Anthropic 端点(kimi/zhipu/deepseek),CC/Responses
  入站交叉组合走 apicompat 双向转换链(responses/chat_completions anthropic-native 转发器)
- count_tokens:anthropic 协议透传原生端点;其余 CN 协议本地 tiktoken 估算
- Coding Plan 额度探测(5h/weekly 滚动窗口)+ payg 余额探测(kimi/deepseek),
  deepseek 双币种 CNY+USD 明细,任一币种达标不停调
- 周期任务 [CNBalance] 并发探测 + 预算随工作量放大;响应式 429 冷却到最早窗口
  重置点;余额不足可恢复临时停调;智谱 CREDIT_LIMIT 不污染窗口解析
- CC→anthropic 流式客户端断开后继续排水上游保住 usage 计量

前端:
- 创建/编辑弹窗 account_mode + api_protocol + base_url 联动预设(含 watcher 竞态防护)
- 用量单元格:kimi/zhipu coding 显示 5h/weekly 窗口,kimi/deepseek payg 显示余额,
  多币种并列展示;探测失败保留快照;挂载自动探测 5min 去抖
- 调度阈值设置面板补 kimi/zhipu 平台(对齐后端 AllowedSchedulingThresholdPlatforms)
2026-08-15 10:37:51 +00:00