172 Commits

This Branch

This Branch

All Branches

Author	SHA1	Message	Date
phamnazage-jpg	047ddca92c	docs(portal): record codegraph workspace integration + OMP AGENTS.md refresh	2026-06-03 13:29:28 +08:00
Hermes Agent	56474264d6	refactor(portal): dedup inline scripts in accounts + batch-import + providers ... All three admin pages had two parallel inline <script> blocks (a modern S1 that used adminRuntime + a legacy S2 that was self-contained). Both had a nested <script> text inside S1 that the browser tolerated only because the second script re-ran any state-affecting calls. Merge into a single inline script per page; fix the nested <script> comment. - providers.html: 100371 -> 62761 chars (-37610, -37%) - accounts.html: 54878 -> 33098 chars (-21780, -40%) - batch-import: 43861 -> 26570 chars (-17291, -39%) Also rename draftProviderIDInput -> providerIDInput in providers.html (the old draft-provider-id input was removed during the earlier workflow merge, leaving the script with a null addEventListener on draft id). All scripts pass node --check. Both test_tksea_portal_assets.sh and verify_frontend_smoke.sh PASS.	2026-06-03 13:14:31 +08:00
phamnazage-jpg	09f7c07de3	feat(portal): make provider/batch-import form fields self-explanatory + auto-fill ... Problem: provider manifest form had all-empty fields with cryptic placeholders, users had to know what model IDs to type. Fix on /portal/admin/providers.html (Provider Manifest 草稿): - DISPLAY NAME: datalist of common vendors (OpenAI / DeepSeek / 硅基流动 / Moonshot / 智谱 / Anthropic / 零一万物 / MiniMax / Qwen / Baichuan / 混元) - PLATFORM: datalist of common platforms (openai / openai-compatible / deepseek / anthropic / gemini / zhipu / moonshot / minimax / qwen / ...) - SMOKE TEST MODEL: datalist of common smoke models + auto-fills with first model from MODELS field if user leaves it empty - BASE URL PLACEHOLDER: datalist of common base URLs (12 presets) - MODELS: chip-row of 11 common models (gpt-5.4, gpt-5.4-mini, deepseek-chat, MiniMax-M2.7-highspeed, kimi-k2.6, glm-4.6, claude-sonnet-4-5, gemini-2.5-pro, qwen3-coder-plus, gpt-4o, o4-mini) + clear button. Click chip → append to MODELS field (dedup). - KEYS textarea: 6 rows + example placeholder (sk-example-1/2/3) Fix on /portal/admin-batch-import.html (发起导入): - HOST ID: datalist of common host_ids + hint about loading pack first - ENTRIES textarea: 6 rows + multi-line hint explaining base_url|api_key|model1,model2 format, optional model, batch import JS change: syncDraftHelperState() in providers.html now auto-fills smoke_test_model with first model if user hasn't filled it yet. Also fixed: 2 duplicate copies of syncDraftHelperState (from earlier batch script restoration) — both now have the new logic. Verification: - bash scripts/test/test_tksea_portal_assets.sh → PASS - bash scripts/test/verify_frontend_smoke.sh → PASS - browser_console click test: gpt-5.4 + deepseek-chat + kimi-k2.6 chips → models='gpt-5.4,deepseek-chat,kimi-k2.6' + smoke='gpt-5.4' auto-fill ✓ - screenshot: /tmp/portal-screenshots/admin-providers-v5.png	2026-06-03 11:24:54 +08:00
phamnazage-jpg	122d6282e1	fix(portal): unify all input/select/textarea/label/button/table styles via global fallback ... Root cause: only admin/index.html had explicit .input / .select / .label classes. 100+ inputs across logical-groups / route-health / accounts / providers / admin-batch-import + public portal had no class → browser default styling → 页面看起来「未统一」。 Fix: - portal.css: add global rules that auto-apply design system styling to any input/select/textarea/label/button/table that doesn't opt out via .raw-input / .field-label. The existing .input / .select / .label / .btn classes still win (same styles, just explicit). - portal.js: detectInitialTheme() now respects HTML's data-theme attribute first (page author intent), then localStorage, then OS preference. This makes admin pages' explicit data-theme="dark" actually stick instead of being overridden. - admin/index.html: h3 标题 8 个 article 统一用 class="card-title" (前 4 个 inline 15px / 后 3 个 inline 16px 已统一) - 6 admin pages: 修复 critical HTML 结构 bug — 之前 batch 处理的 残留让 <link> 和 <style> 嵌套在 <style>:root{} 块内，浏览器 解析时直接忽略，导致所有 stylesheet 不加载、整个页面无样式 Verification: - bash scripts/test/test_tksea_portal_assets.sh → PASS - bash scripts/test/verify_frontend_smoke.sh → PASS - 8 张 screenshot v4 在 /tmp/portal-screenshots/ (各 600KB-1.2MB) - 浏览器实测：3 stylesheets 加载，103 个 input 全部 38px/12px 圆角输入框 35 个 label 全部 12px uppercase slate-400 6 个 select 全部 38px + 自定义箭头	2026-06-03 11:05:10 +08:00
phamnazage-jpg	e804a830a0	docs(portal): record 2026-06-03 frontend visual upgrade + design system runbook ... - EXECUTION_BOARD.md: new 2026-06-03 entry with full evidence trail (asset tests, browser smoke, screenshot list, conclusion=已闭环) - 2026-06-03-FRONTEND-DESIGN-SYSTEM-RUNBOOK.md (new, 10KB): * file structure + design token quick reference * standard page skeleton + component API (stat-card, statusbar, Portal.icons, Portal.toast, Portal.copyText, Portal.theme, Portal.renderModernAdminNav) * test-contract string rules (70+ strings must remain in HTML) * common pitfalls (duplicate <!doctype>, duplicate const AdminCommon, stat-card ID drift, accidental script removal) * submission workflow + screenshot evidence commands	2026-06-03 09:11:18 +08:00
phamnazage-jpg	cc8fc900ca	refactor(portal): migrate 8 pages to portal.css+portal.js design system ... Each page now uses the new page-hero + stat-card + statusbar pattern with the Linear/Vercel-aligned token system, while preserving all admin-common.js nav render contract and 70+ test-contract strings. - public portal: index.html (1816 → 1280 lines) - admin entry: admin/index.html - admin pages: logical-groups / route-health / accounts / providers - batch import: admin/batch-import.html (39-line redirect to admin-batch-import.html for legacy URL compatibility) - admin-batch-import.html: real legacy URL handler page Verified: - bash scripts/test/test_tksea_portal_assets.sh → PASS - bash scripts/test/verify_frontend_smoke.sh → PASS (all 7 admin pages + public portal render with smoke-admin / Smoke Logical Group / Smoke Provider Account / smoke-route-primary visible) - 8 screenshot artifacts at /tmp/portal-screenshots/ (1440×2400 chromium headless, 269KB–1.2MB each = real content)	2026-06-03 09:11:07 +08:00
phamnazage-jpg	3a9061e11d	style(portal): add design system + shared layer (portal.css/portal.js/admin-common.css shim) ... - portal.css: 777-line real design system (Linear/Vercel 信息建筑派) * tokens: spacing 4/8/12/16/24/32/48, type 12/13/14/15/17/20/24/32/44 * colors: ink/paper/accent/success/warn/danger × 50/100/500/900 * teal #14b8a6 1:1 aligned with host sub2api Vue/Tailwind * dark-first; light override for public portal * components: page-hero, stat-card, card, status, pill, btn-primary, toast-host, empty, skeleton, drawer, tabs - portal.js: window.Sub2ApiPortal — toast, lucide 1.75px stroke SVG icon registry (shield/group/activity/route/health/account/provider/ import/check/x/alert/info/copy/edit/trash/plus/refresh/...), copyToClipboard, theme auto/dark/light, drawer, renderModernAdminNav - admin-common.css: 4KB legacy shim — maps old class names (.topnav/.primary/.secondary/.ghost/.danger/.metric/.statusbar/.stat/ .eyebrow/.hero-points/.page-hero__eyebrow/.shell/.fade-in/.topline/ .chip/.tag/.mono/.meta-card/.meta-label/.status-pill/.inline-code/ .tone-*) onto new tokens without breaking admin-common.js nav contract Evidence: - bash scripts/test/test_tksea_portal_assets.sh → PASS (70+ string assertions) - bash scripts/test/verify_frontend_smoke.sh → PASS (chromium headless 7 pages)	2026-06-03 09:10:45 +08:00
phamnazage-jpg	35447be934	feat(deploy): add CRM-only online deployment to remote43 ... - scripts/deploy/deploy_crm_only.sh: 单进程部署 sub2api-cn-relay-manager CRM 控制面到 remote43，不依赖 sub2api host / PG / Redis 容器。 复用 scripts/deploy/remote43_patched_stack_lib.sh 的 env 渲染 (render_remote43_crm_env)，render_crm_only_bootstrap 用 $\{VAR\} 占位符 + sed 注入避开 set -u + unquoted-heredoc 边缘问题。 部署前先 kill 老进程 (再 scp 二进制) 避免 ELF overwrite 失败。 - docs/DEPLOYMENT.md: 加 '在线部署节点' 段，记录 stack / 端口 / 入口 / 验证。 - docs/EXECUTION_BOARD.md: 顶部加 'Latest Online Stack' 段。 - artifacts/online-deploy-20260602/: 本次真实部署的证据 - 01-local-build.txt: 本地 server 二进制 md5 + git head - 02-remote-inspect.txt: 远端 process / port / db tables - 03-crm-api-checks.txt: /healthz /api/packs /api/hosts /metrics 真实响应 - 04-portal-public.txt: sub.tksea.top 公共入口 - 05-quality-gates.txt: gofmt / vet / test -race / integration - manifest.json: 结构化汇总 验证（2026-06-02 21:32-21:43）: - /healthz: HTTP 200 'ok' - /api/packs (Bearer): HTTP 200 '{"packs":[]}' - /api/hosts (Bearer): HTTP 200 '{"hosts":[]}' - /api/packs (no auth): HTTP 401 - /metrics (Prometheus): HTTP 200，含 active_hosts/active_providers/ db_connections_active + Go runtime - sub.tksea.top/portal/: HTTP 200 - sub.tksea.top/portal-admin-api/healthz: HTTP 200 'ok'（反代到 CRM） - go test -race ./internal/... ./tests/integration/...: PASS - gofmt / go vet: 干净	2026-06-02 21:46:39 +08:00
phamnazage-jpg	4ec9dad44f	test: 修 build-broken edge-case 测试 ... - internal/store/sqlite/edge_cases_test.go: 把错误的 sqlite.New 调用换成 实际存在的 sqlite.Open(ctx, dsn)，清掉阻塞 `go test ./internal/...` 的 build 失败 - internal/host/sub2api/edge_cases_test.go: gofmt - internal/worker/runner_extra_test.go: TestRunnerLoggerCalled 加 sync.Mutex 保护 logger 写入的共享状态；测试结束前 cancel 并留 20ms flush 窗口，避免 -race 检测到 goroutine 仍在写 验证: gofmt -l . 干净，go vet ./... 零警告， go test -race -count=1 ./internal/... 全包通过，集成测试通过	2026-06-02 20:38:29 +08:00
phamnazage-jpg	f6600d663a	feat(monitoring): add complete Prometheus + Grafana monitoring stack ... Add production-ready monitoring infrastructure: - 15 alerting rules (4 Critical + 11 Warning) - Grafana dashboard with service health panels - Full documentation with deployment guide Covers: service availability, error rates, latency, routing health, database connections, and log metrics	2026-06-02 19:54:38 +08:00
phamnazage-jpg	fb32deb2b1	docs: 更新验证报告，记录 metrics 端点修复	2026-06-02 07:22:24 +08:00
phamnazage-jpg	5e3bd44e96	fix(metrics): 在主路由中注册 /metrics 端点 ... - 添加 metrics 包导入 - 在主 handler 中注册 GET /metrics 路由	2026-06-02 07:22:01 +08:00
phamnazage-jpg	278991bc9d	docs: 添加生产上线全面验证报告	2026-06-02 07:17:37 +08:00
phamnazage-jpg	8bbdffaf17	test: 补充真实功能验证测试 ... - ErrorMetrics 并发安全测试 - AsyncLogWriter 错误指标真实记录测试 - HTTP Server 超时配置真实验证 - Prometheus 指标 HTTP 端点真实测试 - 日志文件输出真实写入测试	2026-06-02 07:07:53 +08:00
phamnazage-jpg	28f377f2bd	feat(api): M-04 添加版本信息端点 ... - 添加 /version 端点返回版本信息 - 版本变量支持构建时 ldflags 注入 - 返回 version、commit、build_time、go_version	2026-06-02 07:00:15 +08:00
phamnazage-jpg	133da2d442	test: M-03 添加边界测试 ... - 添加大量数据边界测试（100条记录） - 添加特殊字符 PackID 边界测试 - 添加空字段验证边界测试	2026-06-02 06:58:45 +08:00
phamnazage-jpg	b69cb9166e	refactor: M-02 添加 errs 包提供错误处理基础设施	2026-06-02 06:57:13 +08:00
phamnazage-jpg	08f7db1bc2	test: M-01 替换测试代码中的 panic 为 t.Fatal ... - packs_repo_test.go: panic → t.Fatal - providers_repo_test.go: panic → t.Fatal	2026-06-02 06:56:30 +08:00
phamnazage-jpg	21065735dd	fix(docker): H-05 优化 Dockerfile 环境变量配置 ... - 添加清晰的注释说明 SUB2API_CRM_ADMIN_TOKEN 为必需配置 - 分离必需和可选环境变量到不同区块 - 更新 Go 版本到 1.23 以匹配 go.mod	2026-06-02 06:55:36 +08:00
phamnazage-jpg	d688722dd2	feat(metrics): H-04 Prometheus 指标暴露 ... - 创建 internal/metrics 包集成 Prometheus 客户端 - 添加 HTTP 请求指标（总量、延迟直方图） - 添加业务指标（active_hosts、active_providers） - 添加路由指标（decisions、failovers） - 添加数据库指标（connections、operations） - 添加日志指标（flush_errors、dropped_events） - 添加 HTTP Middleware 自动收集请求指标 - 添加 StartServer 方法启动独立 metrics 服务	2026-06-02 06:53:24 +08:00
phamnazage-jpg	8984451845	feat(log): H-03 日志 flush 错误监控 ... - 添加 ErrorMetrics 结构体记录 flush/write/drop 错误数 - 添加 ErrorHandler 回调接口用于自定义错误处理 - 在 AsyncLogWriterOptions 中支持配置错误处理器 - 在 flushBatch 中记录 flush 错误指标并回调错误处理器 - 在 enqueue fallback 路径中记录丢弃事件数 - 添加 Metrics() 方法暴露错误统计	2026-06-02 06:51:14 +08:00
phamnazage-jpg	97502b8a86	test: H-02 补充 migrations 测试 ... - 验证迁移文件正确嵌入 embed.FS - 测试所有迁移文件可读且非空 - 验证迁移文件命名规范（NNNN_前缀） - 测试迁移文件排序一致性 - 验证初始迁移文件包含预期 SQL 内容	2026-06-02 06:50:02 +08:00
phamnazage-jpg	3838d78b00	test: H-01 补充 testutil 测试 ... - 添加 SQLiteTestDSN 函数测试（外键启用/禁用、特殊字符） - 添加路径唯一性验证测试 - 添加 OpenSQLiteStore 功能测试（含并发测试） - 添加 CloseSQLiteStore 测试 - 添加无效 DSN 错误处理测试	2026-06-02 06:49:07 +08:00
phamnazage-jpg	6865b3a33b	ci: B-04 添加 GitHub Actions CI/CD 工作流 ... - 添加完整的 CI 流水线：构建、测试、覆盖率检查 - 集成 golangci-lint 静态代码分析 - 集成 gosec 和 govulncheck 安全扫描 - 添加 Docker 镜像构建验证 - 添加 Release 自动打包（多架构支持） - 设置覆盖率阈值 60%	2026-06-02 06:47:06 +08:00
phamnazage-jpg	cf7dd35e1d	feat(log): B-03 日志轮转配置 - 添加 lumberjack 支持 ... - 添加 lumberjack.v2 依赖实现日志轮转 - 支持配置文件输出（stdout/stderr/file） - 支持文件轮转（100MB/3备份/7天/压缩） - 添加 Config 结构体灵活配置 - 添加完整测试用例 测试验证： - TestInitWithConfig PASS - TestInitWithConfigFileOutput PASS - TestDefaultConfig PASS - 全量日志测试通过	2026-06-01 22:06:56 +08:00
phamnazage-jpg	714c4acbe4	fix(log): 修复测试文件未使用导入	2026-06-01 22:03:14 +08:00
phamnazage-jpg	91fa5d6ab4	fix(review): 完成系统性 Review 修复方案 - Task B-01 HTTP Server 超时配置 ... 本次提交包含： - B-01: HTTP Server 添加超时配置 (ReadTimeout/WriteTimeout/IdleTimeout/MaxHeaderBytes) - 添加结构化日志包 internal/log/ (B-02 部分完成) - 添加 Review 报告文档 - 添加系统性修复方案文档 - 添加最佳实践审核报告文档 - 更新任务清单和执行板 测试验证： - TestServerHasTimeoutConfiguration 通过 关联文档： - docs/2026-06-01-SYSTEMATIC-REVIEW-REPORT.md - docs/2026-06-01-SYSTEMATIC-REPAIR-PLAN.md - docs/2026-06-01-BEST-PRACTICE-AUDIT-REPORT.md	2026-06-01 22:02:01 +08:00
phamnazage-jpg	5fbac6ef0b	test(frontend): add provider admin acceptance coverage ... Add a dedicated acceptance script for providers.html, cover it in the local real-host script regression suite, and document the current frontend review baseline, closure audit, providers action matrix, and remediation task board. This keeps the frontend acceptance boundary explicit: providers.html now has a repeatable verification entry point for its page-level actions, while non-UI provider operations remain documented as backend-only capabilities.	2026-06-01 09:58:20 +08:00
phamnazage-jpg	c588a95c7d	fix(access): verify subscription readiness with real user keys ... When subscription access is requested with an explicit access_api_key, assign the subscription to the real target user, bind that user's API key to the subscription group, and probe readiness with the same key instead of falling back to a managed synthetic user. Update the runtime/reconcile flows, adapter tests, and source-of-truth docs so subscription_ready now reflects user-visible host access rather than managed-key-only closure success.	2026-06-01 09:55:11 +08:00
phamnazage-jpg	702ae19a61	feat(pack): publish provider admin draft provider-admin-publish-1780230107	2026-05-31 20:21:50 +08:00
phamnazage-jpg	fd12838519	docs(testing): record repo upsert closure	2026-05-30 18:54:07 +08:00
phamnazage-jpg	64e14ac30d	test(quality): cover repo upsert fallback branches	2026-05-30 18:54:07 +08:00
phamnazage-jpg	2f2653c76f	docs(testing): record upsert and open branch follow-up	2026-05-30 18:48:20 +08:00
phamnazage-jpg	9e32a24f8c	test(quality): cover upsert validation and redis open branches	2026-05-30 18:48:20 +08:00
phamnazage-jpg	d9216d5712	docs(testing): record narrow branch coverage follow-up	2026-05-30 18:28:03 +08:00
phamnazage-jpg	f895eb9035	test(quality): cover sqlite redis and overlay edge branches	2026-05-30 18:28:03 +08:00
phamnazage-jpg	51472e9951	docs(testing): record hotspot branch coverage uplift	2026-05-30 17:22:25 +08:00
phamnazage-jpg	6bbd55111c	test(quality): cover routing sqlite overlay hotspots	2026-05-30 17:22:11 +08:00
phamnazage-jpg	b33fa10677	test(quality): promote overlay coverage to core gate	2026-05-30 17:13:38 +08:00
phamnazage-jpg	249ad5938f	test(quality): promote sqlite coverage to core gate	2026-05-30 17:04:45 +08:00
phamnazage-jpg	b371c698cc	test(quality): promote routing coverage to core gate	2026-05-30 16:54:05 +08:00
phamnazage-jpg	2a925e2954	test(quality): promote internal app coverage to core gate	2026-05-30 16:48:55 +08:00
phamnazage-jpg	800096542b	docs(testing): record internal app coverage uplift	2026-05-30 16:41:05 +08:00
phamnazage-jpg	cbfaf4a28b	test(app): raise handler coverage above 70 percent	2026-05-30 16:40:55 +08:00
phamnazage-jpg	b5529eb9d4	docs(testing): record unified quality gate rollout	2026-05-30 15:28:58 +08:00
phamnazage-jpg	61a5a36c58	feat(testing): add unified quality gates and coverage baseline	2026-05-30 15:28:32 +08:00
phamnazage-jpg	347389c0a2	chore(skills): add project workflow skills	2026-05-30 14:55:16 +08:00
phamnazage-jpg	17c406b661	docs(skills): capture project learnings as reusable workflows	2026-05-30 14:55:05 +08:00
phamnazage-jpg	0a3828d682	docs(status): record false negative verification	2026-05-30 14:46:20 +08:00
phamnazage-jpg	15b7437edd	feat(status): suppress false negative provider readiness	2026-05-30 14:42:51 +08:00