diff --git a/.github/workflows/backend-ci.yml b/.github/workflows/backend-ci.yml index 01c00bb9..342f48da 100644 --- a/.github/workflows/backend-ci.yml +++ b/.github/workflows/backend-ci.yml @@ -17,6 +17,7 @@ jobs: go-version-file: backend/go.mod check-latest: false cache: true + cache-dependency-path: backend/go.sum - name: Verify Go version run: | go version | grep -q 'go1.26.1' @@ -36,6 +37,7 @@ jobs: go-version-file: backend/go.mod check-latest: false cache: true + cache-dependency-path: backend/go.sum - name: Verify Go version run: | go version | grep -q 'go1.26.1' diff --git a/.gitignore b/.gitignore index 297c1d6f..da112576 100644 --- a/.gitignore +++ b/.gitignore @@ -78,6 +78,7 @@ Desktop.ini # =================== tmp/ temp/ +logs/ *.tmp *.temp *.log @@ -128,8 +129,15 @@ deploy/docker-compose.override.yml vite.config.js docs/* .serena/ + +# =================== +# 压测工具 +# =================== +tools/loadtest/ +# Antigravity Manager +Antigravity-Manager/ +antigravity_projectid_fix.patch .codex/ frontend/coverage/ aicodex output/ - diff --git a/AGENTS.md b/AGENTS.md new file mode 100644 index 00000000..8edfa58b --- /dev/null +++ b/AGENTS.md @@ -0,0 +1,1392 @@ +# Sub2API 开发说明 + +## 版本管理策略 + +### 版本号规则 + +我们在官方版本号后面添加自己的小版本号： + +- 官方版本：`v0.1.68` +- 我们的版本：`v0.1.68.1`、`v0.1.68.2`（递增） + +### 分支策略 + +| 分支 | 说明 | +|------|------| +| `main` | 我们的主分支，包含所有定制功能 | +| `release/custom-X.Y.Z` | 基于官方 `vX.Y.Z` 的发布分支 | +| `upstream/main` | 上游官方仓库 | + +--- + +## 发布流程（基于新官方版本） + +当官方发布新版本（如 `v0.1.69`）时： + +### 1. 同步上游并创建发布分支 + +```bash +# 获取上游最新代码 +git fetch upstream --tags + +# 基于官方标签创建新的发布分支 +git checkout v0.1.69 -b release/custom-0.1.69 + +# 合并我们的 main 分支（包含所有定制功能） +git merge main --no-edit + +# 解决可能的冲突后继续 +``` + +### 2. 更新版本号并打标签 + +```bash +# 更新版本号文件 +echo "0.1.69.1" > backend/cmd/server/VERSION +git add backend/cmd/server/VERSION +git commit -m "chore: bump version to 0.1.69.1" + +# 打上我们自己的标签 +git tag v0.1.69.1 + +# 推送分支和标签 +git push origin release/custom-0.1.69 +git push origin v0.1.69.1 +``` + +### 3. 更新 main 分支 + +```bash +# 将发布分支合并回 main，保持 main 包含最新定制功能 +git checkout main +git merge release/custom-0.1.69 +git push origin main +``` + +--- + +## 热修复发布（在现有版本上修复） + +当需要在当前版本上发布修复时： + +```bash +# 在当前发布分支上修复 +git checkout release/custom-0.1.68 +# ... 进行修复 ... +git commit -m "fix: 修复描述" + +# 递增小版本号 +echo "0.1.68.2" > backend/cmd/server/VERSION +git add backend/cmd/server/VERSION +git commit -m "chore: bump version to 0.1.68.2" + +# 打标签并推送 +git tag v0.1.68.2 +git push origin release/custom-0.1.68 +git push origin v0.1.68.2 + +# 同步修复到 main +git checkout main +git cherry-pick +git push origin main +``` + +--- + +## 服务器部署流程 + +### 前置条件 + +- 本地已配置 SSH 别名 `clicodeplus` 连接到生产服务器（运行服务） +- 本地已配置 SSH 别名 `us-asaki-root` 连接到构建服务器（拉取代码、构建镜像） +- 生产服务器部署目录：`/root/sub2api`（正式）、`/root/sub2api-beta`（测试）、`/root/sub2api-star`（Star） +- 生产服务器使用 Docker Compose 部署 +- **镜像统一在构建服务器上构建**，避免生产服务器因编译占用 CPU/内存影响线上服务 + +### 服务器角色说明 + +| 服务器 | SSH 别名 | 职责 | +|--------|----------|------| +| 构建服务器 | `us-asaki-root` | 拉取代码、`docker build` 构建镜像 | +| 生产服务器 | `clicodeplus` | 加载镜像、运行服务、部署验证 | +| 数据库服务器 | `db-clicodeplus` | PostgreSQL 16 + Redis 7，所有环境共用 | + +> 数据库服务器运维手册：`db-clicodeplus:/root/README.md` + +### 部署环境说明 + +| 环境 | 目录（生产服务器） | 端口 | 数据库 | Redis DB | 容器名 | +|------|------|------|--------|----------|--------| +| 正式 | `/root/sub2api` | 8080 | `sub2api` | 0 | `sub2api` | +| Beta | `/root/sub2api-beta` | 8084 | `beta` | 2 | `sub2api-beta` | +| OpenAI | `/root/sub2api-openai` | 8083 | `openai` | 3 | `sub2api-openai` | +| Star | `/root/sub2api-star` | 8086 | `star` | 4 | `sub2api-star` | + +### 外部数据库与 Redis + +所有环境（正式、Beta、OpenAI、Star）共用 `db.clicodeplus.com` 上的 **PostgreSQL 16** 和 **Redis 7**，不使用容器内数据库或 Redis。 + +**PostgreSQL**（端口 5432，TLS 加密，scram-sha-256 认证）： + +| 环境 | 用户名 | 数据库 | +|------|--------|--------| +| 正式 | `sub2api` | `sub2api` | +| Beta | `beta` | `beta` | +| OpenAI | `openai` | `openai` | +| Star | `star` | `star` | + +**Redis**（端口 6379，密码认证）： + +| 环境 | DB | +|------|-----| +| 正式 | 0 | +| Beta | 2 | +| OpenAI | 3 | +| Star | 4 | + +**配置方式**： +- 数据库通过 `.env` 中的 `DATABASE_HOST`、`DATABASE_SSLMODE`、`POSTGRES_USER`、`POSTGRES_PASSWORD`、`POSTGRES_DB` 配置 +- Redis 通过 `docker-compose.override.yml` 覆盖 `REDIS_HOST`（因主 compose 文件硬编码为 `redis`），密码通过 `.env` 中的 `REDIS_PASSWORD` 配置 +- 各环境的 `docker-compose.override.yml` 已通过 `depends_on: !reset {}` 和 `redis: profiles: [disabled]` 去掉了对容器 Redis 的依赖 + +#### 数据库操作命令 + +通过 SSH 在服务器上执行数据库操作： + +```bash +# 正式环境 - 查询迁移记录 +ssh clicodeplus "source /root/sub2api/deploy/.env && PGPASSWORD=\"\$POSTGRES_PASSWORD\" psql -h \$DATABASE_HOST -U \$POSTGRES_USER -d \$POSTGRES_DB -c 'SELECT * FROM schema_migrations ORDER BY applied_at DESC LIMIT 5;'" + +# Beta 环境 - 查询迁移记录 +ssh clicodeplus "source /root/sub2api-beta/deploy/.env && PGPASSWORD=\"\$POSTGRES_PASSWORD\" psql -h \$DATABASE_HOST -U \$POSTGRES_USER -d \$POSTGRES_DB -c 'SELECT * FROM schema_migrations ORDER BY applied_at DESC LIMIT 5;'" + +# Beta 环境 - 清除指定迁移记录（重新执行迁移） +ssh clicodeplus "source /root/sub2api-beta/deploy/.env && PGPASSWORD=\"\$POSTGRES_PASSWORD\" psql -h \$DATABASE_HOST -U \$POSTGRES_USER -d \$POSTGRES_DB -c \"DELETE FROM schema_migrations WHERE filename LIKE '%049%';\"" + +# Beta 环境 - 更新账号数据 +ssh clicodeplus "source /root/sub2api-beta/deploy/.env && PGPASSWORD=\"\$POSTGRES_PASSWORD\" psql -h \$DATABASE_HOST -U \$POSTGRES_USER -d \$POSTGRES_DB -c \"UPDATE accounts SET credentials = credentials - 'model_mapping' WHERE platform = 'antigravity';\"" +``` + +> **注意**：使用 `source .env` 加载环境变量，避免在命令行中暴露密码。 + +### 部署步骤 + +**重要：每次部署都必须递增版本号！** + +#### 0. 递增版本号并推送（本地操作） + +每次部署前，先在本地递增小版本号并确保推送成功： + +```bash +# 查看当前版本号 +cat backend/cmd/server/VERSION +# 假设当前是 0.1.69.1 + +# 递增版本号 +echo "0.1.69.2" > backend/cmd/server/VERSION +git add backend/cmd/server/VERSION +git commit -m "chore: bump version to 0.1.69.2" +git push origin release/custom-0.1.69 + +# ⚠️ 确认推送成功（必须看到分支更新输出，不能有 rejected 错误） +``` + +> **检查点**：如果有其他未提交的改动，应先 commit 并 push，确保 release 分支上的所有代码都已推送到远程。 + +#### 1. 构建服务器拉取代码 + +```bash +# 拉取最新代码并切换分支 +ssh us-asaki-root "cd /root/sub2api && git fetch origin && git checkout -B release/custom-0.1.69 origin/release/custom-0.1.69" + +# ⚠️ 验证版本号与步骤 0 一致 +ssh us-asaki-root "cat /root/sub2api/backend/cmd/server/VERSION" +``` + +> **首次使用构建服务器？** 需要先初始化仓库，参见下方「构建服务器首次初始化」章节。 + +#### 2. 构建服务器构建镜像 + +```bash +ssh us-asaki-root "cd /root/sub2api && docker build --no-cache -t sub2api:latest -f Dockerfile ." + +# ⚠️ 必须看到构建成功输出，如果失败需要先排查问题 +``` + +> **常见构建问题**： +> - `buildx` 版本过旧导致 API 版本不兼容 → 更新 buildx：`curl -fsSL "https://github.com/docker/buildx/releases/latest/download/buildx-$(curl -fsSL https://api.github.com/repos/docker/buildx/releases/latest | grep tag_name | cut -d'"' -f4).linux-amd64" -o ~/.docker/cli-plugins/docker-buildx && chmod +x ~/.docker/cli-plugins/docker-buildx` +> - 磁盘空间不足 → `docker system prune -f` 清理无用镜像 + +#### 3. 传输镜像到生产服务器并加载 + +```bash +# 导出镜像 → 通过管道传输 → 生产服务器加载 +ssh us-asaki-root "docker save sub2api:latest" | ssh clicodeplus "docker load" + +# ⚠️ 必须看到 "Loaded image: sub2api:latest" 输出 +``` + +#### 4. 生产服务器同步代码、更新标签并重启 + +```bash +# 同步代码（用于版本号确认和 deploy 配置） +ssh clicodeplus "cd /root/sub2api && git fetch fork && git checkout -B release/custom-0.1.69 fork/release/custom-0.1.69" + +# 更新镜像标签并重启 +ssh clicodeplus "docker tag sub2api:latest weishaw/sub2api:latest" +ssh clicodeplus "cd /root/sub2api/deploy && docker compose up -d --force-recreate sub2api" +``` + +#### 5. 验证部署 + +```bash +# 查看启动日志 +ssh clicodeplus "docker logs sub2api --tail 20" + +# 确认版本号（必须与步骤 0 中设置的版本号一致） +ssh clicodeplus "cat /root/sub2api/backend/cmd/server/VERSION" + +# 检查容器状态（必须显示 healthy） +ssh clicodeplus "docker ps | grep sub2api" +``` + +--- + +### 构建服务器首次初始化 + +首次使用 `us-asaki-root` 作为构建服务器时，需要执行以下一次性操作： + +```bash +ssh us-asaki-root + +# 1) 克隆仓库 +cd /root +git clone https://github.com/touwaeriol/sub2api.git sub2api +cd sub2api + +# 2) 验证 Docker 和 buildx 版本 +docker version +docker buildx version +# 如果 buildx 版本过旧（< v0.14），执行更新： +# LATEST=$(curl -fsSL https://api.github.com/repos/docker/buildx/releases/latest | grep tag_name | cut -d'"' -f4) +# curl -fsSL "https://github.com/docker/buildx/releases/download/${LATEST}/buildx-${LATEST}.linux-amd64" -o ~/.docker/cli-plugins/docker-buildx +# chmod +x ~/.docker/cli-plugins/docker-buildx + +# 3) 验证构建能力 +docker build --no-cache -t sub2api:test -f Dockerfile . +docker rmi sub2api:test +``` + +--- + +## Beta 并行部署（不影响现网） + +目标：在同一台服务器上并行启动一个 beta 实例（例如端口 `8084`），**严禁改动/重启**现网实例（默认目录 `/root/sub2api`）。 + +### 设计原则 + +- **新目录**：beta 使用独立目录，例如 `/root/sub2api-beta`。 +- **敏感信息只放 `.env`**：beta 的数据库密码、JWT_SECRET 等只写入 `/root/sub2api-beta/deploy/.env`，不要提交到 git。 +- **独立 Compose Project**：通过 `docker compose -p sub2api-beta ...` 启动，确保 network/volume 隔离。 +- **独立端口**：通过 `.env` 的 `SERVER_PORT` 映射宿主机端口（例如 `8084:8080`）。 + +### 前置检查 + +```bash +# 1) 确保 8084 未被占用 +ssh clicodeplus "ss -ltnp | grep :8084 || echo '8084 is free'" + +# 2) 确认现网容器还在（只读检查） +ssh clicodeplus "docker ps --format 'table {{.Names}}\t{{.Image}}\t{{.Ports}}' | sed -n '1,200p'" +``` + +### 首次部署步骤 + +> **构建服务器说明**：正式和 beta 共用构建服务器上的 `/root/sub2api` 仓库，通过不同的镜像标签区分（`sub2api:latest` 用于正式，`sub2api:beta` 用于测试）。 + +```bash +# 1) 构建服务器构建 beta 镜像（共用 /root/sub2api 仓库，切到目标分支后打 beta 标签） +ssh us-asaki-root "cd /root/sub2api && git fetch origin && git checkout -B release/custom-0.1.71 origin/release/custom-0.1.71" +ssh us-asaki-root "cd /root/sub2api && docker build --no-cache -t sub2api:beta -f Dockerfile ." + +# ⚠️ 构建完成后如需恢复正式分支： +# ssh us-asaki-root "cd /root/sub2api && git checkout release/custom-<正式版本>" + +# 2) 传输镜像到生产服务器 +ssh us-asaki-root "docker save sub2api:beta" | ssh clicodeplus "docker load" +# ⚠️ 必须看到 "Loaded image: sub2api:beta" 输出 + +# 3) 在生产服务器上准备 beta 环境 +ssh clicodeplus + +# 克隆代码（仅用于 deploy 配置和版本号确认，不在此构建） +cd /root +git clone https://github.com/touwaeriol/sub2api.git sub2api-beta +cd /root/sub2api-beta +git checkout release/custom-0.1.71 + +# 4) 准备 beta 的 .env（敏感信息只写这里） +cd /root/sub2api-beta/deploy + +# 推荐：从现网 .env 复制，保证除 DB 名/用户/端口外完全一致 +cp -f /root/sub2api/deploy/.env ./.env + +# 仅修改以下三项（其他保持不变） +perl -pi -e 's/^SERVER_PORT=.*/SERVER_PORT=8084/' ./.env +perl -pi -e 's/^POSTGRES_USER=.*/POSTGRES_USER=beta/' ./.env +perl -pi -e 's/^POSTGRES_DB=.*/POSTGRES_DB=beta/' ./.env + +# 5) 写 compose override（避免与现网容器名冲突，镜像使用构建服务器传输的 sub2api:beta，Redis 使用外部服务） +cat > docker-compose.override.yml <<'YAML' +services: + sub2api: + image: sub2api:beta + container_name: sub2api-beta + environment: + - DATABASE_HOST=${DATABASE_HOST:-postgres} + - DATABASE_SSLMODE=${DATABASE_SSLMODE:-disable} + - REDIS_HOST=db.clicodeplus.com + depends_on: !reset {} + redis: + profiles: + - disabled +YAML + +# 6) 启动 beta（独立 project，确保不影响现网） +cd /root/sub2api-beta/deploy +docker compose -p sub2api-beta --env-file .env -f docker-compose.yml -f docker-compose.override.yml up -d + +# 7) 验证 beta +curl -fsS http://127.0.0.1:8084/health +docker logs sub2api-beta --tail 50 +``` + +### 数据库配置约定（beta） + +- 数据库地址/SSL/密码：与现网一致（从现网 `.env` 复制即可），均指向 `db.clicodeplus.com`。 +- 仅修改： + - `POSTGRES_USER=beta` + - `POSTGRES_DB=beta` + - `REDIS_DB=2` + +注意：需要数据库侧已存在 `beta` 用户与 `beta` 数据库，并授予权限；否则容器会启动失败并不断重启。 + +### 更新 beta（构建服务器构建 + 传输 + 仅重启 beta 容器） + +```bash +# 1) 构建服务器拉取代码并构建镜像（共用 /root/sub2api 仓库） +ssh us-asaki-root "cd /root/sub2api && git fetch origin && git checkout -B release/custom-0.1.71 origin/release/custom-0.1.71" +ssh us-asaki-root "cd /root/sub2api && docker build --no-cache -t sub2api:beta -f Dockerfile ." +# ⚠️ 必须看到构建成功输出 + +# 2) 传输镜像到生产服务器 +ssh us-asaki-root "docker save sub2api:beta" | ssh clicodeplus "docker load" +# ⚠️ 必须看到 "Loaded image: sub2api:beta" 输出 + +# 3) 生产服务器同步代码（用于版本号确认和 deploy 配置） +ssh clicodeplus "set -e; cd /root/sub2api-beta && git fetch --all --tags && git checkout -f release/custom-0.1.71 && git reset --hard origin/release/custom-0.1.71" + +# 4) 重启 beta 容器并验证 +ssh clicodeplus "cd /root/sub2api-beta/deploy && docker compose -p sub2api-beta --env-file .env -f docker-compose.yml -f docker-compose.override.yml up -d --no-deps --force-recreate sub2api" +ssh clicodeplus "sleep 5 && curl -fsS http://127.0.0.1:8084/health" +ssh clicodeplus "cat /root/sub2api-beta/backend/cmd/server/VERSION" +``` + +### 停止/回滚 beta（只影响 beta） + +```bash +ssh clicodeplus "cd /root/sub2api-beta/deploy && docker compose -p sub2api-beta -f docker-compose.yml -f docker-compose.override.yml down" +``` + +--- + +## 服务器首次部署 + +### 1. 构建服务器：克隆代码并配置远程仓库 + +```bash +ssh us-asaki-root +cd /root +git clone https://github.com/Wei-Shaw/sub2api.git +cd sub2api + +# 添加 fork 仓库 +git remote add fork https://github.com/touwaeriol/sub2api.git +``` + +### 2. 构建服务器：切换到定制分支并构建镜像 + +```bash +git fetch fork +git checkout -B release/custom-0.1.69 fork/release/custom-0.1.69 + +cd /root/sub2api +docker build -t sub2api:latest -f Dockerfile . +exit +``` + +### 3. 传输镜像到生产服务器 + +```bash +ssh us-asaki-root "docker save sub2api:latest" | ssh clicodeplus "docker load" +``` + +### 4. 生产服务器：克隆代码并配置环境 + +```bash +ssh clicodeplus +cd /root +git clone https://github.com/Wei-Shaw/sub2api.git +cd sub2api + +# 添加 fork 仓库 +git remote add fork https://github.com/touwaeriol/sub2api.git +git fetch fork +git checkout -B release/custom-0.1.69 fork/release/custom-0.1.69 + +# 配置环境变量 +cd deploy +cp .env.example .env +vim .env # 配置 DATABASE_HOST=db.clicodeplus.com, POSTGRES_PASSWORD, REDIS_PASSWORD, JWT_SECRET 等 + +# 创建 override 文件（Redis 指向外部服务，去掉容器 Redis 依赖） +cat > docker-compose.override.yml <<'YAML' +services: + sub2api: + environment: + - REDIS_HOST=db.clicodeplus.com + depends_on: !reset {} + redis: + profiles: + - disabled +YAML +``` + +### 5. 生产服务器：更新镜像标签并启动服务 + +```bash +docker tag sub2api:latest weishaw/sub2api:latest +cd /root/sub2api/deploy && docker compose up -d +``` + +### 6. 验证部署 + +```bash +# 查看应用日志 +docker logs sub2api --tail 50 + +# 检查健康状态 +curl http://localhost:8080/health + +# 确认版本号 +cat /root/sub2api/backend/cmd/server/VERSION +``` + +### 7. 常用运维命令 + +```bash +# 查看实时日志 +docker logs -f sub2api + +# 重启服务 +docker compose restart sub2api + +# 停止所有服务 +docker compose down + +# 停止并删除数据卷（慎用！会删除数据库数据） +docker compose down -v + +# 查看资源使用情况 +docker stats sub2api +``` + +--- + +## 定制功能说明 + +当前定制分支包含以下功能（相对于官方版本）： + +### UI/UX 定制 + +| 功能 | 说明 | +|------|------| +| 首页优化 | 面向用户的价值主张设计 | +| 移除 GitHub 链接 | 用户菜单中不显示 GitHub 导航 | +| 微信客服按钮 | 首页悬浮微信客服入口 | +| 限流时间精确显示 | 账号限流时间显示精确到秒 | + +### Antigravity 平台增强 + +| 功能 | 说明 | +|------|------| +| Scope 级别限流 | 按配额域（claude/gemini_text/gemini_image）独立限流，避免整个账号被锁定 | +| 模型级别限流 | 按具体模型（如 claude-opus-4-5）独立限流，更精细的限流控制 | +| 限流预检查 | 调度时预检查账号/模型限流状态，避免选中已限流账号 | +| 秒级冷却时间 | 支持 429 响应的秒级精确冷却时间 | +| 身份注入优化 | 模型身份信息注入 + 静默边界防止身份泄露 | +| thoughtSignature 修复 | Gemini 3 函数调用 400 错误修复 | +| max_tokens 自动修正 | 自动修正 max_tokens <= budget_tokens 导致的 400 错误 | + +### 调度算法优化 + +| 功能 | 说明 | +|------|------| +| 分层过滤选择 | 调度算法从全排序改为分层过滤，提升性能 | +| LRU 随机选择 | 相同 LRU 时间时随机选择，避免账号集中 | +| 限流等待阈值配置化 | 可配置的限流等待阈值 | + +### 运维增强 + +| 功能 | 说明 | +|------|------| +| Scope 限流统计 | 运维界面展示 Antigravity 账号 scope 级别限流统计 | +| 账号限流状态显示 | 账号列表显示 scope 和模型级别限流状态 | +| 清除限流按钮增强 | 有 scope/模型限流时也显示清除限流按钮 | + +### 其他修复 + +| 功能 | 说明 | +|------|------| +| .gitattributes | 确保迁移文件使用 LF 换行符（解决 Windows 下 SQL 摘要不一致） | +| 部署配置优化 | DATABASE_HOST 和 DATABASE_SSLMODE 可通过 .env 配置 | + +--- + +## Admin API 接口文档 + +### ⚠️ API 操作流程规范 + +当收到操作正式环境 Web 界面的新需求，但文档中未记录对应 API 接口时，**必须按以下流程执行**： + +1. **探索接口**：通过代码库搜索路由定义（`backend/internal/server/routes/`）、Handler（`backend/internal/handler/admin/`）和请求结构体，确定正确的 API 端点、请求方法、请求体格式 +2. **更新文档**：将新发现的接口补充到本文档的 Admin API 接口文档章节中，包含端点、参数说明和 curl 示例 +3. **执行操作**：根据最新文档中记录的接口完成用户需求 + +> **目的**：避免每次遇到相同需求都重复探索代码库，确保 API 文档持续完善，后续操作可直接查阅文档执行。 + +--- + +### 认证方式 + +所有 Admin API 通过 `x-api-key` 请求头传递 Admin API Key 认证。 + +``` +x-api-key: admin-xxx +``` + +> **使用说明**：Admin API Key 统一存放在项目根目录 `.env` 文件的 `ADMIN_API_KEY` 变量中（该文件已被 `.gitignore` 排除，不会提交到代码库）。操作前先从 `.env` 读取密钥；若密钥失效（返回 401），应提示用户提供新的密钥并更新到 `.env` 中。Token 格式为 `admin-` + 64 位十六进制字符，在管理后台 `设置 > Admin API Key` 中生成。**请勿将实际 token 写入文档或代码中。** + +### 环境地址 + +| 环境 | 基础地址 | 说明 | +|------|----------|------| +| 正式 | `https://clicodeplus.com` | 生产环境 | +| Beta | `http://<服务器IP>:8084` | 仅内网访问 | +| OpenAI | `http://<服务器IP>:8083` | 仅内网访问 | +| Star | `https://hyntoken.com` | 独立环境 | + +> 以下接口文档中，`${BASE}` 代表环境基础地址，`${KEY}` 代表 `.env` 中的 `ADMIN_API_KEY`。操作前执行 `source .env` 或 `export KEY=$ADMIN_API_KEY` 加载。 + +--- + +### 1. 账号管理 + +#### 1.1 获取账号列表 + +``` +GET /api/v1/admin/accounts +``` + +**查询参数**： + +| 参数 | 类型 | 必填 | 说明 | +|------|------|------|------| +| `platform` | string | 否 | 平台筛选：`antigravity` / `anthropic` / `openai` / `gemini` | +| `type` | string | 否 | 账号类型：`oauth` / `api_key` / `cookie` | +| `status` | string | 否 | 状态：`active` / `disabled` / `error` | +| `search` | string | 否 | 搜索关键词（名称、备注） | +| `page` | int | 否 | 页码，默认 1 | +| `page_size` | int | 否 | 每页数量，默认 20 | + +```bash +curl -s "${BASE}/api/v1/admin/accounts?platform=antigravity&page=1&page_size=100" \ + -H "x-api-key: ${KEY}" +``` + +**响应**： +```json +{ + "code": 0, + "message": "success", + "data": { + "items": [{"id": 1, "name": "xxx@gmail.com", "platform": "antigravity", "status": "active", ...}], + "total": 66 + } +} +``` + +#### 1.2 获取账号详情 + +``` +GET /api/v1/admin/accounts/:id +``` + +```bash +curl -s "${BASE}/api/v1/admin/accounts/1" -H "x-api-key: ${KEY}" +``` + +#### 1.3 测试账号连接 + +``` +POST /api/v1/admin/accounts/:id/test +``` + +**请求体**（JSON，可选）： + +| 字段 | 类型 | 必填 | 说明 | +|------|------|------|------| +| `model_id` | string | 否 | 指定测试模型，如 `claude-opus-4-6`；不传则使用默认模型 | + +**响应格式**：SSE（Server-Sent Events）流 + +```bash +curl -N -X POST "${BASE}/api/v1/admin/accounts/1/test" \ + -H "x-api-key: ${KEY}" \ + -H "Content-Type: application/json" \ + -d '{"model_id": "claude-opus-4-6"}' +``` + +**SSE 事件类型**： + +| type | 字段 | 说明 | +|------|------|------| +| `test_start` | `model` | 测试开始，返回测试模型名 | +| `content` | `text` | 模型响应内容（流式文本片段） | +| `test_end` | `success`, `error` | 测试结束，`success=true` 表示成功 | +| `error` | `text` | 错误信息 | + +#### 1.4 清除账号限流 + +``` +POST /api/v1/admin/accounts/:id/clear-rate-limit +``` + +```bash +curl -X POST "${BASE}/api/v1/admin/accounts/1/clear-rate-limit" \ + -H "x-api-key: ${KEY}" +``` + +#### 1.5 清除账号错误状态 + +``` +POST /api/v1/admin/accounts/:id/clear-error +``` + +```bash +curl -X POST "${BASE}/api/v1/admin/accounts/1/clear-error" \ + -H "x-api-key: ${KEY}" +``` + +#### 1.6 获取账号可用模型 + +``` +GET /api/v1/admin/accounts/:id/models +``` + +```bash +curl -s "${BASE}/api/v1/admin/accounts/1/models" -H "x-api-key: ${KEY}" +``` + +#### 1.7 刷新 OAuth Token + +``` +POST /api/v1/admin/accounts/:id/refresh +``` + +```bash +curl -X POST "${BASE}/api/v1/admin/accounts/1/refresh" -H "x-api-key: ${KEY}" +``` + +#### 1.8 刷新账号等级 + +``` +POST /api/v1/admin/accounts/:id/refresh-tier +``` + +```bash +curl -X POST "${BASE}/api/v1/admin/accounts/1/refresh-tier" -H "x-api-key: ${KEY}" +``` + +#### 1.9 获取账号统计 + +``` +GET /api/v1/admin/accounts/:id/stats +``` + +```bash +curl -s "${BASE}/api/v1/admin/accounts/1/stats" -H "x-api-key: ${KEY}" +``` + +#### 1.10 获取账号用量 + +``` +GET /api/v1/admin/accounts/:id/usage +``` + +```bash +curl -s "${BASE}/api/v1/admin/accounts/1/usage" -H "x-api-key: ${KEY}" +``` + +#### 1.11 更新单个账号 + +``` +PUT /api/v1/admin/accounts/:id +``` + +**请求体**（JSON，所有字段均为可选，仅传需要更新的字段）： + +| 字段 | 类型 | 说明 | +|------|------|------| +| `name` | string | 账号名称 | +| `notes` | *string | 备注 | +| `type` | string | 类型：`oauth` / `setup-token` / `apikey` / `upstream` | +| `credentials` | object | 凭证信息 | +| `extra` | object | 额外配置 | +| `proxy_id` | *int64 | 代理 ID | +| `concurrency` | *int | 并发数 | +| `priority` | *int | 优先级（默认 50） | +| `rate_multiplier` | *float64 | 速率倍数 | +| `status` | string | 状态：`active` / `inactive` | +| `group_ids` | *[]int64 | 分组 ID 列表 | +| `expires_at` | *int64 | 过期时间戳 | +| `auto_pause_on_expired` | *bool | 过期后自动暂停 | + +> 使用指针类型（`*`）的字段可以区分"未提供"和"设置为零值"。 + +```bash +# 示例：更新账号优先级为 100 +curl -X PUT "${BASE}/api/v1/admin/accounts/1" \ + -H "x-api-key: ${KEY}" \ + -H "Content-Type: application/json" \ + -d '{"priority": 100}' +``` + +#### 1.12 批量更新账号 + +``` +POST /api/v1/admin/accounts/bulk-update +``` + +**请求体**（JSON）： + +| 字段 | 类型 | 必填 | 说明 | +|------|------|------|------| +| `account_ids` | []int64 | **是** | 要更新的账号 ID 列表 | +| `priority` | *int | 否 | 优先级 | +| `concurrency` | *int | 否 | 并发数 | +| `rate_multiplier` | *float64 | 否 | 速率倍数 | +| `status` | string | 否 | 状态：`active` / `inactive` / `error` | +| `schedulable` | *bool | 否 | 是否可调度 | +| `group_ids` | *[]int64 | 否 | 分组 ID 列表 | +| `proxy_id` | *int64 | 否 | 代理 ID | +| `credentials` | object | 否 | 凭证信息（批量覆盖） | +| `extra` | object | 否 | 额外配置（批量覆盖） | + +```bash +# 示例：批量设置多个账号优先级为 100 +curl -X POST "${BASE}/api/v1/admin/accounts/bulk-update" \ + -H "x-api-key: ${KEY}" \ + -H "Content-Type: application/json" \ + -d '{"account_ids": [1, 2, 3], "priority": 100}' +``` + +#### 1.13 批量测试账号（脚本） + +批量测试指定平台所有账号的指定模型连通性： + +```bash +# 用户需提供：BASE（环境地址）、KEY（admin token）、MODEL（测试模型） +ACCOUNT_IDS=$(curl -s "${BASE}/api/v1/admin/accounts?platform=antigravity&page=1&page_size=100" \ + -H "x-api-key: ${KEY}" | python3 -c " +import json, sys +data = json.load(sys.stdin) +for item in data['data']['items']: + print(f\"{item['id']}|{item['name']}\") +") + +while IFS='|' read -r ID NAME; do + echo "测试账号 ID=${ID} (${NAME})..." + RESPONSE=$(curl -s --max-time 60 -N \ + -X POST "${BASE}/api/v1/admin/accounts/${ID}/test" \ + -H "x-api-key: ${KEY}" \ + -H "Content-Type: application/json" \ + -d "{\"model_id\": \"${MODEL}\"}" 2>&1) + if echo "$RESPONSE" | grep -q '"success":true'; then + echo " ✅ 成功" + elif echo "$RESPONSE" | grep -q '"type":"content"'; then + echo " ✅ 成功（有内容响应）" + else + ERROR_MSG=$(echo "$RESPONSE" | grep -o '"error":"[^"]*"' | tail -1) + echo " ❌ 失败: ${ERROR_MSG}" + fi +done <<< "$ACCOUNT_IDS" +``` + +--- + +### 2. 运维监控 + +#### 2.1 并发统计 + +``` +GET /api/v1/admin/ops/concurrency +``` + +```bash +curl -s "${BASE}/api/v1/admin/ops/concurrency" -H "x-api-key: ${KEY}" +``` + +#### 2.2 账号可用性 + +``` +GET /api/v1/admin/ops/account-availability +``` + +```bash +curl -s "${BASE}/api/v1/admin/ops/account-availability" -H "x-api-key: ${KEY}" +``` + +#### 2.3 实时流量摘要 + +``` +GET /api/v1/admin/ops/realtime-traffic +``` + +```bash +curl -s "${BASE}/api/v1/admin/ops/realtime-traffic" -H "x-api-key: ${KEY}" +``` + +#### 2.4 请求错误列表 + +``` +GET /api/v1/admin/ops/request-errors +``` + +**查询参数**：`page`、`page_size` + +```bash +curl -s "${BASE}/api/v1/admin/ops/request-errors?page=1&page_size=50" \ + -H "x-api-key: ${KEY}" +``` + +#### 2.5 上游错误列表 + +``` +GET /api/v1/admin/ops/upstream-errors +``` + +```bash +curl -s "${BASE}/api/v1/admin/ops/upstream-errors?page=1&page_size=50" \ + -H "x-api-key: ${KEY}" +``` + +#### 2.6 仪表板概览 + +``` +GET /api/v1/admin/ops/dashboard/overview +``` + +```bash +curl -s "${BASE}/api/v1/admin/ops/dashboard/overview" -H "x-api-key: ${KEY}" +``` + +--- + +### 3. 系统设置 + +#### 3.1 获取系统设置 + +``` +GET /api/v1/admin/settings +``` + +```bash +curl -s "${BASE}/api/v1/admin/settings" -H "x-api-key: ${KEY}" +``` + +#### 3.2 更新系统设置 + +``` +PUT /api/v1/admin/settings +``` + +```bash +curl -X PUT "${BASE}/api/v1/admin/settings" \ + -H "x-api-key: ${KEY}" \ + -H "Content-Type: application/json" \ + -d '{ ... }' +``` + +#### 3.3 Admin API Key 状态（脱敏） + +``` +GET /api/v1/admin/settings/admin-api-key +``` + +```bash +curl -s "${BASE}/api/v1/admin/settings/admin-api-key" -H "x-api-key: ${KEY}" +``` + +--- + +### 4. 用户管理 + +#### 4.1 用户列表 + +``` +GET /api/v1/admin/users +``` + +```bash +curl -s "${BASE}/api/v1/admin/users?page=1&page_size=20" -H "x-api-key: ${KEY}" +``` + +#### 4.2 用户详情 + +``` +GET /api/v1/admin/users/:id +``` + +```bash +curl -s "${BASE}/api/v1/admin/users/1" -H "x-api-key: ${KEY}" +``` + +#### 4.3 更新用户余额 + +``` +POST /api/v1/admin/users/:id/balance +``` + +```bash +curl -X POST "${BASE}/api/v1/admin/users/1/balance" \ + -H "x-api-key: ${KEY}" \ + -H "Content-Type: application/json" \ + -d '{"amount": 100, "reason": "充值"}' +``` + +--- + +### 5. 分组管理 + +#### 5.1 分组列表 + +``` +GET /api/v1/admin/groups +``` + +```bash +curl -s "${BASE}/api/v1/admin/groups" -H "x-api-key: ${KEY}" +``` + +#### 5.2 所有分组（不分页） + +``` +GET /api/v1/admin/groups/all +``` + +```bash +curl -s "${BASE}/api/v1/admin/groups/all" -H "x-api-key: ${KEY}" +``` + +--- + +## 注意事项 + +1. **前端必须打包进镜像**：使用 `docker build` 在构建服务器（`us-asaki-root`）上构建，Dockerfile 会自动编译前端并 embed 到后端二进制中，构建完成后通过 `docker save | docker load` 传输到生产服务器（`clicodeplus`） + +2. **镜像标签**：docker-compose.yml 使用 `weishaw/sub2api:latest`，本地构建后需要 `docker tag` 覆盖 + +3. **Windows 换行符问题**：已通过 `.gitattributes` 解决，确保 `*.sql` 文件始终使用 LF + +4. **版本号管理**：每次发布必须更新 `backend/cmd/server/VERSION` 并打标签 + +5. **合并冲突**：合并上游新版本时，重点关注以下文件可能的冲突： + - `backend/internal/service/antigravity_gateway_service.go` + - `backend/internal/service/gateway_service.go` + - `backend/internal/pkg/antigravity/request_transformer.go` + +--- + +## Go 代码规范 + +### 1. 函数设计 + +#### 单一职责原则 +- **函数行数**：单个函数常规不应超过 **30 行**，超过时应拆分为子函数。若某段逻辑确实不可拆分（如复杂的状态机、协议解析等），可以例外，但需添加注释说明原因 +- **嵌套层级**：避免超过 3 层嵌套，使用 early return 减少嵌套 + +```go +// ❌ 不推荐：深层嵌套 +func process(data []Item) { + for _, item := range data { + if item.Valid { + if item.Type == "A" { + if item.Status == "active" { + // 业务逻辑... + } + } + } + } +} + +// ✅ 推荐：early return +func process(data []Item) { + for _, item := range data { + if !item.Valid { + continue + } + if item.Type != "A" { + continue + } + if item.Status != "active" { + continue + } + // 业务逻辑... + } +} +``` + +#### 复杂逻辑提取 +将复杂的条件判断或处理逻辑提取为独立函数： + +```go +// ❌ 不推荐：内联复杂逻辑 +if resp.StatusCode == 429 || resp.StatusCode == 503 { + // 80+ 行处理逻辑... +} + +// ✅ 推荐：提取为独立函数 +result := handleRateLimitResponse(resp, params) +switch result.action { +case actionRetry: + continue +case actionBreak: + return result.resp, nil +} +``` + +### 2. 重复代码消除 + +#### 配置获取模式 +将重复的配置获取逻辑提取为方法： + +```go +// ❌ 不推荐：重复代码 +logBody := s.settingService != nil && s.settingService.cfg != nil && s.settingService.cfg.Gateway.LogUpstreamErrorBody +maxBytes := 2048 +if s.settingService != nil && s.settingService.cfg != nil && s.settingService.cfg.Gateway.LogUpstreamErrorBodyMaxBytes > 0 { + maxBytes = s.settingService.cfg.Gateway.LogUpstreamErrorBodyMaxBytes +} + +// ✅ 推荐：提取为方法 +func (s *Service) getLogConfig() (logBody bool, maxBytes int) { + maxBytes = 2048 + if s.settingService == nil || s.settingService.cfg == nil { + return false, maxBytes + } + cfg := s.settingService.cfg.Gateway + if cfg.LogUpstreamErrorBodyMaxBytes > 0 { + maxBytes = cfg.LogUpstreamErrorBodyMaxBytes + } + return cfg.LogUpstreamErrorBody, maxBytes +} +``` + +### 3. 常量管理 + +#### 避免魔法数字 +所有硬编码的数值都应定义为常量： + +```go +// ❌ 不推荐 +if retryDelay >= 10*time.Second { + resetAt := time.Now().Add(30 * time.Second) +} + +// ✅ 推荐 +const ( + rateLimitThreshold = 10 * time.Second + defaultRateLimitDuration = 30 * time.Second +) + +if retryDelay >= rateLimitThreshold { + resetAt := time.Now().Add(defaultRateLimitDuration) +} +``` + +#### 注释引用常量名 +在注释中引用常量名而非硬编码值： + +```go +// ❌ 不推荐 +// < 10s: 等待后重试 + +// ✅ 推荐 +// < rateLimitThreshold: 等待后重试 +``` + +### 4. 错误处理 + +#### 使用结构化日志 +优先使用 `slog` 进行结构化日志记录： + +```go +// ❌ 不推荐 +log.Printf("%s status=%d model_rate_limit_failed model=%s error=%v", prefix, statusCode, modelName, err) + +// ✅ 推荐 +slog.Error("failed to set model rate limit", + "prefix", prefix, + "status_code", statusCode, + "model", modelName, + "error", err, +) +``` + +### 5. 测试规范 + +#### Mock 函数签名同步 +修改函数签名时，必须同步更新所有测试中的 mock 函数： + +```go +// 如果修改了 handleError 签名 +handleError func(..., groupID int64, sessionHash string) *Result + +// 必须同步更新测试中的 mock +handleError: func(..., groupID int64, sessionHash string) *Result { + return nil +}, +``` + +#### 测试构建标签 +统一使用测试构建标签： + +```go +//go:build unit + +package service +``` + +### 6. 时间格式解析 + +#### 使用标准库 +优先使用 `time.ParseDuration`，支持所有 Go duration 格式： + +```go +// ❌ 不推荐：手动限制格式 +if !strings.HasSuffix(delay, "s") || strings.Contains(delay, "m") { + continue +} + +// ✅ 推荐：使用标准库 +dur, err := time.ParseDuration(delay) // 支持 "0.5s", "4m50s", "1h30m" 等 +``` + +### 7. 接口设计 + +#### 接口隔离原则 +定义最小化接口，只包含必需的方法： + +```go +// ❌ 不推荐：使用过于宽泛的接口 +type AccountRepository interface { + // 20+ 个方法... +} + +// ✅ 推荐：定义最小化接口 +type ModelRateLimiter interface { + SetModelRateLimit(ctx context.Context, id int64, modelKey string, resetAt time.Time) error +} +``` + +### 8. 并发安全 + +#### 共享数据保护 +访问可能被并发修改的数据时，确保线程安全： + +```go +// 如果 Account.Extra 可能被并发修改 +// 需要使用互斥锁或原子操作保护读取 +func (a *Account) GetRateLimitRemainingTime(model string) time.Duration { + a.mu.RLock() + defer a.mu.RUnlock() + // 读取 Extra 字段... +} +``` + +### 9. 命名规范 + +#### 一致的命名风格 +- 常量使用 camelCase：`rateLimitThreshold` +- 类型使用 PascalCase：`AntigravityQuotaScope` +- 同一概念使用统一命名：`Threshold` 或 `Limit`，不要混用 + +```go +// ❌ 不推荐：命名不一致 +antigravitySmartRetryMinWait // 使用 Min +antigravityRateLimitThreshold // 使用 Threshold + +// ✅ 推荐：统一风格 +antigravityMinRetryWait +antigravityRateLimitThreshold +``` + +### 10. 代码审查清单 + +在提交代码前，检查以下项目： + +- [ ] 函数是否超过 30 行？（不可拆分的逻辑除外，需注释说明） +- [ ] 嵌套是否超过 3 层？ +- [ ] 是否有重复代码可以提取？ +- [ ] 是否使用了魔法数字？ +- [ ] Mock 函数签名是否与实际函数一致？ +- [ ] 测试是否覆盖了新增逻辑？ +- [ ] 日志是否包含足够的上下文信息？ +- [ ] 是否考虑了并发安全？ + +--- + +## CI 检查与发布门禁 + +### GitHub Actions 检查项 + +本项目有 4 个 CI 任务，**任何代码推送或发布前都必须全部通过**： + +| Workflow | Job | 说明 | 本地验证命令 | +|----------|-----|------|-------------| +| CI | `test` | 单元测试 + 集成测试 | `cd backend && make test-unit && make test-integration` | +| CI | `golangci-lint` | Go 代码静态检查（golangci-lint v2.7） | `cd backend && golangci-lint run --timeout=5m` | +| Security Scan | `backend-security` | govulncheck + gosec 安全扫描 | `cd backend && govulncheck ./... && gosec -severity high -confidence high ./...` | +| Security Scan | `frontend-security` | pnpm audit 前端依赖安全检查 | `cd frontend && pnpm audit --prod --audit-level=high` | + +### 向上游提交 PR + +PR 目标是上游官方仓库，**只包含通用功能改动**（bug fix、新功能、性能优化等）。 + +**以下文件禁止出现在 PR 中**（属于我们 fork 的定制化内容）： +- `CLAUDE.md`、`AGENTS.md` — 我们的开发文档 +- `backend/cmd/server/VERSION` — 我们的版本号文件 +- UI 定制改动（GitHub 链接移除、微信客服按钮、首页定制等） +- 部署配置（`deploy/` 目录下的定制修改） + +**PR 流程**： +1. 从 `develop` 创建功能分支，只包含要提交给上游的改动 +2. 推送分支后，**等待 4 个 CI job 全部通过** +3. 确认通过后再创建 PR +4. 使用 `gh run list --repo touwaeriol/sub2api --branch ` 检查状态 + +### 自有分支推送（develop / main） + +推送到我们自己的 `develop` 或 `main` 分支时，包含所有改动（定制化 + 通用功能）。 + +**推送前必须在本地执行全部 CI 检查**（不要等 GitHub Actions）： + +```bash +# 确保 Go 工具链可用（macOS homebrew） +export PATH="/opt/homebrew/bin:$HOME/go/bin:$PATH" + +# 1. 单元测试（必须） +cd backend && make test-unit + +# 2. 集成测试（推荐，需要 Docker） +make test-integration + +# 3. golangci-lint 静态检查（必须） +golangci-lint run --timeout=5m + +# 4. gofmt 格式检查（必须） +gofmt -l ./... +# 如果有输出，运行 gofmt -w 修复 +``` + +**推送后确认**： +1. 使用 `gh run list --repo touwaeriol/sub2api --branch ` 检查 GitHub Actions 状态 +2. 确认 CI 和 Security Scan 两个 workflow 的 4 个 job 全部绿色 ✅ +3. 任何 job 失败必须立即修复，**禁止在 CI 未通过的状态下继续后续操作** + +### 发布版本 + +1. 本地执行上述全部 CI 检查通过 +2. 递增 `backend/cmd/server/VERSION`，提交并推送 +3. 推送后确认 GitHub Actions 的 4 个 CI job 全部通过 +4. **CI 未通过时禁止部署** — 必须先修复问题 +5. 使用 `gh run list --repo touwaeriol/sub2api --limit 10` 确认状态 + +### 常见 CI 失败原因及修复 +- **gofmt**：struct 字段对齐不一致 → 运行 `gofmt -w ` 修复 +- **golangci-lint**：未使用的变量/导入 → 删除或使用 `_` 忽略 +- **test 失败**：mock 函数签名不一致 → 同步更新 mock +- **gosec**：安全漏洞 → 根据提示修复或添加例外 + +--- + +## PR 描述格式规范 + +所有 PR 描述使用中英文同步（先中文、后英文），包含以下三个部分： + +### 模板 + +```markdown +## 背景 / Background + +<一两句说明问题现状或触发原因> + + + +--- + +## 目的 / Purpose + +<本次改动要解决的问题或达到的目标> + + + +--- + +## 改动内容 / Changes + +### 后端 / Backend + +- **改动点 1**：说明 +- **改动点 2**：说明 + +--- + +- **Change 1**: description +- **Change 2**: description + +### 前端 / Frontend + +- **改动点 1**：说明 +- **改动点 2**：说明 + +--- + +- **Change 1**: description +- **Change 2**: description + +--- + +## 截图 / Screenshot（可选） + +ASCII 示意图或实际截图 +``` + +### 规范要点 + +- **标题**：使用 conventional commits 格式，如 `feat(scope): description` +- **中英文顺序**：同一段落先中文后英文，用空行分隔，不用 `---` 分割同段内容 +- **改动分类**：按 Backend / Frontend / Config 等模块分组，先列中文要点再列英文要点 +- **截图/示意图**：有 UI 变动时必须附上，可用 ASCII 示意布局 +- **目标分支**：提交到 `touwaeriol/sub2api` 的 `main` 分支 diff --git a/CLAUDE.md b/CLAUDE.md new file mode 100644 index 00000000..b634af05 --- /dev/null +++ b/CLAUDE.md @@ -0,0 +1,1337 @@ +# Sub2API 开发说明 + +## 版本管理策略 + +### 版本号规则 + +我们在官方版本号后面添加自己的小版本号： + +- 官方版本：`v0.1.68` +- 我们的版本：`v0.1.68.1`、`v0.1.68.2`（递增） + +### 分支策略 + +| 分支 | 说明 | +|------|------| +| `main` | 我们的主分支，包含所有定制功能 | +| `release/custom-X.Y.Z` | 基于官方 `vX.Y.Z` 的发布分支 | +| `upstream/main` | 上游官方仓库 | + +--- + +## 发布流程（基于新官方版本） + +当官方发布新版本（如 `v0.1.69`）时： + +### 1. 同步上游并创建发布分支 + +```bash +# 获取上游最新代码 +git fetch upstream --tags + +# 基于官方标签创建新的发布分支 +git checkout v0.1.69 -b release/custom-0.1.69 + +# 合并我们的 main 分支（包含所有定制功能） +git merge main --no-edit + +# 解决可能的冲突后继续 +``` + +### 2. 更新版本号并打标签 + +```bash +# 更新版本号文件 +echo "0.1.69.1" > backend/cmd/server/VERSION +git add backend/cmd/server/VERSION +git commit -m "chore: bump version to 0.1.69.1" + +# 打上我们自己的标签 +git tag v0.1.69.1 + +# 推送分支和标签 +git push origin release/custom-0.1.69 +git push origin v0.1.69.1 +``` + +### 3. 更新 main 分支 + +```bash +# 将发布分支合并回 main，保持 main 包含最新定制功能 +git checkout main +git merge release/custom-0.1.69 +git push origin main +``` + +--- + +## 热修复发布（在现有版本上修复） + +当需要在当前版本上发布修复时： + +```bash +# 在当前发布分支上修复 +git checkout release/custom-0.1.68 +# ... 进行修复 ... +git commit -m "fix: 修复描述" + +# 递增小版本号 +echo "0.1.68.2" > backend/cmd/server/VERSION +git add backend/cmd/server/VERSION +git commit -m "chore: bump version to 0.1.68.2" + +# 打标签并推送 +git tag v0.1.68.2 +git push origin release/custom-0.1.68 +git push origin v0.1.68.2 + +# 同步修复到 main +git checkout main +git cherry-pick +git push origin main +``` + +--- + +## 服务器部署流程 + +### 前置条件 + +- 本地已配置 SSH 别名 `clicodeplus` 连接到生产服务器（运行服务 + 构建镜像） +- 生产服务器部署目录：`/root/sub2api`（正式）、`/root/sub2api-beta`（测试）、`/root/sub2api-star`（Star） +- 生产服务器使用 Docker Compose 部署 +- **镜像在生产服务器本机构建**，使用资源限制的 `limited-builder` 构建器（3 核 CPU、4G 内存），避免构建占满服务器资源影响线上服务 + +### 服务器角色说明 + +| 服务器 | SSH 别名 | 职责 | +|--------|----------|------| +| 生产服务器 | `clicodeplus` | 拉取代码、构建镜像、运行服务、部署验证 | +| 数据库服务器 | `db-clicodeplus` | PostgreSQL 16 + Redis 7，所有环境共用 | + +> 数据库服务器运维手册：`db-clicodeplus:/root/README.md` + +### 构建器说明 + +生产服务器上配置了资源限制的 Docker buildx 构建器 `limited-builder`，**所有构建操作必须使用此构建器**： + +- **构建器名称**：`limited-builder` +- **驱动**：`docker-container`（独立容器运行 BuildKit） +- **资源限制**：3 核 CPU、4G 内存（服务器共 6 核 8G，预留一半给线上服务） +- **容器名**：`buildx_buildkit_limited-builder0` + +```bash +# 构建命令格式（必须指定 --builder） +ssh clicodeplus "cd /root/sub2api && docker buildx build --builder limited-builder --no-cache --load -t sub2api:latest -f Dockerfile ." + +# 查看构建器状态 +ssh clicodeplus "docker buildx inspect limited-builder" + +# 如果构建器容器被意外删除，重新创建： +ssh clicodeplus "docker buildx create --name limited-builder --driver docker-container --driver-opt 'default-load=true' && docker buildx inspect --builder limited-builder --bootstrap && docker update --cpus=3 --memory=4g --memory-swap=4g buildx_buildkit_limited-builder0" +``` + +### 部署环境说明 + +| 环境 | 目录（生产服务器） | 端口 | 数据库 | Redis DB | 容器名 | +|------|------|------|--------|----------|--------| +| 正式 | `/root/sub2api` | 8080 | `sub2api` | 0 | `sub2api` | +| Beta | `/root/sub2api-beta` | 8084 | `beta` | 2 | `sub2api-beta` | +| OpenAI | `/root/sub2api-openai` | 8083 | `openai` | 3 | `sub2api-openai` | +| Star | `/root/sub2api-star` | 8086 | `star` | 4 | `sub2api-star` | + +### 外部数据库与 Redis + +所有环境（正式、Beta、OpenAI、Star）共用 `db.clicodeplus.com` 上的 **PostgreSQL 16** 和 **Redis 7**，不使用容器内数据库或 Redis。 + +**PostgreSQL**（端口 5432，TLS 加密，scram-sha-256 认证）： + +| 环境 | 用户名 | 数据库 | +|------|--------|--------| +| 正式 | `sub2api` | `sub2api` | +| Beta | `beta` | `beta` | +| OpenAI | `openai` | `openai` | +| Star | `star` | `star` | + +**Redis**（端口 6379，密码认证）： + +| 环境 | DB | +|------|-----| +| 正式 | 0 | +| Beta | 2 | +| OpenAI | 3 | +| Star | 4 | + +**配置方式**： +- 数据库通过 `.env` 中的 `DATABASE_HOST`、`DATABASE_SSLMODE`、`POSTGRES_USER`、`POSTGRES_PASSWORD`、`POSTGRES_DB` 配置 +- Redis 通过 `docker-compose.override.yml` 覆盖 `REDIS_HOST`（因主 compose 文件硬编码为 `redis`），密码通过 `.env` 中的 `REDIS_PASSWORD` 配置 +- 各环境的 `docker-compose.override.yml` 已通过 `depends_on: !reset {}` 和 `redis: profiles: [disabled]` 去掉了对容器 Redis 的依赖 + +#### 数据库操作命令 + +通过 SSH 在服务器上执行数据库操作： + +```bash +# 正式环境 - 查询迁移记录 +ssh clicodeplus "source /root/sub2api/deploy/.env && PGPASSWORD=\"\$POSTGRES_PASSWORD\" psql -h \$DATABASE_HOST -U \$POSTGRES_USER -d \$POSTGRES_DB -c 'SELECT * FROM schema_migrations ORDER BY applied_at DESC LIMIT 5;'" + +# Beta 环境 - 查询迁移记录 +ssh clicodeplus "source /root/sub2api-beta/deploy/.env && PGPASSWORD=\"\$POSTGRES_PASSWORD\" psql -h \$DATABASE_HOST -U \$POSTGRES_USER -d \$POSTGRES_DB -c 'SELECT * FROM schema_migrations ORDER BY applied_at DESC LIMIT 5;'" + +# Beta 环境 - 清除指定迁移记录（重新执行迁移） +ssh clicodeplus "source /root/sub2api-beta/deploy/.env && PGPASSWORD=\"\$POSTGRES_PASSWORD\" psql -h \$DATABASE_HOST -U \$POSTGRES_USER -d \$POSTGRES_DB -c \"DELETE FROM schema_migrations WHERE filename LIKE '%049%';\"" + +# Beta 环境 - 更新账号数据 +ssh clicodeplus "source /root/sub2api-beta/deploy/.env && PGPASSWORD=\"\$POSTGRES_PASSWORD\" psql -h \$DATABASE_HOST -U \$POSTGRES_USER -d \$POSTGRES_DB -c \"UPDATE accounts SET credentials = credentials - 'model_mapping' WHERE platform = 'antigravity';\"" +``` + +> **注意**：使用 `source .env` 加载环境变量，避免在命令行中暴露密码。 + +### 部署步骤 + +**重要：每次部署都必须递增版本号！** + +#### 0. 递增版本号并推送（本地操作） + +每次部署前，先在本地递增小版本号并确保推送成功： + +```bash +# 查看当前版本号 +cat backend/cmd/server/VERSION +# 假设当前是 0.1.69.1 + +# 递增版本号 +echo "0.1.69.2" > backend/cmd/server/VERSION +git add backend/cmd/server/VERSION +git commit -m "chore: bump version to 0.1.69.2" +git push origin release/custom-0.1.69 + +# ⚠️ 确认推送成功（必须看到分支更新输出，不能有 rejected 错误） +``` + +> **检查点**：如果有其他未提交的改动，应先 commit 并 push，确保 release 分支上的所有代码都已推送到远程。 + +#### 1. 生产服务器拉取代码 + +```bash +# 拉取最新代码并切换分支 +ssh clicodeplus "cd /root/sub2api && git fetch fork && git checkout -B release/custom-0.1.69 fork/release/custom-0.1.69" + +# ⚠️ 验证版本号与步骤 0 一致 +ssh clicodeplus "cat /root/sub2api/backend/cmd/server/VERSION" +``` + +#### 2. 生产服务器构建镜像（使用 limited-builder） + +```bash +ssh clicodeplus "cd /root/sub2api && docker buildx build --builder limited-builder --no-cache --load -t sub2api:latest -f Dockerfile ." + +# ⚠️ 必须看到构建成功输出，如果失败需要先排查问题 +``` + +> **常见构建问题**： +> - 构建器未启动 → `docker buildx inspect --builder limited-builder --bootstrap` +> - 磁盘空间不足 → `docker system prune -f` 清理无用镜像 +> - 构建器被删除 → 参见上方「构建器说明」重新创建 + +#### 3. 更新镜像标签并重启 + +```bash +# 更新镜像标签并重启 +ssh clicodeplus "docker tag sub2api:latest weishaw/sub2api:latest" +ssh clicodeplus "cd /root/sub2api/deploy && docker compose up -d --force-recreate sub2api" +``` + +#### 4. 验证部署 + +```bash +# 查看启动日志 +ssh clicodeplus "docker logs sub2api --tail 20" + +# 确认版本号（必须与步骤 0 中设置的版本号一致） +ssh clicodeplus "cat /root/sub2api/backend/cmd/server/VERSION" + +# 检查容器状态（必须显示 healthy） +ssh clicodeplus "docker ps | grep sub2api" +``` + +--- + +## Beta 并行部署（不影响现网） + +目标：在同一台服务器上并行启动一个 beta 实例（例如端口 `8084`），**严禁改动/重启**现网实例（默认目录 `/root/sub2api`）。 + +### 设计原则 + +- **新目录**：beta 使用独立目录，例如 `/root/sub2api-beta`。 +- **敏感信息只放 `.env`**：beta 的数据库密码、JWT_SECRET 等只写入 `/root/sub2api-beta/deploy/.env`，不要提交到 git。 +- **独立 Compose Project**：通过 `docker compose -p sub2api-beta ...` 启动，确保 network/volume 隔离。 +- **独立端口**：通过 `.env` 的 `SERVER_PORT` 映射宿主机端口（例如 `8084:8080`）。 + +### 前置检查 + +```bash +# 1) 确保 8084 未被占用 +ssh clicodeplus "ss -ltnp | grep :8084 || echo '8084 is free'" + +# 2) 确认现网容器还在（只读检查） +ssh clicodeplus "docker ps --format 'table {{.Names}}\t{{.Image}}\t{{.Ports}}' | sed -n '1,200p'" +``` + +### 首次部署步骤 + +> **构建说明**：正式和 beta 通过不同的镜像标签区分（`sub2api:latest` 用于正式，`sub2api:beta` 用于测试），均在生产服务器本机使用 `limited-builder` 构建。 + +```bash +# 1) 在生产服务器上拉取代码并构建 beta 镜像 +ssh clicodeplus "cd /root/sub2api-beta && git fetch --all --tags && git checkout -f release/custom-0.1.71 && git reset --hard origin/release/custom-0.1.71" +ssh clicodeplus "cd /root/sub2api-beta && docker buildx build --builder limited-builder --no-cache --load -t sub2api:beta -f Dockerfile ." + +# 2) 在生产服务器上准备 beta 环境 +ssh clicodeplus + +# 克隆代码（仅用于 deploy 配置和版本号确认，不在此构建） +cd /root +git clone https://github.com/touwaeriol/sub2api.git sub2api-beta +cd /root/sub2api-beta +git checkout release/custom-0.1.71 + +# 4) 准备 beta 的 .env（敏感信息只写这里） +cd /root/sub2api-beta/deploy + +# 推荐：从现网 .env 复制，保证除 DB 名/用户/端口外完全一致 +cp -f /root/sub2api/deploy/.env ./.env + +# 仅修改以下三项（其他保持不变） +perl -pi -e 's/^SERVER_PORT=.*/SERVER_PORT=8084/' ./.env +perl -pi -e 's/^POSTGRES_USER=.*/POSTGRES_USER=beta/' ./.env +perl -pi -e 's/^POSTGRES_DB=.*/POSTGRES_DB=beta/' ./.env + +# 5) 写 compose override（避免与现网容器名冲突，镜像使用本机构建的 sub2api:beta，Redis 使用外部服务） +cat > docker-compose.override.yml <<'YAML' +services: + sub2api: + image: sub2api:beta + container_name: sub2api-beta + environment: + - DATABASE_HOST=${DATABASE_HOST:-postgres} + - DATABASE_SSLMODE=${DATABASE_SSLMODE:-disable} + - REDIS_HOST=db.clicodeplus.com + depends_on: !reset {} + redis: + profiles: + - disabled +YAML + +# 6) 启动 beta（独立 project，确保不影响现网） +cd /root/sub2api-beta/deploy +docker compose -p sub2api-beta --env-file .env -f docker-compose.yml -f docker-compose.override.yml up -d + +# 7) 验证 beta +curl -fsS http://127.0.0.1:8084/health +docker logs sub2api-beta --tail 50 +``` + +### 数据库配置约定（beta） + +- 数据库地址/SSL/密码：与现网一致（从现网 `.env` 复制即可），均指向 `db.clicodeplus.com`。 +- 仅修改： + - `POSTGRES_USER=beta` + - `POSTGRES_DB=beta` + - `REDIS_DB=2` + +注意：需要数据库侧已存在 `beta` 用户与 `beta` 数据库，并授予权限；否则容器会启动失败并不断重启。 + +### 更新 beta（本机构建 + 仅重启 beta 容器） + +```bash +# 1) 生产服务器拉取代码并构建镜像 +ssh clicodeplus "cd /root/sub2api-beta && git fetch --all --tags && git checkout -f release/custom-0.1.71 && git reset --hard origin/release/custom-0.1.71" +ssh clicodeplus "cd /root/sub2api-beta && docker buildx build --builder limited-builder --no-cache --load -t sub2api:beta -f Dockerfile ." +# ⚠️ 必须看到构建成功输出 + +# 2) 重启 beta 容器并验证 +ssh clicodeplus "cd /root/sub2api-beta/deploy && docker compose -p sub2api-beta --env-file .env -f docker-compose.yml -f docker-compose.override.yml up -d --no-deps --force-recreate sub2api" +ssh clicodeplus "sleep 5 && curl -fsS http://127.0.0.1:8084/health" +ssh clicodeplus "cat /root/sub2api-beta/backend/cmd/server/VERSION" +``` + +### 停止/回滚 beta（只影响 beta） + +```bash +ssh clicodeplus "cd /root/sub2api-beta/deploy && docker compose -p sub2api-beta -f docker-compose.yml -f docker-compose.override.yml down" +``` + +--- + +## 服务器首次部署 + +### 1. 生产服务器：克隆代码并配置环境 + +```bash +ssh clicodeplus +cd /root +git clone https://github.com/Wei-Shaw/sub2api.git +cd sub2api + +# 添加 fork 仓库 +git remote add fork https://github.com/touwaeriol/sub2api.git +git fetch fork +git checkout -B release/custom-0.1.69 fork/release/custom-0.1.69 + +# 配置环境变量 +cd deploy +cp .env.example .env +vim .env # 配置 DATABASE_HOST=db.clicodeplus.com, POSTGRES_PASSWORD, REDIS_PASSWORD, JWT_SECRET 等 + +# 创建 override 文件（Redis 指向外部服务，去掉容器 Redis 依赖） +cat > docker-compose.override.yml <<'YAML' +services: + sub2api: + environment: + - REDIS_HOST=db.clicodeplus.com + depends_on: !reset {} + redis: + profiles: + - disabled +YAML +``` + +### 2. 生产服务器：创建构建器并构建镜像 + +```bash +# 创建资源限制的构建器（首次执行一次即可） +docker buildx create --name limited-builder --driver docker-container --driver-opt "default-load=true" +docker buildx inspect --builder limited-builder --bootstrap +docker update --cpus=3 --memory=4g --memory-swap=4g buildx_buildkit_limited-builder0 + +# 构建镜像 +cd /root/sub2api +docker buildx build --builder limited-builder --no-cache --load -t sub2api:latest -f Dockerfile . + +# 更新镜像标签并启动 +docker tag sub2api:latest weishaw/sub2api:latest +cd /root/sub2api/deploy && docker compose up -d +``` + +### 3. 验证部署 + +```bash +# 查看应用日志 +docker logs sub2api --tail 50 + +# 检查健康状态 +curl http://localhost:8080/health + +# 确认版本号 +cat /root/sub2api/backend/cmd/server/VERSION +``` + +### 4. 常用运维命令 + +```bash +# 查看实时日志 +docker logs -f sub2api + +# 重启服务 +docker compose restart sub2api + +# 停止所有服务 +docker compose down + +# 停止并删除数据卷（慎用！会删除数据库数据） +docker compose down -v + +# 查看资源使用情况 +docker stats sub2api +``` + +--- + +## 定制功能说明 + +当前定制分支包含以下功能（相对于官方版本）： + +### UI/UX 定制 + +| 功能 | 说明 | +|------|------| +| 首页优化 | 面向用户的价值主张设计 | +| 移除 GitHub 链接 | 用户菜单中不显示 GitHub 导航 | +| 微信客服按钮 | 首页悬浮微信客服入口 | +| 限流时间精确显示 | 账号限流时间显示精确到秒 | + +### Antigravity 平台增强 + +| 功能 | 说明 | +|------|------| +| Scope 级别限流 | 按配额域（claude/gemini_text/gemini_image）独立限流，避免整个账号被锁定 | +| 模型级别限流 | 按具体模型（如 claude-opus-4-5）独立限流，更精细的限流控制 | +| 限流预检查 | 调度时预检查账号/模型限流状态，避免选中已限流账号 | +| 秒级冷却时间 | 支持 429 响应的秒级精确冷却时间 | +| 身份注入优化 | 模型身份信息注入 + 静默边界防止身份泄露 | +| thoughtSignature 修复 | Gemini 3 函数调用 400 错误修复 | +| max_tokens 自动修正 | 自动修正 max_tokens <= budget_tokens 导致的 400 错误 | + +### 调度算法优化 + +| 功能 | 说明 | +|------|------| +| 分层过滤选择 | 调度算法从全排序改为分层过滤，提升性能 | +| LRU 随机选择 | 相同 LRU 时间时随机选择，避免账号集中 | +| 限流等待阈值配置化 | 可配置的限流等待阈值 | + +### 运维增强 + +| 功能 | 说明 | +|------|------| +| Scope 限流统计 | 运维界面展示 Antigravity 账号 scope 级别限流统计 | +| 账号限流状态显示 | 账号列表显示 scope 和模型级别限流状态 | +| 清除限流按钮增强 | 有 scope/模型限流时也显示清除限流按钮 | + +### 其他修复 + +| 功能 | 说明 | +|------|------| +| .gitattributes | 确保迁移文件使用 LF 换行符（解决 Windows 下 SQL 摘要不一致） | +| 部署配置优化 | DATABASE_HOST 和 DATABASE_SSLMODE 可通过 .env 配置 | + +--- + +## Admin API 接口文档 + +### ⚠️ API 操作流程规范 + +当收到操作正式环境 Web 界面的新需求，但文档中未记录对应 API 接口时，**必须按以下流程执行**： + +1. **探索接口**：通过代码库搜索路由定义（`backend/internal/server/routes/`）、Handler（`backend/internal/handler/admin/`）和请求结构体，确定正确的 API 端点、请求方法、请求体格式 +2. **更新文档**：将新发现的接口补充到本文档的 Admin API 接口文档章节中，包含端点、参数说明和 curl 示例 +3. **执行操作**：根据最新文档中记录的接口完成用户需求 + +> **目的**：避免每次遇到相同需求都重复探索代码库，确保 API 文档持续完善，后续操作可直接查阅文档执行。 + +--- + +### 认证方式 + +所有 Admin API 通过 `x-api-key` 请求头传递 Admin API Key 认证。 + +``` +x-api-key: admin-xxx +``` + +> **使用说明**：Admin API Key 统一存放在项目根目录 `.env` 文件的 `ADMIN_API_KEY` 变量中（该文件已被 `.gitignore` 排除，不会提交到代码库）。操作前先从 `.env` 读取密钥；若密钥失效（返回 401），应提示用户提供新的密钥并更新到 `.env` 中。Token 格式为 `admin-` + 64 位十六进制字符，在管理后台 `设置 > Admin API Key` 中生成。**请勿将实际 token 写入文档或代码中。** + +### 环境地址 + +| 环境 | 基础地址 | 说明 | +|------|----------|------| +| 正式 | `https://clicodeplus.com` | 生产环境 | +| Beta | `http://<服务器IP>:8084` | 仅内网访问 | +| OpenAI | `http://<服务器IP>:8083` | 仅内网访问 | +| Star | `https://hyntoken.com` | 独立环境 | + +> 以下接口文档中，`${BASE}` 代表环境基础地址，`${KEY}` 代表 `.env` 中的 `ADMIN_API_KEY`。操作前执行 `source .env` 或 `export KEY=$ADMIN_API_KEY` 加载。 + +--- + +### 1. 账号管理 + +#### 1.1 获取账号列表 + +``` +GET /api/v1/admin/accounts +``` + +**查询参数**： + +| 参数 | 类型 | 必填 | 说明 | +|------|------|------|------| +| `platform` | string | 否 | 平台筛选：`antigravity` / `anthropic` / `openai` / `gemini` | +| `type` | string | 否 | 账号类型：`oauth` / `api_key` / `cookie` | +| `status` | string | 否 | 状态：`active` / `disabled` / `error` | +| `search` | string | 否 | 搜索关键词（名称、备注） | +| `page` | int | 否 | 页码，默认 1 | +| `page_size` | int | 否 | 每页数量，默认 20 | + +```bash +curl -s "${BASE}/api/v1/admin/accounts?platform=antigravity&page=1&page_size=100" \ + -H "x-api-key: ${KEY}" +``` + +**响应**： +```json +{ + "code": 0, + "message": "success", + "data": { + "items": [{"id": 1, "name": "xxx@gmail.com", "platform": "antigravity", "status": "active", ...}], + "total": 66 + } +} +``` + +#### 1.2 获取账号详情 + +``` +GET /api/v1/admin/accounts/:id +``` + +```bash +curl -s "${BASE}/api/v1/admin/accounts/1" -H "x-api-key: ${KEY}" +``` + +#### 1.3 测试账号连接 + +``` +POST /api/v1/admin/accounts/:id/test +``` + +**请求体**（JSON，可选）： + +| 字段 | 类型 | 必填 | 说明 | +|------|------|------|------| +| `model_id` | string | 否 | 指定测试模型，如 `claude-opus-4-6`；不传则使用默认模型 | + +**响应格式**：SSE（Server-Sent Events）流 + +```bash +curl -N -X POST "${BASE}/api/v1/admin/accounts/1/test" \ + -H "x-api-key: ${KEY}" \ + -H "Content-Type: application/json" \ + -d '{"model_id": "claude-opus-4-6"}' +``` + +**SSE 事件类型**： + +| type | 字段 | 说明 | +|------|------|------| +| `test_start` | `model` | 测试开始，返回测试模型名 | +| `content` | `text` | 模型响应内容（流式文本片段） | +| `test_end` | `success`, `error` | 测试结束，`success=true` 表示成功 | +| `error` | `text` | 错误信息 | + +#### 1.4 清除账号限流 + +``` +POST /api/v1/admin/accounts/:id/clear-rate-limit +``` + +```bash +curl -X POST "${BASE}/api/v1/admin/accounts/1/clear-rate-limit" \ + -H "x-api-key: ${KEY}" +``` + +#### 1.5 清除账号错误状态 + +``` +POST /api/v1/admin/accounts/:id/clear-error +``` + +```bash +curl -X POST "${BASE}/api/v1/admin/accounts/1/clear-error" \ + -H "x-api-key: ${KEY}" +``` + +#### 1.6 获取账号可用模型 + +``` +GET /api/v1/admin/accounts/:id/models +``` + +```bash +curl -s "${BASE}/api/v1/admin/accounts/1/models" -H "x-api-key: ${KEY}" +``` + +#### 1.7 刷新 OAuth Token + +``` +POST /api/v1/admin/accounts/:id/refresh +``` + +```bash +curl -X POST "${BASE}/api/v1/admin/accounts/1/refresh" -H "x-api-key: ${KEY}" +``` + +#### 1.8 刷新账号等级 + +``` +POST /api/v1/admin/accounts/:id/refresh-tier +``` + +```bash +curl -X POST "${BASE}/api/v1/admin/accounts/1/refresh-tier" -H "x-api-key: ${KEY}" +``` + +#### 1.9 获取账号统计 + +``` +GET /api/v1/admin/accounts/:id/stats +``` + +```bash +curl -s "${BASE}/api/v1/admin/accounts/1/stats" -H "x-api-key: ${KEY}" +``` + +#### 1.10 获取账号用量 + +``` +GET /api/v1/admin/accounts/:id/usage +``` + +```bash +curl -s "${BASE}/api/v1/admin/accounts/1/usage" -H "x-api-key: ${KEY}" +``` + +#### 1.11 更新单个账号 + +``` +PUT /api/v1/admin/accounts/:id +``` + +**请求体**（JSON，所有字段均为可选，仅传需要更新的字段）： + +| 字段 | 类型 | 说明 | +|------|------|------| +| `name` | string | 账号名称 | +| `notes` | *string | 备注 | +| `type` | string | 类型：`oauth` / `setup-token` / `apikey` / `upstream` | +| `credentials` | object | 凭证信息 | +| `extra` | object | 额外配置 | +| `proxy_id` | *int64 | 代理 ID | +| `concurrency` | *int | 并发数 | +| `priority` | *int | 优先级（默认 50） | +| `rate_multiplier` | *float64 | 速率倍数 | +| `status` | string | 状态：`active` / `inactive` | +| `group_ids` | *[]int64 | 分组 ID 列表 | +| `expires_at` | *int64 | 过期时间戳 | +| `auto_pause_on_expired` | *bool | 过期后自动暂停 | + +> 使用指针类型（`*`）的字段可以区分"未提供"和"设置为零值"。 + +```bash +# 示例：更新账号优先级为 100 +curl -X PUT "${BASE}/api/v1/admin/accounts/1" \ + -H "x-api-key: ${KEY}" \ + -H "Content-Type: application/json" \ + -d '{"priority": 100}' +``` + +#### 1.12 批量更新账号 + +``` +POST /api/v1/admin/accounts/bulk-update +``` + +**请求体**（JSON）： + +| 字段 | 类型 | 必填 | 说明 | +|------|------|------|------| +| `account_ids` | []int64 | **是** | 要更新的账号 ID 列表 | +| `priority` | *int | 否 | 优先级 | +| `concurrency` | *int | 否 | 并发数 | +| `rate_multiplier` | *float64 | 否 | 速率倍数 | +| `status` | string | 否 | 状态：`active` / `inactive` / `error` | +| `schedulable` | *bool | 否 | 是否可调度 | +| `group_ids` | *[]int64 | 否 | 分组 ID 列表 | +| `proxy_id` | *int64 | 否 | 代理 ID | +| `credentials` | object | 否 | 凭证信息（批量覆盖） | +| `extra` | object | 否 | 额外配置（批量覆盖） | + +```bash +# 示例：批量设置多个账号优先级为 100 +curl -X POST "${BASE}/api/v1/admin/accounts/bulk-update" \ + -H "x-api-key: ${KEY}" \ + -H "Content-Type: application/json" \ + -d '{"account_ids": [1, 2, 3], "priority": 100}' +``` + +#### 1.13 批量测试账号（脚本） + +批量测试指定平台所有账号的指定模型连通性： + +```bash +# 用户需提供：BASE（环境地址）、KEY（admin token）、MODEL（测试模型） +ACCOUNT_IDS=$(curl -s "${BASE}/api/v1/admin/accounts?platform=antigravity&page=1&page_size=100" \ + -H "x-api-key: ${KEY}" | python3 -c " +import json, sys +data = json.load(sys.stdin) +for item in data['data']['items']: + print(f\"{item['id']}|{item['name']}\") +") + +while IFS='|' read -r ID NAME; do + echo "测试账号 ID=${ID} (${NAME})..." + RESPONSE=$(curl -s --max-time 60 -N \ + -X POST "${BASE}/api/v1/admin/accounts/${ID}/test" \ + -H "x-api-key: ${KEY}" \ + -H "Content-Type: application/json" \ + -d "{\"model_id\": \"${MODEL}\"}" 2>&1) + if echo "$RESPONSE" | grep -q '"success":true'; then + echo " ✅ 成功" + elif echo "$RESPONSE" | grep -q '"type":"content"'; then + echo " ✅ 成功（有内容响应）" + else + ERROR_MSG=$(echo "$RESPONSE" | grep -o '"error":"[^"]*"' | tail -1) + echo " ❌ 失败: ${ERROR_MSG}" + fi +done <<< "$ACCOUNT_IDS" +``` + +--- + +### 2. 运维监控 + +#### 2.1 并发统计 + +``` +GET /api/v1/admin/ops/concurrency +``` + +```bash +curl -s "${BASE}/api/v1/admin/ops/concurrency" -H "x-api-key: ${KEY}" +``` + +#### 2.2 账号可用性 + +``` +GET /api/v1/admin/ops/account-availability +``` + +```bash +curl -s "${BASE}/api/v1/admin/ops/account-availability" -H "x-api-key: ${KEY}" +``` + +#### 2.3 实时流量摘要 + +``` +GET /api/v1/admin/ops/realtime-traffic +``` + +```bash +curl -s "${BASE}/api/v1/admin/ops/realtime-traffic" -H "x-api-key: ${KEY}" +``` + +#### 2.4 请求错误列表 + +``` +GET /api/v1/admin/ops/request-errors +``` + +**查询参数**：`page`、`page_size` + +```bash +curl -s "${BASE}/api/v1/admin/ops/request-errors?page=1&page_size=50" \ + -H "x-api-key: ${KEY}" +``` + +#### 2.5 上游错误列表 + +``` +GET /api/v1/admin/ops/upstream-errors +``` + +```bash +curl -s "${BASE}/api/v1/admin/ops/upstream-errors?page=1&page_size=50" \ + -H "x-api-key: ${KEY}" +``` + +#### 2.6 仪表板概览 + +``` +GET /api/v1/admin/ops/dashboard/overview +``` + +```bash +curl -s "${BASE}/api/v1/admin/ops/dashboard/overview" -H "x-api-key: ${KEY}" +``` + +--- + +### 3. 系统设置 + +#### 3.1 获取系统设置 + +``` +GET /api/v1/admin/settings +``` + +```bash +curl -s "${BASE}/api/v1/admin/settings" -H "x-api-key: ${KEY}" +``` + +#### 3.2 更新系统设置 + +``` +PUT /api/v1/admin/settings +``` + +```bash +curl -X PUT "${BASE}/api/v1/admin/settings" \ + -H "x-api-key: ${KEY}" \ + -H "Content-Type: application/json" \ + -d '{ ... }' +``` + +#### 3.3 Admin API Key 状态（脱敏） + +``` +GET /api/v1/admin/settings/admin-api-key +``` + +```bash +curl -s "${BASE}/api/v1/admin/settings/admin-api-key" -H "x-api-key: ${KEY}" +``` + +--- + +### 4. 用户管理 + +#### 4.1 用户列表 + +``` +GET /api/v1/admin/users +``` + +```bash +curl -s "${BASE}/api/v1/admin/users?page=1&page_size=20" -H "x-api-key: ${KEY}" +``` + +#### 4.2 用户详情 + +``` +GET /api/v1/admin/users/:id +``` + +```bash +curl -s "${BASE}/api/v1/admin/users/1" -H "x-api-key: ${KEY}" +``` + +#### 4.3 更新用户余额 + +``` +POST /api/v1/admin/users/:id/balance +``` + +```bash +curl -X POST "${BASE}/api/v1/admin/users/1/balance" \ + -H "x-api-key: ${KEY}" \ + -H "Content-Type: application/json" \ + -d '{"amount": 100, "reason": "充值"}' +``` + +--- + +### 5. 分组管理 + +#### 5.1 分组列表 + +``` +GET /api/v1/admin/groups +``` + +```bash +curl -s "${BASE}/api/v1/admin/groups" -H "x-api-key: ${KEY}" +``` + +#### 5.2 所有分组（不分页） + +``` +GET /api/v1/admin/groups/all +``` + +```bash +curl -s "${BASE}/api/v1/admin/groups/all" -H "x-api-key: ${KEY}" +``` + +--- + +## 注意事项 + +1. **前端必须打包进镜像**：使用 `docker buildx build --builder limited-builder` 在生产服务器（`clicodeplus`）本机构建，Dockerfile 会自动编译前端并 embed 到后端二进制中 + +2. **镜像标签**：docker-compose.yml 使用 `weishaw/sub2api:latest`，本地构建后需要 `docker tag` 覆盖 + +3. **Windows 换行符问题**：已通过 `.gitattributes` 解决，确保 `*.sql` 文件始终使用 LF + +4. **版本号管理**：每次发布必须更新 `backend/cmd/server/VERSION` 并打标签 + +5. **合并冲突**：合并上游新版本时，重点关注以下文件可能的冲突： + - `backend/internal/service/antigravity_gateway_service.go` + - `backend/internal/service/gateway_service.go` + - `backend/internal/pkg/antigravity/request_transformer.go` + +--- + +## Go 代码规范 + +### 1. 函数设计 + +#### 单一职责原则 +- **函数行数**：单个函数常规不应超过 **30 行**，超过时应拆分为子函数。若某段逻辑确实不可拆分（如复杂的状态机、协议解析等），可以例外，但需添加注释说明原因 +- **嵌套层级**：避免超过 3 层嵌套，使用 early return 减少嵌套 + +```go +// ❌ 不推荐：深层嵌套 +func process(data []Item) { + for _, item := range data { + if item.Valid { + if item.Type == "A" { + if item.Status == "active" { + // 业务逻辑... + } + } + } + } +} + +// ✅ 推荐：early return +func process(data []Item) { + for _, item := range data { + if !item.Valid { + continue + } + if item.Type != "A" { + continue + } + if item.Status != "active" { + continue + } + // 业务逻辑... + } +} +``` + +#### 复杂逻辑提取 +将复杂的条件判断或处理逻辑提取为独立函数： + +```go +// ❌ 不推荐：内联复杂逻辑 +if resp.StatusCode == 429 || resp.StatusCode == 503 { + // 80+ 行处理逻辑... +} + +// ✅ 推荐：提取为独立函数 +result := handleRateLimitResponse(resp, params) +switch result.action { +case actionRetry: + continue +case actionBreak: + return result.resp, nil +} +``` + +### 2. 重复代码消除 + +#### 配置获取模式 +将重复的配置获取逻辑提取为方法： + +```go +// ❌ 不推荐：重复代码 +logBody := s.settingService != nil && s.settingService.cfg != nil && s.settingService.cfg.Gateway.LogUpstreamErrorBody +maxBytes := 2048 +if s.settingService != nil && s.settingService.cfg != nil && s.settingService.cfg.Gateway.LogUpstreamErrorBodyMaxBytes > 0 { + maxBytes = s.settingService.cfg.Gateway.LogUpstreamErrorBodyMaxBytes +} + +// ✅ 推荐：提取为方法 +func (s *Service) getLogConfig() (logBody bool, maxBytes int) { + maxBytes = 2048 + if s.settingService == nil || s.settingService.cfg == nil { + return false, maxBytes + } + cfg := s.settingService.cfg.Gateway + if cfg.LogUpstreamErrorBodyMaxBytes > 0 { + maxBytes = cfg.LogUpstreamErrorBodyMaxBytes + } + return cfg.LogUpstreamErrorBody, maxBytes +} +``` + +### 3. 常量管理 + +#### 避免魔法数字 +所有硬编码的数值都应定义为常量： + +```go +// ❌ 不推荐 +if retryDelay >= 10*time.Second { + resetAt := time.Now().Add(30 * time.Second) +} + +// ✅ 推荐 +const ( + rateLimitThreshold = 10 * time.Second + defaultRateLimitDuration = 30 * time.Second +) + +if retryDelay >= rateLimitThreshold { + resetAt := time.Now().Add(defaultRateLimitDuration) +} +``` + +#### 注释引用常量名 +在注释中引用常量名而非硬编码值： + +```go +// ❌ 不推荐 +// < 10s: 等待后重试 + +// ✅ 推荐 +// < rateLimitThreshold: 等待后重试 +``` + +### 4. 错误处理 + +#### 使用结构化日志 +优先使用 `slog` 进行结构化日志记录： + +```go +// ❌ 不推荐 +log.Printf("%s status=%d model_rate_limit_failed model=%s error=%v", prefix, statusCode, modelName, err) + +// ✅ 推荐 +slog.Error("failed to set model rate limit", + "prefix", prefix, + "status_code", statusCode, + "model", modelName, + "error", err, +) +``` + +### 5. 测试规范 + +#### Mock 函数签名同步 +修改函数签名时，必须同步更新所有测试中的 mock 函数： + +```go +// 如果修改了 handleError 签名 +handleError func(..., groupID int64, sessionHash string) *Result + +// 必须同步更新测试中的 mock +handleError: func(..., groupID int64, sessionHash string) *Result { + return nil +}, +``` + +#### 测试构建标签 +统一使用测试构建标签： + +```go +//go:build unit + +package service +``` + +### 6. 时间格式解析 + +#### 使用标准库 +优先使用 `time.ParseDuration`，支持所有 Go duration 格式： + +```go +// ❌ 不推荐：手动限制格式 +if !strings.HasSuffix(delay, "s") || strings.Contains(delay, "m") { + continue +} + +// ✅ 推荐：使用标准库 +dur, err := time.ParseDuration(delay) // 支持 "0.5s", "4m50s", "1h30m" 等 +``` + +### 7. 接口设计 + +#### 接口隔离原则 +定义最小化接口，只包含必需的方法： + +```go +// ❌ 不推荐：使用过于宽泛的接口 +type AccountRepository interface { + // 20+ 个方法... +} + +// ✅ 推荐：定义最小化接口 +type ModelRateLimiter interface { + SetModelRateLimit(ctx context.Context, id int64, modelKey string, resetAt time.Time) error +} +``` + +### 8. 并发安全 + +#### 共享数据保护 +访问可能被并发修改的数据时，确保线程安全： + +```go +// 如果 Account.Extra 可能被并发修改 +// 需要使用互斥锁或原子操作保护读取 +func (a *Account) GetRateLimitRemainingTime(model string) time.Duration { + a.mu.RLock() + defer a.mu.RUnlock() + // 读取 Extra 字段... +} +``` + +### 9. 命名规范 + +#### 一致的命名风格 +- 常量使用 camelCase：`rateLimitThreshold` +- 类型使用 PascalCase：`AntigravityQuotaScope` +- 同一概念使用统一命名：`Threshold` 或 `Limit`，不要混用 + +```go +// ❌ 不推荐：命名不一致 +antigravitySmartRetryMinWait // 使用 Min +antigravityRateLimitThreshold // 使用 Threshold + +// ✅ 推荐：统一风格 +antigravityMinRetryWait +antigravityRateLimitThreshold +``` + +### 10. 代码审查清单 + +在提交代码前，检查以下项目： + +- [ ] 函数是否超过 30 行？（不可拆分的逻辑除外，需注释说明） +- [ ] 嵌套是否超过 3 层？ +- [ ] 是否有重复代码可以提取？ +- [ ] 是否使用了魔法数字？ +- [ ] Mock 函数签名是否与实际函数一致？ +- [ ] 测试是否覆盖了新增逻辑？ +- [ ] 日志是否包含足够的上下文信息？ +- [ ] 是否考虑了并发安全？ + +--- + +## CI 检查与发布门禁 + +### GitHub Actions 检查项 + +本项目有 4 个 CI 任务，**任何代码推送或发布前都必须全部通过**： + +| Workflow | Job | 说明 | 本地验证命令 | +|----------|-----|------|-------------| +| CI | `test` | 单元测试 + 集成测试 | `cd backend && make test-unit && make test-integration` | +| CI | `golangci-lint` | Go 代码静态检查（golangci-lint v2.7） | `cd backend && golangci-lint run --timeout=5m` | +| Security Scan | `backend-security` | govulncheck + gosec 安全扫描 | `cd backend && govulncheck ./... && gosec -severity high -confidence high ./...` | +| Security Scan | `frontend-security` | pnpm audit 前端依赖安全检查 | `cd frontend && pnpm audit --prod --audit-level=high` | + +### 向上游提交 PR + +PR 目标是上游官方仓库，**只包含通用功能改动**（bug fix、新功能、性能优化等）。 + +**以下文件禁止出现在 PR 中**（属于我们 fork 的定制化内容）： +- `CLAUDE.md`、`AGENTS.md` — 我们的开发文档 +- `backend/cmd/server/VERSION` — 我们的版本号文件 +- UI 定制改动（GitHub 链接移除、微信客服按钮、首页定制等） +- 部署配置（`deploy/` 目录下的定制修改） + +**PR 流程**： +1. 从 `develop` 创建功能分支，只包含要提交给上游的改动 +2. 推送分支后，**等待 4 个 CI job 全部通过** +3. 确认通过后再创建 PR +4. 使用 `gh run list --repo touwaeriol/sub2api --branch ` 检查状态 + +### 自有分支推送（develop / main） + +推送到我们自己的 `develop` 或 `main` 分支时，包含所有改动（定制化 + 通用功能）。 + +**推送前必须在本地执行全部 CI 检查**（不要等 GitHub Actions）： + +```bash +# 确保 Go 工具链可用（macOS homebrew） +export PATH="/opt/homebrew/bin:$HOME/go/bin:$PATH" + +# 1. 单元测试（必须） +cd backend && make test-unit + +# 2. 集成测试（推荐，需要 Docker） +make test-integration + +# 3. golangci-lint 静态检查（必须） +golangci-lint run --timeout=5m + +# 4. gofmt 格式检查（必须） +gofmt -l ./... +# 如果有输出，运行 gofmt -w 修复 +``` + +**推送后确认**： +1. 使用 `gh run list --repo touwaeriol/sub2api --branch ` 检查 GitHub Actions 状态 +2. 确认 CI 和 Security Scan 两个 workflow 的 4 个 job 全部绿色 ✅ +3. 任何 job 失败必须立即修复，**禁止在 CI 未通过的状态下继续后续操作** + +### 发布版本 + +1. 本地执行上述全部 CI 检查通过 +2. 递增 `backend/cmd/server/VERSION`，提交并推送 +3. 推送后确认 GitHub Actions 的 4 个 CI job 全部通过 +4. **CI 未通过时禁止部署** — 必须先修复问题 +5. 使用 `gh run list --repo touwaeriol/sub2api --limit 10` 确认状态 + +### 常见 CI 失败原因及修复 +- **gofmt**：struct 字段对齐不一致 → 运行 `gofmt -w ` 修复 +- **golangci-lint**：未使用的变量/导入 → 删除或使用 `_` 忽略 +- **test 失败**：mock 函数签名不一致 → 同步更新 mock +- **gosec**：安全漏洞 → 根据提示修复或添加例外 + +--- + +## PR 描述格式规范 + +所有 PR 描述使用中英文同步（先中文、后英文），包含以下三个部分： + +### 模板 + +```markdown +## 背景 / Background + +<一两句说明问题现状或触发原因> + + + +--- + +## 目的 / Purpose + +<本次改动要解决的问题或达到的目标> + + + +--- + +## 改动内容 / Changes + +### 后端 / Backend + +- **改动点 1**：说明 +- **改动点 2**：说明 + +--- + +- **Change 1**: description +- **Change 2**: description + +### 前端 / Frontend + +- **改动点 1**：说明 +- **改动点 2**：说明 + +--- + +- **Change 1**: description +- **Change 2**: description + +--- + +## 截图 / Screenshot（可选） + +ASCII 示意图或实际截图 +``` + +### 规范要点 + +- **标题**：使用 conventional commits 格式，如 `feat(scope): description` +- **中英文顺序**：同一段落先中文后英文，用空行分隔，不用 `---` 分割同段内容 +- **改动分类**：按 Backend / Frontend / Config 等模块分组，先列中文要点再列英文要点 +- **截图/示意图**：有 UI 变动时必须附上，可用 ASCII 示意布局 +- **目标分支**：提交到 `touwaeriol/sub2api` 的 `main` 分支 diff --git a/backend/cmd/server/VERSION b/backend/cmd/server/VERSION index 32844913..8b699e6e 100644 --- a/backend/cmd/server/VERSION +++ b/backend/cmd/server/VERSION @@ -1 +1 @@ -0.1.88 \ No newline at end of file +0.1.95.1 diff --git a/backend/ent/group.go b/backend/ent/group.go index 3db54a64..7ed49905 100644 --- a/backend/ent/group.go +++ b/backend/ent/group.go @@ -62,26 +62,28 @@ type Group struct { SoraVideoPricePerRequestHd *float64 `json:"sora_video_price_per_request_hd,omitempty"` // SoraStorageQuotaBytes holds the value of the "sora_storage_quota_bytes" field. SoraStorageQuotaBytes int64 `json:"sora_storage_quota_bytes,omitempty"` - // 是否仅允许 Claude Code 客户端 + // allow Claude Code client only ClaudeCodeOnly bool `json:"claude_code_only,omitempty"` - // 非 Claude Code 请求降级使用的分组 ID + // fallback group for non-Claude-Code requests FallbackGroupID *int64 `json:"fallback_group_id,omitempty"` - // 无效请求兜底使用的分组 ID + // fallback group for invalid request FallbackGroupIDOnInvalidRequest *int64 `json:"fallback_group_id_on_invalid_request,omitempty"` - // 模型路由配置：模型模式 -> 优先账号ID列表 + // model routing config: pattern -> account ids ModelRouting map[string][]int64 `json:"model_routing,omitempty"` - // 是否启用模型路由配置 + // whether model routing is enabled ModelRoutingEnabled bool `json:"model_routing_enabled,omitempty"` - // 是否注入 MCP XML 调用协议提示词（仅 antigravity 平台） + // whether MCP XML prompt injection is enabled McpXMLInject bool `json:"mcp_xml_inject,omitempty"` - // 支持的模型系列：claude, gemini_text, gemini_image + // supported model scopes: claude, gemini_text, gemini_image SupportedModelScopes []string `json:"supported_model_scopes,omitempty"` - // 分组显示排序，数值越小越靠前 + // group display order, lower comes first SortOrder int `json:"sort_order,omitempty"` // 是否允许 /v1/messages 调度到此 OpenAI 分组 AllowMessagesDispatch bool `json:"allow_messages_dispatch,omitempty"` // 默认映射模型 ID，当账号级映射找不到时使用此值 DefaultMappedModel string `json:"default_mapped_model,omitempty"` + // simulate claude usage as claude-max style (1h cache write) + SimulateClaudeMaxEnabled bool `json:"simulate_claude_max_enabled,omitempty"` // Edges holds the relations/edges for other nodes in the graph. // The values are being populated by the GroupQuery when eager-loading is set. Edges GroupEdges `json:"edges"` @@ -190,7 +192,7 @@ func (*Group) scanValues(columns []string) ([]any, error) { switch columns[i] { case group.FieldModelRouting, group.FieldSupportedModelScopes: values[i] = new([]byte) - case group.FieldIsExclusive, group.FieldClaudeCodeOnly, group.FieldModelRoutingEnabled, group.FieldMcpXMLInject, group.FieldAllowMessagesDispatch: + case group.FieldIsExclusive, group.FieldClaudeCodeOnly, group.FieldModelRoutingEnabled, group.FieldMcpXMLInject, group.FieldAllowMessagesDispatch, group.FieldSimulateClaudeMaxEnabled: values[i] = new(sql.NullBool) case group.FieldRateMultiplier, group.FieldDailyLimitUsd, group.FieldWeeklyLimitUsd, group.FieldMonthlyLimitUsd, group.FieldImagePrice1k, group.FieldImagePrice2k, group.FieldImagePrice4k, group.FieldSoraImagePrice360, group.FieldSoraImagePrice540, group.FieldSoraVideoPricePerRequest, group.FieldSoraVideoPricePerRequestHd: values[i] = new(sql.NullFloat64) @@ -431,6 +433,12 @@ func (_m *Group) assignValues(columns []string, values []any) error { } else if value.Valid { _m.DefaultMappedModel = value.String } + case group.FieldSimulateClaudeMaxEnabled: + if value, ok := values[i].(*sql.NullBool); !ok { + return fmt.Errorf("unexpected type %T for field simulate_claude_max_enabled", values[i]) + } else if value.Valid { + _m.SimulateClaudeMaxEnabled = value.Bool + } default: _m.selectValues.Set(columns[i], values[i]) } @@ -630,6 +638,9 @@ func (_m *Group) String() string { builder.WriteString(", ") builder.WriteString("default_mapped_model=") builder.WriteString(_m.DefaultMappedModel) + builder.WriteString(", ") + builder.WriteString("simulate_claude_max_enabled=") + builder.WriteString(fmt.Sprintf("%v", _m.SimulateClaudeMaxEnabled)) builder.WriteByte(')') return builder.String() } diff --git a/backend/ent/group/group.go b/backend/ent/group/group.go index 2612b6cf..970c7a85 100644 --- a/backend/ent/group/group.go +++ b/backend/ent/group/group.go @@ -79,6 +79,8 @@ const ( FieldAllowMessagesDispatch = "allow_messages_dispatch" // FieldDefaultMappedModel holds the string denoting the default_mapped_model field in the database. FieldDefaultMappedModel = "default_mapped_model" + // FieldSimulateClaudeMaxEnabled holds the string denoting the simulate_claude_max_enabled field in the database. + FieldSimulateClaudeMaxEnabled = "simulate_claude_max_enabled" // EdgeAPIKeys holds the string denoting the api_keys edge name in mutations. EdgeAPIKeys = "api_keys" // EdgeRedeemCodes holds the string denoting the redeem_codes edge name in mutations. @@ -186,6 +188,7 @@ var Columns = []string{ FieldSortOrder, FieldAllowMessagesDispatch, FieldDefaultMappedModel, + FieldSimulateClaudeMaxEnabled, } var ( @@ -259,6 +262,8 @@ var ( DefaultDefaultMappedModel string // DefaultMappedModelValidator is a validator for the "default_mapped_model" field. It is called by the builders before save. DefaultMappedModelValidator func(string) error + // DefaultSimulateClaudeMaxEnabled holds the default value on creation for the "simulate_claude_max_enabled" field. + DefaultSimulateClaudeMaxEnabled bool ) // OrderOption defines the ordering options for the Group queries. @@ -419,6 +424,11 @@ func ByDefaultMappedModel(opts ...sql.OrderTermOption) OrderOption { return sql.OrderByField(FieldDefaultMappedModel, opts...).ToFunc() } +// BySimulateClaudeMaxEnabled orders the results by the simulate_claude_max_enabled field. +func BySimulateClaudeMaxEnabled(opts ...sql.OrderTermOption) OrderOption { + return sql.OrderByField(FieldSimulateClaudeMaxEnabled, opts...).ToFunc() +} + // ByAPIKeysCount orders the results by api_keys count. func ByAPIKeysCount(opts ...sql.OrderTermOption) OrderOption { return func(s *sql.Selector) { diff --git a/backend/ent/group/where.go b/backend/ent/group/where.go index 5dd8759e..62c91d5a 100644 --- a/backend/ent/group/where.go +++ b/backend/ent/group/where.go @@ -205,6 +205,11 @@ func DefaultMappedModel(v string) predicate.Group { return predicate.Group(sql.FieldEQ(FieldDefaultMappedModel, v)) } +// SimulateClaudeMaxEnabled applies equality check predicate on the "simulate_claude_max_enabled" field. It's identical to SimulateClaudeMaxEnabledEQ. +func SimulateClaudeMaxEnabled(v bool) predicate.Group { + return predicate.Group(sql.FieldEQ(FieldSimulateClaudeMaxEnabled, v)) +} + // CreatedAtEQ applies the EQ predicate on the "created_at" field. func CreatedAtEQ(v time.Time) predicate.Group { return predicate.Group(sql.FieldEQ(FieldCreatedAt, v)) @@ -1555,6 +1560,16 @@ func DefaultMappedModelContainsFold(v string) predicate.Group { return predicate.Group(sql.FieldContainsFold(FieldDefaultMappedModel, v)) } +// SimulateClaudeMaxEnabledEQ applies the EQ predicate on the "simulate_claude_max_enabled" field. +func SimulateClaudeMaxEnabledEQ(v bool) predicate.Group { + return predicate.Group(sql.FieldEQ(FieldSimulateClaudeMaxEnabled, v)) +} + +// SimulateClaudeMaxEnabledNEQ applies the NEQ predicate on the "simulate_claude_max_enabled" field. +func SimulateClaudeMaxEnabledNEQ(v bool) predicate.Group { + return predicate.Group(sql.FieldNEQ(FieldSimulateClaudeMaxEnabled, v)) +} + // HasAPIKeys applies the HasEdge predicate on the "api_keys" edge. func HasAPIKeys() predicate.Group { return predicate.Group(func(s *sql.Selector) { diff --git a/backend/ent/group_create.go b/backend/ent/group_create.go index 6db5b974..9418b02f 100644 --- a/backend/ent/group_create.go +++ b/backend/ent/group_create.go @@ -452,6 +452,20 @@ func (_c *GroupCreate) SetNillableDefaultMappedModel(v *string) *GroupCreate { return _c } +// SetSimulateClaudeMaxEnabled sets the "simulate_claude_max_enabled" field. +func (_c *GroupCreate) SetSimulateClaudeMaxEnabled(v bool) *GroupCreate { + _c.mutation.SetSimulateClaudeMaxEnabled(v) + return _c +} + +// SetNillableSimulateClaudeMaxEnabled sets the "simulate_claude_max_enabled" field if the given value is not nil. +func (_c *GroupCreate) SetNillableSimulateClaudeMaxEnabled(v *bool) *GroupCreate { + if v != nil { + _c.SetSimulateClaudeMaxEnabled(*v) + } + return _c +} + // AddAPIKeyIDs adds the "api_keys" edge to the APIKey entity by IDs. func (_c *GroupCreate) AddAPIKeyIDs(ids ...int64) *GroupCreate { _c.mutation.AddAPIKeyIDs(ids...) @@ -649,6 +663,10 @@ func (_c *GroupCreate) defaults() error { v := group.DefaultDefaultMappedModel _c.mutation.SetDefaultMappedModel(v) } + if _, ok := _c.mutation.SimulateClaudeMaxEnabled(); !ok { + v := group.DefaultSimulateClaudeMaxEnabled + _c.mutation.SetSimulateClaudeMaxEnabled(v) + } return nil } @@ -730,6 +748,9 @@ func (_c *GroupCreate) check() error { return &ValidationError{Name: "default_mapped_model", err: fmt.Errorf(`ent: validator failed for field "Group.default_mapped_model": %w`, err)} } } + if _, ok := _c.mutation.SimulateClaudeMaxEnabled(); !ok { + return &ValidationError{Name: "simulate_claude_max_enabled", err: errors.New(`ent: missing required field "Group.simulate_claude_max_enabled"`)} + } return nil } @@ -885,6 +906,10 @@ func (_c *GroupCreate) createSpec() (*Group, *sqlgraph.CreateSpec) { _spec.SetField(group.FieldDefaultMappedModel, field.TypeString, value) _node.DefaultMappedModel = value } + if value, ok := _c.mutation.SimulateClaudeMaxEnabled(); ok { + _spec.SetField(group.FieldSimulateClaudeMaxEnabled, field.TypeBool, value) + _node.SimulateClaudeMaxEnabled = value + } if nodes := _c.mutation.APIKeysIDs(); len(nodes) > 0 { edge := &sqlgraph.EdgeSpec{ Rel: sqlgraph.O2M, @@ -1599,6 +1624,18 @@ func (u *GroupUpsert) UpdateDefaultMappedModel() *GroupUpsert { return u } +// SetSimulateClaudeMaxEnabled sets the "simulate_claude_max_enabled" field. +func (u *GroupUpsert) SetSimulateClaudeMaxEnabled(v bool) *GroupUpsert { + u.Set(group.FieldSimulateClaudeMaxEnabled, v) + return u +} + +// UpdateSimulateClaudeMaxEnabled sets the "simulate_claude_max_enabled" field to the value that was provided on create. +func (u *GroupUpsert) UpdateSimulateClaudeMaxEnabled() *GroupUpsert { + u.SetExcluded(group.FieldSimulateClaudeMaxEnabled) + return u +} + // UpdateNewValues updates the mutable fields using the new values that were set on create. // Using this option is equivalent to using: // @@ -2295,6 +2332,20 @@ func (u *GroupUpsertOne) UpdateDefaultMappedModel() *GroupUpsertOne { }) } +// SetSimulateClaudeMaxEnabled sets the "simulate_claude_max_enabled" field. +func (u *GroupUpsertOne) SetSimulateClaudeMaxEnabled(v bool) *GroupUpsertOne { + return u.Update(func(s *GroupUpsert) { + s.SetSimulateClaudeMaxEnabled(v) + }) +} + +// UpdateSimulateClaudeMaxEnabled sets the "simulate_claude_max_enabled" field to the value that was provided on create. +func (u *GroupUpsertOne) UpdateSimulateClaudeMaxEnabled() *GroupUpsertOne { + return u.Update(func(s *GroupUpsert) { + s.UpdateSimulateClaudeMaxEnabled() + }) +} + // Exec executes the query. func (u *GroupUpsertOne) Exec(ctx context.Context) error { if len(u.create.conflict) == 0 { @@ -3157,6 +3208,20 @@ func (u *GroupUpsertBulk) UpdateDefaultMappedModel() *GroupUpsertBulk { }) } +// SetSimulateClaudeMaxEnabled sets the "simulate_claude_max_enabled" field. +func (u *GroupUpsertBulk) SetSimulateClaudeMaxEnabled(v bool) *GroupUpsertBulk { + return u.Update(func(s *GroupUpsert) { + s.SetSimulateClaudeMaxEnabled(v) + }) +} + +// UpdateSimulateClaudeMaxEnabled sets the "simulate_claude_max_enabled" field to the value that was provided on create. +func (u *GroupUpsertBulk) UpdateSimulateClaudeMaxEnabled() *GroupUpsertBulk { + return u.Update(func(s *GroupUpsert) { + s.UpdateSimulateClaudeMaxEnabled() + }) +} + // Exec executes the query. func (u *GroupUpsertBulk) Exec(ctx context.Context) error { if u.create.err != nil { diff --git a/backend/ent/group_update.go b/backend/ent/group_update.go index b3698596..75955f7d 100644 --- a/backend/ent/group_update.go +++ b/backend/ent/group_update.go @@ -653,6 +653,20 @@ func (_u *GroupUpdate) SetNillableDefaultMappedModel(v *string) *GroupUpdate { return _u } +// SetSimulateClaudeMaxEnabled sets the "simulate_claude_max_enabled" field. +func (_u *GroupUpdate) SetSimulateClaudeMaxEnabled(v bool) *GroupUpdate { + _u.mutation.SetSimulateClaudeMaxEnabled(v) + return _u +} + +// SetNillableSimulateClaudeMaxEnabled sets the "simulate_claude_max_enabled" field if the given value is not nil. +func (_u *GroupUpdate) SetNillableSimulateClaudeMaxEnabled(v *bool) *GroupUpdate { + if v != nil { + _u.SetSimulateClaudeMaxEnabled(*v) + } + return _u +} + // AddAPIKeyIDs adds the "api_keys" edge to the APIKey entity by IDs. func (_u *GroupUpdate) AddAPIKeyIDs(ids ...int64) *GroupUpdate { _u.mutation.AddAPIKeyIDs(ids...) @@ -1149,6 +1163,9 @@ func (_u *GroupUpdate) sqlSave(ctx context.Context) (_node int, err error) { if value, ok := _u.mutation.DefaultMappedModel(); ok { _spec.SetField(group.FieldDefaultMappedModel, field.TypeString, value) } + if value, ok := _u.mutation.SimulateClaudeMaxEnabled(); ok { + _spec.SetField(group.FieldSimulateClaudeMaxEnabled, field.TypeBool, value) + } if _u.mutation.APIKeysCleared() { edge := &sqlgraph.EdgeSpec{ Rel: sqlgraph.O2M, @@ -2081,6 +2098,20 @@ func (_u *GroupUpdateOne) SetNillableDefaultMappedModel(v *string) *GroupUpdateO return _u } +// SetSimulateClaudeMaxEnabled sets the "simulate_claude_max_enabled" field. +func (_u *GroupUpdateOne) SetSimulateClaudeMaxEnabled(v bool) *GroupUpdateOne { + _u.mutation.SetSimulateClaudeMaxEnabled(v) + return _u +} + +// SetNillableSimulateClaudeMaxEnabled sets the "simulate_claude_max_enabled" field if the given value is not nil. +func (_u *GroupUpdateOne) SetNillableSimulateClaudeMaxEnabled(v *bool) *GroupUpdateOne { + if v != nil { + _u.SetSimulateClaudeMaxEnabled(*v) + } + return _u +} + // AddAPIKeyIDs adds the "api_keys" edge to the APIKey entity by IDs. func (_u *GroupUpdateOne) AddAPIKeyIDs(ids ...int64) *GroupUpdateOne { _u.mutation.AddAPIKeyIDs(ids...) @@ -2607,6 +2638,9 @@ func (_u *GroupUpdateOne) sqlSave(ctx context.Context) (_node *Group, err error) if value, ok := _u.mutation.DefaultMappedModel(); ok { _spec.SetField(group.FieldDefaultMappedModel, field.TypeString, value) } + if value, ok := _u.mutation.SimulateClaudeMaxEnabled(); ok { + _spec.SetField(group.FieldSimulateClaudeMaxEnabled, field.TypeBool, value) + } if _u.mutation.APIKeysCleared() { edge := &sqlgraph.EdgeSpec{ Rel: sqlgraph.O2M, diff --git a/backend/ent/migrate/schema.go b/backend/ent/migrate/schema.go index ff1c1b88..03c66cd1 100644 --- a/backend/ent/migrate/schema.go +++ b/backend/ent/migrate/schema.go @@ -410,6 +410,7 @@ var ( {Name: "sort_order", Type: field.TypeInt, Default: 0}, {Name: "allow_messages_dispatch", Type: field.TypeBool, Default: false}, {Name: "default_mapped_model", Type: field.TypeString, Size: 100, Default: ""}, + {Name: "simulate_claude_max_enabled", Type: field.TypeBool, Default: false}, } // GroupsTable holds the schema information for the "groups" table. GroupsTable = &schema.Table{ diff --git a/backend/ent/mutation.go b/backend/ent/mutation.go index 652adcac..8177d14d 100644 --- a/backend/ent/mutation.go +++ b/backend/ent/mutation.go @@ -8252,6 +8252,7 @@ type GroupMutation struct { addsort_order *int allow_messages_dispatch *bool default_mapped_model *string + simulate_claude_max_enabled *bool clearedFields map[string]struct{} api_keys map[int64]struct{} removedapi_keys map[int64]struct{} @@ -10068,6 +10069,42 @@ func (m *GroupMutation) ResetDefaultMappedModel() { m.default_mapped_model = nil } +// SetSimulateClaudeMaxEnabled sets the "simulate_claude_max_enabled" field. +func (m *GroupMutation) SetSimulateClaudeMaxEnabled(b bool) { + m.simulate_claude_max_enabled = &b +} + +// SimulateClaudeMaxEnabled returns the value of the "simulate_claude_max_enabled" field in the mutation. +func (m *GroupMutation) SimulateClaudeMaxEnabled() (r bool, exists bool) { + v := m.simulate_claude_max_enabled + if v == nil { + return + } + return *v, true +} + +// OldSimulateClaudeMaxEnabled returns the old "simulate_claude_max_enabled" field's value of the Group entity. +// If the Group object wasn't provided to the builder, the object is fetched from the database. +// An error is returned if the mutation operation is not UpdateOne, or the database query fails. +func (m *GroupMutation) OldSimulateClaudeMaxEnabled(ctx context.Context) (v bool, err error) { + if !m.op.Is(OpUpdateOne) { + return v, errors.New("OldSimulateClaudeMaxEnabled is only allowed on UpdateOne operations") + } + if m.id == nil || m.oldValue == nil { + return v, errors.New("OldSimulateClaudeMaxEnabled requires an ID field in the mutation") + } + oldValue, err := m.oldValue(ctx) + if err != nil { + return v, fmt.Errorf("querying old value for OldSimulateClaudeMaxEnabled: %w", err) + } + return oldValue.SimulateClaudeMaxEnabled, nil +} + +// ResetSimulateClaudeMaxEnabled resets all changes to the "simulate_claude_max_enabled" field. +func (m *GroupMutation) ResetSimulateClaudeMaxEnabled() { + m.simulate_claude_max_enabled = nil +} + // AddAPIKeyIDs adds the "api_keys" edge to the APIKey entity by ids. func (m *GroupMutation) AddAPIKeyIDs(ids ...int64) { if m.api_keys == nil { @@ -10426,7 +10463,7 @@ func (m *GroupMutation) Type() string { // order to get all numeric fields that were incremented/decremented, call // AddedFields(). func (m *GroupMutation) Fields() []string { - fields := make([]string, 0, 32) + fields := make([]string, 0, 33) if m.created_at != nil { fields = append(fields, group.FieldCreatedAt) } @@ -10523,6 +10560,9 @@ func (m *GroupMutation) Fields() []string { if m.default_mapped_model != nil { fields = append(fields, group.FieldDefaultMappedModel) } + if m.simulate_claude_max_enabled != nil { + fields = append(fields, group.FieldSimulateClaudeMaxEnabled) + } return fields } @@ -10595,6 +10635,8 @@ func (m *GroupMutation) Field(name string) (ent.Value, bool) { return m.AllowMessagesDispatch() case group.FieldDefaultMappedModel: return m.DefaultMappedModel() + case group.FieldSimulateClaudeMaxEnabled: + return m.SimulateClaudeMaxEnabled() } return nil, false } @@ -10668,6 +10710,8 @@ func (m *GroupMutation) OldField(ctx context.Context, name string) (ent.Value, e return m.OldAllowMessagesDispatch(ctx) case group.FieldDefaultMappedModel: return m.OldDefaultMappedModel(ctx) + case group.FieldSimulateClaudeMaxEnabled: + return m.OldSimulateClaudeMaxEnabled(ctx) } return nil, fmt.Errorf("unknown Group field %s", name) } @@ -10901,6 +10945,13 @@ func (m *GroupMutation) SetField(name string, value ent.Value) error { } m.SetDefaultMappedModel(v) return nil + case group.FieldSimulateClaudeMaxEnabled: + v, ok := value.(bool) + if !ok { + return fmt.Errorf("unexpected type %T for field %s", value, name) + } + m.SetSimulateClaudeMaxEnabled(v) + return nil } return fmt.Errorf("unknown Group field %s", name) } @@ -11334,6 +11385,9 @@ func (m *GroupMutation) ResetField(name string) error { case group.FieldDefaultMappedModel: m.ResetDefaultMappedModel() return nil + case group.FieldSimulateClaudeMaxEnabled: + m.ResetSimulateClaudeMaxEnabled() + return nil } return fmt.Errorf("unknown Group field %s", name) } diff --git a/backend/ent/runtime/runtime.go b/backend/ent/runtime/runtime.go index b8facf36..ff8a655b 100644 --- a/backend/ent/runtime/runtime.go +++ b/backend/ent/runtime/runtime.go @@ -463,6 +463,10 @@ func init() { group.DefaultDefaultMappedModel = groupDescDefaultMappedModel.Default.(string) // group.DefaultMappedModelValidator is a validator for the "default_mapped_model" field. It is called by the builders before save. group.DefaultMappedModelValidator = groupDescDefaultMappedModel.Validators[0].(func(string) error) + // groupDescSimulateClaudeMaxEnabled is the schema descriptor for simulate_claude_max_enabled field. + groupDescSimulateClaudeMaxEnabled := groupFields[29].Descriptor() + // group.DefaultSimulateClaudeMaxEnabled holds the default value on creation for the simulate_claude_max_enabled field. + group.DefaultSimulateClaudeMaxEnabled = groupDescSimulateClaudeMaxEnabled.Default.(bool) idempotencyrecordMixin := schema.IdempotencyRecord{}.Mixin() idempotencyrecordMixinFields0 := idempotencyrecordMixin[0].Fields() _ = idempotencyrecordMixinFields0 diff --git a/backend/ent/schema/group.go b/backend/ent/schema/group.go index 0f5a7b14..0842a0f8 100644 --- a/backend/ent/schema/group.go +++ b/backend/ent/schema/group.go @@ -33,8 +33,6 @@ func (Group) Mixin() []ent.Mixin { func (Group) Fields() []ent.Field { return []ent.Field{ - // 唯一约束通过部分索引实现（WHERE deleted_at IS NULL），支持软删除后重用 - // 见迁移文件 016_soft_delete_partial_unique_indexes.sql field.String("name"). MaxLen(100). NotEmpty(), @@ -51,7 +49,6 @@ func (Group) Fields() []ent.Field { MaxLen(20). Default(domain.StatusActive), - // Subscription-related fields (added by migration 003) field.String("platform"). MaxLen(50). Default(domain.PlatformAnthropic), @@ -73,7 +70,6 @@ func (Group) Fields() []ent.Field { field.Int("default_validity_days"). Default(30), - // 图片生成计费配置（antigravity 和 gemini 平台使用） field.Float("image_price_1k"). Optional(). Nillable(). @@ -87,7 +83,6 @@ func (Group) Fields() []ent.Field { Nillable(). SchemaType(map[string]string{dialect.Postgres: "decimal(20,8)"}), - // Sora 按次计费配置（阶段 1） field.Float("sora_image_price_360"). Optional(). Nillable(). @@ -109,45 +104,38 @@ func (Group) Fields() []ent.Field { field.Int64("sora_storage_quota_bytes"). Default(0), - // Claude Code 客户端限制 (added by migration 029) field.Bool("claude_code_only"). Default(false). - Comment("是否仅允许 Claude Code 客户端"), + Comment("allow Claude Code client only"), field.Int64("fallback_group_id"). Optional(). Nillable(). - Comment("非 Claude Code 请求降级使用的分组 ID"), + Comment("fallback group for non-Claude-Code requests"), field.Int64("fallback_group_id_on_invalid_request"). Optional(). Nillable(). - Comment("无效请求兜底使用的分组 ID"), + Comment("fallback group for invalid request"), - // 模型路由配置 (added by migration 040) field.JSON("model_routing", map[string][]int64{}). Optional(). SchemaType(map[string]string{dialect.Postgres: "jsonb"}). - Comment("模型路由配置：模型模式 -> 优先账号ID列表"), - - // 模型路由开关 (added by migration 041) + Comment("model routing config: pattern -> account ids"), field.Bool("model_routing_enabled"). Default(false). - Comment("是否启用模型路由配置"), + Comment("whether model routing is enabled"), - // MCP XML 协议注入开关 (added by migration 042) field.Bool("mcp_xml_inject"). Default(true). - Comment("是否注入 MCP XML 调用协议提示词（仅 antigravity 平台）"), + Comment("whether MCP XML prompt injection is enabled"), - // 支持的模型系列 (added by migration 046) field.JSON("supported_model_scopes", []string{}). Default([]string{"claude", "gemini_text", "gemini_image"}). SchemaType(map[string]string{dialect.Postgres: "jsonb"}). - Comment("支持的模型系列：claude, gemini_text, gemini_image"), + Comment("supported model scopes: claude, gemini_text, gemini_image"), - // 分组排序 (added by migration 052) field.Int("sort_order"). Default(0). - Comment("分组显示排序，数值越小越靠前"), + Comment("group display order, lower comes first"), // OpenAI Messages 调度配置 (added by migration 069) field.Bool("allow_messages_dispatch"). @@ -157,6 +145,9 @@ func (Group) Fields() []ent.Field { MaxLen(100). Default(""). Comment("默认映射模型 ID，当账号级映射找不到时使用此值"), + field.Bool("simulate_claude_max_enabled"). + Default(false). + Comment("simulate claude usage as claude-max style (1h cache write)"), } } @@ -172,14 +163,11 @@ func (Group) Edges() []ent.Edge { edge.From("allowed_users", User.Type). Ref("allowed_groups"). Through("user_allowed_groups", UserAllowedGroup.Type), - // 注意：fallback_group_id 直接作为字段使用，不定义 edge - // 这样允许多个分组指向同一个降级分组（M2O 关系） } } func (Group) Indexes() []ent.Index { return []ent.Index{ - // name 字段已在 Fields() 中声明 Unique()，无需重复索引 index.Fields("status"), index.Fields("platform"), index.Fields("subscription_type"), diff --git a/backend/go.mod b/backend/go.mod index 03637401..267fcf60 100644 --- a/backend/go.mod +++ b/backend/go.mod @@ -87,6 +87,7 @@ require ( github.com/davecgh/go-spew v1.1.2-0.20180830191138-d8f796af33cc // indirect github.com/dgryski/go-rendezvous v0.0.0-20200823014737-9f7001d12a5f // indirect github.com/distribution/reference v0.6.0 // indirect + github.com/dlclark/regexp2 v1.10.0 // indirect github.com/docker/docker v28.5.1+incompatible // indirect github.com/docker/go-connections v0.6.0 // indirect github.com/docker/go-units v0.5.0 // indirect @@ -137,6 +138,8 @@ require ( github.com/opencontainers/image-spec v1.1.1 // indirect github.com/pelletier/go-toml/v2 v2.2.2 // indirect github.com/pkg/errors v0.9.1 // indirect + github.com/pkoukk/tiktoken-go v0.1.8 // indirect + github.com/pkoukk/tiktoken-go-loader v0.0.2 // indirect github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 // indirect github.com/power-devops/perfstat v0.0.0-20210106213030-5aafc221ea8c // indirect github.com/quic-go/qpack v0.6.0 // indirect diff --git a/backend/go.sum b/backend/go.sum index 993a1d54..965f7442 100644 --- a/backend/go.sum +++ b/backend/go.sum @@ -124,6 +124,8 @@ github.com/dgryski/go-rendezvous v0.0.0-20200823014737-9f7001d12a5f h1:lO4WD4F/r github.com/dgryski/go-rendezvous v0.0.0-20200823014737-9f7001d12a5f/go.mod h1:cuUVRXasLTGF7a8hSLbxyZXjz+1KgoB3wDUb6vlszIc= github.com/distribution/reference v0.6.0 h1:0IXCQ5g4/QMHHkarYzh5l+u8T3t73zM5QvfrDyIgxBk= github.com/distribution/reference v0.6.0/go.mod h1:BbU0aIcezP1/5jX/8MP0YiH4SdvB5Y4f/wlDRiLyi3E= +github.com/dlclark/regexp2 v1.10.0 h1:+/GIL799phkJqYW+3YbOd8LCcbHzT0Pbo8zl70MHsq0= +github.com/dlclark/regexp2 v1.10.0/go.mod h1:DHkYz0B9wPfa6wondMfaivmHpzrQ3v9q8cnmRbL6yW8= github.com/docker/docker v28.5.1+incompatible h1:Bm8DchhSD2J6PsFzxC35TZo4TLGR2PdW/E69rU45NhM= github.com/docker/docker v28.5.1+incompatible/go.mod h1:eEKB0N0r5NX/I1kEveEz05bcu8tLC/8azJZsviup8Sk= github.com/docker/go-connections v0.6.0 h1:LlMG9azAe1TqfR7sO+NJttz1gy6KO7VJBh+pMmjSD94= @@ -180,6 +182,7 @@ github.com/google/go-querystring v1.1.0/go.mod h1:Kcdr2DB4koayq7X8pmAG4sNG59So17 github.com/google/gofuzz v1.0.0/go.mod h1:dBl0BpW6vV/+mYPU4Po3pmUjxk6FQPldtuIdl/M65Eg= github.com/google/pprof v0.0.0-20250317173921-a4b03ec1a45e h1:ijClszYn+mADRFY17kjQEVQ1XRhq2/JR1M3sGqeJoxs= github.com/google/pprof v0.0.0-20250317173921-a4b03ec1a45e/go.mod h1:boTsfXsheKC2y+lKOCMpSfarhxDeIzfZG1jqGcPl3cA= +github.com/google/subcommands v1.2.0/go.mod h1:ZjhPrFU+Olkh9WazFPsl27BQ4UPiG37m3yTrtFlrHVk= github.com/google/uuid v1.6.0 h1:NIvaJDMOsjHA8n1jAhLSgzrAzy1Hgr+hNrb57e+94F0= github.com/google/uuid v1.6.0/go.mod h1:TIyPZe4MgqvfeYDBFedMoGGpEw/LqOeaOT+nhxU+yHo= github.com/google/wire v0.7.0 h1:JxUKI6+CVBgCO2WToKy/nQk0sS+amI9z9EjVmdaocj4= @@ -199,6 +202,8 @@ github.com/icholy/digest v1.1.0 h1:HfGg9Irj7i+IX1o1QAmPfIBNu/Q5A5Tu3n/MED9k9H4= github.com/icholy/digest v1.1.0/go.mod h1:QNrsSGQ5v7v9cReDI0+eyjsXGUoRSUZQHeQ5C4XLa0Y= github.com/imroc/req/v3 v3.57.0 h1:LMTUjNRUybUkTPn8oJDq8Kg3JRBOBTcnDhKu7mzupKI= github.com/imroc/req/v3 v3.57.0/go.mod h1:JL62ey1nvSLq81HORNcosvlf7SxZStONNqOprg0Pz00= +github.com/inconshreveable/mousetrap v1.1.0 h1:wN+x4NVGpMsO7ErUn/mUI3vEoE6Jt13X2s0bqwp9tc8= +github.com/inconshreveable/mousetrap v1.1.0/go.mod h1:vpF70FUmC8bwa3OWnCshd2FqLfsEA9PFc4w1p2J65bw= github.com/jackc/pgpassfile v1.0.0 h1:/6Hmqy13Ss2zCq62VdNG8tM1wchn8zjSGOBJ6icpsIM= github.com/jackc/pgpassfile v1.0.0/go.mod h1:CEx0iS5ambNFdcRtxPj5JhEz+xB6uRky5eyVu/W2HEg= github.com/jackc/pgservicefile v0.0.0-20240606120523-5a60cdf6a761 h1:iCEnooe7UlwOQYpKFhBabPMi4aNAfoODPEFNiAnClxo= @@ -281,6 +286,10 @@ github.com/pelletier/go-toml/v2 v2.2.2 h1:aYUidT7k73Pcl9nb2gScu7NSrKCSHIDE89b3+6 github.com/pelletier/go-toml/v2 v2.2.2/go.mod h1:1t835xjRzz80PqgE6HHgN2JOsmgYu/h4qDAS4n929Rs= github.com/pkg/errors v0.9.1 h1:FEBLx1zS214owpjy7qsBeixbURkuhQAwrK5UwLGTwt4= github.com/pkg/errors v0.9.1/go.mod h1:bwawxfHBFNV+L2hUp1rHADufV3IMtnDRdf1r5NINEl0= +github.com/pkoukk/tiktoken-go v0.1.8 h1:85ENo+3FpWgAACBaEUVp+lctuTcYUO7BtmfhlN/QTRo= +github.com/pkoukk/tiktoken-go v0.1.8/go.mod h1:9NiV+i9mJKGj1rYOT+njbv+ZwA/zJxYdewGl6qVatpg= +github.com/pkoukk/tiktoken-go-loader v0.0.2 h1:LUKws63GV3pVHwH1srkBplBv+7URgmOmhSkRxsIvsK4= +github.com/pkoukk/tiktoken-go-loader v0.0.2/go.mod h1:4mIkYyZooFlnenDlormIo6cd5wrlUKNr97wp9nGgEKo= github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4= github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 h1:Jamvg5psRIccs7FGNTlIRMkT8wgtp5eCXdBlqhYGL6U= github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4= @@ -333,6 +342,8 @@ github.com/stretchr/objx v0.4.0/go.mod h1:YvHI0jy2hoMjB+UWwv71VJQ9isScKT/TqJzVSS github.com/stretchr/objx v0.5.0/go.mod h1:Yh+to48EsGEfYuaHDzXPcE3xhTkx73EhmCGUpEOglKo= github.com/stretchr/objx v0.5.2 h1:xuMeJ0Sdp5ZMRXx/aWO6RZxdr3beISkG5/G/aIRr3pY= github.com/stretchr/objx v0.5.2/go.mod h1:FRsXN1f5AsAjCGJKqEizvkpNtU+EGNCLh3NxZ/8L+MA= +github.com/stretchr/testify v1.11.1 h1:7s2iGBzp5EwR7/aIZr8ao5+dra3wiQyKjjFuvgVKu7U= +github.com/stretchr/testify v1.11.1/go.mod h1:wZwfW3scLgRK+23gO65QZefKpKQRnfz6sD981Nm4B6U= github.com/stretchr/testify v1.3.0/go.mod h1:M5WIy9Dh21IEIfnGCwXGc5bZfKNJtfHm1UVUgZn+9EI= github.com/stretchr/testify v1.7.0/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg= github.com/stretchr/testify v1.7.1/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg= @@ -341,8 +352,6 @@ github.com/stretchr/testify v1.8.1/go.mod h1:w2LPCIKwWwSfY2zedu0+kehJoqGctiVI29o github.com/stretchr/testify v1.8.2/go.mod h1:w2LPCIKwWwSfY2zedu0+kehJoqGctiVI29o6fzry7u4= github.com/stretchr/testify v1.8.4/go.mod h1:sz/lmYIOXD/1dqDmKjjqLyZ2RngseejIcXlSw2iwfAo= github.com/stretchr/testify v1.9.0/go.mod h1:r2ic/lqez/lEtzL7wO/rwa5dbSLXVDPFyf8C91i36aY= -github.com/stretchr/testify v1.11.1 h1:7s2iGBzp5EwR7/aIZr8ao5+dra3wiQyKjjFuvgVKu7U= -github.com/stretchr/testify v1.11.1/go.mod h1:wZwfW3scLgRK+23gO65QZefKpKQRnfz6sD981Nm4B6U= github.com/subosito/gotenv v1.6.0 h1:9NlTDc1FTs4qu0DDq7AEtTPNw6SVm7uBMsUCUjABIf8= github.com/subosito/gotenv v1.6.0/go.mod h1:Dk4QP5c2W3ibzajGcXpNraDfq2IrhjMIvMSWPKKo0FU= github.com/tam7t/hpkp v0.0.0-20160821193359-2b70b4024ed5 h1:YqAladjX7xpA6BM04leXMWAEjS0mTZ5kUU9KRBriQJc= @@ -432,11 +441,11 @@ golang.org/x/sys v0.0.0-20210616094352-59db8d763f22/go.mod h1:oPkhp1MJrh7nUepCBc golang.org/x/sys v0.0.0-20220704084225-05e143d24a9e/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg= golang.org/x/sys v0.0.0-20220715151400-c0bba94af5f8/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg= golang.org/x/sys v0.0.0-20220811171246-fbc7d0a398ab/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg= -golang.org/x/sys v0.6.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg= -golang.org/x/sys v0.8.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg= golang.org/x/sys v0.11.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg= golang.org/x/sys v0.41.0 h1:Ivj+2Cp/ylzLiEU89QhWblYnOE9zerudt9Ftecq2C6k= golang.org/x/sys v0.41.0/go.mod h1:OgkHotnGiDImocRcuBABYBEXf8A9a87e/uXjp9XT3ks= +golang.org/x/sys v0.6.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg= +golang.org/x/sys v0.8.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg= golang.org/x/term v0.0.0-20201126162022-7de9c90e9dd1/go.mod h1:bj7SfCRtBDWHUb9snDiAeCFNEtKQo2Wmx5Cou7ajbmo= golang.org/x/term v0.40.0 h1:36e4zGLqU4yhjlmxEaagx2KuYbJq3EwY8K943ZsHcvg= golang.org/x/term v0.40.0/go.mod h1:w2P8uVp06p2iyKKuvXIm7N/y0UCRt3UfJTfZ7oOpglM= diff --git a/backend/internal/handler/admin/account_handler.go b/backend/internal/handler/admin/account_handler.go index 7c4d4638..1f0d0b52 100644 --- a/backend/internal/handler/admin/account_handler.go +++ b/backend/internal/handler/admin/account_handler.go @@ -1338,6 +1338,12 @@ func (h *AccountHandler) BulkUpdate(c *gin.Context) { c.JSON(409, gin.H{ "error": "mixed_channel_warning", "message": mixedErr.Error(), + "details": gin.H{ + "group_id": mixedErr.GroupID, + "group_name": mixedErr.GroupName, + "current_platform": mixedErr.CurrentPlatform, + "other_platform": mixedErr.OtherPlatform, + }, }) return } diff --git a/backend/internal/handler/admin/account_handler_mixed_channel_test.go b/backend/internal/handler/admin/account_handler_mixed_channel_test.go index 24ec5bcf..5b81db2a 100644 --- a/backend/internal/handler/admin/account_handler_mixed_channel_test.go +++ b/backend/internal/handler/admin/account_handler_mixed_channel_test.go @@ -111,7 +111,7 @@ func TestAccountHandlerCreateMixedChannelConflictSimplifiedResponse(t *testing.T var resp map[string]any require.NoError(t, json.Unmarshal(rec.Body.Bytes(), &resp)) require.Equal(t, "mixed_channel_warning", resp["error"]) - require.Contains(t, resp["message"], "mixed_channel_warning") + require.Contains(t, resp["message"], "claude-max") _, hasDetails := resp["details"] _, hasRequireConfirmation := resp["require_confirmation"] require.False(t, hasDetails) @@ -140,7 +140,7 @@ func TestAccountHandlerUpdateMixedChannelConflictSimplifiedResponse(t *testing.T var resp map[string]any require.NoError(t, json.Unmarshal(rec.Body.Bytes(), &resp)) require.Equal(t, "mixed_channel_warning", resp["error"]) - require.Contains(t, resp["message"], "mixed_channel_warning") + require.Contains(t, resp["message"], "claude-max") _, hasDetails := resp["details"] _, hasRequireConfirmation := resp["require_confirmation"] require.False(t, hasDetails) diff --git a/backend/internal/handler/admin/admin_service_stub_test.go b/backend/internal/handler/admin/admin_service_stub_test.go index 84a9f102..b77a2b7f 100644 --- a/backend/internal/handler/admin/admin_service_stub_test.go +++ b/backend/internal/handler/admin/admin_service_stub_test.go @@ -175,6 +175,10 @@ func (s *stubAdminService) GetGroupAPIKeys(ctx context.Context, groupID int64, p return s.apiKeys, int64(len(s.apiKeys)), nil } +func (s *stubAdminService) GetGroupRateMultipliers(_ context.Context, _ int64) ([]service.UserGroupRateEntry, error) { + return nil, nil +} + func (s *stubAdminService) ListAccounts(ctx context.Context, page, pageSize int, platform, accountType, status, search string, groupID int64) ([]service.Account, int64, error) { return s.accounts, int64(len(s.accounts)), nil } diff --git a/backend/internal/handler/admin/group_handler.go b/backend/internal/handler/admin/group_handler.go index 734acaaa..34c94f2a 100644 --- a/backend/internal/handler/admin/group_handler.go +++ b/backend/internal/handler/admin/group_handler.go @@ -46,9 +46,10 @@ type CreateGroupRequest struct { FallbackGroupID *int64 `json:"fallback_group_id"` FallbackGroupIDOnInvalidRequest *int64 `json:"fallback_group_id_on_invalid_request"` // 模型路由配置（仅 anthropic 平台使用） - ModelRouting map[string][]int64 `json:"model_routing"` - ModelRoutingEnabled bool `json:"model_routing_enabled"` - MCPXMLInject *bool `json:"mcp_xml_inject"` + ModelRouting map[string][]int64 `json:"model_routing"` + ModelRoutingEnabled bool `json:"model_routing_enabled"` + MCPXMLInject *bool `json:"mcp_xml_inject"` + SimulateClaudeMaxEnabled *bool `json:"simulate_claude_max_enabled"` // 支持的模型系列（仅 antigravity 平台使用） SupportedModelScopes []string `json:"supported_model_scopes"` // Sora 存储配额 @@ -84,9 +85,10 @@ type UpdateGroupRequest struct { FallbackGroupID *int64 `json:"fallback_group_id"` FallbackGroupIDOnInvalidRequest *int64 `json:"fallback_group_id_on_invalid_request"` // 模型路由配置（仅 anthropic 平台使用） - ModelRouting map[string][]int64 `json:"model_routing"` - ModelRoutingEnabled *bool `json:"model_routing_enabled"` - MCPXMLInject *bool `json:"mcp_xml_inject"` + ModelRouting map[string][]int64 `json:"model_routing"` + ModelRoutingEnabled *bool `json:"model_routing_enabled"` + MCPXMLInject *bool `json:"mcp_xml_inject"` + SimulateClaudeMaxEnabled *bool `json:"simulate_claude_max_enabled"` // 支持的模型系列（仅 antigravity 平台使用） SupportedModelScopes *[]string `json:"supported_model_scopes"` // Sora 存储配额 @@ -207,6 +209,7 @@ func (h *GroupHandler) Create(c *gin.Context) { ModelRouting: req.ModelRouting, ModelRoutingEnabled: req.ModelRoutingEnabled, MCPXMLInject: req.MCPXMLInject, + SimulateClaudeMaxEnabled: req.SimulateClaudeMaxEnabled, SupportedModelScopes: req.SupportedModelScopes, SoraStorageQuotaBytes: req.SoraStorageQuotaBytes, AllowMessagesDispatch: req.AllowMessagesDispatch, @@ -260,6 +263,7 @@ func (h *GroupHandler) Update(c *gin.Context) { ModelRouting: req.ModelRouting, ModelRoutingEnabled: req.ModelRoutingEnabled, MCPXMLInject: req.MCPXMLInject, + SimulateClaudeMaxEnabled: req.SimulateClaudeMaxEnabled, SupportedModelScopes: req.SupportedModelScopes, SoraStorageQuotaBytes: req.SoraStorageQuotaBytes, AllowMessagesDispatch: req.AllowMessagesDispatch, @@ -335,6 +339,27 @@ func (h *GroupHandler) GetGroupAPIKeys(c *gin.Context) { response.Paginated(c, outKeys, total, page, pageSize) } +// GetGroupRateMultipliers handles getting rate multipliers for users in a group +// GET /api/v1/admin/groups/:id/rate-multipliers +func (h *GroupHandler) GetGroupRateMultipliers(c *gin.Context) { + groupID, err := strconv.ParseInt(c.Param("id"), 10, 64) + if err != nil { + response.BadRequest(c, "Invalid group ID") + return + } + + entries, err := h.adminService.GetGroupRateMultipliers(c.Request.Context(), groupID) + if err != nil { + response.ErrorFrom(c, err) + return + } + + if entries == nil { + entries = []service.UserGroupRateEntry{} + } + response.Success(c, entries) +} + // UpdateSortOrderRequest represents the request to update group sort orders type UpdateSortOrderRequest struct { Updates []struct { diff --git a/backend/internal/handler/dto/mappers.go b/backend/internal/handler/dto/mappers.go index 205ccd65..402bf095 100644 --- a/backend/internal/handler/dto/mappers.go +++ b/backend/internal/handler/dto/mappers.go @@ -135,14 +135,15 @@ func GroupFromServiceAdmin(g *service.Group) *AdminGroup { return nil } out := &AdminGroup{ - Group: groupFromServiceBase(g), - ModelRouting: g.ModelRouting, - ModelRoutingEnabled: g.ModelRoutingEnabled, - MCPXMLInject: g.MCPXMLInject, - DefaultMappedModel: g.DefaultMappedModel, - SupportedModelScopes: g.SupportedModelScopes, - AccountCount: g.AccountCount, - SortOrder: g.SortOrder, + Group: groupFromServiceBase(g), + ModelRouting: g.ModelRouting, + ModelRoutingEnabled: g.ModelRoutingEnabled, + MCPXMLInject: g.MCPXMLInject, + DefaultMappedModel: g.DefaultMappedModel, + SimulateClaudeMaxEnabled: g.SimulateClaudeMaxEnabled, + SupportedModelScopes: g.SupportedModelScopes, + AccountCount: g.AccountCount, + SortOrder: g.SortOrder, } if len(g.AccountGroups) > 0 { out.AccountGroups = make([]AccountGroup, 0, len(g.AccountGroups)) diff --git a/backend/internal/handler/dto/types.go b/backend/internal/handler/dto/types.go index d9ccda2d..7f1788a1 100644 --- a/backend/internal/handler/dto/types.go +++ b/backend/internal/handler/dto/types.go @@ -117,6 +117,8 @@ type AdminGroup struct { // MCP XML 协议注入（仅 antigravity 平台使用） MCPXMLInject bool `json:"mcp_xml_inject"` + // Claude usage 模拟开关（仅管理员可见） + SimulateClaudeMaxEnabled bool `json:"simulate_claude_max_enabled"` // OpenAI Messages 调度配置（仅 openai 平台使用） DefaultMappedModel string `json:"default_mapped_model"` diff --git a/backend/internal/handler/gateway_handler.go b/backend/internal/handler/gateway_handler.go index 4441cf07..743624a2 100644 --- a/backend/internal/handler/gateway_handler.go +++ b/backend/internal/handler/gateway_handler.go @@ -439,6 +439,7 @@ func (h *GatewayHandler) Messages(c *gin.Context) { h.submitUsageRecordTask(func(ctx context.Context) { if err := h.gatewayService.RecordUsage(ctx, &service.RecordUsageInput{ Result: result, + ParsedRequest: parsedReq, APIKey: apiKey, User: apiKey.User, Account: account, @@ -630,6 +631,7 @@ func (h *GatewayHandler) Messages(c *gin.Context) { // ===== 用户消息串行队列 END ===== // 转发请求 - 根据账号平台分流 + c.Set("parsed_request", parsedReq) var result *service.ForwardResult requestCtx := c.Request.Context() if fs.SwitchCount > 0 { @@ -741,6 +743,7 @@ func (h *GatewayHandler) Messages(c *gin.Context) { h.submitUsageRecordTask(func(ctx context.Context) { if err := h.gatewayService.RecordUsage(ctx, &service.RecordUsageInput{ Result: result, + ParsedRequest: parsedReq, APIKey: currentAPIKey, User: currentAPIKey.User, Account: account, diff --git a/backend/internal/pkg/antigravity/stream_transformer.go b/backend/internal/pkg/antigravity/stream_transformer.go index deed5f92..ee600c8b 100644 --- a/backend/internal/pkg/antigravity/stream_transformer.go +++ b/backend/internal/pkg/antigravity/stream_transformer.go @@ -18,6 +18,9 @@ const ( BlockTypeFunction ) +// UsageMapHook is a callback that can modify usage data before it's emitted in SSE events. +type UsageMapHook func(usageMap map[string]any) + // StreamingProcessor 流式响应处理器 type StreamingProcessor struct { blockType BlockType @@ -30,6 +33,7 @@ type StreamingProcessor struct { originalModel string webSearchQueries []string groundingChunks []GeminiGroundingChunk + usageMapHook UsageMapHook // 累计 usage inputTokens int @@ -45,6 +49,25 @@ func NewStreamingProcessor(originalModel string) *StreamingProcessor { } } +// SetUsageMapHook sets an optional hook that modifies usage maps before they are emitted. +func (p *StreamingProcessor) SetUsageMapHook(fn UsageMapHook) { + p.usageMapHook = fn +} + +func usageToMap(u ClaudeUsage) map[string]any { + m := map[string]any{ + "input_tokens": u.InputTokens, + "output_tokens": u.OutputTokens, + } + if u.CacheCreationInputTokens > 0 { + m["cache_creation_input_tokens"] = u.CacheCreationInputTokens + } + if u.CacheReadInputTokens > 0 { + m["cache_read_input_tokens"] = u.CacheReadInputTokens + } + return m +} + // ProcessLine 处理 SSE 行，返回 Claude SSE 事件 func (p *StreamingProcessor) ProcessLine(line string) []byte { line = strings.TrimSpace(line) @@ -168,6 +191,13 @@ func (p *StreamingProcessor) emitMessageStart(v1Resp *V1InternalResponse) []byte responseID = "msg_" + generateRandomID() } + var usageValue any = usage + if p.usageMapHook != nil { + usageMap := usageToMap(usage) + p.usageMapHook(usageMap) + usageValue = usageMap + } + message := map[string]any{ "id": responseID, "type": "message", @@ -176,7 +206,7 @@ func (p *StreamingProcessor) emitMessageStart(v1Resp *V1InternalResponse) []byte "model": p.originalModel, "stop_reason": nil, "stop_sequence": nil, - "usage": usage, + "usage": usageValue, } event := map[string]any{ @@ -487,13 +517,20 @@ func (p *StreamingProcessor) emitFinish(finishReason string) []byte { CacheReadInputTokens: p.cacheReadTokens, } + var usageValue any = usage + if p.usageMapHook != nil { + usageMap := usageToMap(usage) + p.usageMapHook(usageMap) + usageValue = usageMap + } + deltaEvent := map[string]any{ "type": "message_delta", "delta": map[string]any{ "stop_reason": stopReason, "stop_sequence": nil, }, - "usage": usage, + "usage": usageValue, } _, _ = result.Write(p.formatSSE("message_delta", deltaEvent)) diff --git a/backend/internal/repository/api_key_repo.go b/backend/internal/repository/api_key_repo.go index 4c7f38a8..a45a83a3 100644 --- a/backend/internal/repository/api_key_repo.go +++ b/backend/internal/repository/api_key_repo.go @@ -164,6 +164,7 @@ func (r *apiKeyRepository) GetByKeyForAuth(ctx context.Context, key string) (*se group.FieldModelRoutingEnabled, group.FieldModelRouting, group.FieldMcpXMLInject, + group.FieldSimulateClaudeMaxEnabled, group.FieldSupportedModelScopes, group.FieldAllowMessagesDispatch, group.FieldDefaultMappedModel, @@ -645,6 +646,7 @@ func groupEntityToService(g *dbent.Group) *service.Group { ModelRouting: g.ModelRouting, ModelRoutingEnabled: g.ModelRoutingEnabled, MCPXMLInject: g.McpXMLInject, + SimulateClaudeMaxEnabled: g.SimulateClaudeMaxEnabled, SupportedModelScopes: g.SupportedModelScopes, SortOrder: g.SortOrder, AllowMessagesDispatch: g.AllowMessagesDispatch, diff --git a/backend/internal/repository/group_repo.go b/backend/internal/repository/group_repo.go index c195f1f1..27d68354 100644 --- a/backend/internal/repository/group_repo.go +++ b/backend/internal/repository/group_repo.go @@ -61,7 +61,8 @@ func (r *groupRepository) Create(ctx context.Context, groupIn *service.Group) er SetMcpXMLInject(groupIn.MCPXMLInject). SetSoraStorageQuotaBytes(groupIn.SoraStorageQuotaBytes). SetAllowMessagesDispatch(groupIn.AllowMessagesDispatch). - SetDefaultMappedModel(groupIn.DefaultMappedModel) + SetDefaultMappedModel(groupIn.DefaultMappedModel). + SetSimulateClaudeMaxEnabled(groupIn.SimulateClaudeMaxEnabled) // 设置模型路由配置 if groupIn.ModelRouting != nil { @@ -129,7 +130,8 @@ func (r *groupRepository) Update(ctx context.Context, groupIn *service.Group) er SetMcpXMLInject(groupIn.MCPXMLInject). SetSoraStorageQuotaBytes(groupIn.SoraStorageQuotaBytes). SetAllowMessagesDispatch(groupIn.AllowMessagesDispatch). - SetDefaultMappedModel(groupIn.DefaultMappedModel) + SetDefaultMappedModel(groupIn.DefaultMappedModel). + SetSimulateClaudeMaxEnabled(groupIn.SimulateClaudeMaxEnabled) // 显式处理可空字段：nil 需要 clear，非 nil 需要 set。 if groupIn.DailyLimitUSD != nil { diff --git a/backend/internal/repository/usage_log_repo.go b/backend/internal/repository/usage_log_repo.go index c91a68e5..b9207f34 100644 --- a/backend/internal/repository/usage_log_repo.go +++ b/backend/internal/repository/usage_log_repo.go @@ -1873,7 +1873,7 @@ func (r *usageLogRepository) GetGroupStatsWithFilters(ctx context.Context, start query := ` SELECT COALESCE(ul.group_id, 0) as group_id, - COALESCE(g.name, '') as group_name, + COALESCE(g.name, '(无分组)') as group_name, COUNT(*) as requests, COALESCE(SUM(ul.input_tokens + ul.output_tokens + ul.cache_creation_tokens + ul.cache_read_tokens), 0) as total_tokens, COALESCE(SUM(ul.total_cost), 0) as cost, diff --git a/backend/internal/repository/user_group_rate_repo.go b/backend/internal/repository/user_group_rate_repo.go index e3b11096..b57a3bb9 100644 --- a/backend/internal/repository/user_group_rate_repo.go +++ b/backend/internal/repository/user_group_rate_repo.go @@ -95,6 +95,35 @@ func (r *userGroupRateRepository) GetByUserIDs(ctx context.Context, userIDs []in return result, nil } +// GetByGroupID 获取指定分组下所有用户的专属倍率 +func (r *userGroupRateRepository) GetByGroupID(ctx context.Context, groupID int64) ([]service.UserGroupRateEntry, error) { + query := ` + SELECT ugr.user_id, u.email, ugr.rate_multiplier + FROM user_group_rate_multipliers ugr + JOIN users u ON u.id = ugr.user_id + WHERE ugr.group_id = $1 + ORDER BY ugr.user_id + ` + rows, err := r.sql.QueryContext(ctx, query, groupID) + if err != nil { + return nil, err + } + defer func() { _ = rows.Close() }() + + var result []service.UserGroupRateEntry + for rows.Next() { + var entry service.UserGroupRateEntry + if err := rows.Scan(&entry.UserID, &entry.UserEmail, &entry.RateMultiplier); err != nil { + return nil, err + } + result = append(result, entry) + } + if err := rows.Err(); err != nil { + return nil, err + } + return result, nil +} + // GetByUserAndGroup 获取用户在特定分组的专属倍率 func (r *userGroupRateRepository) GetByUserAndGroup(ctx context.Context, userID, groupID int64) (*float64, error) { query := `SELECT rate_multiplier FROM user_group_rate_multipliers WHERE user_id = $1 AND group_id = $2` diff --git a/backend/internal/server/routes/admin.go b/backend/internal/server/routes/admin.go index 9fdb233b..dc5e8269 100644 --- a/backend/internal/server/routes/admin.go +++ b/backend/internal/server/routes/admin.go @@ -228,6 +228,7 @@ func registerGroupRoutes(admin *gin.RouterGroup, h *handler.Handlers) { groups.PUT("/:id", h.Admin.Group.Update) groups.DELETE("/:id", h.Admin.Group.Delete) groups.GET("/:id/stats", h.Admin.Group.GetStats) + groups.GET("/:id/rate-multipliers", h.Admin.Group.GetGroupRateMultipliers) groups.GET("/:id/api-keys", h.Admin.Group.GetGroupAPIKeys) } } diff --git a/backend/internal/service/admin_service.go b/backend/internal/service/admin_service.go index dec4ed33..f2e7bd9b 100644 --- a/backend/internal/service/admin_service.go +++ b/backend/internal/service/admin_service.go @@ -42,6 +42,7 @@ type AdminService interface { UpdateGroup(ctx context.Context, id int64, input *UpdateGroupInput) (*Group, error) DeleteGroup(ctx context.Context, id int64) error GetGroupAPIKeys(ctx context.Context, groupID int64, page, pageSize int) ([]APIKey, int64, error) + GetGroupRateMultipliers(ctx context.Context, groupID int64) ([]UserGroupRateEntry, error) UpdateGroupSortOrders(ctx context.Context, updates []GroupSortOrderUpdate) error // API Key management (admin) @@ -138,9 +139,10 @@ type CreateGroupInput struct { // 无效请求兜底分组 ID（仅 anthropic 平台使用） FallbackGroupIDOnInvalidRequest *int64 // 模型路由配置（仅 anthropic 平台使用） - ModelRouting map[string][]int64 - ModelRoutingEnabled bool // 是否启用模型路由 - MCPXMLInject *bool + ModelRouting map[string][]int64 + ModelRoutingEnabled bool // 是否启用模型路由 + MCPXMLInject *bool + SimulateClaudeMaxEnabled *bool // 支持的模型系列（仅 antigravity 平台使用） SupportedModelScopes []string // Sora 存储配额 @@ -177,9 +179,10 @@ type UpdateGroupInput struct { // 无效请求兜底分组 ID（仅 anthropic 平台使用） FallbackGroupIDOnInvalidRequest *int64 // 模型路由配置（仅 anthropic 平台使用） - ModelRouting map[string][]int64 - ModelRoutingEnabled *bool // 是否启用模型路由 - MCPXMLInject *bool + ModelRouting map[string][]int64 + ModelRoutingEnabled *bool // 是否启用模型路由 + MCPXMLInject *bool + SimulateClaudeMaxEnabled *bool // 支持的模型系列（仅 antigravity 平台使用） SupportedModelScopes *[]string // Sora 存储配额 @@ -363,6 +366,10 @@ type ProxyExitInfoProber interface { ProbeProxy(ctx context.Context, proxyURL string) (*ProxyExitInfo, int64, error) } +type groupExistenceBatchReader interface { + ExistsByIDs(ctx context.Context, ids []int64) (map[int64]bool, error) +} + type proxyQualityTarget struct { Target string URL string @@ -439,10 +446,6 @@ type userGroupRateBatchReader interface { GetByUserIDs(ctx context.Context, userIDs []int64) (map[int64]map[int64]float64, error) } -type groupExistenceBatchReader interface { - ExistsByIDs(ctx context.Context, ids []int64) (map[int64]bool, error) -} - // NewAdminService creates a new AdminService func NewAdminService( userRepo UserRepository, @@ -860,6 +863,13 @@ func (s *adminServiceImpl) CreateGroup(ctx context.Context, input *CreateGroupIn if input.MCPXMLInject != nil { mcpXMLInject = *input.MCPXMLInject } + simulateClaudeMaxEnabled := false + if input.SimulateClaudeMaxEnabled != nil { + if platform != PlatformAnthropic && *input.SimulateClaudeMaxEnabled { + return nil, fmt.Errorf("simulate_claude_max_enabled only supported for anthropic groups") + } + simulateClaudeMaxEnabled = *input.SimulateClaudeMaxEnabled + } // 如果指定了复制账号的源分组，先获取账号 ID 列表 var accountIDsToCopy []int64 @@ -916,6 +926,7 @@ func (s *adminServiceImpl) CreateGroup(ctx context.Context, input *CreateGroupIn FallbackGroupIDOnInvalidRequest: fallbackOnInvalidRequest, ModelRouting: input.ModelRouting, MCPXMLInject: mcpXMLInject, + SimulateClaudeMaxEnabled: simulateClaudeMaxEnabled, SupportedModelScopes: input.SupportedModelScopes, SoraStorageQuotaBytes: input.SoraStorageQuotaBytes, AllowMessagesDispatch: input.AllowMessagesDispatch, @@ -1127,6 +1138,15 @@ func (s *adminServiceImpl) UpdateGroup(ctx context.Context, id int64, input *Upd if input.MCPXMLInject != nil { group.MCPXMLInject = *input.MCPXMLInject } + if input.SimulateClaudeMaxEnabled != nil { + if group.Platform != PlatformAnthropic && *input.SimulateClaudeMaxEnabled { + return nil, fmt.Errorf("simulate_claude_max_enabled only supported for anthropic groups") + } + group.SimulateClaudeMaxEnabled = *input.SimulateClaudeMaxEnabled + } + if group.Platform != PlatformAnthropic { + group.SimulateClaudeMaxEnabled = false + } // 支持的模型系列（仅 antigravity 平台使用） if input.SupportedModelScopes != nil { @@ -1244,6 +1264,13 @@ func (s *adminServiceImpl) GetGroupAPIKeys(ctx context.Context, groupID int64, p return keys, result.Total, nil } +func (s *adminServiceImpl) GetGroupRateMultipliers(ctx context.Context, groupID int64) ([]UserGroupRateEntry, error) { + if s.userGroupRateRepo == nil { + return nil, nil + } + return s.userGroupRateRepo.GetByGroupID(ctx, groupID) +} + func (s *adminServiceImpl) UpdateGroupSortOrders(ctx context.Context, updates []GroupSortOrderUpdate) error { return s.groupRepo.UpdateSortOrders(ctx, updates) } diff --git a/backend/internal/service/admin_service_bulk_update_test.go b/backend/internal/service/admin_service_bulk_update_test.go index 4845d87c..e90ec93a 100644 --- a/backend/internal/service/admin_service_bulk_update_test.go +++ b/backend/internal/service/admin_service_bulk_update_test.go @@ -43,6 +43,16 @@ func (s *accountRepoStubForBulkUpdate) BindGroups(_ context.Context, accountID i return nil } +func (s *accountRepoStubForBulkUpdate) ListByGroup(_ context.Context, groupID int64) ([]Account, error) { + if err, ok := s.listByGroupErr[groupID]; ok { + return nil, err + } + if rows, ok := s.listByGroupData[groupID]; ok { + return rows, nil + } + return nil, nil +} + func (s *accountRepoStubForBulkUpdate) GetByIDs(_ context.Context, ids []int64) ([]*Account, error) { s.getByIDsCalled = true s.getByIDsIDs = append([]int64{}, ids...) @@ -63,16 +73,6 @@ func (s *accountRepoStubForBulkUpdate) GetByID(_ context.Context, id int64) (*Ac return nil, errors.New("account not found") } -func (s *accountRepoStubForBulkUpdate) ListByGroup(_ context.Context, groupID int64) ([]Account, error) { - if err, ok := s.listByGroupErr[groupID]; ok { - return nil, err - } - if rows, ok := s.listByGroupData[groupID]; ok { - return rows, nil - } - return nil, nil -} - // TestAdminService_BulkUpdateAccounts_AllSuccessIDs 验证批量更新成功时返回 success_ids/failed_ids。 func TestAdminService_BulkUpdateAccounts_AllSuccessIDs(t *testing.T) { repo := &accountRepoStubForBulkUpdate{} diff --git a/backend/internal/service/admin_service_group_test.go b/backend/internal/service/admin_service_group_test.go index ef77a980..0e6fe084 100644 --- a/backend/internal/service/admin_service_group_test.go +++ b/backend/internal/service/admin_service_group_test.go @@ -785,3 +785,57 @@ func TestAdminService_UpdateGroup_InvalidRequestFallbackAllowsAntigravity(t *tes require.NotNil(t, repo.updated) require.Equal(t, fallbackID, *repo.updated.FallbackGroupIDOnInvalidRequest) } + +func TestAdminService_CreateGroup_SimulateClaudeMaxRequiresAnthropic(t *testing.T) { + repo := &groupRepoStubForAdmin{} + svc := &adminServiceImpl{groupRepo: repo} + + enabled := true + _, err := svc.CreateGroup(context.Background(), &CreateGroupInput{ + Name: "openai-group", + Platform: PlatformOpenAI, + SimulateClaudeMaxEnabled: &enabled, + }) + require.Error(t, err) + require.Contains(t, err.Error(), "simulate_claude_max_enabled only supported for anthropic groups") + require.Nil(t, repo.created) +} + +func TestAdminService_UpdateGroup_SimulateClaudeMaxRequiresAnthropic(t *testing.T) { + existingGroup := &Group{ + ID: 1, + Name: "openai-group", + Platform: PlatformOpenAI, + Status: StatusActive, + } + repo := &groupRepoStubForAdmin{getByID: existingGroup} + svc := &adminServiceImpl{groupRepo: repo} + + enabled := true + _, err := svc.UpdateGroup(context.Background(), 1, &UpdateGroupInput{ + SimulateClaudeMaxEnabled: &enabled, + }) + require.Error(t, err) + require.Contains(t, err.Error(), "simulate_claude_max_enabled only supported for anthropic groups") + require.Nil(t, repo.updated) +} + +func TestAdminService_UpdateGroup_ClearsSimulateClaudeMaxWhenPlatformChanges(t *testing.T) { + existingGroup := &Group{ + ID: 1, + Name: "anthropic-group", + Platform: PlatformAnthropic, + Status: StatusActive, + SimulateClaudeMaxEnabled: true, + } + repo := &groupRepoStubForAdmin{getByID: existingGroup} + svc := &adminServiceImpl{groupRepo: repo} + + group, err := svc.UpdateGroup(context.Background(), 1, &UpdateGroupInput{ + Platform: PlatformOpenAI, + }) + require.NoError(t, err) + require.NotNil(t, group) + require.NotNil(t, repo.updated) + require.False(t, repo.updated.SimulateClaudeMaxEnabled) +} diff --git a/backend/internal/service/admin_service_list_users_test.go b/backend/internal/service/admin_service_list_users_test.go index 8b50530a..579fa981 100644 --- a/backend/internal/service/admin_service_list_users_test.go +++ b/backend/internal/service/admin_service_list_users_test.go @@ -68,6 +68,10 @@ func (s *userGroupRateRepoStubForListUsers) SyncUserGroupRates(_ context.Context panic("unexpected SyncUserGroupRates call") } +func (s *userGroupRateRepoStubForListUsers) GetByGroupID(_ context.Context, groupID int64) ([]UserGroupRateEntry, error) { + panic("unexpected GetByGroupID call") +} + func (s *userGroupRateRepoStubForListUsers) DeleteByGroupID(_ context.Context, groupID int64) error { panic("unexpected DeleteByGroupID call") } diff --git a/backend/internal/service/antigravity_gateway_service.go b/backend/internal/service/antigravity_gateway_service.go index f63802b8..bfca7a82 100644 --- a/backend/internal/service/antigravity_gateway_service.go +++ b/backend/internal/service/antigravity_gateway_service.go @@ -1673,7 +1673,7 @@ func (s *AntigravityGatewayService) Forward(ctx context.Context, c *gin.Context, var clientDisconnect bool if claudeReq.Stream { // 客户端要求流式，直接透传转换 - streamRes, err := s.handleClaudeStreamingResponse(c, resp, startTime, originalModel) + streamRes, err := s.handleClaudeStreamingResponse(c, resp, startTime, originalModel, account.ID) if err != nil { logger.LegacyPrintf("service.antigravity_gateway", "%s status=stream_error error=%v", prefix, err) return nil, err @@ -1683,7 +1683,7 @@ func (s *AntigravityGatewayService) Forward(ctx context.Context, c *gin.Context, clientDisconnect = streamRes.clientDisconnect } else { // 客户端要求非流式，收集流式响应后转换返回 - streamRes, err := s.handleClaudeStreamToNonStreaming(c, resp, startTime, originalModel) + streamRes, err := s.handleClaudeStreamToNonStreaming(c, resp, startTime, originalModel, account.ID) if err != nil { logger.LegacyPrintf("service.antigravity_gateway", "%s status=stream_collect_error error=%v", prefix, err) return nil, err @@ -1692,6 +1692,9 @@ func (s *AntigravityGatewayService) Forward(ctx context.Context, c *gin.Context, firstTokenMs = streamRes.firstTokenMs } + // Claude Max cache billing: 同步 ForwardResult.Usage 与客户端响应体一致 + applyClaudeMaxCacheBillingPolicyToUsage(usage, parsedRequestFromGinContext(c), claudeMaxGroupFromGinContext(c), originalModel, account.ID) + return &ForwardResult{ RequestID: requestID, Usage: *usage, @@ -3595,7 +3598,7 @@ func (s *AntigravityGatewayService) writeGoogleError(c *gin.Context, status int, // handleClaudeStreamToNonStreaming 收集上游流式响应，转换为 Claude 非流式格式返回 // 用于处理客户端非流式请求但上游只支持流式的情况 -func (s *AntigravityGatewayService) handleClaudeStreamToNonStreaming(c *gin.Context, resp *http.Response, startTime time.Time, originalModel string) (*antigravityStreamResult, error) { +func (s *AntigravityGatewayService) handleClaudeStreamToNonStreaming(c *gin.Context, resp *http.Response, startTime time.Time, originalModel string, accountID int64) (*antigravityStreamResult, error) { scanner := bufio.NewScanner(resp.Body) maxLineSize := defaultMaxLineSize if s.settingService.cfg != nil && s.settingService.cfg.Gateway.MaxLineSize > 0 { @@ -3753,6 +3756,9 @@ returnResponse: return nil, s.writeClaudeError(c, http.StatusBadGateway, "upstream_error", "Failed to parse upstream response") } + // Claude Max cache billing simulation (non-streaming) + claudeResp = applyClaudeMaxNonStreamingRewrite(c, claudeResp, agUsage, originalModel, accountID) + c.Data(http.StatusOK, "application/json", claudeResp) // 转换为 service.ClaudeUsage @@ -3767,7 +3773,7 @@ returnResponse: } // handleClaudeStreamingResponse 处理 Claude 流式响应（Gemini SSE → Claude SSE 转换） -func (s *AntigravityGatewayService) handleClaudeStreamingResponse(c *gin.Context, resp *http.Response, startTime time.Time, originalModel string) (*antigravityStreamResult, error) { +func (s *AntigravityGatewayService) handleClaudeStreamingResponse(c *gin.Context, resp *http.Response, startTime time.Time, originalModel string, accountID int64) (*antigravityStreamResult, error) { c.Header("Content-Type", "text/event-stream") c.Header("Cache-Control", "no-cache") c.Header("Connection", "keep-alive") @@ -3780,6 +3786,8 @@ func (s *AntigravityGatewayService) handleClaudeStreamingResponse(c *gin.Context } processor := antigravity.NewStreamingProcessor(originalModel) + setupClaudeMaxStreamingHook(c, processor, originalModel, accountID) + var firstTokenMs *int // 使用 Scanner 并限制单行大小，避免 ReadString 无上限导致 OOM scanner := bufio.NewScanner(resp.Body) diff --git a/backend/internal/service/antigravity_gateway_service_test.go b/backend/internal/service/antigravity_gateway_service_test.go index 6e0a7305..b2e2fc38 100644 --- a/backend/internal/service/antigravity_gateway_service_test.go +++ b/backend/internal/service/antigravity_gateway_service_test.go @@ -922,7 +922,7 @@ func TestHandleClaudeStreamingResponse_NormalComplete(t *testing.T) { fmt.Fprintln(pw, "") }() - result, err := svc.handleClaudeStreamingResponse(c, resp, time.Now(), "claude-sonnet-4-5") + result, err := svc.handleClaudeStreamingResponse(c, resp, time.Now(), "claude-sonnet-4-5", 0) _ = pr.Close() require.NoError(t, err) @@ -999,7 +999,7 @@ func TestHandleClaudeStreamingResponse_ThoughtsTokenCount(t *testing.T) { fmt.Fprintln(pw, "") }() - result, err := svc.handleClaudeStreamingResponse(c, resp, time.Now(), "gemini-2.5-pro") + result, err := svc.handleClaudeStreamingResponse(c, resp, time.Now(), "gemini-2.5-pro", 0) _ = pr.Close() require.NoError(t, err) @@ -1202,7 +1202,7 @@ func TestHandleClaudeStreamingResponse_ClientDisconnect(t *testing.T) { fmt.Fprintln(pw, "") }() - result, err := svc.handleClaudeStreamingResponse(c, resp, time.Now(), "claude-sonnet-4-5") + result, err := svc.handleClaudeStreamingResponse(c, resp, time.Now(), "claude-sonnet-4-5", 0) _ = pr.Close() require.NoError(t, err) @@ -1234,7 +1234,7 @@ func TestHandleClaudeStreamingResponse_EmptyStream(t *testing.T) { fmt.Fprintln(pw, "") }() - _, err := svc.handleClaudeStreamingResponse(c, resp, time.Now(), "claude-sonnet-4-5") + _, err := svc.handleClaudeStreamingResponse(c, resp, time.Now(), "claude-sonnet-4-5", 0) _ = pr.Close() // 应当返回 UpstreamFailoverError 而非 nil，以便上层触发 failover @@ -1266,7 +1266,7 @@ func TestHandleClaudeStreamingResponse_ContextCanceled(t *testing.T) { resp := &http.Response{StatusCode: http.StatusOK, Body: cancelReadCloser{}, Header: http.Header{}} - result, err := svc.handleClaudeStreamingResponse(c, resp, time.Now(), "claude-sonnet-4-5") + result, err := svc.handleClaudeStreamingResponse(c, resp, time.Now(), "claude-sonnet-4-5", 0) require.NoError(t, err) require.NotNil(t, result) diff --git a/backend/internal/service/api_key_auth_cache.go b/backend/internal/service/api_key_auth_cache.go index e8ad5c9c..258b842b 100644 --- a/backend/internal/service/api_key_auth_cache.go +++ b/backend/internal/service/api_key_auth_cache.go @@ -59,9 +59,10 @@ type APIKeyAuthGroupSnapshot struct { // Model routing is used by gateway account selection, so it must be part of auth cache snapshot. // Only anthropic groups use these fields; others may leave them empty. - ModelRouting map[string][]int64 `json:"model_routing,omitempty"` - ModelRoutingEnabled bool `json:"model_routing_enabled"` - MCPXMLInject bool `json:"mcp_xml_inject"` + ModelRouting map[string][]int64 `json:"model_routing,omitempty"` + ModelRoutingEnabled bool `json:"model_routing_enabled"` + MCPXMLInject bool `json:"mcp_xml_inject"` + SimulateClaudeMaxEnabled bool `json:"simulate_claude_max_enabled"` // 支持的模型系列（仅 antigravity 平台使用） SupportedModelScopes []string `json:"supported_model_scopes,omitempty"` diff --git a/backend/internal/service/api_key_auth_cache_impl.go b/backend/internal/service/api_key_auth_cache_impl.go index f727ab10..d874ccf2 100644 --- a/backend/internal/service/api_key_auth_cache_impl.go +++ b/backend/internal/service/api_key_auth_cache_impl.go @@ -244,6 +244,7 @@ func (s *APIKeyService) snapshotFromAPIKey(apiKey *APIKey) *APIKeyAuthSnapshot { ModelRouting: apiKey.Group.ModelRouting, ModelRoutingEnabled: apiKey.Group.ModelRoutingEnabled, MCPXMLInject: apiKey.Group.MCPXMLInject, + SimulateClaudeMaxEnabled: apiKey.Group.SimulateClaudeMaxEnabled, SupportedModelScopes: apiKey.Group.SupportedModelScopes, AllowMessagesDispatch: apiKey.Group.AllowMessagesDispatch, DefaultMappedModel: apiKey.Group.DefaultMappedModel, @@ -303,6 +304,7 @@ func (s *APIKeyService) snapshotToAPIKey(key string, snapshot *APIKeyAuthSnapsho ModelRouting: snapshot.Group.ModelRouting, ModelRoutingEnabled: snapshot.Group.ModelRoutingEnabled, MCPXMLInject: snapshot.Group.MCPXMLInject, + SimulateClaudeMaxEnabled: snapshot.Group.SimulateClaudeMaxEnabled, SupportedModelScopes: snapshot.Group.SupportedModelScopes, AllowMessagesDispatch: snapshot.Group.AllowMessagesDispatch, DefaultMappedModel: snapshot.Group.DefaultMappedModel, diff --git a/backend/internal/service/claude_max_cache_billing_policy.go b/backend/internal/service/claude_max_cache_billing_policy.go new file mode 100644 index 00000000..2381915e --- /dev/null +++ b/backend/internal/service/claude_max_cache_billing_policy.go @@ -0,0 +1,450 @@ +package service + +import ( + "encoding/json" + "strings" + + "github.com/Wei-Shaw/sub2api/internal/pkg/claude" + "github.com/Wei-Shaw/sub2api/internal/pkg/logger" + "github.com/tidwall/gjson" +) + +type claudeMaxCacheBillingOutcome struct { + Simulated bool +} + +func applyClaudeMaxCacheBillingPolicyToUsage(usage *ClaudeUsage, parsed *ParsedRequest, group *Group, model string, accountID int64) claudeMaxCacheBillingOutcome { + var out claudeMaxCacheBillingOutcome + if usage == nil || !shouldApplyClaudeMaxBillingRulesForUsage(group, model, parsed) { + return out + } + + resolvedModel := strings.TrimSpace(model) + if resolvedModel == "" && parsed != nil { + resolvedModel = strings.TrimSpace(parsed.Model) + } + + if hasCacheCreationTokens(*usage) { + // Upstream already returned cache creation usage; keep original usage. + return out + } + + if !shouldSimulateClaudeMaxUsageForUsage(*usage, parsed) { + return out + } + beforeInputTokens := usage.InputTokens + out.Simulated = safelyProjectUsageToClaudeMax1H(usage, parsed) + if out.Simulated { + logger.LegacyPrintf("service.gateway", "simulate_claude_max_usage: model=%s account=%d input_tokens:%d->%d cache_creation_1h=%d", + resolvedModel, + accountID, + beforeInputTokens, + usage.InputTokens, + usage.CacheCreation1hTokens, + ) + } + return out +} + +func isClaudeFamilyModel(model string) bool { + normalized := strings.ToLower(strings.TrimSpace(claude.NormalizeModelID(model))) + if normalized == "" { + return false + } + return strings.Contains(normalized, "claude-") +} + +func shouldApplyClaudeMaxBillingRules(input *RecordUsageInput) bool { + if input == nil || input.Result == nil || input.APIKey == nil || input.APIKey.Group == nil { + return false + } + return shouldApplyClaudeMaxBillingRulesForUsage(input.APIKey.Group, input.Result.Model, input.ParsedRequest) +} + +func shouldApplyClaudeMaxBillingRulesForUsage(group *Group, model string, parsed *ParsedRequest) bool { + if group == nil { + return false + } + if !group.SimulateClaudeMaxEnabled || group.Platform != PlatformAnthropic { + return false + } + + resolvedModel := model + if resolvedModel == "" && parsed != nil { + resolvedModel = parsed.Model + } + if !isClaudeFamilyModel(resolvedModel) { + return false + } + return true +} + +func hasCacheCreationTokens(usage ClaudeUsage) bool { + return usage.CacheCreationInputTokens > 0 || usage.CacheCreation5mTokens > 0 || usage.CacheCreation1hTokens > 0 +} + +func shouldSimulateClaudeMaxUsage(input *RecordUsageInput) bool { + if input == nil || input.Result == nil { + return false + } + if !shouldApplyClaudeMaxBillingRules(input) { + return false + } + return shouldSimulateClaudeMaxUsageForUsage(input.Result.Usage, input.ParsedRequest) +} + +func shouldSimulateClaudeMaxUsageForUsage(usage ClaudeUsage, parsed *ParsedRequest) bool { + if usage.InputTokens <= 0 { + return false + } + if hasCacheCreationTokens(usage) { + return false + } + if !hasClaudeCacheSignals(parsed) { + return false + } + return true +} + +func safelyProjectUsageToClaudeMax1H(usage *ClaudeUsage, parsed *ParsedRequest) (changed bool) { + defer func() { + if r := recover(); r != nil { + logger.LegacyPrintf("service.gateway", "simulate_claude_max_usage skipped: panic=%v", r) + changed = false + } + }() + return projectUsageToClaudeMax1H(usage, parsed) +} + +func projectUsageToClaudeMax1H(usage *ClaudeUsage, parsed *ParsedRequest) bool { + if usage == nil { + return false + } + totalWindowTokens := usage.InputTokens + usage.CacheCreation5mTokens + usage.CacheCreation1hTokens + if totalWindowTokens <= 1 { + return false + } + + simulatedInputTokens := computeClaudeMaxProjectedInputTokens(totalWindowTokens, parsed) + if simulatedInputTokens <= 0 { + simulatedInputTokens = 1 + } + if simulatedInputTokens >= totalWindowTokens { + simulatedInputTokens = totalWindowTokens - 1 + } + + cacheCreation1hTokens := totalWindowTokens - simulatedInputTokens + if usage.InputTokens == simulatedInputTokens && + usage.CacheCreation5mTokens == 0 && + usage.CacheCreation1hTokens == cacheCreation1hTokens && + usage.CacheCreationInputTokens == cacheCreation1hTokens { + return false + } + + usage.InputTokens = simulatedInputTokens + usage.CacheCreation5mTokens = 0 + usage.CacheCreation1hTokens = cacheCreation1hTokens + usage.CacheCreationInputTokens = cacheCreation1hTokens + return true +} + +type claudeCacheProjection struct { + HasBreakpoint bool + BreakpointCount int + TotalEstimatedTokens int + TailEstimatedTokens int +} + +func computeClaudeMaxProjectedInputTokens(totalWindowTokens int, parsed *ParsedRequest) int { + if totalWindowTokens <= 1 { + return totalWindowTokens + } + + projection := analyzeClaudeCacheProjection(parsed) + if !projection.HasBreakpoint || projection.TotalEstimatedTokens <= 0 || projection.TailEstimatedTokens <= 0 { + return totalWindowTokens + } + + totalEstimate := int64(projection.TotalEstimatedTokens) + tailEstimate := int64(projection.TailEstimatedTokens) + if tailEstimate > totalEstimate { + tailEstimate = totalEstimate + } + + scaled := (int64(totalWindowTokens)*tailEstimate + totalEstimate/2) / totalEstimate + if scaled <= 0 { + scaled = 1 + } + if scaled >= int64(totalWindowTokens) { + scaled = int64(totalWindowTokens - 1) + } + return int(scaled) +} + +func hasClaudeCacheSignals(parsed *ParsedRequest) bool { + if parsed == nil { + return false + } + if hasTopLevelEphemeralCacheControl(parsed) { + return true + } + return countExplicitCacheBreakpoints(parsed) > 0 +} + +func hasTopLevelEphemeralCacheControl(parsed *ParsedRequest) bool { + if parsed == nil || len(parsed.Body) == 0 { + return false + } + cacheType := strings.TrimSpace(gjson.GetBytes(parsed.Body, "cache_control.type").String()) + return strings.EqualFold(cacheType, "ephemeral") +} + +func analyzeClaudeCacheProjection(parsed *ParsedRequest) claudeCacheProjection { + var projection claudeCacheProjection + if parsed == nil { + return projection + } + + total := 0 + lastBreakpointAt := -1 + + switch system := parsed.System.(type) { + case string: + total += claudeMaxMessageOverheadTokens + estimateClaudeTextTokens(system) + case []any: + for _, raw := range system { + block, ok := raw.(map[string]any) + if !ok { + total += claudeMaxUnknownContentTokens + continue + } + total += estimateClaudeBlockTokens(block) + if hasEphemeralCacheControl(block) { + lastBreakpointAt = total + projection.BreakpointCount++ + projection.HasBreakpoint = true + } + } + } + + for _, rawMsg := range parsed.Messages { + total += claudeMaxMessageOverheadTokens + msg, ok := rawMsg.(map[string]any) + if !ok { + total += claudeMaxUnknownContentTokens + continue + } + content, exists := msg["content"] + if !exists { + continue + } + msgTokens, msgLastBreak, msgBreakCount := estimateClaudeContentTokens(content) + total += msgTokens + if msgBreakCount > 0 { + lastBreakpointAt = total - msgTokens + msgLastBreak + projection.BreakpointCount += msgBreakCount + projection.HasBreakpoint = true + } + } + + if total <= 0 { + total = 1 + } + projection.TotalEstimatedTokens = total + + if projection.HasBreakpoint && lastBreakpointAt >= 0 { + tail := total - lastBreakpointAt + if tail <= 0 { + tail = 1 + } + projection.TailEstimatedTokens = tail + return projection + } + + if hasTopLevelEphemeralCacheControl(parsed) { + tail := estimateLastUserMessageTokens(parsed) + if tail <= 0 { + tail = 1 + } + projection.HasBreakpoint = true + projection.BreakpointCount = 1 + projection.TailEstimatedTokens = tail + } + return projection +} + +func countExplicitCacheBreakpoints(parsed *ParsedRequest) int { + if parsed == nil { + return 0 + } + total := 0 + if system, ok := parsed.System.([]any); ok { + for _, raw := range system { + if block, ok := raw.(map[string]any); ok && hasEphemeralCacheControl(block) { + total++ + } + } + } + for _, rawMsg := range parsed.Messages { + msg, ok := rawMsg.(map[string]any) + if !ok { + continue + } + content, ok := msg["content"].([]any) + if !ok { + continue + } + for _, raw := range content { + if block, ok := raw.(map[string]any); ok && hasEphemeralCacheControl(block) { + total++ + } + } + } + return total +} + +func hasEphemeralCacheControl(block map[string]any) bool { + if block == nil { + return false + } + raw, ok := block["cache_control"] + if !ok || raw == nil { + return false + } + switch cc := raw.(type) { + case map[string]any: + cacheType, _ := cc["type"].(string) + return strings.EqualFold(strings.TrimSpace(cacheType), "ephemeral") + case map[string]string: + return strings.EqualFold(strings.TrimSpace(cc["type"]), "ephemeral") + default: + return false + } +} + +func estimateClaudeContentTokens(content any) (tokens int, lastBreakAt int, breakpointCount int) { + switch value := content.(type) { + case string: + return estimateClaudeTextTokens(value), -1, 0 + case []any: + total := 0 + lastBreak := -1 + breaks := 0 + for _, raw := range value { + block, ok := raw.(map[string]any) + if !ok { + total += claudeMaxUnknownContentTokens + continue + } + total += estimateClaudeBlockTokens(block) + if hasEphemeralCacheControl(block) { + lastBreak = total + breaks++ + } + } + return total, lastBreak, breaks + default: + return estimateStructuredTokens(value), -1, 0 + } +} + +func estimateClaudeBlockTokens(block map[string]any) int { + if block == nil { + return claudeMaxUnknownContentTokens + } + tokens := claudeMaxBlockOverheadTokens + blockType, _ := block["type"].(string) + switch blockType { + case "text": + if text, ok := block["text"].(string); ok { + tokens += estimateClaudeTextTokens(text) + } + case "tool_result": + if content, ok := block["content"]; ok { + nested, _, _ := estimateClaudeContentTokens(content) + tokens += nested + } + case "tool_use": + if name, ok := block["name"].(string); ok { + tokens += estimateClaudeTextTokens(name) + } + if input, ok := block["input"]; ok { + tokens += estimateStructuredTokens(input) + } + default: + if text, ok := block["text"].(string); ok { + tokens += estimateClaudeTextTokens(text) + } else if content, ok := block["content"]; ok { + nested, _, _ := estimateClaudeContentTokens(content) + tokens += nested + } + } + if tokens <= claudeMaxBlockOverheadTokens { + tokens += claudeMaxUnknownContentTokens + } + return tokens +} + +func estimateLastUserMessageTokens(parsed *ParsedRequest) int { + if parsed == nil || len(parsed.Messages) == 0 { + return 0 + } + for i := len(parsed.Messages) - 1; i >= 0; i-- { + msg, ok := parsed.Messages[i].(map[string]any) + if !ok { + continue + } + role, _ := msg["role"].(string) + if !strings.EqualFold(strings.TrimSpace(role), "user") { + continue + } + tokens, _, _ := estimateClaudeContentTokens(msg["content"]) + return claudeMaxMessageOverheadTokens + tokens + } + return 0 +} + +func estimateStructuredTokens(v any) int { + if v == nil { + return 0 + } + raw, err := json.Marshal(v) + if err != nil { + return claudeMaxUnknownContentTokens + } + return estimateClaudeTextTokens(string(raw)) +} + +func estimateClaudeTextTokens(text string) int { + if tokens, ok := estimateTokensByThirdPartyTokenizer(text); ok { + return tokens + } + return estimateClaudeTextTokensHeuristic(text) +} + +func estimateClaudeTextTokensHeuristic(text string) int { + normalized := strings.Join(strings.Fields(strings.TrimSpace(text)), " ") + if normalized == "" { + return 0 + } + asciiChars := 0 + nonASCIIChars := 0 + for _, r := range normalized { + if r <= 127 { + asciiChars++ + } else { + nonASCIIChars++ + } + } + tokens := nonASCIIChars + if asciiChars > 0 { + tokens += (asciiChars + 3) / 4 + } + if words := len(strings.Fields(normalized)); words > tokens { + tokens = words + } + if tokens <= 0 { + return 1 + } + return tokens +} diff --git a/backend/internal/service/claude_max_simulation_test.go b/backend/internal/service/claude_max_simulation_test.go new file mode 100644 index 00000000..3d2ae2e6 --- /dev/null +++ b/backend/internal/service/claude_max_simulation_test.go @@ -0,0 +1,156 @@ +package service + +import ( + "strings" + "testing" +) + +func TestProjectUsageToClaudeMax1H_Conservation(t *testing.T) { + usage := &ClaudeUsage{ + InputTokens: 1200, + CacheCreationInputTokens: 0, + CacheCreation5mTokens: 0, + CacheCreation1hTokens: 0, + } + parsed := &ParsedRequest{ + Model: "claude-sonnet-4-5", + Messages: []any{ + map[string]any{ + "role": "user", + "content": []any{ + map[string]any{ + "type": "text", + "text": strings.Repeat("cached context ", 200), + "cache_control": map[string]any{"type": "ephemeral"}, + }, + map[string]any{ + "type": "text", + "text": "summarize quickly", + }, + }, + }, + }, + } + + changed := projectUsageToClaudeMax1H(usage, parsed) + if !changed { + t.Fatalf("expected usage to be projected") + } + + total := usage.InputTokens + usage.CacheCreation5mTokens + usage.CacheCreation1hTokens + if total != 1200 { + t.Fatalf("total tokens changed: got=%d want=%d", total, 1200) + } + if usage.CacheCreation5mTokens != 0 { + t.Fatalf("cache_creation_5m should be 0, got=%d", usage.CacheCreation5mTokens) + } + if usage.InputTokens <= 0 || usage.InputTokens >= 1200 { + t.Fatalf("simulated input out of range, got=%d", usage.InputTokens) + } + if usage.InputTokens > 100 { + t.Fatalf("simulated input should stay near cache breakpoint tail, got=%d", usage.InputTokens) + } + if usage.CacheCreation1hTokens <= 0 { + t.Fatalf("cache_creation_1h should be > 0, got=%d", usage.CacheCreation1hTokens) + } + if usage.CacheCreationInputTokens != usage.CacheCreation1hTokens { + t.Fatalf("cache_creation_input_tokens mismatch: got=%d want=%d", usage.CacheCreationInputTokens, usage.CacheCreation1hTokens) + } +} + +func TestComputeClaudeMaxProjectedInputTokens_Deterministic(t *testing.T) { + parsed := &ParsedRequest{ + Model: "claude-opus-4-5", + Messages: []any{ + map[string]any{ + "role": "user", + "content": []any{ + map[string]any{ + "type": "text", + "text": "build context", + "cache_control": map[string]any{"type": "ephemeral"}, + }, + map[string]any{ + "type": "text", + "text": "what is failing now", + }, + }, + }, + }, + } + + got1 := computeClaudeMaxProjectedInputTokens(4096, parsed) + got2 := computeClaudeMaxProjectedInputTokens(4096, parsed) + if got1 != got2 { + t.Fatalf("non-deterministic input tokens: %d != %d", got1, got2) + } +} + +func TestShouldSimulateClaudeMaxUsage(t *testing.T) { + group := &Group{ + Platform: PlatformAnthropic, + SimulateClaudeMaxEnabled: true, + } + input := &RecordUsageInput{ + Result: &ForwardResult{ + Model: "claude-sonnet-4-5", + Usage: ClaudeUsage{ + InputTokens: 3000, + CacheCreationInputTokens: 0, + CacheCreation5mTokens: 0, + CacheCreation1hTokens: 0, + }, + }, + ParsedRequest: &ParsedRequest{ + Messages: []any{ + map[string]any{ + "role": "user", + "content": []any{ + map[string]any{ + "type": "text", + "text": "cached", + "cache_control": map[string]any{"type": "ephemeral"}, + }, + map[string]any{ + "type": "text", + "text": "tail", + }, + }, + }, + }, + }, + APIKey: &APIKey{Group: group}, + } + + if !shouldSimulateClaudeMaxUsage(input) { + t.Fatalf("expected simulate=true for claude group with cache signal") + } + + input.ParsedRequest = &ParsedRequest{ + Messages: []any{ + map[string]any{"role": "user", "content": "no cache signal"}, + }, + } + if shouldSimulateClaudeMaxUsage(input) { + t.Fatalf("expected simulate=false when request has no cache signal") + } + + input.ParsedRequest = &ParsedRequest{ + Messages: []any{ + map[string]any{ + "role": "user", + "content": []any{ + map[string]any{ + "type": "text", + "text": "cached", + "cache_control": map[string]any{"type": "ephemeral"}, + }, + }, + }, + }, + } + input.Result.Usage.CacheCreationInputTokens = 100 + if shouldSimulateClaudeMaxUsage(input) { + t.Fatalf("expected simulate=false when cache creation already exists") + } +} diff --git a/backend/internal/service/claude_tokenizer.go b/backend/internal/service/claude_tokenizer.go new file mode 100644 index 00000000..61f5e961 --- /dev/null +++ b/backend/internal/service/claude_tokenizer.go @@ -0,0 +1,41 @@ +package service + +import ( + "sync" + + tiktoken "github.com/pkoukk/tiktoken-go" + tiktokenloader "github.com/pkoukk/tiktoken-go-loader" +) + +var ( + claudeTokenizerOnce sync.Once + claudeTokenizer *tiktoken.Tiktoken +) + +func getClaudeTokenizer() *tiktoken.Tiktoken { + claudeTokenizerOnce.Do(func() { + // Use offline loader to avoid runtime dictionary download. + tiktoken.SetBpeLoader(tiktokenloader.NewOfflineLoader()) + // Use a high-capacity tokenizer as the default approximation for Claude payloads. + enc, err := tiktoken.GetEncoding(tiktoken.MODEL_O200K_BASE) + if err != nil { + enc, err = tiktoken.GetEncoding(tiktoken.MODEL_CL100K_BASE) + } + if err == nil { + claudeTokenizer = enc + } + }) + return claudeTokenizer +} + +func estimateTokensByThirdPartyTokenizer(text string) (int, bool) { + enc := getClaudeTokenizer() + if enc == nil { + return 0, false + } + tokens := len(enc.EncodeOrdinary(text)) + if tokens <= 0 { + return 0, false + } + return tokens, true +} diff --git a/backend/internal/service/concurrency_service.go b/backend/internal/service/concurrency_service.go index 217b83d6..386d5ed0 100644 --- a/backend/internal/service/concurrency_service.go +++ b/backend/internal/service/concurrency_service.go @@ -343,8 +343,9 @@ func (s *ConcurrencyService) StartSlotCleanupWorker(accountRepo AccountRepositor }() } -// GetAccountConcurrencyBatch gets current concurrency counts for multiple accounts -// Returns a map of accountID -> current concurrency count +// GetAccountConcurrencyBatch gets current concurrency counts for multiple accounts. +// Uses a detached context with timeout to prevent HTTP request cancellation from +// causing the entire batch to fail (which would show all concurrency as 0). func (s *ConcurrencyService) GetAccountConcurrencyBatch(ctx context.Context, accountIDs []int64) (map[int64]int, error) { if len(accountIDs) == 0 { return map[int64]int{}, nil @@ -356,5 +357,11 @@ func (s *ConcurrencyService) GetAccountConcurrencyBatch(ctx context.Context, acc } return result, nil } - return s.cache.GetAccountConcurrencyBatch(ctx, accountIDs) + + // Use a detached context so that a cancelled HTTP request doesn't cause + // the Redis pipeline to fail and return all-zero concurrency counts. + redisCtx, cancel := context.WithTimeout(context.Background(), 3*time.Second) + defer cancel() + + return s.cache.GetAccountConcurrencyBatch(redisCtx, accountIDs) } diff --git a/backend/internal/service/error_passthrough_runtime_test.go b/backend/internal/service/error_passthrough_runtime_test.go index 7032d15b..2b7bbf60 100644 --- a/backend/internal/service/error_passthrough_runtime_test.go +++ b/backend/internal/service/error_passthrough_runtime_test.go @@ -220,7 +220,7 @@ func TestApplyErrorPassthroughRule_SkipMonitoringSetsContextKey(t *testing.T) { v, exists := c.Get(OpsSkipPassthroughKey) assert.True(t, exists, "OpsSkipPassthroughKey should be set when skip_monitoring=true") boolVal, ok := v.(bool) - assert.True(t, ok, "value should be bool") + assert.True(t, ok, "value should be a bool") assert.True(t, boolVal) } diff --git a/backend/internal/service/gateway_claude_max_response_helpers.go b/backend/internal/service/gateway_claude_max_response_helpers.go new file mode 100644 index 00000000..a5f5f3d2 --- /dev/null +++ b/backend/internal/service/gateway_claude_max_response_helpers.go @@ -0,0 +1,196 @@ +package service + +import ( + "context" + "encoding/json" + + "github.com/Wei-Shaw/sub2api/internal/pkg/antigravity" + "github.com/gin-gonic/gin" + "github.com/tidwall/sjson" +) + +type claudeMaxResponseRewriteContext struct { + Parsed *ParsedRequest + Group *Group +} + +type claudeMaxResponseRewriteContextKeyType struct{} + +var claudeMaxResponseRewriteContextKey = claudeMaxResponseRewriteContextKeyType{} + +func withClaudeMaxResponseRewriteContext(ctx context.Context, c *gin.Context, parsed *ParsedRequest) context.Context { + if ctx == nil { + ctx = context.Background() + } + value := claudeMaxResponseRewriteContext{ + Parsed: parsed, + Group: claudeMaxGroupFromGinContext(c), + } + return context.WithValue(ctx, claudeMaxResponseRewriteContextKey, value) +} + +func claudeMaxResponseRewriteContextFromContext(ctx context.Context) claudeMaxResponseRewriteContext { + if ctx == nil { + return claudeMaxResponseRewriteContext{} + } + value, _ := ctx.Value(claudeMaxResponseRewriteContextKey).(claudeMaxResponseRewriteContext) + return value +} + +func claudeMaxGroupFromGinContext(c *gin.Context) *Group { + if c == nil { + return nil + } + raw, exists := c.Get("api_key") + if !exists { + return nil + } + apiKey, ok := raw.(*APIKey) + if !ok || apiKey == nil { + return nil + } + return apiKey.Group +} + +func parsedRequestFromGinContext(c *gin.Context) *ParsedRequest { + if c == nil { + return nil + } + raw, exists := c.Get("parsed_request") + if !exists { + return nil + } + parsed, _ := raw.(*ParsedRequest) + return parsed +} + +func applyClaudeMaxSimulationToUsage(ctx context.Context, usage *ClaudeUsage, model string, accountID int64) claudeMaxCacheBillingOutcome { + var out claudeMaxCacheBillingOutcome + if usage == nil { + return out + } + rewriteCtx := claudeMaxResponseRewriteContextFromContext(ctx) + return applyClaudeMaxCacheBillingPolicyToUsage(usage, rewriteCtx.Parsed, rewriteCtx.Group, model, accountID) +} + +func applyClaudeMaxSimulationToUsageJSONMap(ctx context.Context, usageObj map[string]any, model string, accountID int64) claudeMaxCacheBillingOutcome { + var out claudeMaxCacheBillingOutcome + if usageObj == nil { + return out + } + usage := claudeUsageFromJSONMap(usageObj) + out = applyClaudeMaxSimulationToUsage(ctx, &usage, model, accountID) + if out.Simulated { + rewriteClaudeUsageJSONMap(usageObj, usage) + } + return out +} + +func rewriteClaudeUsageJSONBytes(body []byte, usage ClaudeUsage) []byte { + updated := body + var err error + + updated, err = sjson.SetBytes(updated, "usage.input_tokens", usage.InputTokens) + if err != nil { + return body + } + updated, err = sjson.SetBytes(updated, "usage.cache_creation_input_tokens", usage.CacheCreationInputTokens) + if err != nil { + return body + } + updated, err = sjson.SetBytes(updated, "usage.cache_creation.ephemeral_5m_input_tokens", usage.CacheCreation5mTokens) + if err != nil { + return body + } + updated, err = sjson.SetBytes(updated, "usage.cache_creation.ephemeral_1h_input_tokens", usage.CacheCreation1hTokens) + if err != nil { + return body + } + return updated +} + +func claudeUsageFromJSONMap(usageObj map[string]any) ClaudeUsage { + var usage ClaudeUsage + if usageObj == nil { + return usage + } + + usage.InputTokens = usageIntFromAny(usageObj["input_tokens"]) + usage.OutputTokens = usageIntFromAny(usageObj["output_tokens"]) + usage.CacheCreationInputTokens = usageIntFromAny(usageObj["cache_creation_input_tokens"]) + usage.CacheReadInputTokens = usageIntFromAny(usageObj["cache_read_input_tokens"]) + + if ccObj, ok := usageObj["cache_creation"].(map[string]any); ok { + usage.CacheCreation5mTokens = usageIntFromAny(ccObj["ephemeral_5m_input_tokens"]) + usage.CacheCreation1hTokens = usageIntFromAny(ccObj["ephemeral_1h_input_tokens"]) + } + return usage +} + +func rewriteClaudeUsageJSONMap(usageObj map[string]any, usage ClaudeUsage) { + if usageObj == nil { + return + } + usageObj["input_tokens"] = usage.InputTokens + usageObj["cache_creation_input_tokens"] = usage.CacheCreationInputTokens + + ccObj, _ := usageObj["cache_creation"].(map[string]any) + if ccObj == nil { + ccObj = make(map[string]any, 2) + usageObj["cache_creation"] = ccObj + } + ccObj["ephemeral_5m_input_tokens"] = usage.CacheCreation5mTokens + ccObj["ephemeral_1h_input_tokens"] = usage.CacheCreation1hTokens +} + +func usageIntFromAny(v any) int { + switch value := v.(type) { + case int: + return value + case int64: + return int(value) + case float64: + return int(value) + case json.Number: + if n, err := value.Int64(); err == nil { + return int(n) + } + } + return 0 +} + +// setupClaudeMaxStreamingHook 为 Antigravity 流式路径设置 SSE usage 改写 hook。 +func setupClaudeMaxStreamingHook(c *gin.Context, processor *antigravity.StreamingProcessor, originalModel string, accountID int64) { + group := claudeMaxGroupFromGinContext(c) + parsed := parsedRequestFromGinContext(c) + if !shouldApplyClaudeMaxBillingRulesForUsage(group, originalModel, parsed) { + return + } + processor.SetUsageMapHook(func(usageMap map[string]any) { + svcUsage := claudeUsageFromJSONMap(usageMap) + outcome := applyClaudeMaxCacheBillingPolicyToUsage(&svcUsage, parsed, group, originalModel, accountID) + if outcome.Simulated { + rewriteClaudeUsageJSONMap(usageMap, svcUsage) + } + }) +} + +// applyClaudeMaxNonStreamingRewrite 为 Antigravity 非流式路径改写响应体中的 usage。 +func applyClaudeMaxNonStreamingRewrite(c *gin.Context, claudeResp []byte, agUsage *antigravity.ClaudeUsage, originalModel string, accountID int64) []byte { + group := claudeMaxGroupFromGinContext(c) + parsed := parsedRequestFromGinContext(c) + if !shouldApplyClaudeMaxBillingRulesForUsage(group, originalModel, parsed) { + return claudeResp + } + svcUsage := &ClaudeUsage{ + InputTokens: agUsage.InputTokens, + OutputTokens: agUsage.OutputTokens, + CacheCreationInputTokens: agUsage.CacheCreationInputTokens, + CacheReadInputTokens: agUsage.CacheReadInputTokens, + } + outcome := applyClaudeMaxCacheBillingPolicyToUsage(svcUsage, parsed, group, originalModel, accountID) + if outcome.Simulated { + return rewriteClaudeUsageJSONBytes(claudeResp, *svcUsage) + } + return claudeResp +} diff --git a/backend/internal/service/gateway_record_usage_claude_max_test.go b/backend/internal/service/gateway_record_usage_claude_max_test.go new file mode 100644 index 00000000..3cd86938 --- /dev/null +++ b/backend/internal/service/gateway_record_usage_claude_max_test.go @@ -0,0 +1,199 @@ +package service + +import ( + "context" + "testing" + "time" + + "github.com/Wei-Shaw/sub2api/internal/config" + "github.com/stretchr/testify/require" +) + +type usageLogRepoRecordUsageStub struct { + UsageLogRepository + + last *UsageLog + inserted bool + err error +} + +func (s *usageLogRepoRecordUsageStub) Create(_ context.Context, log *UsageLog) (bool, error) { + copied := *log + s.last = &copied + return s.inserted, s.err +} + +func newGatewayServiceForRecordUsageTest(repo UsageLogRepository) *GatewayService { + return &GatewayService{ + usageLogRepo: repo, + billingService: NewBillingService(&config.Config{}, nil), + cfg: &config.Config{RunMode: config.RunModeSimple}, + deferredService: &DeferredService{}, + } +} + +func TestRecordUsage_SimulateClaudeMaxEnabled_ProjectsUsageAndSkipsTTLOverride(t *testing.T) { + repo := &usageLogRepoRecordUsageStub{inserted: true} + svc := newGatewayServiceForRecordUsageTest(repo) + + groupID := int64(11) + input := &RecordUsageInput{ + Result: &ForwardResult{ + RequestID: "req-sim-1", + Model: "claude-sonnet-4", + Duration: time.Second, + Usage: ClaudeUsage{ + InputTokens: 160, + }, + }, + ParsedRequest: &ParsedRequest{ + Model: "claude-sonnet-4", + Messages: []any{ + map[string]any{ + "role": "user", + "content": []any{ + map[string]any{ + "type": "text", + "text": "long cached context for prior turns", + "cache_control": map[string]any{"type": "ephemeral"}, + }, + map[string]any{ + "type": "text", + "text": "please summarize the logs and provide root cause analysis", + }, + }, + }, + }, + }, + APIKey: &APIKey{ + ID: 1, + GroupID: &groupID, + Group: &Group{ + ID: groupID, + Platform: PlatformAnthropic, + RateMultiplier: 1, + SimulateClaudeMaxEnabled: true, + }, + }, + User: &User{ID: 2}, + Account: &Account{ + ID: 3, + Platform: PlatformAnthropic, + Type: AccountTypeOAuth, + Extra: map[string]any{ + "cache_ttl_override_enabled": true, + "cache_ttl_override_target": "5m", + }, + }, + } + + err := svc.RecordUsage(context.Background(), input) + require.NoError(t, err) + require.NotNil(t, repo.last) + + log := repo.last + require.Equal(t, 80, log.InputTokens) + require.Equal(t, 80, log.CacheCreationTokens) + require.Equal(t, 0, log.CacheCreation5mTokens) + require.Equal(t, 80, log.CacheCreation1hTokens) + require.False(t, log.CacheTTLOverridden, "simulate outcome should skip account ttl override") +} + +func TestRecordUsage_SimulateClaudeMaxDisabled_AppliesTTLOverride(t *testing.T) { + repo := &usageLogRepoRecordUsageStub{inserted: true} + svc := newGatewayServiceForRecordUsageTest(repo) + + groupID := int64(12) + input := &RecordUsageInput{ + Result: &ForwardResult{ + RequestID: "req-sim-2", + Model: "claude-sonnet-4", + Duration: time.Second, + Usage: ClaudeUsage{ + InputTokens: 40, + CacheCreationInputTokens: 120, + CacheCreation1hTokens: 120, + }, + }, + APIKey: &APIKey{ + ID: 2, + GroupID: &groupID, + Group: &Group{ + ID: groupID, + Platform: PlatformAnthropic, + RateMultiplier: 1, + SimulateClaudeMaxEnabled: false, + }, + }, + User: &User{ID: 3}, + Account: &Account{ + ID: 4, + Platform: PlatformAnthropic, + Type: AccountTypeOAuth, + Extra: map[string]any{ + "cache_ttl_override_enabled": true, + "cache_ttl_override_target": "5m", + }, + }, + } + + err := svc.RecordUsage(context.Background(), input) + require.NoError(t, err) + require.NotNil(t, repo.last) + + log := repo.last + require.Equal(t, 120, log.CacheCreationTokens) + require.Equal(t, 120, log.CacheCreation5mTokens) + require.Equal(t, 0, log.CacheCreation1hTokens) + require.True(t, log.CacheTTLOverridden) +} + +func TestRecordUsage_SimulateClaudeMaxEnabled_ExistingCacheCreationBypassesSimulation(t *testing.T) { + repo := &usageLogRepoRecordUsageStub{inserted: true} + svc := newGatewayServiceForRecordUsageTest(repo) + + groupID := int64(13) + input := &RecordUsageInput{ + Result: &ForwardResult{ + RequestID: "req-sim-3", + Model: "claude-sonnet-4", + Duration: time.Second, + Usage: ClaudeUsage{ + InputTokens: 20, + CacheCreationInputTokens: 120, + CacheCreation5mTokens: 120, + }, + }, + APIKey: &APIKey{ + ID: 3, + GroupID: &groupID, + Group: &Group{ + ID: groupID, + Platform: PlatformAnthropic, + RateMultiplier: 1, + SimulateClaudeMaxEnabled: true, + }, + }, + User: &User{ID: 4}, + Account: &Account{ + ID: 5, + Platform: PlatformAnthropic, + Type: AccountTypeOAuth, + Extra: map[string]any{ + "cache_ttl_override_enabled": true, + "cache_ttl_override_target": "5m", + }, + }, + } + + err := svc.RecordUsage(context.Background(), input) + require.NoError(t, err) + require.NotNil(t, repo.last) + + log := repo.last + require.Equal(t, 20, log.InputTokens) + require.Equal(t, 120, log.CacheCreation5mTokens) + require.Equal(t, 0, log.CacheCreation1hTokens) + require.Equal(t, 120, log.CacheCreationTokens) + require.False(t, log.CacheTTLOverridden, "existing cache_creation with SimulateClaudeMax enabled should skip account ttl override") +} diff --git a/backend/internal/service/gateway_response_usage_sync_test.go b/backend/internal/service/gateway_response_usage_sync_test.go new file mode 100644 index 00000000..445ee8ad --- /dev/null +++ b/backend/internal/service/gateway_response_usage_sync_test.go @@ -0,0 +1,170 @@ +package service + +import ( + "bytes" + "context" + "encoding/json" + "net/http" + "net/http/httptest" + "testing" + + "github.com/Wei-Shaw/sub2api/internal/config" + "github.com/gin-gonic/gin" + "github.com/stretchr/testify/require" + "github.com/tidwall/gjson" +) + +func TestHandleNonStreamingResponse_UsageAlignedWithClaudeMaxSimulation(t *testing.T) { + gin.SetMode(gin.TestMode) + + svc := &GatewayService{ + cfg: &config.Config{}, + rateLimitService: &RateLimitService{}, + } + + account := &Account{ + ID: 11, + Platform: PlatformAnthropic, + Type: AccountTypeOAuth, + Extra: map[string]any{ + "cache_ttl_override_enabled": true, + "cache_ttl_override_target": "5m", + }, + } + group := &Group{ + ID: 99, + Platform: PlatformAnthropic, + SimulateClaudeMaxEnabled: true, + } + parsed := &ParsedRequest{ + Model: "claude-sonnet-4", + Messages: []any{ + map[string]any{ + "role": "user", + "content": []any{ + map[string]any{ + "type": "text", + "text": "long cached context", + "cache_control": map[string]any{"type": "ephemeral"}, + }, + map[string]any{ + "type": "text", + "text": "new user question", + }, + }, + }, + }, + } + + upstreamBody := []byte(`{"id":"msg_1","model":"claude-sonnet-4","usage":{"input_tokens":120,"output_tokens":8}}`) + resp := &http.Response{ + StatusCode: http.StatusOK, + Header: http.Header{"Content-Type": []string{"application/json"}}, + Body: ioNopCloserBytes(upstreamBody), + } + + rec := httptest.NewRecorder() + c, _ := gin.CreateTestContext(rec) + c.Request = httptest.NewRequest(http.MethodPost, "/v1/messages", bytes.NewReader(nil)) + c.Set("api_key", &APIKey{Group: group}) + requestCtx := withClaudeMaxResponseRewriteContext(context.Background(), c, parsed) + + usage, err := svc.handleNonStreamingResponse(requestCtx, resp, c, account, "claude-sonnet-4", "claude-sonnet-4") + require.NoError(t, err) + require.NotNil(t, usage) + + var rendered struct { + Usage ClaudeUsage `json:"usage"` + } + require.NoError(t, json.Unmarshal(rec.Body.Bytes(), &rendered)) + rendered.Usage.CacheCreation5mTokens = int(gjson.GetBytes(rec.Body.Bytes(), "usage.cache_creation.ephemeral_5m_input_tokens").Int()) + rendered.Usage.CacheCreation1hTokens = int(gjson.GetBytes(rec.Body.Bytes(), "usage.cache_creation.ephemeral_1h_input_tokens").Int()) + + require.Equal(t, rendered.Usage.InputTokens, usage.InputTokens) + require.Equal(t, rendered.Usage.OutputTokens, usage.OutputTokens) + require.Equal(t, rendered.Usage.CacheCreationInputTokens, usage.CacheCreationInputTokens) + require.Equal(t, rendered.Usage.CacheCreation5mTokens, usage.CacheCreation5mTokens) + require.Equal(t, rendered.Usage.CacheCreation1hTokens, usage.CacheCreation1hTokens) + require.Equal(t, rendered.Usage.CacheReadInputTokens, usage.CacheReadInputTokens) + + require.Greater(t, usage.CacheCreation1hTokens, 0) + require.Equal(t, 0, usage.CacheCreation5mTokens) + require.Less(t, usage.InputTokens, 120) +} + +func TestHandleNonStreamingResponse_ClaudeMaxDisabled_NoSimulationIntercept(t *testing.T) { + gin.SetMode(gin.TestMode) + + svc := &GatewayService{ + cfg: &config.Config{}, + rateLimitService: &RateLimitService{}, + } + + account := &Account{ + ID: 12, + Platform: PlatformAnthropic, + Type: AccountTypeOAuth, + Extra: map[string]any{ + "cache_ttl_override_enabled": true, + "cache_ttl_override_target": "5m", + }, + } + group := &Group{ + ID: 100, + Platform: PlatformAnthropic, + SimulateClaudeMaxEnabled: false, + } + parsed := &ParsedRequest{ + Model: "claude-sonnet-4", + Messages: []any{ + map[string]any{ + "role": "user", + "content": []any{ + map[string]any{ + "type": "text", + "text": "long cached context", + "cache_control": map[string]any{"type": "ephemeral"}, + }, + map[string]any{ + "type": "text", + "text": "new user question", + }, + }, + }, + }, + } + + upstreamBody := []byte(`{"id":"msg_2","model":"claude-sonnet-4","usage":{"input_tokens":120,"output_tokens":8}}`) + resp := &http.Response{ + StatusCode: http.StatusOK, + Header: http.Header{"Content-Type": []string{"application/json"}}, + Body: ioNopCloserBytes(upstreamBody), + } + + rec := httptest.NewRecorder() + c, _ := gin.CreateTestContext(rec) + c.Request = httptest.NewRequest(http.MethodPost, "/v1/messages", bytes.NewReader(nil)) + c.Set("api_key", &APIKey{Group: group}) + requestCtx := withClaudeMaxResponseRewriteContext(context.Background(), c, parsed) + + usage, err := svc.handleNonStreamingResponse(requestCtx, resp, c, account, "claude-sonnet-4", "claude-sonnet-4") + require.NoError(t, err) + require.NotNil(t, usage) + + require.Equal(t, 120, usage.InputTokens) + require.Equal(t, 0, usage.CacheCreationInputTokens) + require.Equal(t, 0, usage.CacheCreation5mTokens) + require.Equal(t, 0, usage.CacheCreation1hTokens) +} + +func ioNopCloserBytes(b []byte) *readCloserFromBytes { + return &readCloserFromBytes{Reader: bytes.NewReader(b)} +} + +type readCloserFromBytes struct { + *bytes.Reader +} + +func (r *readCloserFromBytes) Close() error { + return nil +} diff --git a/backend/internal/service/gateway_service.go b/backend/internal/service/gateway_service.go index 8a433a36..85a8e2b7 100644 --- a/backend/internal/service/gateway_service.go +++ b/backend/internal/service/gateway_service.go @@ -56,6 +56,12 @@ const ( claudeMimicDebugInfoKey = "claude_mimic_debug_info" ) +const ( + claudeMaxMessageOverheadTokens = 3 + claudeMaxBlockOverheadTokens = 1 + claudeMaxUnknownContentTokens = 4 +) + // ForceCacheBillingContextKey 强制缓存计费上下文键 // 用于粘性会话切换时，将 input_tokens 转为 cache_read_input_tokens 计费 type forceCacheBillingKeyType struct{} @@ -4424,6 +4430,7 @@ func (s *GatewayService) Forward(ctx context.Context, c *gin.Context, account *A } // 处理正常响应 + ctx = withClaudeMaxResponseRewriteContext(ctx, c, parsed) // 触发上游接受回调（提前释放串行锁，不等流完成） if parsed.OnUpstreamAccepted != nil { @@ -6027,6 +6034,7 @@ func (s *GatewayService) handleStreamingResponse(ctx context.Context, resp *http needModelReplace := originalModel != mappedModel clientDisconnected := false // 客户端断开标志，断开后继续读取上游以获取完整usage + skipAccountTTLOverride := false pendingEventLines := make([]string, 0, 4) @@ -6087,17 +6095,25 @@ func (s *GatewayService) handleStreamingResponse(ctx context.Context, resp *http if msg, ok := event["message"].(map[string]any); ok { if u, ok := msg["usage"].(map[string]any); ok { eventChanged = reconcileCachedTokens(u) || eventChanged + claudeMaxOutcome := applyClaudeMaxSimulationToUsageJSONMap(ctx, u, originalModel, account.ID) + if claudeMaxOutcome.Simulated { + skipAccountTTLOverride = true + } } } } if eventType == "message_delta" { if u, ok := event["usage"].(map[string]any); ok { eventChanged = reconcileCachedTokens(u) || eventChanged + claudeMaxOutcome := applyClaudeMaxSimulationToUsageJSONMap(ctx, u, originalModel, account.ID) + if claudeMaxOutcome.Simulated { + skipAccountTTLOverride = true + } } } // Cache TTL Override: 重写 SSE 事件中的 cache_creation 分类 - if account.IsCacheTTLOverrideEnabled() { + if account.IsCacheTTLOverrideEnabled() && !skipAccountTTLOverride { overrideTarget := account.GetCacheTTLOverrideTarget() if eventType == "message_start" { if msg, ok := event["message"].(map[string]any); ok { @@ -6524,8 +6540,13 @@ func (s *GatewayService) handleNonStreamingResponse(ctx context.Context, resp *h } } + claudeMaxOutcome := applyClaudeMaxSimulationToUsage(ctx, &response.Usage, originalModel, account.ID) + if claudeMaxOutcome.Simulated { + body = rewriteClaudeUsageJSONBytes(body, response.Usage) + } + // Cache TTL Override: 重写 non-streaming 响应中的 cache_creation 分类 - if account.IsCacheTTLOverrideEnabled() { + if account.IsCacheTTLOverrideEnabled() && !claudeMaxOutcome.Simulated { overrideTarget := account.GetCacheTTLOverrideTarget() if applyCacheTTLOverride(&response.Usage, overrideTarget) { // 同步更新 body JSON 中的嵌套 cache_creation 对象 @@ -6591,6 +6612,7 @@ func (s *GatewayService) getUserGroupRateMultiplier(ctx context.Context, userID, // RecordUsageInput 记录使用量的输入参数 type RecordUsageInput struct { Result *ForwardResult + ParsedRequest *ParsedRequest APIKey *APIKey User *User Account *Account @@ -6707,9 +6729,19 @@ func (s *GatewayService) RecordUsage(ctx context.Context, input *RecordUsageInpu result.Usage.InputTokens = 0 } + // Claude Max cache billing policy (group-level): + // - GatewayService 路径: Forward 已改写 usage（含 cache tokens）→ apply 见到 cache tokens 跳过 → simulatedClaudeMax=true（通过第二条件） + // - Antigravity 路径: Forward 中 hook 改写了客户端 SSE，但 ForwardResult.Usage 是原始值 → apply 实际执行模拟 → simulatedClaudeMax=true + var apiKeyGroup *Group + if apiKey != nil { + apiKeyGroup = apiKey.Group + } + claudeMaxOutcome := applyClaudeMaxCacheBillingPolicyToUsage(&result.Usage, input.ParsedRequest, apiKeyGroup, result.Model, account.ID) + simulatedClaudeMax := claudeMaxOutcome.Simulated || + (shouldApplyClaudeMaxBillingRulesForUsage(apiKeyGroup, result.Model, input.ParsedRequest) && hasCacheCreationTokens(result.Usage)) // Cache TTL Override: 确保计费时 token 分类与账号设置一致 cacheTTLOverridden := false - if account.IsCacheTTLOverrideEnabled() { + if account.IsCacheTTLOverrideEnabled() && !simulatedClaudeMax { applyCacheTTLOverride(&result.Usage, account.GetCacheTTLOverrideTarget()) cacheTTLOverridden = (result.Usage.CacheCreation5mTokens + result.Usage.CacheCreation1hTokens) > 0 } diff --git a/backend/internal/service/group.go b/backend/internal/service/group.go index 537b5a3b..c9851bd8 100644 --- a/backend/internal/service/group.go +++ b/backend/internal/service/group.go @@ -50,6 +50,9 @@ type Group struct { // MCP XML 协议注入开关（仅 antigravity 平台使用） MCPXMLInject bool + // Claude usage 模拟开关：将无写缓存 usage 模拟为 claude-max 风格 + SimulateClaudeMaxEnabled bool + // 支持的模型系列（仅 antigravity 平台使用） // 可选值: claude, gemini_text, gemini_image SupportedModelScopes []string diff --git a/backend/internal/service/ops_concurrency.go b/backend/internal/service/ops_concurrency.go index a571dd4d..c03108c4 100644 --- a/backend/internal/service/ops_concurrency.go +++ b/backend/internal/service/ops_concurrency.go @@ -64,12 +64,9 @@ func (s *OpsService) getAccountsLoadMapBestEffort(ctx context.Context, accounts if acc.ID <= 0 { continue } - c := acc.Concurrency - if c <= 0 { - c = 1 - } - if prev, ok := unique[acc.ID]; !ok || c > prev { - unique[acc.ID] = c + lf := acc.EffectiveLoadFactor() + if prev, ok := unique[acc.ID]; !ok || lf > prev { + unique[acc.ID] = lf } } diff --git a/backend/internal/service/ops_metrics_collector.go b/backend/internal/service/ops_metrics_collector.go index f93481e7..6c337071 100644 --- a/backend/internal/service/ops_metrics_collector.go +++ b/backend/internal/service/ops_metrics_collector.go @@ -391,7 +391,7 @@ func (c *OpsMetricsCollector) collectConcurrencyQueueDepth(parentCtx context.Con } batch = append(batch, AccountWithConcurrency{ ID: acc.ID, - MaxConcurrency: acc.Concurrency, + MaxConcurrency: acc.EffectiveLoadFactor(), }) } if len(batch) == 0 { diff --git a/backend/internal/service/user_group_rate.go b/backend/internal/service/user_group_rate.go index 9eb5f067..9908546e 100644 --- a/backend/internal/service/user_group_rate.go +++ b/backend/internal/service/user_group_rate.go @@ -2,6 +2,13 @@ package service import "context" +// UserGroupRateEntry 分组下用户专属倍率条目 +type UserGroupRateEntry struct { + UserID int64 `json:"user_id"` + UserEmail string `json:"user_email"` + RateMultiplier float64 `json:"rate_multiplier"` +} + // UserGroupRateRepository 用户专属分组倍率仓储接口 // 允许管理员为特定用户设置分组的专属计费倍率，覆盖分组默认倍率 type UserGroupRateRepository interface { @@ -13,6 +20,9 @@ type UserGroupRateRepository interface { // 如果未设置专属倍率，返回 nil GetByUserAndGroup(ctx context.Context, userID, groupID int64) (*float64, error) + // GetByGroupID 获取指定分组下所有用户的专属倍率 + GetByGroupID(ctx context.Context, groupID int64) ([]UserGroupRateEntry, error) + // SyncUserGroupRates 同步用户的分组专属倍率 // rates: map[groupID]*rateMultiplier，nil 表示删除该分组的专属倍率 SyncUserGroupRates(ctx context.Context, userID int64, rates map[int64]*float64) error diff --git a/backend/migrations/056_add_sonnet46_to_model_mapping.sql b/backend/migrations/056_add_sonnet46_to_model_mapping.sql new file mode 100644 index 00000000..aa7657d7 --- /dev/null +++ b/backend/migrations/056_add_sonnet46_to_model_mapping.sql @@ -0,0 +1,42 @@ +-- Add claude-sonnet-4-6 to model_mapping for all Antigravity accounts +-- +-- Background: +-- Antigravity now supports claude-sonnet-4-6 +-- +-- Strategy: +-- Directly overwrite the entire model_mapping with updated mappings +-- This ensures consistency with DefaultAntigravityModelMapping in constants.go + +UPDATE accounts +SET credentials = jsonb_set( + credentials, + '{model_mapping}', + '{ + "claude-opus-4-6-thinking": "claude-opus-4-6-thinking", + "claude-opus-4-6": "claude-opus-4-6-thinking", + "claude-opus-4-5-thinking": "claude-opus-4-6-thinking", + "claude-opus-4-5-20251101": "claude-opus-4-6-thinking", + "claude-sonnet-4-6": "claude-sonnet-4-6", + "claude-sonnet-4-5": "claude-sonnet-4-5", + "claude-sonnet-4-5-thinking": "claude-sonnet-4-5-thinking", + "claude-sonnet-4-5-20250929": "claude-sonnet-4-5", + "claude-haiku-4-5": "claude-sonnet-4-5", + "claude-haiku-4-5-20251001": "claude-sonnet-4-5", + "gemini-2.5-flash": "gemini-2.5-flash", + "gemini-2.5-flash-lite": "gemini-2.5-flash-lite", + "gemini-2.5-flash-thinking": "gemini-2.5-flash-thinking", + "gemini-2.5-pro": "gemini-2.5-pro", + "gemini-3-flash": "gemini-3-flash", + "gemini-3-pro-high": "gemini-3-pro-high", + "gemini-3-pro-low": "gemini-3-pro-low", + "gemini-3-pro-image": "gemini-3-pro-image", + "gemini-3-flash-preview": "gemini-3-flash", + "gemini-3-pro-preview": "gemini-3-pro-high", + "gemini-3-pro-image-preview": "gemini-3-pro-image", + "gpt-oss-120b-medium": "gpt-oss-120b-medium", + "tab_flash_lite_preview": "tab_flash_lite_preview" + }'::jsonb +) +WHERE platform = 'antigravity' + AND deleted_at IS NULL + AND credentials->'model_mapping' IS NOT NULL; diff --git a/backend/migrations/057_add_gemini31_pro_to_model_mapping.sql b/backend/migrations/057_add_gemini31_pro_to_model_mapping.sql new file mode 100644 index 00000000..6305e717 --- /dev/null +++ b/backend/migrations/057_add_gemini31_pro_to_model_mapping.sql @@ -0,0 +1,45 @@ +-- Add gemini-3.1-pro-high, gemini-3.1-pro-low, gemini-3.1-pro-preview to model_mapping +-- +-- Background: +-- Antigravity now supports gemini-3.1-pro-high and gemini-3.1-pro-low +-- +-- Strategy: +-- Directly overwrite the entire model_mapping with updated mappings +-- This ensures consistency with DefaultAntigravityModelMapping in constants.go + +UPDATE accounts +SET credentials = jsonb_set( + credentials, + '{model_mapping}', + '{ + "claude-opus-4-6-thinking": "claude-opus-4-6-thinking", + "claude-opus-4-6": "claude-opus-4-6-thinking", + "claude-opus-4-5-thinking": "claude-opus-4-6-thinking", + "claude-opus-4-5-20251101": "claude-opus-4-6-thinking", + "claude-sonnet-4-6": "claude-sonnet-4-6", + "claude-sonnet-4-5": "claude-sonnet-4-5", + "claude-sonnet-4-5-thinking": "claude-sonnet-4-5-thinking", + "claude-sonnet-4-5-20250929": "claude-sonnet-4-5", + "claude-haiku-4-5": "claude-sonnet-4-5", + "claude-haiku-4-5-20251001": "claude-sonnet-4-5", + "gemini-2.5-flash": "gemini-2.5-flash", + "gemini-2.5-flash-lite": "gemini-2.5-flash-lite", + "gemini-2.5-flash-thinking": "gemini-2.5-flash-thinking", + "gemini-2.5-pro": "gemini-2.5-pro", + "gemini-3-flash": "gemini-3-flash", + "gemini-3-pro-high": "gemini-3-pro-high", + "gemini-3-pro-low": "gemini-3-pro-low", + "gemini-3-pro-image": "gemini-3-pro-image", + "gemini-3-flash-preview": "gemini-3-flash", + "gemini-3-pro-preview": "gemini-3-pro-high", + "gemini-3-pro-image-preview": "gemini-3-pro-image", + "gemini-3.1-pro-high": "gemini-3.1-pro-high", + "gemini-3.1-pro-low": "gemini-3.1-pro-low", + "gemini-3.1-pro-preview": "gemini-3.1-pro-high", + "gpt-oss-120b-medium": "gpt-oss-120b-medium", + "tab_flash_lite_preview": "tab_flash_lite_preview" + }'::jsonb +) +WHERE platform = 'antigravity' + AND deleted_at IS NULL + AND credentials->'model_mapping' IS NOT NULL; diff --git a/backend/migrations/060_add_group_simulate_claude_max.sql b/backend/migrations/060_add_group_simulate_claude_max.sql new file mode 100644 index 00000000..55662dfd --- /dev/null +++ b/backend/migrations/060_add_group_simulate_claude_max.sql @@ -0,0 +1,3 @@ +ALTER TABLE groups + ADD COLUMN IF NOT EXISTS simulate_claude_max_enabled BOOLEAN NOT NULL DEFAULT FALSE; + diff --git a/deploy/docker-compose.yml b/deploy/docker-compose.yml index e5c97bf8..8715d75d 100644 --- a/deploy/docker-compose.yml +++ b/deploy/docker-compose.yml @@ -47,13 +47,15 @@ services: # ======================================================================= # Database Configuration (PostgreSQL) + # Default: uses local postgres container + # External DB: set DATABASE_HOST and DATABASE_SSLMODE in .env # ======================================================================= - - DATABASE_HOST=postgres - - DATABASE_PORT=5432 + - DATABASE_HOST=${DATABASE_HOST:-postgres} + - DATABASE_PORT=${DATABASE_PORT:-5432} - DATABASE_USER=${POSTGRES_USER:-sub2api} - DATABASE_PASSWORD=${POSTGRES_PASSWORD:?POSTGRES_PASSWORD is required} - DATABASE_DBNAME=${POSTGRES_DB:-sub2api} - - DATABASE_SSLMODE=disable + - DATABASE_SSLMODE=${DATABASE_SSLMODE:-disable} - DATABASE_MAX_OPEN_CONNS=${DATABASE_MAX_OPEN_CONNS:-50} - DATABASE_MAX_IDLE_CONNS=${DATABASE_MAX_IDLE_CONNS:-10} - DATABASE_CONN_MAX_LIFETIME_MINUTES=${DATABASE_CONN_MAX_LIFETIME_MINUTES:-30} @@ -139,8 +141,6 @@ services: # Examples: http://host:port, socks5://host:port - UPDATE_PROXY_URL=${UPDATE_PROXY_URL:-} depends_on: - postgres: - condition: service_healthy redis: condition: service_healthy networks: diff --git a/frontend/public/wechat-qr.jpg b/frontend/public/wechat-qr.jpg new file mode 100644 index 00000000..659068d8 Binary files /dev/null and b/frontend/public/wechat-qr.jpg differ diff --git a/frontend/src/api/admin/groups.ts b/frontend/src/api/admin/groups.ts index 3d18ba87..81bf4e55 100644 --- a/frontend/src/api/admin/groups.ts +++ b/frontend/src/api/admin/groups.ts @@ -153,6 +153,27 @@ export async function getGroupApiKeys( return data } +/** + * Rate multiplier entry for a user in a group + */ +export interface GroupRateMultiplierEntry { + user_id: number + user_email: string + rate_multiplier: number +} + +/** + * Get rate multipliers for users in a group + * @param id - Group ID + * @returns List of user rate multiplier entries + */ +export async function getGroupRateMultipliers(id: number): Promise { + const { data } = await apiClient.get( + `/admin/groups/${id}/rate-multipliers` + ) + return data +} + /** * Update group sort orders * @param updates - Array of { id, sort_order } objects @@ -178,6 +199,7 @@ export const groupsAPI = { toggleStatus, getStats, getGroupApiKeys, + getGroupRateMultipliers, updateSortOrder } diff --git a/frontend/src/components/account/AccountStatusIndicator.vue b/frontend/src/components/account/AccountStatusIndicator.vue index 220b5c8b..0f46ed93 100644 --- a/frontend/src/components/account/AccountStatusIndicator.vue +++ b/frontend/src/components/account/AccountStatusIndicator.vue @@ -81,15 +81,15 @@ v-if="activeModelRateLimits.length > 0" :class="[ activeModelRateLimits.length <= 4 - ? 'flex flex-col gap-1' + ? 'flex flex-col gap-0.5' : activeModelRateLimits.length <= 8 ? 'columns-2 gap-x-2' : 'columns-3 gap-x-2' ]" > -
+
{{ formatScopeName(item.model) }} diff --git a/frontend/src/components/account/AccountUsageCell.vue b/frontend/src/components/account/AccountUsageCell.vue index e83eaead..28d4106c 100644 --- a/frontend/src/components/account/AccountUsageCell.vue +++ b/frontend/src/components/account/AccountUsageCell.vue @@ -361,12 +361,13 @@ + + diff --git a/frontend/src/components/common/WechatServiceButton.vue b/frontend/src/components/common/WechatServiceButton.vue new file mode 100644 index 00000000..9ee8d3d5 --- /dev/null +++ b/frontend/src/components/common/WechatServiceButton.vue @@ -0,0 +1,104 @@ + + + + + diff --git a/frontend/src/components/layout/AppHeader.vue b/frontend/src/components/layout/AppHeader.vue index 76bf684f..5729d880 100644 --- a/frontend/src/components/layout/AppHeader.vue +++ b/frontend/src/components/layout/AppHeader.vue @@ -121,23 +121,6 @@ {{ t('nav.apiKeys') }} - - - - - - {{ t('nav.github') }} -
diff --git a/frontend/src/i18n/locales/en.ts b/frontend/src/i18n/locales/en.ts index 9fd0c006..6a512c58 100644 --- a/frontend/src/i18n/locales/en.ts +++ b/frontend/src/i18n/locales/en.ts @@ -1411,6 +1411,14 @@ export default { failedToUpdate: 'Failed to update group', failedToDelete: 'Failed to delete group', nameRequired: 'Please enter group name', + rateMultipliers: 'Rate Multipliers', + rateMultipliersTitle: 'Group Rate Multipliers', + addUserRate: 'Add User Rate Multiplier', + searchUserPlaceholder: 'Search user email...', + noRateMultipliers: 'No user rate multipliers configured', + rateUpdated: 'Rate multiplier updated', + rateDeleted: 'Rate multiplier removed', + rateAdded: 'Rate multiplier added', platforms: { all: 'All Platforms', anthropic: 'Anthropic', @@ -1508,6 +1516,14 @@ export default { enabled: 'Enabled', disabled: 'Disabled' }, + claudeMaxSimulation: { + title: 'Claude Max Usage Simulation', + tooltip: + 'When enabled, for Claude models without upstream cache-write usage, the system deterministically maps tokens to a small input plus 1h cache creation while keeping total tokens unchanged.', + enabled: 'Enabled (simulate 1h cache)', + disabled: 'Disabled', + hint: 'Only token categories in usage billing logs are adjusted. No per-request mapping state is persisted.' + }, supportedScopes: { title: 'Supported Model Families', tooltip: 'Select the model families this group supports. Unchecked families will not be routed to this group.', diff --git a/frontend/src/i18n/locales/zh.ts b/frontend/src/i18n/locales/zh.ts index d139cd34..2349efb2 100644 --- a/frontend/src/i18n/locales/zh.ts +++ b/frontend/src/i18n/locales/zh.ts @@ -1510,6 +1510,14 @@ export default { failedToCreate: '创建分组失败', failedToUpdate: '更新分组失败', nameRequired: '请输入分组名称', + rateMultipliers: '专属倍率', + rateMultipliersTitle: '分组专属倍率管理', + addUserRate: '添加用户专属倍率', + searchUserPlaceholder: '搜索用户邮箱...', + noRateMultipliers: '暂无用户设置了专属倍率', + rateUpdated: '专属倍率已更新', + rateDeleted: '专属倍率已删除', + rateAdded: '专属倍率已添加', subscription: { title: '订阅设置', type: '计费类型', diff --git a/frontend/src/types/index.ts b/frontend/src/types/index.ts index 5764134d..e9f70fec 100644 --- a/frontend/src/types/index.ts +++ b/frontend/src/types/index.ts @@ -402,6 +402,8 @@ export interface AdminGroup extends Group { // MCP XML 协议注入（仅 antigravity 平台使用） mcp_xml_inject: boolean + // Claude usage 模拟开关（仅 anthropic 平台使用） + simulate_claude_max_enabled: boolean // 支持的模型系列（仅 antigravity 平台使用） supported_model_scopes?: string[] @@ -496,6 +498,7 @@ export interface CreateGroupRequest { fallback_group_id?: number | null fallback_group_id_on_invalid_request?: number | null mcp_xml_inject?: boolean + simulate_claude_max_enabled?: boolean supported_model_scopes?: string[] // 从指定分组复制账号 copy_accounts_from_group_ids?: number[] @@ -524,6 +527,7 @@ export interface UpdateGroupRequest { fallback_group_id?: number | null fallback_group_id_on_invalid_request?: number | null mcp_xml_inject?: boolean + simulate_claude_max_enabled?: boolean supported_model_scopes?: string[] copy_accounts_from_group_ids?: number[] } diff --git a/frontend/src/utils/__tests__/usageLoadQueue.spec.ts b/frontend/src/utils/__tests__/usageLoadQueue.spec.ts new file mode 100644 index 00000000..24cebec8 --- /dev/null +++ b/frontend/src/utils/__tests__/usageLoadQueue.spec.ts @@ -0,0 +1,87 @@ +import { describe, expect, it, vi } from 'vitest' +import { enqueueUsageRequest } from '../usageLoadQueue' + +function delay(ms: number) { + return new Promise((r) => setTimeout(r, ms)) +} + +describe('usageLoadQueue', () => { + it('同组请求串行执行，间隔 >= 1s', async () => { + const timestamps: number[] = [] + const makeFn = () => async () => { + timestamps.push(Date.now()) + return 'ok' + } + + const p1 = enqueueUsageRequest('anthropic', 'oauth', 1, makeFn()) + const p2 = enqueueUsageRequest('anthropic', 'oauth', 1, makeFn()) + const p3 = enqueueUsageRequest('anthropic', 'oauth', 1, makeFn()) + + await Promise.all([p1, p2, p3]) + + expect(timestamps).toHaveLength(3) + // 随机 1-1.5s 间隔，至少 950ms（留一点误差） + expect(timestamps[1] - timestamps[0]).toBeGreaterThanOrEqual(950) + expect(timestamps[1] - timestamps[0]).toBeLessThan(1600) + expect(timestamps[2] - timestamps[1]).toBeGreaterThanOrEqual(950) + expect(timestamps[2] - timestamps[1]).toBeLessThan(1600) + }) + + it('不同组请求并行执行', async () => { + const timestamps: Record = {} + const makeTracked = (key: string) => async () => { + timestamps[key] = Date.now() + return key + } + + const p1 = enqueueUsageRequest('anthropic', 'oauth', 1, makeTracked('group1')) + const p2 = enqueueUsageRequest('anthropic', 'oauth', 2, makeTracked('group2')) + const p3 = enqueueUsageRequest('gemini', 'oauth', 1, makeTracked('group3')) + + await Promise.all([p1, p2, p3]) + + // 不同组应几乎同时启动（差距 < 50ms） + const values = Object.values(timestamps) + const spread = Math.max(...values) - Math.min(...values) + expect(spread).toBeLessThan(50) + }) + + it('请求失败时 reject，后续任务继续执行', async () => { + const results: string[] = [] + + const p1 = enqueueUsageRequest('anthropic', 'oauth', 99, async () => { + throw new Error('fail') + }) + const p2 = enqueueUsageRequest('anthropic', 'oauth', 99, async () => { + results.push('second') + return 'ok' + }) + + await expect(p1).rejects.toThrow('fail') + await p2 + expect(results).toEqual(['second']) + }) + + it('返回值正确透传', async () => { + const result = await enqueueUsageRequest('test', 'oauth', null, async () => { + return { usage: 42 } + }) + expect(result).toEqual({ usage: 42 }) + }) + + it('proxy_id 为 null 的账号归为同一组', async () => { + const order: number[] = [] + const makeFn = (n: number) => async () => { + order.push(n) + return n + } + + const p1 = enqueueUsageRequest('anthropic', 'oauth', null, makeFn(1)) + const p2 = enqueueUsageRequest('anthropic', 'oauth', null, makeFn(2)) + + await Promise.all([p1, p2]) + + // 同组串行，按入队顺序执行 + expect(order).toEqual([1, 2]) + }) +}) diff --git a/frontend/src/utils/usageLoadQueue.ts b/frontend/src/utils/usageLoadQueue.ts new file mode 100644 index 00000000..97549b15 --- /dev/null +++ b/frontend/src/utils/usageLoadQueue.ts @@ -0,0 +1,72 @@ +/** + * Usage request queue that throttles API calls by group. + * + * Accounts sharing the same upstream (platform + type + proxy) are placed + * into a single serial queue with a configurable delay between requests, + * preventing upstream 429 rate-limit errors. + * + * Different groups run in parallel since they hit different upstreams. + */ + +const GROUP_DELAY_MIN_MS = 1000 +const GROUP_DELAY_MAX_MS = 1500 + +type Task = { + fn: () => Promise + resolve: (value: T) => void + reject: (reason: unknown) => void +} + +const queues = new Map[]>() +const running = new Set() + +function buildGroupKey(platform: string, type: string, proxyId: number | null): string { + return `${platform}:${type}:${proxyId ?? 'direct'}` +} + +async function drain(groupKey: string) { + if (running.has(groupKey)) return + running.add(groupKey) + + const queue = queues.get(groupKey) + while (queue && queue.length > 0) { + const task = queue.shift()! + try { + const result = await task.fn() + task.resolve(result) + } catch (err) { + task.reject(err) + } + // Wait a random 1–1.5s before next request in the same group + if (queue.length > 0) { + const jitter = GROUP_DELAY_MIN_MS + Math.random() * (GROUP_DELAY_MAX_MS - GROUP_DELAY_MIN_MS) + await new Promise((r) => setTimeout(r, jitter)) + } + } + + running.delete(groupKey) + queues.delete(groupKey) +} + +/** + * Enqueue a usage fetch call. Returns a promise that resolves when the + * request completes (after waiting its turn in the group queue). + */ +export function enqueueUsageRequest( + platform: string, + type: string, + proxyId: number | null, + fn: () => Promise +): Promise { + const key = buildGroupKey(platform, type, proxyId) + + return new Promise((resolve, reject) => { + let queue = queues.get(key) + if (!queue) { + queue = [] + queues.set(key, queue) + } + queue.push({ fn, resolve, reject } as Task) + drain(key) + }) +} diff --git a/frontend/src/views/HomeView.vue b/frontend/src/views/HomeView.vue index 6a3753f1..babcf046 100644 --- a/frontend/src/views/HomeView.vue +++ b/frontend/src/views/HomeView.vue @@ -122,8 +122,11 @@ > {{ siteName }} -

- {{ siteSubtitle }} +

+ {{ t('home.heroSubtitle') }} +

+

+ {{ t('home.heroDescription') }}

@@ -177,7 +180,7 @@
-
+
@@ -204,6 +207,63 @@
+ +
+

+ {{ t('home.painPoints.title') }} +

+
+ +
+
+ + + +
+

{{ t('home.painPoints.items.expensive.title') }}

+

{{ t('home.painPoints.items.expensive.desc') }}

+
+ +
+
+ + + +
+

{{ t('home.painPoints.items.complex.title') }}

+

{{ t('home.painPoints.items.complex.desc') }}

+
+ +
+
+ + + +
+

{{ t('home.painPoints.items.unstable.title') }}

+

{{ t('home.painPoints.items.unstable.desc') }}

+
+ +
+
+ + + +
+

{{ t('home.painPoints.items.noControl.title') }}

+

{{ t('home.painPoints.items.noControl.desc') }}

+
+
+
+ + +
+

+ {{ t('home.solutions.title') }} +

+

{{ t('home.solutions.subtitle') }}

+
+
@@ -369,6 +429,77 @@ >
+ + +
+

+ {{ t('home.comparison.title') }} +

+
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
{{ t('home.comparison.headers.feature') }}{{ t('home.comparison.headers.official') }}{{ t('home.comparison.headers.us') }}
{{ t('home.comparison.items.pricing.feature') }}{{ t('home.comparison.items.pricing.official') }}{{ t('home.comparison.items.pricing.us') }}
{{ t('home.comparison.items.models.feature') }}{{ t('home.comparison.items.models.official') }}{{ t('home.comparison.items.models.us') }}
{{ t('home.comparison.items.management.feature') }}{{ t('home.comparison.items.management.official') }}{{ t('home.comparison.items.management.us') }}
{{ t('home.comparison.items.stability.feature') }}{{ t('home.comparison.items.stability.official') }}{{ t('home.comparison.items.stability.us') }}
{{ t('home.comparison.items.control.feature') }}{{ t('home.comparison.items.control.official') }}{{ t('home.comparison.items.control.us') }}
+
+
+ + +
+

+ {{ t('home.cta.title') }} +

+

+ {{ t('home.cta.description') }} +

+ + {{ t('home.cta.button') }} + + + + {{ t('home.goToDashboard') }} + + +
@@ -380,27 +511,20 @@

© {{ currentYear }} {{ siteName }}. {{ t('home.footer.allRightsReserved') }}

-
- - {{ t('home.docs') }} - - - GitHub - -
+ + {{ t('home.docs') }} + + + + @@ -410,6 +534,7 @@ import { useI18n } from 'vue-i18n' import { useAuthStore, useAppStore } from '@/stores' import LocaleSwitcher from '@/components/common/LocaleSwitcher.vue' import Icon from '@/components/icons/Icon.vue' +import WechatServiceButton from '@/components/common/WechatServiceButton.vue' const { t } = useI18n() @@ -419,7 +544,6 @@ const appStore = useAppStore() // Site settings - directly from appStore (already initialized from injected config) const siteName = computed(() => appStore.cachedPublicSettings?.site_name || appStore.siteName || 'Sub2API') const siteLogo = computed(() => appStore.cachedPublicSettings?.site_logo || appStore.siteLogo || '') -const siteSubtitle = computed(() => appStore.cachedPublicSettings?.site_subtitle || 'AI API Gateway Platform') const docUrl = computed(() => appStore.cachedPublicSettings?.doc_url || appStore.docUrl || '') const homeContent = computed(() => appStore.cachedPublicSettings?.home_content || '') @@ -432,9 +556,6 @@ const isHomeContentUrl = computed(() => { // Theme const isDark = ref(document.documentElement.classList.contains('dark')) -// GitHub URL -const githubUrl = 'https://github.com/Wei-Shaw/sub2api' - // Auth state const isAuthenticated = computed(() => authStore.isAuthenticated) const isAdmin = computed(() => authStore.isAdmin) diff --git a/frontend/src/views/admin/AccountsView.vue b/frontend/src/views/admin/AccountsView.vue index f5aff935..fd9f4f30 100644 --- a/frontend/src/views/admin/AccountsView.vue +++ b/frontend/src/views/admin/AccountsView.vue @@ -263,7 +263,7 @@ - + @@ -1176,6 +1176,16 @@ const handleResetQuota = async (a: Account) => { console.error('Failed to reset quota:', error) } } +const handleResetQuota = async (a: Account) => { + try { + const updated = await adminAPI.accounts.resetAccountQuota(a.id) + patchAccountInList(updated) + enterAutoRefreshSilentWindow() + appStore.showSuccess(t('common.success')) + } catch (error) { + console.error('Failed to reset quota:', error) + } +} const handleDelete = (a: Account) => { deletingAcc.value = a; showDeleteDialog.value = true } const confirmDelete = async () => { if(!deletingAcc.value) return; try { await adminAPI.accounts.delete(deletingAcc.value.id); showDeleteDialog.value = false; deletingAcc.value = null; reload() } catch (error) { console.error('Failed to delete account:', error) } } const handleToggleSchedulable = async (a: Account) => { diff --git a/frontend/src/views/admin/GroupsView.vue b/frontend/src/views/admin/GroupsView.vue index 01b98c0c..815b2dd6 100644 --- a/frontend/src/views/admin/GroupsView.vue +++ b/frontend/src/views/admin/GroupsView.vue @@ -181,6 +181,13 @@ {{ t('common.edit') }} + + + {{ + createForm.simulate_claude_max_enabled + ? t('admin.groups.claudeMaxSimulation.enabled') + : t('admin.groups.claudeMaxSimulation.disabled') + }} + + +

+ {{ t('admin.groups.claudeMaxSimulation.hint') }} +

+ +
+ +
+
+ +
+ +
+
+

+ {{ t('admin.groups.claudeMaxSimulation.tooltip') }} +

+
+
+
+
+
+
+ + + {{ + editForm.simulate_claude_max_enabled + ? t('admin.groups.claudeMaxSimulation.enabled') + : t('admin.groups.claudeMaxSimulation.disabled') + }} + +
+

+ {{ t('admin.groups.claudeMaxSimulation.hint') }} +

+
+
+ + + @@ -1796,6 +1915,7 @@ import EmptyState from '@/components/common/EmptyState.vue' import Select from '@/components/common/Select.vue' import PlatformIcon from '@/components/common/PlatformIcon.vue' import Icon from '@/components/icons/Icon.vue' +import GroupRateMultipliersModal from '@/components/admin/group/GroupRateMultipliersModal.vue' import { VueDraggable } from 'vue-draggable-plus' import { createStableObjectKeyResolver } from '@/utils/stableObjectKey' import { useKeyedDebouncedSearch } from '@/composables/useKeyedDebouncedSearch' @@ -1970,6 +2090,8 @@ const submitting = ref(false) const sortSubmitting = ref(false) const editingGroup = ref(null) const deletingGroup = ref(null) +const showRateMultipliersModal = ref(false) +const rateMultipliersGroup = ref(null) const sortableGroups = ref([]) const createForm = reactive({ @@ -1994,6 +2116,8 @@ const createForm = reactive({ sora_storage_quota_gb: null as number | null, // Claude Code 客户端限制（仅 anthropic 平台使用） claude_code_only: false, + // Claude Max usage 模拟开关（仅 anthropic 平台） + simulate_claude_max_enabled: false, fallback_group_id: null as number | null, fallback_group_id_on_invalid_request: null as number | null, // OpenAI Messages 调度配置（仅 openai 平台使用） @@ -2238,6 +2362,8 @@ const editForm = reactive({ sora_storage_quota_gb: null as number | null, // Claude Code 客户端限制（仅 anthropic 平台使用） claude_code_only: false, + // Claude Max usage 模拟开关（仅 anthropic 平台） + simulate_claude_max_enabled: false, fallback_group_id: null as number | null, fallback_group_id_on_invalid_request: null as number | null, // OpenAI Messages 调度配置（仅 openai 平台使用） @@ -2340,6 +2466,7 @@ const closeCreateModal = () => { createForm.sora_video_price_per_request_hd = null createForm.sora_storage_quota_gb = null createForm.claude_code_only = false + createForm.simulate_claude_max_enabled = false createForm.fallback_group_id = null createForm.fallback_group_id_on_invalid_request = null createForm.allow_messages_dispatch = false @@ -2362,6 +2489,8 @@ const handleCreateGroup = async () => { const requestData = { ...createRest, sora_storage_quota_bytes: createQuotaGb ? Math.round(createQuotaGb * 1024 * 1024 * 1024) : 0, + simulate_claude_max_enabled: + createForm.platform === 'anthropic' ? createForm.simulate_claude_max_enabled : false, model_routing: convertRoutingRulesToApiFormat(createModelRoutingRules.value) } await adminAPI.groups.create(requestData) @@ -2402,6 +2531,7 @@ const handleEdit = async (group: AdminGroup) => { editForm.sora_video_price_per_request_hd = group.sora_video_price_per_request_hd editForm.sora_storage_quota_gb = group.sora_storage_quota_bytes ? Number((group.sora_storage_quota_bytes / (1024 * 1024 * 1024)).toFixed(2)) : null editForm.claude_code_only = group.claude_code_only || false + editForm.simulate_claude_max_enabled = group.simulate_claude_max_enabled || false editForm.fallback_group_id = group.fallback_group_id editForm.fallback_group_id_on_invalid_request = group.fallback_group_id_on_invalid_request editForm.allow_messages_dispatch = group.allow_messages_dispatch || false @@ -2423,6 +2553,7 @@ const closeEditModal = () => { showEditModal.value = false editingGroup.value = null editModelRoutingRules.value = [] + editForm.simulate_claude_max_enabled = false editForm.copy_accounts_from_group_ids = [] } @@ -2440,6 +2571,8 @@ const handleUpdateGroup = async () => { const payload = { ...editRest, sora_storage_quota_bytes: editQuotaGb ? Math.round(editQuotaGb * 1024 * 1024 * 1024) : 0, + simulate_claude_max_enabled: + editForm.platform === 'anthropic' ? editForm.simulate_claude_max_enabled : false, fallback_group_id: editForm.fallback_group_id === null ? 0 : editForm.fallback_group_id, fallback_group_id_on_invalid_request: editForm.fallback_group_id_on_invalid_request === null @@ -2459,6 +2592,11 @@ const handleUpdateGroup = async () => { } } +const handleRateMultipliers = (group: AdminGroup) => { + rateMultipliersGroup.value = group + showRateMultipliersModal.value = true +} + const handleDelete = (group: AdminGroup) => { deletingGroup.value = group showDeleteDialog.value = true @@ -2500,6 +2638,25 @@ watch( createForm.allow_messages_dispatch = false createForm.default_mapped_model = '' } + if (newVal !== 'anthropic') { + createForm.simulate_claude_max_enabled = false + } + } +) + +watch( + () => editForm.platform, + (newVal) => { + if (!['anthropic', 'antigravity'].includes(newVal)) { + editForm.fallback_group_id_on_invalid_request = null + } + if (newVal !== 'openai') { + editForm.allow_messages_dispatch = false + editForm.default_mapped_model = '' + } + if (newVal !== 'anthropic') { + editForm.simulate_claude_max_enabled = false + } } ) diff --git a/frontend/src/views/admin/ops/components/OpsConcurrencyCard.vue b/frontend/src/views/admin/ops/components/OpsConcurrencyCard.vue index c7370ab5..ca640ade 100644 --- a/frontend/src/views/admin/ops/components/OpsConcurrencyCard.vue +++ b/frontend/src/views/admin/ops/components/OpsConcurrencyCard.vue @@ -122,6 +122,7 @@ const platformRows = computed((): SummaryRow[] => { available_accounts: availableAccounts, rate_limited_accounts: safeNumber(avail.rate_limit_count), + error_accounts: safeNumber(avail.error_count), total_concurrency: totalConcurrency, used_concurrency: usedConcurrency, @@ -161,7 +162,6 @@ const groupRows = computed((): SummaryRow[] => { total_accounts: totalAccounts, available_accounts: availableAccounts, rate_limited_accounts: safeNumber(avail.rate_limit_count), - error_accounts: safeNumber(avail.error_count), total_concurrency: totalConcurrency, used_concurrency: usedConcurrency, @@ -329,6 +329,7 @@ function formatDuration(seconds: number): string { } + watch( () => realtimeEnabled.value, async (enabled) => { diff --git a/stress_test_gemini_session.sh b/stress_test_gemini_session.sh new file mode 100644 index 00000000..1f2aca57 --- /dev/null +++ b/stress_test_gemini_session.sh @@ -0,0 +1,127 @@ +#!/bin/bash + +# Gemini 粘性会话压力测试脚本 +# 测试目标：验证不同会话分配不同账号，同一会话保持同一账号 + +BASE_URL="http://host.clicodeplus.com:8080" +API_KEY="sk-32ad0a3197e528c840ea84f0dc6b2056dd3fead03526b5c605a60709bd408f7e" +MODEL="gemini-2.5-flash" + +# 创建临时目录存放结果 +RESULT_DIR="/tmp/gemini_stress_test_$(date +%s)" +mkdir -p "$RESULT_DIR" + +echo "==========================================" +echo "Gemini 粘性会话压力测试" +echo "结果目录: $RESULT_DIR" +echo "==========================================" + +# 函数：发送请求并记录 +send_request() { + local session_id=$1 + local round=$2 + local system_prompt=$3 + local contents=$4 + local output_file="$RESULT_DIR/session_${session_id}_round_${round}.json" + + local request_body=$(cat < "$output_file" 2>&1 + + echo "[Session $session_id Round $round] 完成" +} + +# 会话1：数学计算器（累加序列） +run_session_1() { + local sys_prompt="你是一个数学计算器，只返回计算结果数字，不要任何解释" + + # Round 1: 1+1=? + send_request 1 1 "$sys_prompt" '[{"role":"user","parts":[{"text":"1+1=?"}]}]' + + # Round 2: 继续 2+2=?（累加历史） + send_request 1 2 "$sys_prompt" '[{"role":"user","parts":[{"text":"1+1=?"}]},{"role":"model","parts":[{"text":"2"}]},{"role":"user","parts":[{"text":"2+2=?"}]}]' + + # Round 3: 继续 3+3=? + send_request 1 3 "$sys_prompt" '[{"role":"user","parts":[{"text":"1+1=?"}]},{"role":"model","parts":[{"text":"2"}]},{"role":"user","parts":[{"text":"2+2=?"}]},{"role":"model","parts":[{"text":"4"}]},{"role":"user","parts":[{"text":"3+3=?"}]}]' + + # Round 4: 批量计算 10+10, 20+20, 30+30 + send_request 1 4 "$sys_prompt" '[{"role":"user","parts":[{"text":"1+1=?"}]},{"role":"model","parts":[{"text":"2"}]},{"role":"user","parts":[{"text":"2+2=?"}]},{"role":"model","parts":[{"text":"4"}]},{"role":"user","parts":[{"text":"3+3=?"}]},{"role":"model","parts":[{"text":"6"}]},{"role":"user","parts":[{"text":"计算: 10+10=? 20+20=? 30+30=?"}]}]' +} + +# 会话2：英文翻译器（不同系统提示词 = 不同会话） +run_session_2() { + local sys_prompt="你是一个英文翻译器，将中文翻译成英文，只返回翻译结果" + + send_request 2 1 "$sys_prompt" '[{"role":"user","parts":[{"text":"你好"}]}]' + send_request 2 2 "$sys_prompt" '[{"role":"user","parts":[{"text":"你好"}]},{"role":"model","parts":[{"text":"Hello"}]},{"role":"user","parts":[{"text":"世界"}]}]' + send_request 2 3 "$sys_prompt" '[{"role":"user","parts":[{"text":"你好"}]},{"role":"model","parts":[{"text":"Hello"}]},{"role":"user","parts":[{"text":"世界"}]},{"role":"model","parts":[{"text":"World"}]},{"role":"user","parts":[{"text":"早上好"}]}]' +} + +# 会话3：日文翻译器 +run_session_3() { + local sys_prompt="你是一个日文翻译器，将中文翻译成日文，只返回翻译结果" + + send_request 3 1 "$sys_prompt" '[{"role":"user","parts":[{"text":"你好"}]}]' + send_request 3 2 "$sys_prompt" '[{"role":"user","parts":[{"text":"你好"}]},{"role":"model","parts":[{"text":"こんにちは"}]},{"role":"user","parts":[{"text":"谢谢"}]}]' + send_request 3 3 "$sys_prompt" '[{"role":"user","parts":[{"text":"你好"}]},{"role":"model","parts":[{"text":"こんにちは"}]},{"role":"user","parts":[{"text":"谢谢"}]},{"role":"model","parts":[{"text":"ありがとう"}]},{"role":"user","parts":[{"text":"再见"}]}]' +} + +# 会话4：乘法计算器（另一个数学会话，但系统提示词不同） +run_session_4() { + local sys_prompt="你是一个乘法专用计算器，只计算乘法，返回数字结果" + + send_request 4 1 "$sys_prompt" '[{"role":"user","parts":[{"text":"2*3=?"}]}]' + send_request 4 2 "$sys_prompt" '[{"role":"user","parts":[{"text":"2*3=?"}]},{"role":"model","parts":[{"text":"6"}]},{"role":"user","parts":[{"text":"4*5=?"}]}]' + send_request 4 3 "$sys_prompt" '[{"role":"user","parts":[{"text":"2*3=?"}]},{"role":"model","parts":[{"text":"6"}]},{"role":"user","parts":[{"text":"4*5=?"}]},{"role":"model","parts":[{"text":"20"}]},{"role":"user","parts":[{"text":"计算: 10*10=? 20*20=?"}]}]' +} + +# 会话5：诗人（完全不同的角色） +run_session_5() { + local sys_prompt="你是一位诗人，用简短的诗句回应每个话题，每次只写一句诗" + + send_request 5 1 "$sys_prompt" '[{"role":"user","parts":[{"text":"春天"}]}]' + send_request 5 2 "$sys_prompt" '[{"role":"user","parts":[{"text":"春天"}]},{"role":"model","parts":[{"text":"春风拂面花满枝"}]},{"role":"user","parts":[{"text":"夏天"}]}]' + send_request 5 3 "$sys_prompt" '[{"role":"user","parts":[{"text":"春天"}]},{"role":"model","parts":[{"text":"春风拂面花满枝"}]},{"role":"user","parts":[{"text":"夏天"}]},{"role":"model","parts":[{"text":"蝉鸣蛙声伴荷香"}]},{"role":"user","parts":[{"text":"秋天"}]}]' +} + +echo "" +echo "开始并发测试 5 个独立会话..." +echo "" + +# 并发运行所有会话 +run_session_1 & +run_session_2 & +run_session_3 & +run_session_4 & +run_session_5 & + +# 等待所有后台任务完成 +wait + +echo "" +echo "==========================================" +echo "所有请求完成，结果保存在: $RESULT_DIR" +echo "==========================================" + +# 显示结果摘要 +echo "" +echo "响应摘要:" +for f in "$RESULT_DIR"/*.json; do + filename=$(basename "$f") + response=$(cat "$f" | head -c 200) + echo "[$filename]: ${response}..." +done + +echo "" +echo "请检查服务器日志确认账号分配情况"