GitLab

Add a system-wide "Backend Mode" that disables user self-registration
and self-service while keeping admin panel and API gateway fully
functional. When enabled, only admin can log in; all user-facing
routes return 403.

Backend:
- New setting key `backend_mode_enabled` with atomic cached reads (60s TTL)
- BackendModeUserGuard middleware blocks non-admin authenticated routes
- BackendModeAuthGuard middleware blocks registration/password-reset auth routes
- Login/Login2FA/RefreshToken handlers reject non-admin when enabled
- TokenPairWithUser struct for role-aware token refresh
- 20 unit tests (middleware + service layer)

Frontend:
- Router guards redirect unauthenticated users to /login
- Admin toggle in Settings page
- Login page hides register link and footer in backend mode
- 9 unit tests for router guard logic
- i18n support (en/zh)

27 files changed, 833 insertions(+), 17 deletions(-)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

将 /v1/chat/completions 端点从 ResponseWriter 劫持模式重构为独立的
类型安全转换路径，与 Anthropic Messages 端点架构对齐：

- 在 apicompat 包新增 Chat Completions 完整类型定义和双向转换器
- 新增 ForwardAsChatCompletions service 方法，走 Responses API 上游
- Handler 改为独立的账号选择/failover 循环，不再劫持 Responses handler
- 提取 handleCompatErrorResponse 为 Chat Completions 和 Messages 共用
- 删除旧的 forwardChatCompletions 直传路径及相关死代码

feat: 添加 OpenAI Chat Completions 兼容端点（基于 #648，修复编译错误和运行时 panic）

fix: 补充缺失的组级和账户级运维告警指标

1. 修复 WriteFilteredHeaders API 不兼容（2处）:
   将 s.cfg.Security.ResponseHeaders 改为 s.responseHeaderFilter，
   因为 main 分支已将函数签名改为接受 *responseheaders.CompiledHeaderFilter

2. 修复 writer 生命周期导致的 nil pointer panic:
   ChatCompletions handler 替换了 c.Writer 但未恢复，导致
   OpsErrorLogger 中间件的 defer 释放 opsCaptureWriter 后，
   Logger 中间件调用 c.Writer.Status() 触发空指针解引用。
   通过保存并恢复 originalWriter 修复。

3. 为 chatCompletionsResponseWriter 添加防御性 Status() 和
   Written() 方法，包含 nil 安全检查

4. 恢复 gateway.go 中被误删的 net/http import

基于 @yulate 在 PR #648 (commit 0bb6a392) 的工作，解决了与最新
main 分支的合并冲突。

原始功能（@yulate）:
- 添加 /v1/chat/completions 和 /chat/completions 兼容端点
- 将 Chat Completions 请求转换为 Responses API 格式并转换回来
- 添加 API Key 直连转发支持
- 包含单元测试
Co-authored-by: yulate <yulate@users.noreply.github.com>

fix(openai): convert string input to array for Codex OAuth responses endpoint (fix #919)

The ChatGPT backend-api codex/responses endpoint requires `input` to be
an array, but the OpenAI Responses API spec allows it to be a plain string.
When a client sends a string input, sub2api now converts it to the expected
message array format. Empty/whitespace-only strings become an empty array
to avoid triggering a 400 "Input must be a list" error.

新增以下运维告警指标类型:
- group_available_accounts: 组内可用账户数
- group_available_ratio: 组内可用账户比例
- group_rate_limit_ratio: 组内限速账户比例
- account_rate_limited_count: 限速账户数
- account_error_count: 错误账户数
- account_error_ratio: 错误账户比例
- overload_account_count: 过载账户数

包含比例和计数类指标的评估逻辑，并注册新的百分比类指标用于阈值校验。

feat: 允许管理员为持有有效订阅的用户绑定订阅类型分组

 fix: Linux.do OAuth 注册支持邀请码两步流程 (fix #836)

feat: cleanup stale concurrency slots on startup

fix: OpenAI临时性400错误支持池模式同账号重试 & HelpTooltip层级修复

fix: use shared max_line_size config for OpenAI Responses SSE scanner

fix(frontend): stack quota badges vertically in capacity column

feat(frontend): pass locale to iframe embedded pages via lang parameter

feat: OpenAI 账号信息增强 & 批量操作支持

1. 识别OpenAI "An error occurred while processing your request" 临时性400错误
   并触发failover，同时在池模式下标记RetryableOnSameAccount，允许同账号重试
2. ForwardAsAnthropic路径同步支持临时性400错误的识别和同账号重试
3. HelpTooltip组件使用Teleport渲染到body，修复在dialog内被裁切的问题

Two SSE scanners in openai_gateway_messages.go were hardcoded to 1MB
while all other scanners use defaultMaxLineSize (500MB) with config
override. This caused Responses API streams to fail on large payloads.

之前管理员无法通过 API 密钥管理将用户绑定到订阅类型分组（直接返回错误）。
现在改为检查用户是否持有该分组的有效订阅，有则允许绑定，无则拒绝。

- admin_service: 新增 userSubRepo 依赖，替换硬拒绝为订阅校验
- admin_service: 区分 ErrSubscriptionNotFound 和内部错误，避免 DB 故障被误报
- wire_gen/api_contract_test: 同步新增参数
- UserApiKeysModal: 管理员分组下拉不再过滤订阅类型分组
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

- 提取 refreshSingleAccount 私有方法复用单账号刷新逻辑
- 新增 BatchClearError handler (POST /admin/accounts/batch-clear-error)
- 新增 BatchRefresh handler (POST /admin/accounts/batch-refresh)
- 前端 AccountBulkActionsBar 添加批量重置状态/刷新令牌按钮
- AccountsView 添加 handler 支持 partial success 反馈
- i18n 中英文补充批量操作相关翻译

提取 DecodeIDToken（跳过过期校验）供导入场景使用，
ParseIDToken 复用它并保留原有过期检查行为。
导入 OpenAI/Sora OAuth 账号时自动补充缺失的 email、
plan_type、chatgpt_account_id 等字段，不覆盖已有值。

OAuth 授权和 token 刷新时从 id_token 的 OpenAI auth claim 中
提取 chatgpt_plan_type（plus/team/pro/free），存入 credentials，
账号管理页面 PlatformTypeBadge 显示订阅类型。

QuotaBadge components (daily/weekly/total) were wrapped in a
horizontal flex container, making them visually inconsistent with
other capacity badges (concurrency, window cost, sessions, RPM)
which stack vertically. Remove the wrapper so all badges align
consistently.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

When the service restarts, concurrency slots from the old process
remain in Redis, causing phantom occupancy. On startup, scan all
concurrency sorted sets and remove members with non-current process
prefix, then clear orphaned wait queue counters.

Uses Go-side SCAN to discover keys (compatible with Redis client
prefix hooks in tests), then passes them to a Lua script for
atomic member-level cleanup.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Embedded pages (purchase subscription, custom pages) now receive the
current user locale through a `lang` URL parameter, allowing iframe
content to match the user's language preference.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

fix(billing): 修复 OpenAI fast 档位计费并补齐展示

- 打通 service_tier 在 OpenAI HTTP、WS、passthrough 与 usage 记录中的传递
- 修正 priority/flex 计费逻辑，并将 fast 归一化为 priority
- 在用户端和管理端补齐服务档位与计费明细展示
- 补齐前后端测试，并修复 WS 限流信号重复持久化导致的全量回归失败
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

fix(openai): remove misplaced passthrough check from isModelSupportedByAccount

feat(frontend): 为后台账号管理和 IP 管理增加拖筐选中能力