chore: bump version to 0.0.53 [skip ci]

优化上传逻辑

.github/workflows/Auto-docker-publish.yml

@@ -27,6 +27,9 @@ permissions:
 jobs:
   version-and-publish:
     runs-on: ubuntu-latest
+    outputs:
+      new_version: ${{ steps.bump_version.outputs.new_version }}
+      tag: ${{ steps.bump_version.outputs.tag }}
     steps:
       - uses: actions/checkout@v4
         with:
@@ -149,3 +152,117 @@ jobs:
             VERSION=${{ steps.bump_version.outputs.new_version }}
           cache-from: type=gha
           cache-to: type=gha,mode=max
+
+  # 单文件可执行构建：把前端打包进二进制（go:embed），交叉编译 Windows /
+  # Linux / macOS 的 amd64 / arm64 产物，作为 GitHub Release 附件发布。
+  build-frontend:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-node@v4
+        with:
+          node-version: '20'
+          cache: 'npm'
+          cache-dependency-path: web/package-lock.json
+      - name: Install
+        working-directory: web
+        run: npm ci
+      - name: Build SPA
+        working-directory: web
+        run: npm run build
+      - name: Upload web/dist
+        uses: actions/upload-artifact@v4
+        with:
+          name: web-dist
+          path: web/dist
+          retention-days: 1
+
+  # 先创建（幂等）空的 GitHub Release，供后续 build-binaries 并行上传附件，
+  # 也避免矩阵各 job 并发 upload 时 release 尚不存在而互相竞争。
+  publish-create-release:
+    needs: [version-and-publish]
+    runs-on: ubuntu-latest
+    permissions:
+      contents: write
+    steps:
+      - uses: actions/checkout@v4
+      - name: Create release
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          RELEASE_TAG: ${{ needs.version-and-publish.outputs.tag }}
+        run: |
+          set -eux
+          # tag 已由 version-and-publish 推送；若 release 已存在则忽略（--verify-tag 幂等）
+          gh release create "$RELEASE_TAG" \
+            --title "MMTL ${{ needs.version-and-publish.outputs.new_version }}" \
+            --notes "自动化发布 ${{ needs.version-and-publish.outputs.new_version }}" \
+            --verify-tag --latest || true
+
+  build-binaries:
+    needs: [version-and-publish, build-frontend, publish-create-release]
+    runs-on: ubuntu-latest
+    permissions:
+      contents: write
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - goos: linux
+            goarch: amd64
+            ext: ""
+          - goos: linux
+            goarch: arm64
+            ext: ""
+          - goos: windows
+            goarch: amd64
+            ext: .exe
+          - goos: windows
+            goarch: arm64
+            ext: .exe
+          - goos: darwin
+            goarch: amd64
+            ext: ""
+          - goos: darwin
+            goarch: arm64
+            ext: ""
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-go@v5
+        with:
+          go-version: '1.25'
+          cache: true
+      - name: Download web/dist
+        uses: actions/download-artifact@v4
+        with:
+          name: web-dist
+          path: web/dist
+      - name: Build binary
+        run: |
+          CGO_ENABLED=0 GOOS=${{ matrix.goos }} GOARCH=${{ matrix.goarch }} \
+            go build -trimpath -ldflags="-s -w -X main.version=${{ needs.version-and-publish.outputs.tag }}" \
+            -o "dist/mmtl-${{ matrix.goos }}-${{ matrix.goarch }}${{ matrix.ext }}" ./cmd/server
+      - name: Package
+        run: |
+          mkdir -p package/mmtl
+          cp "dist/mmtl-${{ matrix.goos }}-${{ matrix.goarch }}${{ matrix.ext }}" package/mmtl/mmtl${{ matrix.ext }}
+          cp README.md package/mmtl/ 2>/dev/null || true
+          if [ "${{ matrix.goos }}" = "windows" ]; then
+            (cd package && zip -r "../mmtl_${{ matrix.goos }}_${{ matrix.goarch }}.zip" mmtl)
+          else
+            tar -czf "mmtl_${{ matrix.goos }}_${{ matrix.goarch }}.tar.gz" -C package mmtl
+          fi
+      - name: Upload to GitHub Release
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          RELEASE_TAG: ${{ needs.version-and-publish.outputs.tag }}
+        run: |
+          set -eux
+          PKG="mmtl_${{ matrix.goos }}_${{ matrix.goarch }}.zip"
+          TAR="mmtl_${{ matrix.goos }}_${{ matrix.goarch }}.tar.gz"
+          # 并发上传到同一 release 各自文件，--clobber 幂等覆盖
+          if [ -f "$PKG" ]; then
+            for i in 1 2 3; do gh release upload "$RELEASE_TAG" "$PKG" --clobber && break || sleep 5; done
+          fi
+          if [ -f "$TAR" ]; then
+            for i in 1 2 3; do gh release upload "$RELEASE_TAG" "$TAR" --clobber && break || sleep 5; done
+          fi

.github/workflows/ci.yml

@@ -18,6 +18,19 @@ jobs:
           go-version: '1.25'
           cache: true
 
+      # The binary embeds the SPA (web/dist) via go:embed, so the dist must exist
+      # before the Go toolchain touches the `web` package.
+      - uses: actions/setup-node@v4
+        with:
+          node-version: '20'
+          cache: 'npm'
+          cache-dependency-path: web/package-lock.json
+      - name: Build SPA
+        working-directory: web
+        run: |
+          npm ci
+          npm run build
+
       - name: go vet
         run: go vet ./...
 

README.md

@@ -360,15 +360,17 @@ docker compose -f docker-compose.search.yml up -d --no-deps mmtl
 
 本地开发需要 Go、Node.js 和 npm。
 
+后端会将 `web/dist` 通过 `go:embed` 编进二进制，因此**在编译 / 运行后端之前要先构建前端**，否则 `web` 包会因为缺少嵌入资源而编译失败。
+
 ```bash
+# 前端依赖与构建（必须先做，产物被 go:embed 打进二进制）
+npm --prefix web ci
+npm --prefix web run build
+
 # 后端测试
 go test ./...
 
-# 前端依赖与构建
-npm --prefix web install
-npm --prefix web run build
-
-# 本地运行后端
+# 本地运行后端（二进制自带前端界面，无需额外 web 目录）
 go run ./cmd/server
 
 # 本地运行前端开发服务器
@@ -387,6 +389,21 @@ http://127.0.0.1:3000
 http://127.0.0.1:8080/api/health
 ```
 
+### 交叉编译单文件发布物
+
+CI（`.github/workflows/Auto-docker-publish.yml`）每次发布会自动为 Windows / Linux(含 Debian) / macOS 交叉编译 amd64 + arm64 的单文件可执行程序，并上传到对应的 GitHub Release。你可以在 Releases 页面下载 `.zip`（Windows）或 `.tar.gz`（Linux / macOS）附件，解压后直接运行其中的 `mmtl`（Windows 为 `mmtl.exe`），无需额外携带前端目录。
+
+本地手动交叉编译某个平台：
+
+```bash
+# 先构建前端
+npm --prefix web ci && npm --prefix web run build
+
+# 例如：构建 Linux amd64 单文件
+CGO_ENABLED=0 GOOS=linux GOARCH=amd64 \
+  go build -trimpath -ldflags="-s -w" -o mmtl-linux-amd64 ./cmd/server
+```
+
 ## 贡献与反馈
 
 提交 Bug、功能建议或 Pull Request 前，请先阅读 [贡献规范](CONTRIBUTING.md)。

VERSION

@@ -1 +1 @@
-0.0.20
+0.0.53

cmd/server/main_test.go

@@ -52,7 +52,7 @@ func TestServeSPANoCachesIndexAndServesRoutes(t *testing.T) {
 	}
 
 	router := gin.New()
-	serveSPA(router, webDir)
+	serveSPA(router, os.DirFS(webDir))
 
 	for _, path := range []string{"/", "/login", "/library/e1c3507e-2878-40ae-a0e1-6b6e44b7fa7a", "/media/abc"} {
 		req := httptest.NewRequest(http.MethodGet, path, nil)
@@ -93,7 +93,7 @@ func TestServeSPAServesAssetsImmutableAndBypassesAPIRoutes(t *testing.T) {
 	}
 
 	router := gin.New()
-	serveSPA(router, webDir)
+	serveSPA(router, os.DirFS(webDir))
 
 	assetReq := httptest.NewRequest(http.MethodGet, "/assets/app.js", nil)
 	assetResp := httptest.NewRecorder()
@@ -155,7 +155,7 @@ func TestServeSPAServesAssetsImmutableAndBypassesAPIRoutes(t *testing.T) {
 func TestServeSPAMissingIndexReportsExplicit404(t *testing.T) {
 	gin.SetMode(gin.TestMode)
 	router := gin.New()
-	serveSPA(router, t.TempDir())
+	serveSPA(router, os.DirFS(t.TempDir()))
 
 	req := httptest.NewRequest(http.MethodGet, "/", nil)
 	w := httptest.NewRecorder()

cmd/server/router.go

@@ -1,6 +1,8 @@
 package main
 
 import (
+	"io/fs"
+	"mime"
 	"net/http"
 	"os"
 	"path/filepath"
@@ -13,6 +15,8 @@ import (
 	"github.com/ShukeBta/MMTL/internal/handler"
 	"github.com/ShukeBta/MMTL/internal/middleware"
 	"github.com/ShukeBta/MMTL/internal/service"
+
+	"github.com/ShukeBta/MMTL/web"
 )
 
 func buildRouter(cfg *config.Config, logger *zap.Logger, svc *service.Container) *gin.Engine {
@@ -29,31 +33,41 @@ func buildRouter(cfg *config.Config, logger *zap.Logger, svc *service.Container)
 
 	handler.Register(r, cfg, logger, svc)
 
-	if cfg.App.WebDir != "" {
-		serveSPA(r, cfg.App.WebDir)
+	// Prefer a directory on disk when configured explicitly (e.g. the Docker image
+	// mounts web/dist from the build stage, or an operator overrides app.web_dir
+	// with a custom skin). Otherwise fall back to the SPA embedded into the binary,
+	// which is what makes the cross-platform single-file artifacts work.
+	uiFS := webui.DistFS()
+	if dir := cfg.App.WebDir; dir != "" {
+		disk := os.DirFS(dir)
+		if index, err := fs.Stat(disk, "index.html"); err == nil && !index.IsDir() {
+			uiFS = disk
+		}
 	}
+	serveSPA(r, uiFS)
 	return r
 }
 
 // serveSPA serves the React build artifacts and falls back to index.html for
-// non-API, non-asset paths so client-side routing keeps working.
-func serveSPA(r *gin.Engine, webDir string) {
+// non-API, non-asset paths so client-side routing keeps working. The UI tree
+// comes from root, which is either the compiled-in SPA or an on-disk web dir.
+func serveSPA(r *gin.Engine, root fs.FS) {
 	assets := r.Group("/assets")
 	assets.Use(func(c *gin.Context) {
 		c.Header("Cache-Control", "public, max-age=31536000, immutable")
 		c.Next()
 	})
-	assets.Static("/", filepath.Join(webDir, "assets"))
+	assets.GET("/*filepath", serveFSDir(root, "assets"))
 	brand := r.Group("/brand")
 	brand.Use(func(c *gin.Context) {
 		setNoCacheHeaders(c)
 		c.Next()
 	})
-	brand.Static("/", filepath.Join(webDir, "brand"))
+	brand.GET("/*filepath", serveFSDir(root, "brand"))
 	for _, rootFile := range []string{"/favicon.ico", "/favicon.svg", "/artwork-cache-sw.js"} {
-		filePath := filepath.Join(webDir, strings.TrimPrefix(rootFile, "/"))
-		r.GET(rootFile, serveNoCacheFile(filePath))
-		r.HEAD(rootFile, serveNoCacheFile(filePath))
+		name := strings.TrimPrefix(rootFile, "/")
+		r.GET(rootFile, serveFSFile(root, name))
+		r.HEAD(rootFile, serveFSFile(root, name))
 	}
 	r.NoRoute(func(c *gin.Context) {
 		path := c.Request.URL.Path
@@ -61,28 +75,50 @@ func serveSPA(r *gin.Engine, webDir string) {
 			c.Status(http.StatusNotFound)
 			return
 		}
-		serveSPAIndex(c, filepath.Join(webDir, "index.html"))
+		setNoCacheHeaders(c)
+		data, err := fs.ReadFile(root, "index.html")
+		if err != nil {
+			c.String(http.StatusNotFound, "MMTL web UI not found")
+			return
+		}
+		c.Data(http.StatusOK, "text/html; charset=utf-8", data)
 	})
 }
 
-func serveNoCacheFile(filePath string) gin.HandlerFunc {
+// serveFSDir serves a static subdirectory of root. A missing asset returns 404.
+func serveFSDir(root fs.FS, dir string) gin.HandlerFunc {
+	sub, err := fs.Sub(root, dir)
+	if err != nil {
+		return func(c *gin.Context) { c.Status(http.StatusNotFound) }
+	}
+	handler := http.StripPrefix("/"+dir, http.FileServerFS(sub))
 	return func(c *gin.Context) {
-		setNoCacheHeaders(c)
-		if _, err := os.Stat(filePath); err != nil {
-			c.Status(http.StatusNotFound)
-			return
-		}
-		c.File(filePath)
+		handler.ServeHTTP(c.Writer, c.Request)
 	}
 }
 
-func serveSPAIndex(c *gin.Context, indexPath string) {
-	setNoCacheHeaders(c)
-	if _, err := os.Stat(indexPath); err != nil {
-		c.String(http.StatusNotFound, "MMTL web UI not found: %s", indexPath)
-		return
+// serveFSFile serves a single root-level file (favicon / service worker) with
+// no-cache headers. It reads from root, which may be the embedded SPA or disk.
+func serveFSFile(root fs.FS, name string) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		setNoCacheHeaders(c)
+		data, err := fs.ReadFile(root, name)
+		if err != nil {
+			c.Status(http.StatusNotFound)
+			return
+		}
+		c.Data(http.StatusOK, mimeTypeByName(name), data)
+	}
+}
+
+// mimeTypeByName returns an HTTP content type guessed from a file extension.
+func mimeTypeByName(name string) string {
+	switch mime.TypeByExtension(filepath.Ext(name)) {
+	case "":
+		return "application/octet-stream"
+	default:
+		return mime.TypeByExtension(filepath.Ext(name))
 	}
-	c.File(indexPath)
 }
 
 func setNoCacheHeaders(c *gin.Context) {

go.mod

@@ -3,6 +3,7 @@ module github.com/ShukeBta/MMTL
 go 1.25.0
 
 require (
+	github.com/aliyun/alibabacloud-oss-go-sdk-v2 v1.5.2
 	github.com/fsnotify/fsnotify v1.7.0
 	github.com/gin-gonic/gin v1.9.1
 	github.com/glebarez/sqlite v1.11.0
@@ -16,6 +17,7 @@ require (
 	go.uber.org/zap v1.27.0
 	golang.org/x/crypto v0.21.0
 	golang.org/x/sys v0.20.0
+	golang.org/x/time v0.15.0
 	gorm.io/driver/postgres v1.5.7
 	gorm.io/gorm v1.30.0
 )
@@ -66,13 +68,13 @@ require (
 	github.com/tklauser/numcpus v0.6.1 // indirect
 	github.com/twitchyliquid64/golang-asm v0.15.1 // indirect
 	github.com/ugorji/go/codec v1.2.11 // indirect
+	github.com/ulikunitz/xz v0.5.12 // indirect
 	github.com/yusufpapurcu/wmi v1.2.4 // indirect
 	go.uber.org/multierr v1.10.0 // indirect
 	golang.org/x/arch v0.3.0 // indirect
 	golang.org/x/exp v0.0.0-20230905200255-921286631fa9 // indirect
 	golang.org/x/net v0.21.0 // indirect
 	golang.org/x/text v0.20.0 // indirect
-	golang.org/x/time v0.15.0 // indirect
 	google.golang.org/protobuf v1.31.0 // indirect
 	gopkg.in/ini.v1 v1.67.0 // indirect
 	gopkg.in/yaml.v3 v3.0.1 // indirect

go.sum

@@ -1,3 +1,5 @@
+github.com/aliyun/alibabacloud-oss-go-sdk-v2 v1.5.2 h1:40yUSXwdkWN851BHCq6uiDhleh7A4+0yIBS+IUAqZVY=
+github.com/aliyun/alibabacloud-oss-go-sdk-v2 v1.5.2/go.mod h1:FTzydeQVmR24FI0D6XWUOMKckjXehM/jgMn1xC+DA9M=
 github.com/bsm/ginkgo/v2 v2.12.0 h1:Ny8MWAHyOepLGlLKYmXG4IEkioBysk6GpaRTLC8zwWs=
 github.com/bsm/ginkgo/v2 v2.12.0/go.mod h1:SwYbGRRDovPVboqFv0tPTcG1sN61LM1Z4ARdbAV9g4c=
 github.com/bsm/gomega v1.27.10 h1:yeMWxP2pV2fG3FgAODIY8EiRE3dy0aeFYt4l7wh6yKA=
@@ -150,6 +152,8 @@ github.com/twitchyliquid64/golang-asm v0.15.1 h1:SU5vSMR7hnwNxj24w34ZyCi/FmDZTkS
 github.com/twitchyliquid64/golang-asm v0.15.1/go.mod h1:a1lVb/DtPvCB8fslRZhAngC2+aY1QWCk3Cedj/Gdt08=
 github.com/ugorji/go/codec v1.2.11 h1:BMaWp1Bb6fHwEtbplGBGJ498wD+LKlNSl25MjdZY4dU=
 github.com/ugorji/go/codec v1.2.11/go.mod h1:UNopzCgEMSXjBc6AOMqYvWC1ktqTAfzJZUZgYf6w6lg=
+github.com/ulikunitz/xz v0.5.12 h1:37Nm15o69RwBkXM0J6A5OlE67RZTfzUxTj8fB3dfcsc=
+github.com/ulikunitz/xz v0.5.12/go.mod h1:nbz6k7qbPmH4IRqmfOplQw/tblSgqTqBwxkY0oWt/14=
 github.com/yusufpapurcu/wmi v1.2.4 h1:zFUKzehAFReQwLys1b/iSMl+JQGSCSjtVqQn9bBrPo0=
 github.com/yusufpapurcu/wmi v1.2.4/go.mod h1:SBZ9tNy3G9/m5Oi98Zks0QjeHVDvuK0qfxQmPyzfmi0=
 go.uber.org/goleak v1.3.0 h1:2K3zAYmnTNqV73imy9J1T3WC+gmCePx2hEGkimedGto=

internal/config/defaults.go

@@ -3,8 +3,8 @@ package config
 import "github.com/spf13/viper"
 
 const (
-	defaultDatabaseMaxOpenConns = 4
-	defaultDatabaseMaxIdleConns = 2
+	defaultDatabaseMaxOpenConns = 16
+	defaultDatabaseMaxIdleConns = 4
 	defaultLicenseServerURL     = "https://mgosever.3jzs.com"
 	defaultLicensePublicKey     = "MCowBQYDK2VwAyEABRXnXy+urjrbKit6Yu/HiezWgP0NdsZW3tsegJWRrtI="
 )

internal/config/save.go

@@ -0,0 +1,41 @@
+package config
+
+import (
+	"fmt"
+	"os"
+
+	"gopkg.in/yaml.v3"
+)
+
+// SaveDatabaseConfig updates or creates config.yaml with the specified database configuration.
+func SaveDatabaseConfig(dbType, dsn string) error {
+	configPath := "config.yaml"
+	data := make(map[string]any)
+
+	content, err := os.ReadFile(configPath)
+	if err == nil {
+		if err := yaml.Unmarshal(content, &data); err != nil {
+			data = make(map[string]any)
+		}
+	} else if !os.IsNotExist(err) {
+		return fmt.Errorf("read config.yaml: %w", err)
+	}
+
+	dbSection, ok := data["database"].(map[string]any)
+	if !ok {
+		dbSection = make(map[string]any)
+	}
+	dbSection["type"] = dbType
+	dbSection["dsn"] = dsn
+	data["database"] = dbSection
+
+	out, err := yaml.Marshal(data)
+	if err != nil {
+		return fmt.Errorf("marshal config.yaml: %w", err)
+	}
+
+	if err := os.WriteFile(configPath, out, 0644); err != nil {
+		return fmt.Errorf("write config.yaml: %w", err)
+	}
+	return nil
+}

internal/config/save_test.go

@@ -0,0 +1,36 @@
+package config
+
+import (
+	"os"
+	"path/filepath"
+	"testing"
+)
+
+func TestSaveDatabaseConfig(t *testing.T) {
+	dir := t.TempDir()
+	wd, _ := os.Getwd()
+	defer func() { _ = os.Chdir(wd) }()
+	if err := os.Chdir(dir); err != nil {
+		t.Fatalf("chdir: %v", err)
+	}
+
+	dsn := "postgres://admin:pass@127.0.0.1:5432/mmtl?sslmode=disable"
+	if err := SaveDatabaseConfig("postgres", dsn); err != nil {
+		t.Fatalf("SaveDatabaseConfig error: %v", err)
+	}
+
+	if _, err := os.Stat(filepath.Join(dir, "config.yaml")); err != nil {
+		t.Fatalf("expected config.yaml to exist: %v", err)
+	}
+
+	loaded, err := Load()
+	if err != nil {
+		t.Fatalf("Load error: %v", err)
+	}
+	if loaded.Database.Type != "postgres" {
+		t.Fatalf("expected database.type=postgres, got %s", loaded.Database.Type)
+	}
+	if loaded.Database.DSN != dsn {
+		t.Fatalf("expected dsn=%s, got %s", dsn, loaded.Database.DSN)
+	}
+}

internal/database/database_admin.go

@@ -0,0 +1,223 @@
+package database
+
+import (
+	"context"
+	"fmt"
+	"net/url"
+	"strings"
+	"time"
+
+	"go.uber.org/zap"
+	"gorm.io/driver/postgres"
+	"gorm.io/gorm"
+	"gorm.io/gorm/logger"
+
+	"github.com/ShukeBta/MMTL/internal/config"
+	"github.com/ShukeBta/MMTL/internal/model"
+)
+
+// DatabaseStatus describes the currently active database engine and runtime metrics.
+type DatabaseStatus struct {
+	Type         string           `json:"type"`
+	DSN          string           `json:"dsn,omitempty"`
+	DBPath       string           `json:"db_path,omitempty"`
+	OpenConns    int              `json:"open_conns"`
+	InUse        int              `json:"in_use"`
+	Idle         int              `json:"idle"`
+	MaxOpenConns int              `json:"max_open_conns"`
+	TableCounts  map[string]int64 `json:"table_counts"`
+}
+
+// PostgresTestResult returns latency and version info after testing connection.
+type PostgresTestResult struct {
+	Success   bool   `json:"success"`
+	LatencyMS int64  `json:"latency_ms"`
+	Version   string `json:"version,omitempty"`
+	Message   string `json:"message,omitempty"`
+	Error     string `json:"error,omitempty"`
+}
+
+// DatabaseMigrationResult returns row counts and execution duration of migration.
+type DatabaseMigrationResult struct {
+	Success    bool             `json:"success"`
+	TotalRows  int64            `json:"total_rows"`
+	TableRows  map[string]int64 `json:"table_rows"`
+	DurationMS int64            `json:"duration_ms"`
+	Message    string           `json:"message,omitempty"`
+	Error      string           `json:"error,omitempty"`
+}
+
+// InspectDatabaseStatus queries the currently active database for metrics and table rows.
+func InspectDatabaseStatus(db *gorm.DB, cfg *config.Config) *DatabaseStatus {
+	st := &DatabaseStatus{
+		Type:        "sqlite",
+		TableCounts: make(map[string]int64),
+	}
+	if cfg != nil {
+		st.DBPath = cfg.Database.DBPath
+		if cfg.Database.Type == "postgres" || (cfg.Database.Type == "auto" && strings.TrimSpace(cfg.Database.DSN) != "") {
+			st.Type = "postgres"
+			st.DSN = MaskDSN(cfg.Database.DSN)
+		}
+	}
+	if isPostgres(db) {
+		st.Type = "postgres"
+	}
+
+	if db != nil {
+		if sqlDB, err := db.DB(); err == nil {
+			stats := sqlDB.Stats()
+			st.OpenConns = stats.OpenConnections
+			st.InUse = stats.InUse
+			st.Idle = stats.Idle
+			st.MaxOpenConns = stats.MaxOpenConnections
+		}
+
+		// Count rows for major model tables
+		for _, m := range model.AllModels() {
+			if tbl, err := modelTableName(db, m); err == nil {
+				if db.Migrator().HasTable(tbl) {
+					var count int64
+					if err := db.Raw("SELECT COUNT(1) FROM " + quoteIdent(tbl)).Scan(&count).Error; err == nil {
+						st.TableCounts[tbl] = count
+					}
+				}
+			}
+		}
+	}
+	return st
+}
+
+// TestPostgres establishes a temporary connection to verify reachability and permissions.
+func TestPostgres(dsn string) (*PostgresTestResult, error) {
+	dsn = strings.TrimSpace(dsn)
+	if dsn == "" {
+		return &PostgresTestResult{
+			Success: false,
+			Error:   "PostgreSQL DSN 不能为空",
+		}, nil
+	}
+
+	start := time.Now()
+	testDB, err := gorm.Open(postgres.Open(dsn), &gorm.Config{
+		Logger: logger.Default.LogMode(logger.Silent),
+	})
+	if err != nil {
+		return &PostgresTestResult{
+			Success: false,
+			Error:   fmt.Sprintf("连接失败: %v", err),
+		}, nil
+	}
+
+	sqlDB, err := testDB.DB()
+	if err != nil {
+		return &PostgresTestResult{
+			Success: false,
+			Error:   fmt.Sprintf("获取底层连接失败: %v", err),
+		}, nil
+	}
+	defer sqlDB.Close()
+
+	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+	defer cancel()
+
+	if err := sqlDB.PingContext(ctx); err != nil {
+		return &PostgresTestResult{
+			Success: false,
+			Error:   fmt.Sprintf("Ping 超时或失败: %v", err),
+		}, nil
+	}
+
+	var version string
+	if err := testDB.WithContext(ctx).Raw("SELECT version()").Scan(&version).Error; err != nil {
+		version = "PostgreSQL (unknown version)"
+	}
+
+	latency := time.Since(start).Milliseconds()
+	return &PostgresTestResult{
+		Success:   true,
+		LatencyMS: latency,
+		Version:   version,
+		Message:   "连接成功",
+	}, nil
+}
+
+// MigrateCurrentToPostgres performs schema initialization and full table data copy into target PostgreSQL.
+func MigrateCurrentToPostgres(src *gorm.DB, targetDSN string, batchSize int, log *zap.Logger) (*DatabaseMigrationResult, error) {
+	targetDSN = strings.TrimSpace(targetDSN)
+	if targetDSN == "" {
+		return nil, fmt.Errorf("target PostgreSQL DSN cannot be empty")
+	}
+	if src == nil {
+		return nil, fmt.Errorf("current database is not available")
+	}
+
+	started := time.Now()
+	targetDB, err := gorm.Open(postgres.Open(targetDSN), &gorm.Config{
+		Logger: newGormLogger(log),
+	})
+	if err != nil {
+		return nil, fmt.Errorf("open target PostgreSQL: %w", err)
+	}
+	targetSQLDB, err := targetDB.DB()
+	if err == nil {
+		defer targetSQLDB.Close()
+	}
+
+	// 1. 初始化目标库 Schema、类型与索引
+	if err := AutoMigrate(targetDB); err != nil {
+		return nil, fmt.Errorf("auto migrate target PostgreSQL: %w", err)
+	}
+
+	// 2. 安全重置目标数据库的初始默认数据
+	if err := resetBootstrapTargetBeforeSQLiteMigrationIfSafe(src, targetDB, log); err != nil {
+		return nil, fmt.Errorf("reset target bootstrap data: %w", err)
+	}
+
+	// 3. 执行数据批量复制
+	tableRows, totalRows, err := copyModelTables(src, targetDB, batchSize)
+	if err != nil {
+		return nil, fmt.Errorf("copy tables: %w", err)
+	}
+
+	// 4. 标记迁移完成
+	if err := markSQLiteMigrationComplete(targetDB); err != nil {
+		return nil, fmt.Errorf("mark migration complete: %w", err)
+	}
+
+	duration := time.Since(started).Milliseconds()
+	return &DatabaseMigrationResult{
+		Success:    true,
+		TotalRows:  totalRows,
+		TableRows:  tableRows,
+		DurationMS: duration,
+		Message:    fmt.Sprintf("成功迁移 %d 条记录至 PostgreSQL", totalRows),
+	}, nil
+}
+
+// MaskDSN masks the password in a connection string for safe API responses.
+func MaskDSN(rawDSN string) string {
+	rawDSN = strings.TrimSpace(rawDSN)
+	if rawDSN == "" {
+		return ""
+	}
+	if u, err := url.Parse(rawDSN); err == nil && u.User != nil {
+		if pass, hasPassword := u.User.Password(); hasPassword && pass != "" {
+			rawUserPass := u.User.String()
+			user := u.User.Username()
+			maskedUserPass := user + ":******"
+			return strings.Replace(rawDSN, rawUserPass+"@", maskedUserPass+"@", 1)
+		}
+	}
+	// Fallback for keyword-style DSN (e.g. host=... password=...)
+	if strings.Contains(rawDSN, "password=") {
+		parts := strings.Fields(rawDSN)
+		for i, p := range parts {
+			if strings.HasPrefix(p, "password=") {
+				parts[i] = "password=******"
+			}
+		}
+		return strings.Join(parts, " ")
+	}
+	return rawDSN
+}

internal/database/database_admin_test.go

@@ -0,0 +1,71 @@
+package database
+
+import (
+	"testing"
+
+	"github.com/glebarez/sqlite"
+	"gorm.io/gorm"
+
+	"github.com/ShukeBta/MMTL/internal/config"
+	"github.com/ShukeBta/MMTL/internal/model"
+)
+
+func TestMaskDSN(t *testing.T) {
+	cases := []struct {
+		in   string
+		want string
+	}{
+		{
+			in:   "postgres://admin:secret123@localhost:5432/mmtl?sslmode=disable",
+			want: "postgres://admin:******@localhost:5432/mmtl?sslmode=disable",
+		},
+		{
+			in:   "host=localhost port=5432 user=admin password=secret dbname=mmtl sslmode=disable",
+			want: "host=localhost port=5432 user=admin password=****** dbname=mmtl sslmode=disable",
+		},
+		{
+			in:   "sqlite://data/mmtl.db",
+			want: "sqlite://data/mmtl.db",
+		},
+		{
+			in:   "",
+			want: "",
+		},
+	}
+
+	for _, c := range cases {
+		got := MaskDSN(c.in)
+		if got != c.want {
+			t.Errorf("MaskDSN(%q) = %q, want %q", c.in, got, c.want)
+		}
+	}
+}
+
+func TestInspectDatabaseStatus(t *testing.T) {
+	db, err := gorm.Open(sqlite.Open(":memory:"), &gorm.Config{})
+	if err != nil {
+		t.Fatal(err)
+	}
+	if err := db.AutoMigrate(&model.User{}, &model.Media{}); err != nil {
+		t.Fatal(err)
+	}
+	_ = db.Create(&model.User{Username: "testuser", PasswordHash: "h", Role: "user"}).Error
+
+	cfg := &config.Config{}
+	cfg.Database.Type = "sqlite"
+	cfg.Database.DBPath = "./data/mmtl.db"
+
+	st := InspectDatabaseStatus(db, cfg)
+	if st == nil {
+		t.Fatal("expected non-nil DatabaseStatus")
+	}
+	if st.Type != "sqlite" {
+		t.Fatalf("expected sqlite, got %s", st.Type)
+	}
+	if st.DBPath != "./data/mmtl.db" {
+		t.Fatalf("expected db_path, got %s", st.DBPath)
+	}
+	if st.TableCounts["users"] != 1 {
+		t.Fatalf("expected 1 user, got %d", st.TableCounts["users"])
+	}
+}

internal/database/database_test.go

@@ -159,7 +159,7 @@ func TestCopyModelTablesMigratesExistingSQLiteRows(t *testing.T) {
 		t.Fatal(err)
 	}
 
-	copied, err := copyModelTables(src, dst, 2)
+	_, copied, err := copyModelTables(src, dst, 2)
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -222,7 +222,7 @@ func TestCopyModelTablesResumesPartialSQLiteMigration(t *testing.T) {
 		t.Fatal(err)
 	}
 
-	copied, err := copyModelTables(src, dst, 2)
+	_, copied, err := copyModelTables(src, dst, 2)
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -240,7 +240,7 @@ func TestCopyModelTablesResumesPartialSQLiteMigration(t *testing.T) {
 		t.Fatalf("genres = %q, want %q", got.Genres, media.Genres)
 	}
 
-	copied, err = copyModelTables(src, dst, 2)
+	_, copied, err = copyModelTables(src, dst, 2)
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -332,7 +332,7 @@ func TestSQLiteMigrationFallsBackToDataDirDefaultPath(t *testing.T) {
 	if err := resetBootstrapTargetBeforeSQLiteMigrationIfSafe(src2, dst, nil); err != nil {
 		t.Fatal(err)
 	}
-	copied, err := copyModelTables(src2, dst, 2)
+	_, copied, err := copyModelTables(src2, dst, 2)
 	if err != nil {
 		t.Fatal(err)
 	}
@@ -408,13 +408,13 @@ func TestOpenSQLiteMigrationSourceUsesFallbackSourcePath(t *testing.T) {
 			_ = sqlDB2.Close()
 		}
 	}()
-	copied, err := copyModelTables(src2, dst, 2)
-	if err != nil {
-		t.Fatal(err)
-	}
-	if copied != 2 {
-		t.Fatalf("copied rows = %d, want 2", copied)
-	}
+		_, copied, err := copyModelTables(src2, dst, 2)
+		if err != nil {
+			t.Fatal(err)
+		}
+		if copied != 2 {
+			t.Fatalf("copied rows = %d, want 2", copied)
+		}
 	var userCount int64
 	if err := dst.Model(&model.User{}).Where("username = ?", "real-admin").Count(&userCount).Error; err != nil {
 		t.Fatal(err)

internal/database/sqlite_migration.go

@@ -48,7 +48,7 @@ func MigrateSQLiteToCurrentIfNeeded(cfg *config.Config, target *gorm.DB, log *za
 	if err := resetBootstrapTargetBeforeSQLiteMigrationIfSafe(src, target, log); err != nil {
 		return err
 	}
-	copied, err := copyModelTables(src, target, 500)
+	_, copied, err := copyModelTables(src, target, 500)
 	if err != nil {
 		return err
 	}

internal/database/sqlite_migration_copy.go

@@ -13,52 +13,53 @@ import (
 	"github.com/ShukeBta/MMTL/internal/model"
 )
 
-func copyModelTables(src, target *gorm.DB, batchSize int) (int64, error) {
+func copyModelTables(src, target *gorm.DB, batchSize int) (map[string]int64, int64, error) {
 	if batchSize <= 0 {
 		batchSize = 500
 	}
-	var copied int64
+	tableCounts := make(map[string]int64)
+	var totalCopied int64
 	for _, m := range model.AllModels() {
 		table, err := modelTableName(src, m)
 		if err != nil {
-			return copied, err
+			return tableCounts, totalCopied, err
 		}
 		primaryColumns, err := modelPrimaryColumns(src, m)
 		if err != nil {
-			return copied, fmt.Errorf("inspect model %T primary keys: %w", m, err)
+			return tableCounts, totalCopied, fmt.Errorf("inspect model %T primary keys: %w", m, err)
 		}
 		exists, err := sqliteTableExists(src, table)
 		if err != nil {
-			return copied, err
+			return tableCounts, totalCopied, err
 		}
 		if !exists {
 			continue
 		}
 		var sourceCount int64
 		if err := src.Raw("SELECT COUNT(1) FROM " + quoteIdent(table)).Scan(&sourceCount).Error; err != nil {
-			return copied, fmt.Errorf("count sqlite table %s: %w", table, err)
+			return tableCounts, totalCopied, fmt.Errorf("count sqlite table %s: %w", table, err)
 		}
 		if sourceCount == 0 {
 			continue
 		}
 		var targetCount int64
 		if err := target.Raw("SELECT COUNT(1) FROM " + quoteIdent(table)).Scan(&targetCount).Error; err != nil {
-			return copied, fmt.Errorf("count target table %s: %w", table, err)
+			return tableCounts, totalCopied, fmt.Errorf("count target table %s: %w", table, err)
 		}
 		modelType := reflect.TypeOf(m)
 		if modelType.Kind() != reflect.Ptr {
-			return copied, fmt.Errorf("model %T is not a pointer", m)
+			return tableCounts, totalCopied, fmt.Errorf("model %T is not a pointer", m)
 		}
 		sliceType := reflect.SliceOf(modelType.Elem())
 		slicePtr := reflect.New(sliceType)
 		if err := src.Unscoped().Find(slicePtr.Interface()).Error; err != nil {
-			return copied, fmt.Errorf("read sqlite table %s: %w", table, err)
+			return tableCounts, totalCopied, fmt.Errorf("read sqlite table %s: %w", table, err)
 		}
 		filtered := slicePtr.Elem()
 		if targetCount > 0 {
 			primaryKeySet, err := targetPrimaryKeySet(target, table, primaryColumns)
 			if err != nil {
-				return copied, err
+				return tableCounts, totalCopied, err
 			}
 			filtered = filterRowsMissingInTarget(target, table, primaryColumns, filtered, primaryKeySet)
 		}
@@ -68,11 +69,13 @@ func copyModelTables(src, target *gorm.DB, batchSize int) (int64, error) {
 		filteredPtr := reflect.New(filtered.Type())
 		filteredPtr.Elem().Set(filtered)
 		if err := target.Clauses(clause.OnConflict{DoNothing: true}).CreateInBatches(filteredPtr.Interface(), batchSize).Error; err != nil {
-			return copied, fmt.Errorf("copy sqlite table %s: %w", table, err)
+			return tableCounts, totalCopied, fmt.Errorf("copy sqlite table %s: %w", table, err)
 		}
-		copied += int64(filtered.Len())
+		copiedForTable := int64(filtered.Len())
+		tableCounts[table] = copiedForTable
+		totalCopied += copiedForTable
 	}
-	return copied, nil
+	return tableCounts, totalCopied, nil
 }
 
 func modelPrimaryColumns(db *gorm.DB, m any) ([]string, error) {

internal/database/sqlite_runtime.go

@@ -4,6 +4,7 @@ import (
 	"context"
 	"fmt"
 	"path/filepath"
+	"strings"
 
 	"gorm.io/gorm"
 
@@ -32,16 +33,37 @@ func installSQLiteWriteGate(db *gorm.DB) {
 			gate.Unlock()
 		}
 	}
+	rawLock := func(tx *gorm.DB) {
+		if tx.Statement != nil && isReadOnlySQL(tx.Statement.SQL.String()) {
+			return
+		}
+		lock(tx)
+	}
 	_ = db.Callback().Create().Before("gorm:create").Register("mmtl:sqlite_write_lock", lock)
 	_ = db.Callback().Create().After("gorm:create").Register("mmtl:sqlite_write_unlock", unlock)
 	_ = db.Callback().Update().Before("gorm:update").Register("mmtl:sqlite_write_lock", lock)
 	_ = db.Callback().Update().After("gorm:update").Register("mmtl:sqlite_write_unlock", unlock)
 	_ = db.Callback().Delete().Before("gorm:delete").Register("mmtl:sqlite_write_lock", lock)
 	_ = db.Callback().Delete().After("gorm:delete").Register("mmtl:sqlite_write_unlock", unlock)
-	_ = db.Callback().Raw().Before("gorm:raw").Register("mmtl:sqlite_write_lock", lock)
+	_ = db.Callback().Raw().Before("gorm:raw").Register("mmtl:sqlite_write_lock", rawLock)
 	_ = db.Callback().Raw().After("gorm:raw").Register("mmtl:sqlite_write_unlock", unlock)
 }
 
+func isReadOnlySQL(sql string) bool {
+	trimmed := strings.TrimSpace(sql)
+	if len(trimmed) == 0 {
+		return false
+	}
+	upper := strings.ToUpper(trimmed)
+	if strings.HasPrefix(upper, "SELECT") || strings.HasPrefix(upper, "EXPLAIN") {
+		return true
+	}
+	if strings.HasPrefix(upper, "WITH") && !strings.Contains(upper, "INSERT") && !strings.Contains(upper, "UPDATE") && !strings.Contains(upper, "DELETE") {
+		return true
+	}
+	return false
+}
+
 // sqliteWriteGate serializes in-process SQLite writes while respecting the
 // statement context, so request cancellation can break out of a queued write.
 type sqliteWriteGate struct {
@@ -84,7 +106,7 @@ func buildSQLiteDSN(cfg *config.Config) string {
 	}
 	dsn := dbPath + "?_pragma=foreign_keys(1)"
 	if cfg.Database.WALMode {
-		dsn += "&_pragma=journal_mode(WAL)"
+		dsn += "&_pragma=journal_mode(WAL)&_pragma=synchronous(NORMAL)"
 	}
 	if cfg.Database.BusyTimeout > 0 {
 		dsn += fmt.Sprintf("&_pragma=busy_timeout(%d)", cfg.Database.BusyTimeout)
@@ -92,6 +114,7 @@ func buildSQLiteDSN(cfg *config.Config) string {
 	if cfg.Database.CacheSize != 0 {
 		dsn += fmt.Sprintf("&_pragma=cache_size(%d)", cfg.Database.CacheSize)
 	}
+	dsn += "&_pragma=temp_store(MEMORY)&_pragma=mmap_size(268435456)"
 	return dsn
 }
 

internal/handler/admin_database.go

@@ -0,0 +1,145 @@
+package handler
+
+import (
+	"fmt"
+	"net/http"
+	"net/url"
+	"strings"
+
+	"github.com/gin-gonic/gin"
+
+	"github.com/ShukeBta/MMTL/internal/service"
+)
+
+type DatabaseConnectionPayload struct {
+	Type     string `json:"type"`
+	DSN      string `json:"dsn"`
+	Host     string `json:"host"`
+	Port     int    `json:"port"`
+	User     string `json:"user"`
+	Password string `json:"password"`
+	DBName   string `json:"dbname"`
+	SSLMode  string `json:"sslmode"`
+}
+
+func (p *DatabaseConnectionPayload) BuildDSN() string {
+	raw := strings.TrimSpace(p.DSN)
+	if raw != "" {
+		return raw
+	}
+	host := strings.TrimSpace(p.Host)
+	if host == "" {
+		return ""
+	}
+	port := p.Port
+	if port <= 0 {
+		port = 5432
+	}
+	user := strings.TrimSpace(p.User)
+	dbname := strings.TrimSpace(p.DBName)
+	if dbname == "" {
+		dbname = "mmtl"
+	}
+	sslmode := strings.TrimSpace(p.SSLMode)
+	if sslmode == "" {
+		sslmode = "disable"
+	}
+
+	userInfo := url.User(user)
+	if p.Password != "" {
+		userInfo = url.UserPassword(user, p.Password)
+	}
+
+	u := url.URL{
+		Scheme:   "postgres",
+		User:     userInfo,
+		Host:     fmt.Sprintf("%s:%d", host, port),
+		Path:     "/" + dbname,
+		RawQuery: "sslmode=" + url.QueryEscape(sslmode),
+	}
+	return u.String()
+}
+
+func getDatabaseStatusHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if svc.Database == nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": "database service unavailable"})
+			return
+		}
+		status := svc.Database.GetStatus(c.Request.Context())
+		c.JSON(http.StatusOK, status)
+	}
+}
+
+func testDatabaseHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		var req DatabaseConnectionPayload
+		if err := c.ShouldBindJSON(&req); err != nil {
+			c.JSON(http.StatusBadRequest, gin.H{"error": "invalid payload: " + err.Error()})
+			return
+		}
+		dsn := req.BuildDSN()
+		if dsn == "" {
+			c.JSON(http.StatusBadRequest, gin.H{"error": "请提供有效的 PostgreSQL 连接信息或 DSN"})
+			return
+		}
+		res, err := svc.Database.TestPostgres(c.Request.Context(), dsn)
+		if err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, res)
+	}
+}
+
+func migrateDatabaseHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		var req DatabaseConnectionPayload
+		if err := c.ShouldBindJSON(&req); err != nil {
+			c.JSON(http.StatusBadRequest, gin.H{"error": "invalid payload: " + err.Error()})
+			return
+		}
+		dsn := req.BuildDSN()
+		if dsn == "" {
+			c.JSON(http.StatusBadRequest, gin.H{"error": "请提供目标 PostgreSQL 连接信息或 DSN"})
+			return
+		}
+		res, err := svc.Database.MigrateToPostgres(c.Request.Context(), dsn)
+		if err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": "迁移失败: " + err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, res)
+	}
+}
+
+func saveDatabaseConfigHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		var req DatabaseConnectionPayload
+		if err := c.ShouldBindJSON(&req); err != nil {
+			c.JSON(http.StatusBadRequest, gin.H{"error": "invalid payload: " + err.Error()})
+			return
+		}
+		dbType := strings.ToLower(strings.TrimSpace(req.Type))
+		if dbType == "" {
+			dbType = "postgres"
+		}
+		var dsn string
+		if dbType == "postgres" {
+			dsn = req.BuildDSN()
+			if dsn == "" {
+				c.JSON(http.StatusBadRequest, gin.H{"error": "请提供有效的 PostgreSQL 连接信息或 DSN"})
+				return
+			}
+		}
+
+		if err := svc.Database.SaveConfig(c.Request.Context(), dbType, dsn); err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{
+			"message": "数据库配置已成功保存至配置文件，重启服务后将以新数据库运行",
+			"type":    dbType,
+		})
+	}
+}

internal/handler/admin_database_test.go

@@ -0,0 +1,101 @@
+package handler
+
+import (
+	"bytes"
+	"encoding/json"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+
+	"github.com/gin-gonic/gin"
+
+	"github.com/ShukeBta/MMTL/internal/config"
+	"github.com/ShukeBta/MMTL/internal/service"
+)
+
+func TestBuildDSN(t *testing.T) {
+	cases := []struct {
+		payload DatabaseConnectionPayload
+		want    string
+	}{
+		{
+			payload: DatabaseConnectionPayload{
+				DSN: "postgres://myuser:mypass@10.0.0.1:5432/mydb?sslmode=require",
+			},
+			want: "postgres://myuser:mypass@10.0.0.1:5432/mydb?sslmode=require",
+		},
+		{
+			payload: DatabaseConnectionPayload{
+				Host:     "127.0.0.1",
+				Port:     5432,
+				User:     "postgres",
+				Password: "secretpassword",
+				DBName:   "mmtl_prod",
+				SSLMode:  "disable",
+			},
+			want: "postgres://postgres:secretpassword@127.0.0.1:5432/mmtl_prod?sslmode=disable",
+		},
+	}
+
+	for _, c := range cases {
+		got := c.payload.BuildDSN()
+		if got != c.want {
+			t.Errorf("BuildDSN() = %q, want %q", got, c.want)
+		}
+	}
+}
+
+func TestGetDatabaseStatusHandler(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	cfg := &config.Config{}
+	cfg.Database.Type = "sqlite"
+	cfg.Database.DBPath = "./data/mmtl.db"
+
+	svc := &service.Container{
+		Database: service.NewDatabaseAdminService(cfg, nil, nil, nil),
+	}
+
+	r := gin.New()
+	r.GET("/api/admin/database/status", getDatabaseStatusHandler(svc))
+
+	req := httptest.NewRequest(http.MethodGet, "/api/admin/database/status", nil)
+	rec := httptest.NewRecorder()
+	r.ServeHTTP(rec, req)
+
+	if rec.Code != http.StatusOK {
+		t.Fatalf("expected status 200, got %d: %s", rec.Code, rec.Body.String())
+	}
+
+	var resp map[string]any
+	if err := json.Unmarshal(rec.Body.Bytes(), &resp); err != nil {
+		t.Fatalf("unmarshal response: %v", err)
+	}
+	if resp["type"] != "sqlite" {
+		t.Fatalf("expected type=sqlite, got %v", resp["type"])
+	}
+}
+
+func TestSaveDatabaseConfigHandler(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	dir := t.TempDir()
+	cfg := &config.Config{}
+	cfg.App.DataDir = dir
+	cfg.Database.Type = "sqlite"
+
+	svc := &service.Container{
+		Database: service.NewDatabaseAdminService(cfg, nil, nil, nil),
+	}
+
+	r := gin.New()
+	r.POST("/api/admin/database/save-config", saveDatabaseConfigHandler(svc))
+
+	body := bytes.NewBufferString(`{"type":"postgres","host":"localhost","port":5432,"user":"admin","password":"pwd","dbname":"mmtl"}`)
+	req := httptest.NewRequest(http.MethodPost, "/api/admin/database/save-config", body)
+	req.Header.Set("Content-Type", "application/json")
+	rec := httptest.NewRecorder()
+	r.ServeHTTP(rec, req)
+
+	if rec.Code != http.StatusOK {
+		t.Fatalf("expected status 200, got %d: %s", rec.Code, rec.Body.String())
+	}
+}

internal/handler/media.go

@@ -16,12 +16,13 @@ import (
 )
 
 type createLibraryReq struct {
-	Name     string                     `json:"name" binding:"required"`
-	Path     string                     `json:"path"`
-	Paths    []string                   `json:"paths"`
-	Roots    []service.LibraryRootInput `json:"roots"`
-	Type     string                     `json:"type"`
-	CoverURL string                     `json:"cover_url"`
+	Name              string                     `json:"name"`
+	Path              string                     `json:"path"`
+	Paths             []string                   `json:"paths"`
+	Roots             []service.LibraryRootInput `json:"roots"`
+	Type              string                     `json:"type"`
+	CoverURL          string                     `json:"cover_url"`
+	CreatePerSubfolder bool                      `json:"create_per_subfolder"`
 }
 
 func listLibrariesHandler(svc *service.Container) gin.HandlerFunc {
@@ -32,7 +33,7 @@ func listLibrariesHandler(svc *service.Container) gin.HandlerFunc {
 			return
 		}
 		role, _ := c.Get(middleware.CtxUserRole)
-		includeHidden := role == "admin" && (c.Query("include_hidden") == "1" || c.Query("all") == "1")
+		includeHidden := role == "admin" && (c.Query("include_hidden") == "1" || c.Query("include_hidden") == "true" || c.Query("all") == "1")
 		if !includeHidden {
 			libs = service.FilterDisplayCloudLibraries(c.Request.Context(), svc.Repo, libs)
 			visibility := mediaVisibilityForRequest(c, svc)
@@ -44,6 +45,22 @@ func listLibrariesHandler(svc *service.Container) gin.HandlerFunc {
 			}
 			libs = filtered
 		}
+		withPreview := c.Query("with_preview") == "1" || c.Query("with_preview") == "true"
+		if withPreview {
+			limit, _ := strconv.Atoi(c.DefaultQuery("preview_limit", c.DefaultQuery("limit", "10")))
+			if limit <= 0 {
+				limit = 10
+			} else if limit > 100 {
+				limit = 100
+			}
+			previews, err := svc.Media.ListLibrariesWithPreview(c.Request.Context(), libs, mediaVisibilityForRequest(c, svc), limit)
+			if err != nil {
+				c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+				return
+			}
+			c.JSON(http.StatusOK, previews)
+			return
+		}
 		c.JSON(http.StatusOK, libs)
 	}
 }
@@ -60,7 +77,7 @@ func getLibraryHandler(svc *service.Container) gin.HandlerFunc {
 			return
 		}
 		role, _ := c.Get(middleware.CtxUserRole)
-		includeHidden := role == "admin" && (c.Query("include_hidden") == "1" || c.Query("all") == "1")
+		includeHidden := role == "admin" && (c.Query("include_hidden") == "1" || c.Query("include_hidden") == "true" || c.Query("all") == "1")
 		if !includeHidden {
 			libs := service.FilterDisplayCloudLibraries(c.Request.Context(), svc.Repo, []model.Library{*lib})
 			if len(libs) == 0 || !service.LibraryVisibleForUser(c.Request.Context(), svc.Repo, libs[0], mediaVisibilityForRequest(c, svc)) {
@@ -88,9 +105,38 @@ func createLibraryHandler(svc *service.Container) gin.HandlerFunc {
 			}
 		}
 		if len(roots) == 0 && strings.TrimSpace(req.Path) != "" {
-			roots = append(roots, service.LibraryRootInput{Path: req.Path})
+roots = append(roots, service.LibraryRootInput{Path: req.Path})
+	}
+	var l *model.Library
+	if req.CreatePerSubfolder {
+		parent := ""
+		if len(roots) > 0 {
+			parent = roots[0].Path
+		} else if strings.TrimSpace(req.Path) != "" {
+			parent = req.Path
 		}
-		l, err := svc.Media.CreateLibraryWithRootsAndCover(c.Request.Context(), req.Name, req.Type, req.CoverURL, roots)
+		created, err := svc.Media.CreateLibrariesPerSubfolder(c.Request.Context(), parent, req.Type, req.CoverURL)
+		if err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+		uid, _ := c.Get("ctx_user_id")
+		for i := range created {
+			lib := &created[i]
+			svc.Audit.Record(c.Request.Context(), toString(uid), "library.create", lib.ID, c.ClientIP(), lib.Path)
+			if svc.Watcher != nil {
+				go func() { _ = svc.Watcher.Refresh(context.Background()) }()
+			}
+			for _, root := range lib.Roots {
+				if root.Enabled {
+					queueLibraryRootScan(svc, lib.ID, root.ID)
+				}
+			}
+		}
+		c.JSON(http.StatusCreated, gin.H{"libraries": created})
+		return
+	}
+	l, err := svc.Media.CreateLibraryWithRootsAndCover(c.Request.Context(), req.Name, req.Type, req.CoverURL, roots)
 		if err != nil {
 			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
 			return
@@ -117,7 +163,9 @@ func createLibraryHandler(svc *service.Container) gin.HandlerFunc {
 }
 
 type updateLibraryReq struct {
-	CoverURL string `json:"cover_url"`
+	CoverURL        *string `json:"cover_url"`
+	SortOrder       *int    `json:"sort_order"`
+	CarouselEnabled *bool   `json:"carousel_enabled"`
 }
 
 func updateLibraryHandler(svc *service.Container) gin.HandlerFunc {
@@ -127,9 +175,17 @@ func updateLibraryHandler(svc *service.Container) gin.HandlerFunc {
 			c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
 			return
 		}
-		if err := svc.Media.UpdateLibraryCover(c.Request.Context(), c.Param("id"), req.CoverURL); err != nil {
-			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
-			return
+		if req.CoverURL != nil {
+			if err := svc.Media.UpdateLibraryCover(c.Request.Context(), c.Param("id"), *req.CoverURL); err != nil {
+				c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+				return
+			}
+		}
+		if req.SortOrder != nil || req.CarouselEnabled != nil {
+			if err := svc.Media.UpdateLibraryFields(c.Request.Context(), c.Param("id"), req.SortOrder, req.CarouselEnabled); err != nil {
+				c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+				return
+			}
 		}
 		lib, err := svc.Repo.Library.FindByID(c.Request.Context(), c.Param("id"))
 		if err != nil || lib == nil {
@@ -140,6 +196,25 @@ func updateLibraryHandler(svc *service.Container) gin.HandlerFunc {
 	}
 }
 
+type reorderLibrariesReq struct {
+	IDs []string `json:"ids" binding:"required"`
+}
+
+func reorderLibrariesHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		var req reorderLibrariesReq
+		if err := c.ShouldBindJSON(&req); err != nil {
+			c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+			return
+		}
+		if err := svc.Media.ReorderLibraries(c.Request.Context(), req.IDs); err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"updated": len(req.IDs)})
+	}
+}
+
 func deleteLibraryHandler(svc *service.Container) gin.HandlerFunc {
 	return func(c *gin.Context) {
 		id := c.Param("id")

internal/handler/routes_admin.go

@@ -20,6 +20,29 @@ func registerAdminRoutes(api *gin.RouterGroup, cfg *config.Config, svc *service.
 	registerAdminAPIConfigRoutes(admin, svc)
 	registerAdminRecognitionWordRoutes(admin, svc)
 	registerAdminStrmRoutes(admin, svc)
+	registerAdminScraperRoutes(admin, svc)
+	registerAdminDatabaseRoutes(admin, svc)
+
+	// FFmpeg/FFprobe 工具：状态查询 + 一键下载安装（自动匹配当前平台）。
+	admin.GET("/tools/ffmpeg/status", ffToolsStatusHandler(svc))
+	admin.POST("/tools/ffmpeg/install", ffToolsInstallHandler(svc))
+}
+
+func registerAdminScraperRoutes(admin *gin.RouterGroup, svc *service.Container) {
+	admin.GET("/scraper/queue", listScrapeQueueHandler(svc))
+	admin.POST("/scraper/queue/:id/cancel", cancelScrapeTaskHandler(svc))
+	admin.POST("/scraper/queue/:id/retry", retryScrapeTaskHandler(svc))
+	admin.DELETE("/scraper/queue/:id", deleteScrapeTaskHandler(svc))
+	admin.POST("/scraper/queue/batch", batchActionScrapeTasksHandler(svc))
+	admin.POST("/scraper/queue/clear-done", clearDoneScrapeTasksHandler(svc))
+	admin.POST("/scraper/queue/clear-finished", clearFinishedScrapeTasksHandler(svc))
+	admin.POST("/scraper/queue/clear-canceled", clearCanceledScrapeTasksHandler(svc))
+	admin.POST("/scraper/queue/retry-failed", retryAllFailedScrapeTasksHandler(svc))
+	admin.POST("/scraper/queue/cancel-pending", cancelPendingScrapeTasksHandler(svc))
+	admin.POST("/scraper/queue/enqueue-library/:id", enqueueLibraryScrapeHandler(svc))
+	admin.POST("/scraper/queue/enqueue-all", enqueueAllScrapeHandler(svc))
+	admin.POST("/media/repair-rescrape", enqueueAllScrapeHandler(svc))
+	admin.POST("/libraries/:id/repair-rescrape", enqueueLibraryScrapeHandler(svc))
 }
 
 func registerAdminStrmRoutes(admin *gin.RouterGroup, svc *service.Container) {
@@ -43,18 +66,27 @@ func registerAdminStrmRoutes(admin *gin.RouterGroup, svc *service.Container) {
 	admin.POST("/strm/paths/:id/sync", startStrmSyncHandler(svc))
 	admin.POST("/strm/paths/:id/cancel", cancelStrmSyncHandler(svc))
 	admin.GET("/strm/records", listStrmSyncRecordsHandler(svc))
+	admin.DELETE("/strm/records/:id", deleteStrmSyncRecordHandler(svc))
+	admin.DELETE("/strm/records", clearStrmSyncRecordsHandler(svc))
 	admin.GET("/strm/local-dirs", listStrmLocalDirsHandler(svc))
 
 	admin.GET("/strm/downloads", downloadQueueHandler(svc))
 	admin.POST("/strm/downloads/:id/cancel", cancelStrmDownloadHandler(svc))
 	admin.POST("/strm/downloads/:id/retry", retryStrmDownloadHandler(svc))
+	admin.DELETE("/strm/downloads/:id", deleteStrmDownloadHandler(svc))
+	admin.POST("/strm/downloads/batch", batchActionDownloadsHandler(svc))
 	admin.POST("/strm/downloads/clear-done", clearDoneDownloadsHandler(svc))
 	admin.POST("/strm/downloads/clear-finished", clearFinishedDownloadsHandler(svc))
+	admin.POST("/strm/downloads/clear-canceled", clearCanceledDownloadsHandler(svc))
 	admin.POST("/strm/downloads/retry-failed", retryAllFailedDownloadsHandler(svc))
 	admin.POST("/strm/downloads/cancel-pending", cancelPendingDownloadsHandler(svc))
 	admin.GET("/strm/uploads", uploadQueueHandler(svc))
 	admin.POST("/strm/uploads/:id/cancel", cancelStrmUploadHandler(svc))
 	admin.POST("/strm/uploads/:id/retry", retryStrmUploadHandler(svc))
+	admin.DELETE("/strm/uploads/:id", deleteStrmUploadHandler(svc))
+	admin.POST("/strm/uploads/batch", batchActionUploadsHandler(svc))
+	admin.POST("/strm/uploads/cancel-pending", cancelPendingUploadsHandler(svc))
+	admin.POST("/strm/uploads/clear-canceled", clearCanceledUploadsHandler(svc))
 }
 
 func registerAdminUserRoutes(admin *gin.RouterGroup, svc *service.Container) {
@@ -111,3 +143,10 @@ func registerAdminRecognitionWordRoutes(admin *gin.RouterGroup, svc *service.Con
 	admin.POST("/recognition-words/sync", syncRecognitionWordsHandler(svc))
 	admin.POST("/recognition-words/test", testRecognitionWordsHandler(svc))
 }
+
+func registerAdminDatabaseRoutes(admin *gin.RouterGroup, svc *service.Container) {
+	admin.GET("/database/status", getDatabaseStatusHandler(svc))
+	admin.POST("/database/test", testDatabaseHandler(svc))
+	admin.POST("/database/migrate", migrateDatabaseHandler(svc))
+	admin.POST("/database/save-config", saveDatabaseConfigHandler(svc))
+}

internal/handler/routes_authenticated_core.go

@@ -21,6 +21,7 @@ func registerAuthedLibraryRoutes(authed *gin.RouterGroup, svc *service.Container
 	authed.POST("/libraries", middleware.AdminRequired(), createLibraryHandler(svc))
 	authed.GET("/libraries/:id", getLibraryHandler(svc))
 	authed.PATCH("/libraries/:id", middleware.AdminRequired(), updateLibraryHandler(svc))
+	authed.PUT("/libraries/reorder", middleware.AdminRequired(), reorderLibrariesHandler(svc))
 	authed.DELETE("/libraries/:id", middleware.AdminRequired(), deleteLibraryHandler(svc))
 	authed.GET("/libraries/:id/roots", middleware.AdminRequired(), listLibraryRootsHandler(svc))
 	authed.POST("/libraries/:id/roots", middleware.AdminRequired(), createLibraryRootHandler(svc))
@@ -38,6 +39,7 @@ func registerAuthedLibraryRoutes(authed *gin.RouterGroup, svc *service.Container
 
 func registerAuthedMediaRoutes(authed *gin.RouterGroup, svc *service.Container) {
 	authed.GET("/media/:id", getMediaHandler(svc))
+	authed.GET("/media/:id/episodes", listMediaEpisodesHandler(svc))
 	authed.GET("/media", searchMediaHandler(svc))
 	authed.PATCH("/media/:id/metadata", middleware.AdminRequired(), updateMediaMetadataHandler(svc))
 	authed.POST("/media/:id/scrape", middleware.AdminRequired(), scrapeOneHandler(svc))

internal/handler/scraper_queue.go

@@ -0,0 +1,167 @@
+package handler
+
+import (
+	"net/http"
+	"strconv"
+
+	"github.com/gin-gonic/gin"
+
+	"github.com/ShukeBta/MMTL/internal/service"
+)
+
+func listScrapeQueueHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		page, _ := strconv.Atoi(c.DefaultQuery("page", "1"))
+		pageSize, _ := strconv.Atoi(c.DefaultQuery("page_size", "50"))
+		snap, err := svc.Scraper.ScrapeQueueSnapshot(c.Request.Context(), c.Query("status"), page, pageSize)
+		if err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, snap)
+	}
+}
+
+func cancelScrapeTaskHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if err := svc.Scraper.CancelScrapeTask(c.Request.Context(), c.Param("id")); err != nil {
+			c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"ok": true})
+	}
+}
+
+func retryScrapeTaskHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if err := svc.Scraper.RetryScrapeTask(c.Request.Context(), c.Param("id")); err != nil {
+			c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"ok": true})
+	}
+}
+
+func deleteScrapeTaskHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if err := svc.Scraper.DeleteScrapeTask(c.Request.Context(), c.Param("id")); err != nil {
+			c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"ok": true})
+	}
+}
+
+func batchActionScrapeTasksHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		var req queueBatchReq
+		if err := c.ShouldBindJSON(&req); err != nil {
+			c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+			return
+		}
+		n, err := svc.Scraper.BatchActionScrapeTasks(c.Request.Context(), req.Action, req.IDs)
+		if err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"affected": n, "action": req.Action})
+	}
+}
+
+func clearDoneScrapeTasksHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		n, err := svc.Scraper.ClearDoneScrapeTasks(c.Request.Context())
+		if err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"deleted": n})
+	}
+}
+
+func clearFinishedScrapeTasksHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		n, err := svc.Scraper.ClearFinishedScrapeTasks(c.Request.Context())
+		if err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"deleted": n})
+	}
+}
+
+func clearCanceledScrapeTasksHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		n, err := svc.Scraper.ClearCanceledScrapeTasks(c.Request.Context())
+		if err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"deleted": n})
+	}
+}
+
+func retryAllFailedScrapeTasksHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		n, err := svc.Scraper.RetryAllFailedScrapeTasks(c.Request.Context())
+		if err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"retried": n})
+	}
+}
+
+func cancelPendingScrapeTasksHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		n, err := svc.Scraper.CancelPendingScrapeTasks(c.Request.Context())
+		if err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"canceled": n})
+	}
+}
+
+type enqueueScrapeReq struct {
+	EpisodeImages  bool `json:"episode_images"`
+	EpisodeArtwork bool `json:"episode_artwork"`
+	RefreshMatched bool `json:"refresh_matched"`
+	IncludeMatched bool `json:"include_matched"`
+}
+
+func (r enqueueScrapeReq) toOptions() service.ScrapeOptions {
+	epArtwork := r.EpisodeImages || r.EpisodeArtwork
+	return service.ScrapeOptions{
+		EpisodeArtwork: &epArtwork,
+		IncludeMatched: r.IncludeMatched || r.RefreshMatched,
+		RetryNoMatch:   true,
+	}
+}
+
+func enqueueLibraryScrapeHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		var req enqueueScrapeReq
+		_ = c.ShouldBindJSON(&req)
+		libID := c.Param("id")
+		n, err := svc.Scraper.EnqueueLibrary(c.Request.Context(), libID, req.toOptions())
+		if err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"enqueued": n})
+	}
+}
+
+func enqueueAllScrapeHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		var req enqueueScrapeReq
+		_ = c.ShouldBindJSON(&req)
+		n, err := svc.Scraper.EnqueueAll(c.Request.Context(), req.toOptions())
+		if err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"enqueued": n})
+	}
+}

internal/handler/series.go

@@ -128,3 +128,19 @@ func listLibrarySeriesEpisodesHandler(svc *service.Container) gin.HandlerFunc {
 		c.JSON(http.StatusOK, gin.H{"items": items, "total": len(items)})
 	}
 }
+
+func listMediaEpisodesHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		id := c.Param("id")
+		if id == "" {
+			c.JSON(http.StatusBadRequest, gin.H{"error": "id is required"})
+			return
+		}
+		items, err := svc.Media.ListMediaEpisodes(c.Request.Context(), id, mediaVisibilityForRequest(c, svc))
+		if err != nil {
+			writeInternalOrCanceled(c, err)
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"items": items, "total": len(items)})
+	}
+}

internal/handler/streaming.go

@@ -2,7 +2,6 @@
 package handler
 
 import (
-	"context"
 	"errors"
 	"io"
 	"net/http"
@@ -135,28 +134,12 @@ func scrapeOneHandler(svc *service.Container) gin.HandlerFunc {
 			return
 		}
 		options.IncludeMatched = true
-		m, err := svc.Repo.Media.FindByID(c.Request.Context(), c.Param("id"))
-		if err != nil || m == nil {
-			c.JSON(http.StatusNotFound, gin.H{"error": "not found"})
-			return
-		}
-		task := startScrapeHTTPTask(svc, "手动刮削媒体", m.Title, m.Path)
-		if err := svc.Scraper.EnrichOneWithOptions(c.Request.Context(), m, options); err != nil {
-			finishHTTPTask(task, err, "scrape", "手动刮削媒体失败", nil, nil)
+		task, err := svc.Scraper.EnqueueMedia(c.Request.Context(), c.Param("id"), options)
+		if err != nil {
 			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
 			return
 		}
-		reclassified := reclassifyMediaAfterScrape(c.Request.Context(), svc, m.ID)
-		refreshed, _ := svc.Repo.Media.FindByID(c.Request.Context(), m.ID)
-		metrics := map[string]int64{"processed": 1}
-		if refreshed != nil && refreshed.ScrapeStatus == "matched" {
-			metrics["matched"] = 1
-		}
-		if reclassified > 0 {
-			metrics["reclassified"] = int64(reclassified)
-		}
-		finishHTTPTask(task, nil, "completed", "手动刮削媒体结束", metrics, nil)
-		c.JSON(http.StatusOK, refreshed)
+		c.JSON(http.StatusOK, task)
 	}
 }
 
@@ -170,40 +153,12 @@ func scrapeLibraryHandler(svc *service.Container) gin.HandlerFunc {
 			return
 		}
 		options.IncludeMatched = true
-		var task *service.TaskHandle
-		if lib, err := svc.Repo.Library.FindByID(c.Request.Context(), libID); err == nil && lib != nil {
-			task = startScrapeHTTPTask(svc, "手动刮削媒体库", lib.Name, lib.Path)
-		} else {
-			task = startScrapeHTTPTask(svc, "手动刮削媒体库", libID, "")
+		n, err := svc.Scraper.EnqueueLibrary(c.Request.Context(), libID, options)
+		if err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
 		}
-		// Run in the background so HTTP returns instantly; the WS hub
-		// pushes per-item progress on the "scrape" topic.
-		go func(libID string, task *service.TaskHandle, options service.ScrapeOptions) {
-			result, err := svc.Scraper.EnrichLibraryDetailedWithOptions(context.Background(), libID, options)
-			reclassified := 0
-			if result.Processed > 0 {
-				reclassified = reclassifyLibraryAfterScrape(context.Background(), svc, libID)
-			}
-			metrics := map[string]int64{
-				"matched":    int64(result.Matched),
-				"processed":  int64(result.Processed),
-				"candidates": int64(result.Candidates),
-			}
-			if reclassified > 0 {
-				metrics["reclassified"] = int64(reclassified)
-			}
-			if result.Failed > 0 {
-				metrics["errors"] = int64(result.Failed)
-			}
-			stage := "completed"
-			message := "手动刮削媒体库结束"
-			if err != nil {
-				stage = "scrape"
-				message = "手动刮削媒体库失败"
-			}
-			finishHTTPTask(task, err, stage, message, metrics, nil)
-		}(libID, task, options)
-		c.JSON(http.StatusAccepted, gin.H{"status": "scraping"})
+		c.JSON(http.StatusOK, gin.H{"status": "queued", "enqueued": n})
 	}
 }
 

internal/handler/strm.go

@@ -178,6 +178,7 @@ type strmSyncPathReq struct {
 	DeleteDir      *bool  `json:"delete_dir"`
 	Cron           string `json:"cron"`
 	EnableCron     *bool  `json:"enable_cron"`
+	SyncMode       string `json:"sync_mode"`
 	Enabled        *bool  `json:"enabled"`
 }
 
@@ -261,7 +262,16 @@ func deleteStrmSyncPathHandler(svc *service.Container) gin.HandlerFunc {
 
 func startStrmSyncHandler(svc *service.Container) gin.HandlerFunc {
 	return func(c *gin.Context) {
-		if err := svc.Strm.StartSync(c.Request.Context(), c.Param("id")); err != nil {
+		mode := c.Query("mode")
+		if mode == "" {
+			var body struct {
+				Mode string `json:"mode"`
+			}
+			if err := c.ShouldBindJSON(&body); err == nil && body.Mode != "" {
+				mode = body.Mode
+			}
+		}
+		if err := svc.Strm.StartSync(c.Request.Context(), c.Param("id"), mode); err != nil {
 			c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
 			return
 		}
@@ -290,6 +300,31 @@ func listStrmSyncRecordsHandler(svc *service.Container) gin.HandlerFunc {
 	}
 }
 
+func deleteStrmSyncRecordHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if c.Param("id") == "" {
+			c.JSON(http.StatusBadRequest, gin.H{"error": "缺少记录 ID"})
+			return
+		}
+		if err := svc.Strm.DeleteSyncRecord(c.Request.Context(), c.Param("id")); err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"ok": true})
+	}
+}
+
+func clearStrmSyncRecordsHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		deleted, err := svc.Strm.ClearSyncRecords(c.Request.Context(), c.Query("path_id"))
+		if err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"ok": true, "deleted": deleted})
+	}
+}
+
 // ─── 下载/上传队列 ─────────────────────────────────────────────────────────────
 
 func downloadQueueHandler(svc *service.Container) gin.HandlerFunc {
@@ -358,6 +393,63 @@ func retryStrmUploadHandler(svc *service.Container) gin.HandlerFunc {
 	}
 }
 
+func deleteStrmDownloadHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if err := svc.Strm.DeleteDownloadTask(c.Request.Context(), c.Param("id")); err != nil {
+			c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"ok": true})
+	}
+}
+
+func deleteStrmUploadHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if err := svc.Strm.DeleteUploadTask(c.Request.Context(), c.Param("id")); err != nil {
+			c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"ok": true})
+	}
+}
+
+type queueBatchReq struct {
+	Action string   `json:"action" binding:"required"`
+	IDs    []string `json:"ids" binding:"required"`
+}
+
+func batchActionDownloadsHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		var req queueBatchReq
+		if err := c.ShouldBindJSON(&req); err != nil {
+			c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+			return
+		}
+		n, err := svc.Strm.BatchActionDownloadTasks(c.Request.Context(), req.Action, req.IDs)
+		if err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"affected": n, "action": req.Action})
+	}
+}
+
+func batchActionUploadsHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		var req queueBatchReq
+		if err := c.ShouldBindJSON(&req); err != nil {
+			c.JSON(http.StatusBadRequest, gin.H{"error": err.Error()})
+			return
+		}
+		n, err := svc.Strm.BatchActionUploadTasks(c.Request.Context(), req.Action, req.IDs)
+		if err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"affected": n, "action": req.Action})
+	}
+}
+
 // ─── 下载队列批量操作 ─────────────────────────────────────────────────────────
 
 func clearDoneDownloadsHandler(svc *service.Container) gin.HandlerFunc {
@@ -382,6 +474,28 @@ func clearFinishedDownloadsHandler(svc *service.Container) gin.HandlerFunc {
 	}
 }
 
+func clearCanceledDownloadsHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		n, err := svc.Strm.ClearCanceledDownloadTasks(c.Request.Context())
+		if err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"deleted": n})
+	}
+}
+
+func clearCanceledUploadsHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		n, err := svc.Strm.ClearCanceledUploadTasks(c.Request.Context())
+		if err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"deleted": n})
+	}
+}
+
 func retryAllFailedDownloadsHandler(svc *service.Container) gin.HandlerFunc {
 	return func(c *gin.Context) {
 		n, err := svc.Strm.RetryAllFailedDownloadTasks(c.Request.Context())
@@ -404,6 +518,17 @@ func cancelPendingDownloadsHandler(svc *service.Container) gin.HandlerFunc {
 	}
 }
 
+func cancelPendingUploadsHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		n, err := svc.Strm.CancelPendingUploadTasks(c.Request.Context())
+		if err != nil {
+			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, gin.H{"canceled": n})
+	}
+}
+
 // ─── 公开播放端点 ──────────────────────────────────────────────────────────────
 
 // strmPlayHandler 处理 strm 文件指向的播放请求（Emby/Infuse 直接请求，无 JWT）。
@@ -458,6 +583,7 @@ func strmSyncPathFromReq(req strmSyncPathReq) *model.StrmSyncPath {
 		DeleteDir:      boolValue(req.DeleteDir, false),
 		Cron:           strings.TrimSpace(req.Cron),
 		EnableCron:     boolValue(req.EnableCron, false),
+		SyncMode:       strings.TrimSpace(req.SyncMode),
 		Enabled:        boolValue(req.Enabled, true),
 	}
 }

internal/handler/strm_routes_test.go

@@ -50,10 +50,16 @@ func TestStrmAdminRoutesAreRegistered(t *testing.T) {
 		"GET /api/admin/strm/downloads",
 		"POST /api/admin/strm/downloads/:id/cancel",
 		"POST /api/admin/strm/downloads/:id/retry",
-		"GET /api/admin/strm/uploads",
-		"POST /api/admin/strm/uploads/:id/cancel",
-		"POST /api/admin/strm/uploads/:id/retry",
-		"GET /api/strm/play/:provider/:file",
+			"POST /api/admin/strm/downloads/clear-finished",
+			"POST /api/admin/strm/downloads/clear-canceled",
+			"POST /api/admin/strm/downloads/retry-failed",
+			"POST /api/admin/strm/downloads/cancel-pending",
+			"GET /api/admin/strm/uploads",
+			"POST /api/admin/strm/uploads/:id/cancel",
+			"POST /api/admin/strm/uploads/:id/retry",
+			"POST /api/admin/strm/uploads/cancel-pending",
+			"POST /api/admin/strm/uploads/clear-canceled",
+			"GET /api/strm/play/:provider/:file",
 	} {
 		if !routes[want] {
 			t.Fatalf("%s route is not registered", want)

internal/handler/subtitles.go

@@ -31,7 +31,7 @@ func serveSubtitleHandler(svc *service.Container) gin.HandlerFunc {
 			return
 		}
 		c.Header("Content-Type", "text/vtt; charset=utf-8")
-		c.Header("Cache-Control", "public, max-age=3600")
+		c.Header("Cache-Control", "no-cache, no-store, must-revalidate")
 		if err := svc.Subtitle.Serve(c.Request.Context(), c.Param("id"), path, c.Writer); err != nil {
 			c.JSON(http.StatusNotFound, gin.H{"error": err.Error()})
 			return

internal/handler/system_handler.go

@@ -1,144 +0,0 @@
-// Package handler — system tools detection.
-package handler
-
-import (
-	"net/http"
-	"os/exec"
-	"strings"
-
-	"github.com/gin-gonic/gin"
-	"go.uber.org/zap"
-
-	"github.com/ShukeBta/MMTL/internal/config"
-	"github.com/ShukeBta/MMTL/internal/service"
-)
-
-// SystemHandler handles system-related endpoints.
-type SystemHandler struct {
-	cfg *config.Config
-	log *zap.Logger
-	svc *service.Container
-}
-
-// NewSystemHandler is the constructor.
-func NewSystemHandler(cfg *config.Config, log *zap.Logger, svc *service.Container) *SystemHandler {
-	return &SystemHandler{cfg: cfg, log: log, svc: svc}
-}
-
-// ToolStatus represents the detection status of a system tool.
-type ToolStatus struct {
-	Name        string `json:"name"`
-	DisplayName string `json:"display_name"`
-	ConfigKey   string `json:"config_key"`
-	Path        string `json:"path,omitempty"`
-	Detected    bool   `json:"detected"`
-	Version     string `json:"version,omitempty"`
-}
-
-// GetToolsStatus returns the status of system tools.
-func (h *SystemHandler) GetToolsStatus(c *gin.Context) {
-	tools := []ToolStatus{
-		{Name: "ffprobe", DisplayName: "FFprobe", ConfigKey: "app.ffprobe_path"},
-		{Name: "ffmpeg", DisplayName: "FFmpeg", ConfigKey: "app.ffmpeg_path"},
-	}
-
-	for i := range tools {
-		// Check configured path first
-		var configuredPath string
-		switch tools[i].ConfigKey {
-		case "app.ffprobe_path":
-			configuredPath = h.cfg.App.FFprobePath
-			if configuredPath == "" {
-				configuredPath = "ffprobe"
-			}
-		case "app.ffmpeg_path":
-			configuredPath = h.cfg.App.FFmpegPath
-			if configuredPath == "" {
-				configuredPath = "ffmpeg"
-			}
-		}
-
-		// Try to find the tool
-		path, err := exec.LookPath(configuredPath)
-		if err == nil {
-			tools[i].Detected = true
-			tools[i].Path = path
-			// Try to get version
-			tools[i].Version = getToolVersion(path)
-		}
-	}
-
-	c.JSON(http.StatusOK, gin.H{
-		"tools": tools,
-	})
-}
-
-// getToolVersion attempts to get the version of a tool.
-func getToolVersion(path string) string {
-	out, err := exec.Command(path, "-version").Output()
-	if err != nil {
-		return ""
-	}
-
-	// Extract first line as version info
-	lines := strings.Split(string(out), "\n")
-	if len(lines) > 0 {
-		return strings.TrimSpace(lines[0])
-	}
-	return ""
-}
-
-// InstallTools attempts to auto-install system tools (ffmpeg/ffprobe)
-func (h *SystemHandler) InstallTools(c *gin.Context) {
-	h.log.Info("Received tools auto-install request")
-
-	// Call service layer to auto-install
-	ffprobePath, ffmpegPath := service.AutoInstallFFmpeg(h.log, h.cfg)
-
-	result := gin.H{
-		"installed": ffprobePath != "" || ffmpegPath != "",
-	}
-
-	if ffprobePath != "" {
-		result["ffprobe_path"] = ffprobePath
-		result["ffprobe_installed"] = true
-	}
-	if ffmpegPath != "" {
-		result["ffmpeg_path"] = ffmpegPath
-		result["ffmpeg_installed"] = true
-	}
-
-	// Re-detect tool status
-	tools := []ToolStatus{
-		{Name: "ffprobe", DisplayName: "FFprobe", ConfigKey: "app.ffprobe_path"},
-		{Name: "ffmpeg", DisplayName: "FFmpeg", ConfigKey: "app.ffmpeg_path"},
-	}
-
-	for i := range tools {
-		var configuredPath string
-		switch tools[i].ConfigKey {
-		case "app.ffprobe_path":
-			configuredPath = h.cfg.App.FFprobePath
-			if configuredPath == "" {
-				configuredPath = "ffprobe"
-			}
-		case "app.ffmpeg_path":
-			configuredPath = h.cfg.App.FFmpegPath
-			if configuredPath == "" {
-				configuredPath = "ffmpeg"
-			}
-		}
-
-		path, err := exec.LookPath(configuredPath)
-		if err == nil {
-			tools[i].Detected = true
-			tools[i].Path = path
-			tools[i].Version = getToolVersion(path)
-		}
-	}
-
-	result["tools"] = tools
-
-	h.log.Info("Tool installation completed", zap.Any("result", result))
-	c.JSON(http.StatusOK, result)
-}

internal/handler/tools_ffmpeg.go

@@ -0,0 +1,38 @@
+// Package handler — FFmpeg/FFprobe 工具安装端点。
+package handler
+
+import (
+	"net/http"
+
+	"github.com/gin-gonic/gin"
+
+	"github.com/ShukeBta/MMTL/internal/service"
+)
+
+// ffToolsStatusHandler 返回 ffmpeg/ffprobe 当前安装状态
+// （GET /api/admin/tools/ffmpeg/status）。
+func ffToolsStatusHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if svc == nil || svc.FFTools == nil {
+			c.JSON(http.StatusServiceUnavailable, gin.H{"error": "FFmpeg 工具服务不可用"})
+			return
+		}
+		c.JSON(http.StatusOK, svc.FFTools.Status(c.Request.Context()))
+	}
+}
+
+// ffToolsInstallHandler 触发后台下载安装（POST /api/admin/tools/ffmpeg/install）。
+// 自动匹配当前运行环境（OS+架构），安装到 data/tools/ffmpeg/ 并把路径写入设置。
+func ffToolsInstallHandler(svc *service.Container) gin.HandlerFunc {
+	return func(c *gin.Context) {
+		if svc == nil || svc.FFTools == nil {
+			c.JSON(http.StatusServiceUnavailable, gin.H{"error": "FFmpeg 工具服务不可用"})
+			return
+		}
+		if err := svc.FFTools.StartInstall(c.Request.Context()); err != nil {
+			c.JSON(http.StatusConflict, gin.H{"error": err.Error()})
+			return
+		}
+		c.JSON(http.StatusOK, svc.FFTools.Status(c.Request.Context()))
+	}
+}

internal/handler/watch_history.go

@@ -163,7 +163,7 @@ func historyDeleteHandler(svc *service.Container) gin.HandlerFunc {
 			c.JSON(http.StatusBadRequest, gin.H{"error": "status must be completed or incomplete"})
 			return
 		}
-		res := q.Delete(&model.PlaybackHistory{})
+			res := q.Unscoped().Delete(&model.PlaybackHistory{})
 		if err := res.Error; err != nil {
 			c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
 			return

internal/model/library_media.go

@@ -3,12 +3,14 @@ package model
 // Library 表示一个逻辑媒体库。Path 保留为兼容字段，指向第一条 LibraryRoot。
 type Library struct {
 	Base
-	Name     string        `gorm:"size:128;not null" json:"name"`
-	Path     string        `gorm:"size:1024;not null" json:"path"`
-	Type     string        `gorm:"size:16;not null;default:movie" json:"type"` // movie / tv / anime / music
-	CoverURL string        `gorm:"size:1024" json:"cover_url,omitempty"`
-	Enabled  bool          `gorm:"default:true" json:"enabled"`
-	Roots    []LibraryRoot `gorm:"foreignKey:LibraryID" json:"roots,omitempty"`
+	Name            string        `gorm:"size:128;not null" json:"name"`
+	Path            string        `gorm:"size:1024;not null" json:"path"`
+	Type            string        `gorm:"size:16;not null;default:movie" json:"type"` // movie / tv / anime / music
+	CoverURL        string        `gorm:"size:1024" json:"cover_url,omitempty"`
+	Enabled         bool          `gorm:"default:true" json:"enabled"`
+	SortOrder       int           `gorm:"index;default:0" json:"sort_order"`            // 手动拖拽排序用，越小越靠前
+	CarouselEnabled bool          `gorm:"default:false" json:"carousel_enabled"`        // 是否参与首页海报轮播（默认不参与）
+	Roots           []LibraryRoot `gorm:"foreignKey:LibraryID" json:"roots,omitempty"`
 }
 
 // LibraryRoot 是逻辑媒体库下的一条真实物理/挂载路径。

internal/model/model.go

@@ -54,7 +54,9 @@ func AllModels() []interface{} {
 		&StrmAccount{},
 		&StrmSyncPath{},
 		&StrmSyncRecord{},
-		&StrmDownloadTask{},
-		&StrmUploadTask{},
-	}
+			&StrmDownloadTask{},
+			&StrmUploadTask{},
+			&StrmDirCache{},
+			&ScrapeTask{},
+		}
 }

internal/model/scrape_task.go

@@ -0,0 +1,34 @@
+package model
+
+import "time"
+
+const (
+	ScrapeTaskPending  = "pending"
+	ScrapeTaskRunning  = "running"
+	ScrapeTaskDone     = "done"
+	ScrapeTaskFailed   = "failed"
+	ScrapeTaskCanceled = "canceled"
+)
+
+// ScrapeTask 表示一条持久化的媒体刮削任务。
+type ScrapeTask struct {
+	Base
+	MediaID        string     `gorm:"index;size:36" json:"media_id"`
+	LibraryID      string     `gorm:"index;size:36" json:"library_id"`
+	LibraryName    string     `gorm:"size:128" json:"library_name"`
+	MediaTitle     string     `gorm:"size:255;not null" json:"media_title"`
+	MediaPath      string     `gorm:"size:1024;not null" json:"media_path"`
+	MediaType      string     `gorm:"size:16" json:"media_type"`                   // movie / tv / anime / adult
+	Provider       string     `gorm:"size:32" json:"provider"`                     // tmdb / douban / bangumi / thetvdb / metatube
+	MatchedTitle   string     `gorm:"size:255" json:"matched_title"`
+	MatchedYear    int        `json:"matched_year"`
+	PosterURL      string     `gorm:"size:1024" json:"poster_url"`
+	BackdropURL    string     `gorm:"size:1024" json:"backdrop_url"`
+	Status         string     `gorm:"index;size:16;default:pending" json:"status"` // pending / running / done / failed / canceled
+	Error          string     `gorm:"type:text" json:"error"`
+	RetryCount     int        `gorm:"default:0" json:"retry_count"`
+	EpisodeImages  bool       `gorm:"default:true" json:"episode_images"`
+	RefreshMatched bool       `gorm:"default:false" json:"refresh_matched"`
+	StartedAt      *time.Time `json:"started_at,omitempty"`
+	FinishedAt     *time.Time `json:"finished_at,omitempty"`
+}

internal/model/strm.go

@@ -48,12 +48,19 @@ type StrmSyncPath struct {
 	DeleteDir       bool       `json:"delete_dir"`                        // 清理多余文件时删除空目录
 	Cron            string     `gorm:"size:128" json:"cron"`              // 5 段 cron 表达式（可选）
 	EnableCron      bool       `json:"enable_cron"`                       // 是否按 Cron 定时同步
+	SyncMode        string     `gorm:"size:32;default:'incremental'" json:"sync_mode"` // 默认同步模式：incremental / full
 	Enabled         bool       `gorm:"default:true" json:"enabled"`
 	LastSyncAt      *time.Time `json:"last_sync_at"`
 	LastSyncStatus  string     `gorm:"size:16" json:"last_sync_status"` // idle/running/ok/error/canceled
 	LastSyncMessage string     `gorm:"size:1024" json:"last_sync_message"`
 }
 
+// STRM 同步类型。
+const (
+	StrmSyncTypeIncremental = "incremental"
+	StrmSyncTypeFull        = "full"
+)
+
 // StrmSyncRecord 是一次同步执行的记录。
 const (
 	StrmSyncRecordPending  = "pending"
@@ -66,6 +73,7 @@ const (
 type StrmSyncRecord struct {
 	Base
 	SyncPathID string     `gorm:"size:36;index" json:"sync_path_id"`
+	SyncType   string     `gorm:"size:32;default:'incremental'" json:"sync_type"` // incremental / full
 	Status     string     `gorm:"size:16;index" json:"status"`
 	Total      int64      `json:"total"`    // 远端发现的文件总数
 	NewStrm    int64      `json:"new_strm"` // 本次新建/更新的 strm 数
@@ -123,3 +131,12 @@ type StrmUploadTask struct {
 	StartedAt  *time.Time `json:"started_at"`
 	FinishedAt *time.Time `json:"finished_at"`
 }
+
+// StrmDirCache 缓存远端网盘目录 ID 与相对路径映射（支持 115 增量同步秒级寻址）。
+type StrmDirCache struct {
+	Base
+	SyncPathID string `gorm:"size:36;index:idx_strm_dir_cache,priority:1" json:"sync_path_id"`
+	DirID      string `gorm:"size:128;index:idx_strm_dir_cache,priority:2" json:"dir_id"`
+	Path       string `gorm:"size:1024" json:"path"` // 相对根目录的路径
+}
+

internal/repository/api_config_repository.go

@@ -62,9 +62,9 @@ func (r *ApiConfigRepository) Update(ctx context.Context, c *model.ApiConfig) er
 	}).Error
 }
 
-// Delete removes an API config.
+// Delete 物理删除 API 配置。
 func (r *ApiConfigRepository) Delete(ctx context.Context, provider string) error {
-	return r.db.WithContext(ctx).Where("provider = ?", provider).Delete(&model.ApiConfig{}).Error
+	return r.db.WithContext(ctx).Unscoped().Where("provider = ?", provider).Delete(&model.ApiConfig{}).Error
 }
 
 // UpdateTestResult 更新测试结果。

internal/repository/favorite_repository.go

@@ -23,7 +23,7 @@ func (r *FavoriteRepository) Toggle(ctx context.Context, userID, mediaID string)
 	if err != nil {
 		return false, err
 	}
-	return false, r.db.WithContext(ctx).Delete(&f).Error
+		return false, r.db.WithContext(ctx).Unscoped().Delete(&f).Error
 }
 
 // ListByUser returns all favourite media IDs for a user.

internal/repository/library_repository.go

@@ -15,6 +15,11 @@ type LibraryRepository struct{ db *gorm.DB }
 
 // Create persists a new library row.
 func (r *LibraryRepository) Create(ctx context.Context, l *model.Library) error {
+	if l != nil && l.SortOrder == 0 {
+		var maxSort int
+		_ = r.db.WithContext(ctx).Model(&model.Library{}).Select("COALESCE(MAX(sort_order), -1)").Scan(&maxSort)
+		l.SortOrder = maxSort + 1
+	}
 	return r.db.WithContext(ctx).Create(l).Error
 }
 
@@ -23,6 +28,11 @@ func (r *LibraryRepository) CreateWithRoots(ctx context.Context, l *model.Librar
 		return r.Create(ctx, l)
 	}
 	return r.db.WithContext(ctx).Transaction(func(tx *gorm.DB) error {
+		if l != nil && l.SortOrder == 0 {
+			var maxSort int
+			_ = tx.Model(&model.Library{}).Select("COALESCE(MAX(sort_order), -1)").Scan(&maxSort)
+			l.SortOrder = maxSort + 1
+		}
 		if err := tx.Create(l).Error; err != nil {
 			return err
 		}
@@ -50,7 +60,7 @@ func (r *LibraryRepository) CreateWithRoots(ctx context.Context, l *model.Librar
 // List returns all enabled+disabled libraries.
 func (r *LibraryRepository) List(ctx context.Context) ([]model.Library, error) {
 	var ls []model.Library
-	q := r.db.WithContext(ctx).Order("created_at asc")
+	q := r.db.WithContext(ctx).Order("sort_order asc, created_at asc")
 	if r.hasLibraryRootsTable() {
 		q = q.Preload("Roots", func(db *gorm.DB) *gorm.DB {
 			return db.Order("sort_order asc, created_at asc")
@@ -60,6 +70,23 @@ func (r *LibraryRepository) List(ctx context.Context) ([]model.Library, error) {
 	return ls, err
 }
 
+// SetSortOrder assigns sort_order to libraries, preserving position order for
+// any library not present in the provided map.
+func (r *LibraryRepository) SetSortOrder(ctx context.Context, ids []string) error {
+	if len(ids) == 0 {
+		return nil
+	}
+	return r.db.WithContext(ctx).Transaction(func(tx *gorm.DB) error {
+		for i, id := range ids {
+			if err := tx.Model(&model.Library{}).Where("id = ?", id).
+				Update("sort_order", i).Error; err != nil {
+				return err
+			}
+		}
+		return nil
+	})
+}
+
 // FindByID returns the library, or (nil, nil) when missing.
 func (r *LibraryRepository) FindByID(ctx context.Context, id string) (*model.Library, error) {
 	var l model.Library
@@ -79,10 +106,9 @@ func (r *LibraryRepository) FindByID(ctx context.Context, id string) (*model.Lib
 	return &l, nil
 }
 
-// Delete removes a library and (soft) cascades to its media via repository
-// callers; we do not run CASCADE here to keep this method narrow.
+// Delete 物理删除媒体库。
 func (r *LibraryRepository) Delete(ctx context.Context, id string) error {
-	return r.db.WithContext(ctx).Delete(&model.Library{}, "id = ?", id).Error
+	return r.db.WithContext(ctx).Unscoped().Delete(&model.Library{}, "id = ?", id).Error
 }
 
 func (r *LibraryRepository) ListRoots(ctx context.Context, libraryID string) ([]model.LibraryRoot, error) {
@@ -149,7 +175,7 @@ func (r *LibraryRepository) DeleteRoot(ctx context.Context, libraryID, rootID st
 	if !r.hasLibraryRootsTable() {
 		return nil
 	}
-	return r.db.WithContext(ctx).Where("library_id = ?", libraryID).Delete(&model.LibraryRoot{}, "id = ?", rootID).Error
+	return r.db.WithContext(ctx).Unscoped().Where("library_id = ?", libraryID).Delete(&model.LibraryRoot{}, "id = ?", rootID).Error
 }
 
 func (r *LibraryRepository) hasLibraryRootsTable() bool {

internal/repository/media_repository.go

@@ -114,14 +114,43 @@ func (r *MediaRepository) ListByLibrariesFiltered(ctx context.Context, libraryID
 	return items, total, err
 }
 
+type libraryCountRow struct {
+	LibraryID string `gorm:"column:library_id"`
+	Total     int64  `gorm:"column:total"`
+}
+
+// CountByLibraries returns a map of library_id -> total media count for the given library IDs.
+func (r *MediaRepository) CountByLibraries(ctx context.Context, libraryIDs []string, filter MediaQueryFilter) (map[string]int64, error) {
+	out := make(map[string]int64, len(libraryIDs))
+	if len(libraryIDs) == 0 {
+		return out, nil
+	}
+	var rows []libraryCountRow
+	q := r.db.WithContext(ctx).Model(&model.Media{}).
+		Select("library_id, count(*) as total")
+	if len(libraryIDs) == 1 {
+		q = q.Where("library_id = ?", libraryIDs[0])
+	} else {
+		q = q.Where("library_id IN ?", libraryIDs)
+	}
+	q = applyMediaQueryFilter(q, filter)
+	if err := q.Group("library_id").Scan(&rows).Error; err != nil {
+		return nil, err
+	}
+	for _, row := range rows {
+		out[row.LibraryID] = row.Total
+	}
+	return out, nil
+}
+
 // DeleteByLibrary purges all media tied to a library.
 func (r *MediaRepository) DeleteByLibrary(ctx context.Context, libraryID string) error {
-	// FTS 行由 media 表上的触发器同步清理（软删/硬删都覆盖）。
-	return r.db.WithContext(ctx).Where("library_id = ?", libraryID).Delete(&model.Media{}).Error
+	// FTS 行由 media 表上的触发器同步清理（物理删除触发 FTS 清理）。
+	return r.db.WithContext(ctx).Unscoped().Where("library_id = ?", libraryID).Delete(&model.Media{}).Error
 }
 
 func (r *MediaRepository) DeleteByLibraryRoot(ctx context.Context, libraryID, rootID string) error {
-	return r.db.WithContext(ctx).
+	return r.db.WithContext(ctx).Unscoped().
 		Where("library_id = ? AND library_root_id = ?", libraryID, rootID).
 		Delete(&model.Media{}).Error
 }

internal/repository/permission_repository.go

@@ -51,9 +51,9 @@ func (r *PermissionRepository) Upsert(ctx context.Context, p *model.UserPermissi
 	})
 }
 
-// Delete removes a permission record.
+// Delete 物理删除权限记录。
 func (r *PermissionRepository) Delete(ctx context.Context, userID string) error {
 	return withSQLiteBusyRetry(ctx, func() error {
-		return r.db.WithContext(ctx).Where("user_id = ?", userID).Delete(&model.UserPermission{}).Error
+		return r.db.WithContext(ctx).Unscoped().Where("user_id = ?", userID).Delete(&model.UserPermission{}).Error
 	})
 }

internal/repository/play_profile_repo.go

@@ -59,9 +59,9 @@ func (r *PlayProfileRepository) Update(ctx context.Context, id string, patch map
 		Where("id = ?", id).Updates(patch).Error
 }
 
-// Delete soft-deletes a profile.
+// Delete 物理删除播放档案。
 func (r *PlayProfileRepository) Delete(ctx context.Context, id string) error {
-	return r.db.WithContext(ctx).Delete(&model.PlayProfile{}, "id = ?", id).Error
+	return r.db.WithContext(ctx).Unscoped().Delete(&model.PlayProfile{}, "id = ?", id).Error
 }
 
 // ClearDefaultsFor resets is_default for all of a user's profiles.

internal/repository/refresh_token_repository.go

@@ -72,10 +72,10 @@ func (r *RefreshTokenRepository) RevokeOldestActiveByUserID(ctx context.Context,
 	})
 }
 
-// DeleteExpired removes all expired refresh tokens.
+// DeleteExpired 物理清理所有过期的 refresh tokens。
 func (r *RefreshTokenRepository) DeleteExpired(ctx context.Context) error {
 	return withSQLiteBusyRetry(ctx, func() error {
-		return r.db.WithContext(ctx).Where("expires_at < ?", time.Now()).Delete(&model.RefreshToken{}).Error
+		return r.db.WithContext(ctx).Unscoped().Where("expires_at < ?", time.Now()).Delete(&model.RefreshToken{}).Error
 	})
 }
 

internal/repository/repository.go

@@ -30,8 +30,10 @@ type Container struct {
 	StrmSyncPath   *StrmSyncPathRepository
 	StrmSyncRecord *StrmSyncRecordRepository
 	StrmDownload   *StrmDownloadTaskRepository
-	StrmUpload     *StrmUploadTaskRepository
-}
+		StrmUpload     *StrmUploadTaskRepository
+		StrmDirCache   *StrmDirCacheRepository
+		ScrapeTask     *ScrapeTaskRepository
+	}
 
 // New 将每个 repository 连接到单个 *gorm.DB。
 func New(db *gorm.DB) *Container {
@@ -58,5 +60,7 @@ func New(db *gorm.DB) *Container {
 		StrmSyncRecord: &StrmSyncRecordRepository{db: db},
 		StrmDownload:   &StrmDownloadTaskRepository{db: db},
 		StrmUpload:     &StrmUploadTaskRepository{db: db},
+		StrmDirCache:   &StrmDirCacheRepository{db: db},
+		ScrapeTask:     &ScrapeTaskRepository{db: db},
 	}
 }

internal/repository/scrape_task_repository.go

@@ -0,0 +1,274 @@
+package repository
+
+import (
+	"context"
+	"errors"
+	"strings"
+	"sync"
+	"time"
+
+	"gorm.io/gorm"
+
+	"github.com/ShukeBta/MMTL/internal/model"
+)
+
+var scrapeClaimMu sync.Mutex
+
+// ScrapeTaskRepository persists model.ScrapeTask.
+type ScrapeTaskRepository struct{ db *gorm.DB }
+
+func (r *ScrapeTaskRepository) Create(ctx context.Context, t *model.ScrapeTask) error {
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Create(t).Error
+	})
+}
+
+func (r *ScrapeTaskRepository) CreateBatch(ctx context.Context, tasks []model.ScrapeTask) error {
+	if len(tasks) == 0 {
+		return nil
+	}
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).CreateInBatches(tasks, 100).Error
+	})
+}
+
+func (r *ScrapeTaskRepository) FindByID(ctx context.Context, id string) (*model.ScrapeTask, error) {
+	var t model.ScrapeTask
+	err := r.db.WithContext(ctx).Where("id = ?", id).First(&t).Error
+	if errors.Is(err, gorm.ErrRecordNotFound) {
+		return nil, nil
+	}
+	return &t, err
+}
+
+func (r *ScrapeTaskRepository) FindActiveByMediaID(ctx context.Context, mediaID string) (*model.ScrapeTask, error) {
+	var t model.ScrapeTask
+	err := r.db.WithContext(ctx).
+		Where("media_id = ? AND status IN ?", mediaID, []string{model.ScrapeTaskPending, model.ScrapeTaskRunning}).
+		First(&t).Error
+	if errors.Is(err, gorm.ErrRecordNotFound) {
+		return nil, nil
+	}
+	return &t, err
+}
+
+func (r *ScrapeTaskRepository) List(ctx context.Context, status string, page, pageSize int) ([]model.ScrapeTask, int64, error) {
+	if page < 1 {
+		page = 1
+	}
+	if pageSize < 1 || pageSize > 200 {
+		pageSize = 50
+	}
+	q := r.db.WithContext(ctx).Model(&model.ScrapeTask{})
+	if strings.TrimSpace(status) != "" && status != "all" {
+		q = q.Where("status = ?", strings.TrimSpace(status))
+	}
+	var total int64
+	if err := q.Count(&total).Error; err != nil {
+		return nil, 0, err
+	}
+	var rows []model.ScrapeTask
+	err := q.Order("created_at desc").
+		Offset((page - 1) * pageSize).
+		Limit(pageSize).
+		Find(&rows).Error
+	return rows, total, err
+}
+
+func (r *ScrapeTaskRepository) CountByStatus(ctx context.Context) (map[string]int64, error) {
+	var rows []struct {
+		Status string
+		Count  int64
+	}
+	err := r.db.WithContext(ctx).Model(&model.ScrapeTask{}).
+		Select("status, count(*) as count").
+		Group("status").Scan(&rows).Error
+	if err != nil {
+		return nil, err
+	}
+	out := map[string]int64{}
+	for _, row := range rows {
+		out[row.Status] = row.Count
+	}
+	return out, nil
+}
+
+// ClaimPending picks pending scrape tasks and marks them running.
+func (r *ScrapeTaskRepository) ClaimPending(ctx context.Context, limit int) ([]model.ScrapeTask, error) {
+	scrapeClaimMu.Lock()
+	defer scrapeClaimMu.Unlock()
+
+	var rows []model.ScrapeTask
+	err := withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Transaction(func(tx *gorm.DB) error {
+			if err := tx.Where("status = ?", model.ScrapeTaskPending).
+				Order("created_at asc").Limit(limit).Find(&rows).Error; err != nil {
+				return err
+			}
+			if len(rows) == 0 {
+				return nil
+			}
+			ids := make([]string, 0, len(rows))
+			now := time.Now()
+			for i := range rows {
+				ids = append(ids, rows[i].ID)
+				rows[i].Status = model.ScrapeTaskRunning
+				rows[i].StartedAt = &now
+			}
+			return tx.Model(&model.ScrapeTask{}).Where("id IN ?", ids).
+				Updates(map[string]any{"status": model.ScrapeTaskRunning, "started_at": now}).Error
+		})
+	})
+	if err != nil {
+		return nil, err
+	}
+	return rows, nil
+}
+
+func (r *ScrapeTaskRepository) Update(ctx context.Context, t *model.ScrapeTask) error {
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Model(&model.ScrapeTask{}).Where("id = ?", t.ID).Updates(map[string]any{
+			"status":        t.Status,
+			"error":         t.Error,
+			"provider":      t.Provider,
+			"matched_title": t.MatchedTitle,
+			"matched_year":  t.MatchedYear,
+			"poster_url":    t.PosterURL,
+			"backdrop_url":  t.BackdropURL,
+			"retry_count":   t.RetryCount,
+			"started_at":    t.StartedAt,
+			"finished_at":   t.FinishedAt,
+			"updated_at":    time.Now(),
+		}).Error
+	})
+}
+
+func (r *ScrapeTaskRepository) Delete(ctx context.Context, id string) error {
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Unscoped().Where("id = ?", id).Delete(&model.ScrapeTask{}).Error
+	})
+}
+
+func (r *ScrapeTaskRepository) DeleteBatch(ctx context.Context, ids []string) (int64, error) {
+	if len(ids) == 0 {
+		return 0, nil
+	}
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Unscoped().Where("id IN ?", ids).Delete(&model.ScrapeTask{})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
+}
+
+func (r *ScrapeTaskRepository) RetryBatch(ctx context.Context, ids []string) (int64, error) {
+	if len(ids) == 0 {
+		return 0, nil
+	}
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Model(&model.ScrapeTask{}).
+			Where("id IN ? AND status IN ?", ids, []string{model.ScrapeTaskFailed, model.ScrapeTaskCanceled}).
+			Updates(map[string]any{
+				"status":      model.ScrapeTaskPending,
+				"error":       "",
+				"retry_count": 0,
+				"started_at":  nil,
+				"finished_at": nil,
+				"updated_at":  time.Now(),
+			})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
+}
+
+func (r *ScrapeTaskRepository) CancelBatch(ctx context.Context, ids []string) (int64, error) {
+	if len(ids) == 0 {
+		return 0, nil
+	}
+	now := time.Now()
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Model(&model.ScrapeTask{}).
+			Where("id IN ? AND status IN ?", ids, []string{model.ScrapeTaskPending, model.ScrapeTaskRunning}).
+			Updates(map[string]any{
+				"status":      model.ScrapeTaskCanceled,
+				"error":       "已批量取消",
+				"finished_at": now,
+				"updated_at":  now,
+			})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
+}
+
+func (r *ScrapeTaskRepository) ClearDone(ctx context.Context) (int64, error) {
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Unscoped().Where("status = ?", model.ScrapeTaskDone).Delete(&model.ScrapeTask{})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
+}
+
+func (r *ScrapeTaskRepository) ClearFinished(ctx context.Context) (int64, error) {
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Unscoped().Where("status IN ?", []string{model.ScrapeTaskDone, model.ScrapeTaskFailed, model.ScrapeTaskCanceled}).
+			Delete(&model.ScrapeTask{})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
+}
+
+func (r *ScrapeTaskRepository) ClearCanceled(ctx context.Context) (int64, error) {
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Unscoped().Where("status = ?", model.ScrapeTaskCanceled).Delete(&model.ScrapeTask{})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
+}
+
+func (r *ScrapeTaskRepository) RetryAllFailed(ctx context.Context) (int64, error) {
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Model(&model.ScrapeTask{}).
+			Where("status = ?", model.ScrapeTaskFailed).
+			Updates(map[string]any{
+				"status":      model.ScrapeTaskPending,
+				"error":       "",
+				"retry_count": 0,
+				"started_at":  nil,
+				"finished_at": nil,
+				"updated_at":  time.Now(),
+			})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
+}
+
+func (r *ScrapeTaskRepository) CancelPending(ctx context.Context) (int64, error) {
+	now := time.Now()
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Model(&model.ScrapeTask{}).
+			Where("status IN ?", []string{model.ScrapeTaskPending, model.ScrapeTaskRunning}).
+			Updates(map[string]any{
+				"status":      model.ScrapeTaskCanceled,
+				"error":       "已批量取消",
+				"finished_at": now,
+				"updated_at":  now,
+			})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
+}

internal/repository/setting_repository.go

@@ -29,9 +29,9 @@ func (r *SettingRepository) Set(ctx context.Context, key, value string) error {
 	return r.db.WithContext(ctx).Save(&s).Error
 }
 
-// Delete removes a setting key.
+// Delete 物理删除设置键。
 func (r *SettingRepository) Delete(ctx context.Context, key string) error {
-	return r.db.WithContext(ctx).Where("key = ?", key).Delete(&model.Setting{}).Error
+	return r.db.WithContext(ctx).Unscoped().Where("key = ?", key).Delete(&model.Setting{}).Error
 }
 
 // All returns every key/value pair (used by the admin UI).

internal/repository/storage_config_repo.go

@@ -66,9 +66,9 @@ func (r *StorageConfigRepository) Upsert(ctx context.Context, c *model.StorageCo
 	}).Error
 }
 
-// Delete removes a storage config by ID.
+// Delete 物理删除存储配置。
 func (r *StorageConfigRepository) Delete(ctx context.Context, id string) error {
-	return r.db.WithContext(ctx).Where("id = ?", id).Delete(&model.StorageConfig{}).Error
+	return r.db.WithContext(ctx).Unscoped().Where("id = ?", id).Delete(&model.StorageConfig{}).Error
 }
 
 // FindByID returns a storage config by ID.

internal/repository/strm_repository.go

@@ -3,6 +3,7 @@ package repository
 import (
 	"context"
 	"errors"
+	"sync"
 	"time"
 
 	"gorm.io/gorm"
@@ -10,13 +11,17 @@ import (
 	"github.com/ShukeBta/MMTL/internal/model"
 )
 
+var strmClaimMu sync.Mutex
+
 // ─── StrmAccount ───────────────────────────────────────────────────────────────
 
 // StrmAccountRepository persists model.StrmAccount.
 type StrmAccountRepository struct{ db *gorm.DB }
 
 func (r *StrmAccountRepository) Create(ctx context.Context, a *model.StrmAccount) error {
-	return r.db.WithContext(ctx).Create(a).Error
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Create(a).Error
+	})
 }
 
 func (r *StrmAccountRepository) FindByID(ctx context.Context, id string) (*model.StrmAccount, error) {
@@ -38,20 +43,24 @@ func (r *StrmAccountRepository) List(ctx context.Context) ([]model.StrmAccount,
 }
 
 func (r *StrmAccountRepository) Update(ctx context.Context, a *model.StrmAccount) error {
-	return r.db.WithContext(ctx).Model(&model.StrmAccount{}).Where("id = ?", a.ID).Updates(map[string]any{
-		"name":             a.Name,
-		"provider":         a.Provider,
-		"config":           a.Config,
-		"enabled":          a.Enabled,
-		"last_test_at":     a.LastTestAt,
-		"last_test_result": a.LastTestResult,
-		"last_test_ok":     a.LastTestOK,
-		"updated_at":       time.Now(),
-	}).Error
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Model(&model.StrmAccount{}).Where("id = ?", a.ID).Updates(map[string]any{
+			"name":             a.Name,
+			"provider":         a.Provider,
+			"config":           a.Config,
+			"enabled":          a.Enabled,
+			"last_test_at":     a.LastTestAt,
+			"last_test_result": a.LastTestResult,
+			"last_test_ok":     a.LastTestOK,
+			"updated_at":       time.Now(),
+		}).Error
+	})
 }
 
 func (r *StrmAccountRepository) Delete(ctx context.Context, id string) error {
-	return r.db.WithContext(ctx).Where("id = ?", id).Delete(&model.StrmAccount{}).Error
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Unscoped().Where("id = ?", id).Delete(&model.StrmAccount{}).Error
+	})
 }
 
 // ─── StrmSyncPath ──────────────────────────────────────────────────────────────
@@ -60,7 +69,9 @@ func (r *StrmAccountRepository) Delete(ctx context.Context, id string) error {
 type StrmSyncPathRepository struct{ db *gorm.DB }
 
 func (r *StrmSyncPathRepository) Create(ctx context.Context, p *model.StrmSyncPath) error {
-	return r.db.WithContext(ctx).Create(p).Error
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Create(p).Error
+	})
 }
 
 func (r *StrmSyncPathRepository) FindByID(ctx context.Context, id string) (*model.StrmSyncPath, error) {
@@ -82,33 +93,38 @@ func (r *StrmSyncPathRepository) List(ctx context.Context) ([]model.StrmSyncPath
 }
 
 func (r *StrmSyncPathRepository) Update(ctx context.Context, p *model.StrmSyncPath) error {
-	return r.db.WithContext(ctx).Model(&model.StrmSyncPath{}).Where("id = ?", p.ID).Updates(map[string]any{
-		"name":              p.Name,
-		"account_id":        p.AccountID,
-		"provider":          p.Provider,
-		"remote_path":       p.RemotePath,
-		"local_path":        p.LocalPath,
-		"strm_base_url":     p.StrmBaseURL,
-		"video_ext":         p.VideoExt,
-		"meta_ext":          p.MetaExt,
-		"exclude_name":      p.ExcludeName,
-		"min_video_size_mb": p.MinVideoSizeMB,
-		"add_path":          p.AddPath,
-		"download_meta":     p.DownloadMeta,
-		"upload_meta":       p.UploadMeta,
-		"delete_dir":        p.DeleteDir,
-		"cron":              p.Cron,
-		"enable_cron":       p.EnableCron,
-		"enabled":           p.Enabled,
-		"last_sync_at":      p.LastSyncAt,
-		"last_sync_status":  p.LastSyncStatus,
-		"last_sync_message": p.LastSyncMessage,
-		"updated_at":        time.Now(),
-	}).Error
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Model(&model.StrmSyncPath{}).Where("id = ?", p.ID).Updates(map[string]any{
+			"name":              p.Name,
+			"account_id":        p.AccountID,
+			"provider":          p.Provider,
+			"remote_path":       p.RemotePath,
+			"local_path":        p.LocalPath,
+			"strm_base_url":     p.StrmBaseURL,
+			"video_ext":         p.VideoExt,
+			"meta_ext":          p.MetaExt,
+			"exclude_name":      p.ExcludeName,
+			"min_video_size_mb": p.MinVideoSizeMB,
+			"add_path":          p.AddPath,
+			"download_meta":     p.DownloadMeta,
+			"upload_meta":       p.UploadMeta,
+			"delete_dir":        p.DeleteDir,
+			"cron":              p.Cron,
+			"enable_cron":       p.EnableCron,
+			"sync_mode":         p.SyncMode,
+			"enabled":           p.Enabled,
+			"last_sync_at":      p.LastSyncAt,
+			"last_sync_status":  p.LastSyncStatus,
+			"last_sync_message": p.LastSyncMessage,
+			"updated_at":        time.Now(),
+		}).Error
+	})
 }
 
 func (r *StrmSyncPathRepository) Delete(ctx context.Context, id string) error {
-	return r.db.WithContext(ctx).Where("id = ?", id).Delete(&model.StrmSyncPath{}).Error
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Unscoped().Where("id = ?", id).Delete(&model.StrmSyncPath{}).Error
+	})
 }
 
 // ─── StrmSyncRecord ────────────────────────────────────────────────────────────
@@ -117,23 +133,28 @@ func (r *StrmSyncPathRepository) Delete(ctx context.Context, id string) error {
 type StrmSyncRecordRepository struct{ db *gorm.DB }
 
 func (r *StrmSyncRecordRepository) Create(ctx context.Context, rec *model.StrmSyncRecord) error {
-	return r.db.WithContext(ctx).Create(rec).Error
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Create(rec).Error
+	})
 }
 
 func (r *StrmSyncRecordRepository) Update(ctx context.Context, rec *model.StrmSyncRecord) error {
-	return r.db.WithContext(ctx).Model(&model.StrmSyncRecord{}).Where("id = ?", rec.ID).Updates(map[string]any{
-		"status":      rec.Status,
-		"total":       rec.Total,
-		"new_strm":    rec.NewStrm,
-		"new_meta":    rec.NewMeta,
-		"uploaded":    rec.Uploaded,
-		"pruned":      rec.Pruned,
-		"skipped":     rec.Skipped,
-		"message":     rec.Message,
-		"started_at":  rec.StartedAt,
-		"finished_at": rec.FinishedAt,
-		"updated_at":  time.Now(),
-	}).Error
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Model(&model.StrmSyncRecord{}).Where("id = ?", rec.ID).Updates(map[string]any{
+			"sync_type":   rec.SyncType,
+			"status":      rec.Status,
+			"total":       rec.Total,
+			"new_strm":    rec.NewStrm,
+			"new_meta":    rec.NewMeta,
+			"uploaded":    rec.Uploaded,
+			"pruned":      rec.Pruned,
+			"skipped":     rec.Skipped,
+			"message":     rec.Message,
+			"started_at":  rec.StartedAt,
+			"finished_at": rec.FinishedAt,
+			"updated_at":  time.Now(),
+		}).Error
+	})
 }
 
 func (r *StrmSyncRecordRepository) List(ctx context.Context, syncPathID string, limit int) ([]model.StrmSyncRecord, error) {
@@ -149,13 +170,45 @@ func (r *StrmSyncRecordRepository) List(ctx context.Context, syncPathID string,
 	return rows, err
 }
 
+// Delete 删除单条同步记录（物理删除）。
+func (r *StrmSyncRecordRepository) Delete(ctx context.Context, id string) error {
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Unscoped().Where("id = ?", id).Delete(&model.StrmSyncRecord{}).Error
+	})
+}
+
+// DeleteBySyncPathID 删除某同步目录下的全部同步记录（删除同步目录时级联清理）。
+func (r *StrmSyncRecordRepository) DeleteBySyncPathID(ctx context.Context, syncPathID string) (int64, error) {
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Unscoped().Where("sync_path_id = ?", syncPathID).Delete(&model.StrmSyncRecord{})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
+}
+
 // ─── StrmDownloadTask ──────────────────────────────────────────────────────────
 
 // StrmDownloadTaskRepository persists model.StrmDownloadTask.
 type StrmDownloadTaskRepository struct{ db *gorm.DB }
 
 func (r *StrmDownloadTaskRepository) Create(ctx context.Context, t *model.StrmDownloadTask) error {
-	return r.db.WithContext(ctx).Create(t).Error
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Create(t).Error
+	})
+}
+
+func (r *StrmDownloadTaskRepository) CreateInBatches(ctx context.Context, tasks []*model.StrmDownloadTask, batchSize int) error {
+	if len(tasks) == 0 {
+		return nil
+	}
+	if batchSize <= 0 {
+		batchSize = 100
+	}
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).CreateInBatches(tasks, batchSize).Error
+	})
 }
 
 func (r *StrmDownloadTaskRepository) FindByID(ctx context.Context, id string) (*model.StrmDownloadTask, error) {
@@ -205,24 +258,29 @@ func (r *StrmDownloadTaskRepository) CountByStatus(ctx context.Context) (map[str
 // ClaimPendingDownload picks the oldest pending task and marks it running.
 // Returns (nil, nil) when the queue is empty.
 func (r *StrmDownloadTaskRepository) ClaimPendingDownload(ctx context.Context, limit int) ([]model.StrmDownloadTask, error) {
+	strmClaimMu.Lock()
+	defer strmClaimMu.Unlock()
+
 	var rows []model.StrmDownloadTask
-	err := r.db.WithContext(ctx).Transaction(func(tx *gorm.DB) error {
-		if err := tx.Where("status = ? AND (next_try_at IS NULL OR next_try_at <= ?)", model.StrmTaskPending, time.Now()).
-			Order("created_at asc").Limit(limit).Find(&rows).Error; err != nil {
-			return err
-		}
-		if len(rows) == 0 {
-			return nil
-		}
-		ids := make([]string, 0, len(rows))
-		now := time.Now()
-		for i := range rows {
-			ids = append(ids, rows[i].ID)
-			rows[i].Status = model.StrmTaskRunning
-			rows[i].StartedAt = &now
-		}
-		return tx.Model(&model.StrmDownloadTask{}).Where("id IN ?", ids).
-			Updates(map[string]any{"status": model.StrmTaskRunning, "started_at": now}).Error
+	err := withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Transaction(func(tx *gorm.DB) error {
+			if err := tx.Where("status = ? AND (next_try_at IS NULL OR next_try_at <= ?)", model.StrmTaskPending, time.Now()).
+				Order("created_at asc").Limit(limit).Find(&rows).Error; err != nil {
+				return err
+			}
+			if len(rows) == 0 {
+				return nil
+			}
+			ids := make([]string, 0, len(rows))
+			now := time.Now()
+			for i := range rows {
+				ids = append(ids, rows[i].ID)
+				rows[i].Status = model.StrmTaskRunning
+				rows[i].StartedAt = &now
+			}
+			return tx.Model(&model.StrmDownloadTask{}).Where("id IN ?", ids).
+				Updates(map[string]any{"status": model.StrmTaskRunning, "started_at": now}).Error
+		})
 	})
 	if err != nil {
 		return nil, err
@@ -231,62 +289,157 @@ func (r *StrmDownloadTaskRepository) ClaimPendingDownload(ctx context.Context, l
 }
 
 func (r *StrmDownloadTaskRepository) Update(ctx context.Context, t *model.StrmDownloadTask) error {
-	return r.db.WithContext(ctx).Model(&model.StrmDownloadTask{}).Where("id = ?", t.ID).Updates(map[string]any{
-		"status":      t.Status,
-		"error":       t.Error,
-		"retry_count": t.RetryCount,
-		"next_try_at": t.NextTryAt,
-		"started_at":  t.StartedAt,
-		"finished_at": t.FinishedAt,
-		"updated_at":  time.Now(),
-	}).Error
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Model(&model.StrmDownloadTask{}).Where("id = ?", t.ID).Updates(map[string]any{
+			"status":      t.Status,
+			"error":       t.Error,
+			"retry_count": t.RetryCount,
+			"next_try_at": t.NextTryAt,
+			"started_at":  t.StartedAt,
+			"finished_at": t.FinishedAt,
+			"updated_at":  time.Now(),
+		}).Error
+	})
 }
 
 func (r *StrmDownloadTaskRepository) Delete(ctx context.Context, id string) error {
-	return r.db.WithContext(ctx).Where("id = ?", id).Delete(&model.StrmDownloadTask{}).Error
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Unscoped().Where("id = ?", id).Delete(&model.StrmDownloadTask{}).Error
+	})
+}
+
+// DeleteBatch 批量删除指定 ID 的下载任务。
+func (r *StrmDownloadTaskRepository) DeleteBatch(ctx context.Context, ids []string) (int64, error) {
+	if len(ids) == 0 {
+		return 0, nil
+	}
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Unscoped().Where("id IN ?", ids).Delete(&model.StrmDownloadTask{})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
+}
+
+// RetryBatch 批量重试指定 ID 的失败/已取消下载任务。
+func (r *StrmDownloadTaskRepository) RetryBatch(ctx context.Context, ids []string) (int64, error) {
+	if len(ids) == 0 {
+		return 0, nil
+	}
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Model(&model.StrmDownloadTask{}).
+			Where("id IN ? AND status IN ?", ids, []string{model.StrmTaskFailed, model.StrmTaskCanceled}).
+			Updates(map[string]any{
+				"status":      model.StrmTaskPending,
+				"error":       "",
+				"retry_count": 0,
+				"next_try_at": nil,
+				"started_at":  nil,
+				"finished_at": nil,
+				"updated_at":  time.Now(),
+			})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
+}
+
+// CancelBatch 批量取消指定 ID 的排队/进行中下载任务。
+func (r *StrmDownloadTaskRepository) CancelBatch(ctx context.Context, ids []string) (int64, error) {
+	if len(ids) == 0 {
+		return 0, nil
+	}
+	now := time.Now()
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Model(&model.StrmDownloadTask{}).
+			Where("id IN ? AND status IN ?", ids, []string{model.StrmTaskPending, model.StrmTaskRunning}).
+			Updates(map[string]any{
+				"status":      model.StrmTaskCanceled,
+				"error":       "已批量取消",
+				"finished_at": now,
+				"updated_at":  now,
+			})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
 }
 
 // ClearDone 清空全部已完成下载任务。
 func (r *StrmDownloadTaskRepository) ClearDone(ctx context.Context) (int64, error) {
-	res := r.db.WithContext(ctx).Where("status = ?", model.StrmTaskDone).Delete(&model.StrmDownloadTask{})
-	return res.RowsAffected, res.Error
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Unscoped().Where("status = ?", model.StrmTaskDone).Delete(&model.StrmDownloadTask{})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
 }
 
 // ClearFinished 清空全部已完成与失败下载任务。
 func (r *StrmDownloadTaskRepository) ClearFinished(ctx context.Context) (int64, error) {
-	res := r.db.WithContext(ctx).Where("status IN ?", []string{model.StrmTaskDone, model.StrmTaskFailed}).
-		Delete(&model.StrmDownloadTask{})
-	return res.RowsAffected, res.Error
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Unscoped().Where("status IN ?", []string{model.StrmTaskDone, model.StrmTaskFailed, model.StrmTaskCanceled}).
+			Delete(&model.StrmDownloadTask{})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
+}
+
+// ClearCanceled 清空全部已取消下载任务。
+func (r *StrmDownloadTaskRepository) ClearCanceled(ctx context.Context) (int64, error) {
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Unscoped().Where("status = ?", model.StrmTaskCanceled).Delete(&model.StrmDownloadTask{})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
 }
 
 // RetryAllFailed 把所有失败任务重置回待处理，清空错误与重试计数。
 func (r *StrmDownloadTaskRepository) RetryAllFailed(ctx context.Context) (int64, error) {
-	res := r.db.WithContext(ctx).Model(&model.StrmDownloadTask{}).
-		Where("status = ?", model.StrmTaskFailed).
-		Updates(map[string]any{
-			"status":      model.StrmTaskPending,
-			"error":       "",
-			"retry_count": 0,
-			"next_try_at": nil,
-			"started_at":  nil,
-			"finished_at": nil,
-			"updated_at":  time.Now(),
-		})
-	return res.RowsAffected, res.Error
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Model(&model.StrmDownloadTask{}).
+			Where("status = ?", model.StrmTaskFailed).
+			Updates(map[string]any{
+				"status":      model.StrmTaskPending,
+				"error":       "",
+				"retry_count": 0,
+				"next_try_at": nil,
+				"started_at":  nil,
+				"finished_at": nil,
+				"updated_at":  time.Now(),
+			})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
 }
 
-// CancelPending 批量取消所有排队中的任务。
+// CancelPending 批量取消所有排队中和进行中的任务。
 func (r *StrmDownloadTaskRepository) CancelPending(ctx context.Context) (int64, error) {
 	now := time.Now()
-	res := r.db.WithContext(ctx).Model(&model.StrmDownloadTask{}).
-		Where("status = ?", model.StrmTaskPending).
-		Updates(map[string]any{
-			"status":      model.StrmTaskCanceled,
-			"error":       "已批量取消",
-			"finished_at": now,
-			"updated_at":  now,
-		})
-	return res.RowsAffected, res.Error
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Model(&model.StrmDownloadTask{}).
+			Where("status IN ?", []string{model.StrmTaskPending, model.StrmTaskRunning}).
+			Updates(map[string]any{
+				"status":      model.StrmTaskCanceled,
+				"error":       "已批量取消",
+				"finished_at": now,
+				"updated_at":  now,
+			})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
 }
 
 // CountActive 统计某同步目录下目标仍在排队/进行的任务数（用于去重）。
@@ -299,10 +452,28 @@ func (r *StrmDownloadTaskRepository) CountActive(ctx context.Context, syncPathID
 	return count
 }
 
+// GetActiveLocalPathMap 一次性获取某同步目录下正在排队或执行中的 local_path 集合，供同步时 O(1) 内存去重。
+func (r *StrmDownloadTaskRepository) GetActiveLocalPathMap(ctx context.Context, syncPathID string) (map[string]bool, error) {
+	var paths []string
+	err := r.db.WithContext(ctx).Model(&model.StrmDownloadTask{}).
+		Where("sync_path_id = ? AND status IN ?", syncPathID, []string{model.StrmTaskPending, model.StrmTaskRunning}).
+		Pluck("local_path", &paths).Error
+	if err != nil {
+		return nil, err
+	}
+	out := make(map[string]bool, len(paths))
+	for _, p := range paths {
+		out[p] = true
+	}
+	return out, nil
+}
+
 func (r *StrmDownloadTaskRepository) DeleteFinishedOlderThan(ctx context.Context, before time.Time) error {
-	return r.db.WithContext(ctx).Where("status IN ? AND finished_at < ?",
-		[]string{model.StrmTaskDone, model.StrmTaskFailed, model.StrmTaskCanceled}, before).
-		Delete(&model.StrmDownloadTask{}).Error
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Unscoped().Where("status IN ? AND finished_at < ?",
+			[]string{model.StrmTaskDone, model.StrmTaskFailed, model.StrmTaskCanceled}, before).
+			Delete(&model.StrmDownloadTask{}).Error
+	})
 }
 
 // ─── StrmUploadTask ────────────────────────────────────────────────────────────
@@ -311,7 +482,21 @@ func (r *StrmDownloadTaskRepository) DeleteFinishedOlderThan(ctx context.Context
 type StrmUploadTaskRepository struct{ db *gorm.DB }
 
 func (r *StrmUploadTaskRepository) Create(ctx context.Context, t *model.StrmUploadTask) error {
-	return r.db.WithContext(ctx).Create(t).Error
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Create(t).Error
+	})
+}
+
+func (r *StrmUploadTaskRepository) CreateInBatches(ctx context.Context, tasks []*model.StrmUploadTask, batchSize int) error {
+	if len(tasks) == 0 {
+		return nil
+	}
+	if batchSize <= 0 {
+		batchSize = 100
+	}
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).CreateInBatches(tasks, batchSize).Error
+	})
 }
 
 func (r *StrmUploadTaskRepository) FindByID(ctx context.Context, id string) (*model.StrmUploadTask, error) {
@@ -380,24 +565,29 @@ func (r *StrmUploadTaskRepository) CountByStatus(ctx context.Context) (map[strin
 // ClaimPendingUpload picks the oldest pending task and marks it running.
 // Returns (nil, nil) when the queue is empty.
 func (r *StrmUploadTaskRepository) ClaimPendingUpload(ctx context.Context, limit int) ([]model.StrmUploadTask, error) {
+	strmClaimMu.Lock()
+	defer strmClaimMu.Unlock()
+
 	var rows []model.StrmUploadTask
-	err := r.db.WithContext(ctx).Transaction(func(tx *gorm.DB) error {
-		if err := tx.Where("status = ? AND (next_try_at IS NULL OR next_try_at <= ?)", model.StrmTaskPending, time.Now()).
-			Order("created_at asc").Limit(limit).Find(&rows).Error; err != nil {
-			return err
-		}
-		if len(rows) == 0 {
-			return nil
-		}
-		ids := make([]string, 0, len(rows))
-		now := time.Now()
-		for i := range rows {
-			ids = append(ids, rows[i].ID)
-			rows[i].Status = model.StrmTaskRunning
-			rows[i].StartedAt = &now
-		}
-		return tx.Model(&model.StrmUploadTask{}).Where("id IN ?", ids).
-			Updates(map[string]any{"status": model.StrmTaskRunning, "started_at": now}).Error
+	err := withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Transaction(func(tx *gorm.DB) error {
+			if err := tx.Where("status = ? AND (next_try_at IS NULL OR next_try_at <= ?)", model.StrmTaskPending, time.Now()).
+				Order("created_at asc").Limit(limit).Find(&rows).Error; err != nil {
+				return err
+			}
+			if len(rows) == 0 {
+				return nil
+			}
+			ids := make([]string, 0, len(rows))
+			now := time.Now()
+			for i := range rows {
+				ids = append(ids, rows[i].ID)
+				rows[i].Status = model.StrmTaskRunning
+				rows[i].StartedAt = &now
+			}
+			return tx.Model(&model.StrmUploadTask{}).Where("id IN ?", ids).
+				Updates(map[string]any{"status": model.StrmTaskRunning, "started_at": now}).Error
+		})
 	})
 	if err != nil {
 		return nil, err
@@ -406,19 +596,113 @@ func (r *StrmUploadTaskRepository) ClaimPendingUpload(ctx context.Context, limit
 }
 
 func (r *StrmUploadTaskRepository) Update(ctx context.Context, t *model.StrmUploadTask) error {
-	return r.db.WithContext(ctx).Model(&model.StrmUploadTask{}).Where("id = ?", t.ID).Updates(map[string]any{
-		"status":      t.Status,
-		"error":       t.Error,
-		"retry_count": t.RetryCount,
-		"next_try_at": t.NextTryAt,
-		"started_at":  t.StartedAt,
-		"finished_at": t.FinishedAt,
-		"updated_at":  time.Now(),
-	}).Error
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Model(&model.StrmUploadTask{}).Where("id = ?", t.ID).Updates(map[string]any{
+			"status":      t.Status,
+			"error":       t.Error,
+			"retry_count": t.RetryCount,
+			"next_try_at": t.NextTryAt,
+			"started_at":  t.StartedAt,
+			"finished_at": t.FinishedAt,
+			"updated_at":  time.Now(),
+		}).Error
+	})
 }
 
 func (r *StrmUploadTaskRepository) Delete(ctx context.Context, id string) error {
-	return r.db.WithContext(ctx).Where("id = ?", id).Delete(&model.StrmUploadTask{}).Error
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Unscoped().Where("id = ?", id).Delete(&model.StrmUploadTask{}).Error
+	})
+}
+
+// DeleteBatch 批量删除指定 ID 的上传任务。
+func (r *StrmUploadTaskRepository) DeleteBatch(ctx context.Context, ids []string) (int64, error) {
+	if len(ids) == 0 {
+		return 0, nil
+	}
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Unscoped().Where("id IN ?", ids).Delete(&model.StrmUploadTask{})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
+}
+
+// RetryBatch 批量重试指定 ID 的失败/已取消上传任务。
+func (r *StrmUploadTaskRepository) RetryBatch(ctx context.Context, ids []string) (int64, error) {
+	if len(ids) == 0 {
+		return 0, nil
+	}
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Model(&model.StrmUploadTask{}).
+			Where("id IN ? AND status IN ?", ids, []string{model.StrmTaskFailed, model.StrmTaskCanceled}).
+			Updates(map[string]any{
+				"status":      model.StrmTaskPending,
+				"error":       "",
+				"retry_count": 0,
+				"next_try_at": nil,
+				"started_at":  nil,
+				"finished_at": nil,
+				"updated_at":  time.Now(),
+			})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
+}
+
+// CancelBatch 批量取消指定 ID 的排队/进行中上传任务。
+func (r *StrmUploadTaskRepository) CancelBatch(ctx context.Context, ids []string) (int64, error) {
+	if len(ids) == 0 {
+		return 0, nil
+	}
+	now := time.Now()
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Model(&model.StrmUploadTask{}).
+			Where("id IN ? AND status IN ?", ids, []string{model.StrmTaskPending, model.StrmTaskRunning}).
+			Updates(map[string]any{
+				"status":      model.StrmTaskCanceled,
+				"error":       "已批量取消",
+				"finished_at": now,
+				"updated_at":  now,
+			})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
+}
+
+// ClearCanceled 清空全部已取消上传任务。
+func (r *StrmUploadTaskRepository) ClearCanceled(ctx context.Context) (int64, error) {
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Unscoped().Where("status = ?", model.StrmTaskCanceled).Delete(&model.StrmUploadTask{})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
+}
+
+// CancelPending 批量取消所有排队中和进行中的任务。
+func (r *StrmUploadTaskRepository) CancelPending(ctx context.Context) (int64, error) {
+	now := time.Now()
+	var count int64
+	err := withSQLiteBusyRetry(ctx, func() error {
+		res := r.db.WithContext(ctx).Model(&model.StrmUploadTask{}).
+			Where("status IN ?", []string{model.StrmTaskPending, model.StrmTaskRunning}).
+			Updates(map[string]any{
+				"status":      model.StrmTaskCanceled,
+				"error":       "已批量取消",
+				"finished_at": now,
+				"updated_at":  now,
+			})
+		count = res.RowsAffected
+		return res.Error
+	})
+	return count, err
 }
 
 // CountActive 统计某同步目录下目标仍在排队/进行的任务数（用于去重）。
@@ -431,8 +715,67 @@ func (r *StrmUploadTaskRepository) CountActive(ctx context.Context, syncPathID,
 	return count
 }
 
-func (r *StrmUploadTaskRepository) DeleteFinishedOlderThan(ctx context.Context, before time.Time) error {
-	return r.db.WithContext(ctx).Where("status IN ? AND finished_at < ?",
-		[]string{model.StrmTaskDone, model.StrmTaskFailed, model.StrmTaskCanceled}, before).
-		Delete(&model.StrmUploadTask{}).Error
+// GetActiveLocalPathMap 一次性获取某同步目录下正在排队或执行中的 local_path 集合，供同步时 O(1) 内存去重。
+func (r *StrmUploadTaskRepository) GetActiveLocalPathMap(ctx context.Context, syncPathID string) (map[string]bool, error) {
+	var paths []string
+	err := r.db.WithContext(ctx).Model(&model.StrmUploadTask{}).
+		Where("sync_path_id = ? AND status IN ?", syncPathID, []string{model.StrmTaskPending, model.StrmTaskRunning}).
+		Pluck("local_path", &paths).Error
+	if err != nil {
+		return nil, err
+	}
+	out := make(map[string]bool, len(paths))
+	for _, p := range paths {
+		out[p] = true
+	}
+	return out, nil
 }
+
+func (r *StrmUploadTaskRepository) DeleteFinishedOlderThan(ctx context.Context, before time.Time) error {
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Unscoped().Where("status IN ? AND finished_at < ?",
+			[]string{model.StrmTaskDone, model.StrmTaskFailed, model.StrmTaskCanceled}, before).
+			Delete(&model.StrmUploadTask{}).Error
+	})
+}
+
+// ─── StrmDirCache ─────────────────────────────────────────────────────────────
+
+// StrmDirCacheRepository persists model.StrmDirCache.
+type StrmDirCacheRepository struct{ db *gorm.DB }
+
+func (r *StrmDirCacheRepository) ListBySyncPathID(ctx context.Context, syncPathID string) ([]model.StrmDirCache, error) {
+	var rows []model.StrmDirCache
+	err := r.db.WithContext(ctx).Where("sync_path_id = ?", syncPathID).Find(&rows).Error
+	return rows, err
+}
+
+func (r *StrmDirCacheRepository) Set(ctx context.Context, syncPathID, dirID, path string) error {
+	return withSQLiteBusyRetry(ctx, func() error {
+		var row model.StrmDirCache
+		err := r.db.WithContext(ctx).Where("sync_path_id = ? AND dir_id = ?", syncPathID, dirID).First(&row).Error
+		if errors.Is(err, gorm.ErrRecordNotFound) {
+			row = model.StrmDirCache{
+				SyncPathID: syncPathID,
+				DirID:      dirID,
+				Path:       path,
+			}
+			return r.db.WithContext(ctx).Create(&row).Error
+		}
+		if err != nil {
+			return err
+		}
+		return r.db.WithContext(ctx).Model(&model.StrmDirCache{}).Where("id = ?", row.ID).Updates(map[string]any{
+			"path":       path,
+			"updated_at": time.Now(),
+		}).Error
+	})
+}
+
+func (r *StrmDirCacheRepository) DeleteBySyncPathID(ctx context.Context, syncPathID string) error {
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Unscoped().Where("sync_path_id = ?", syncPathID).Delete(&model.StrmDirCache{}).Error
+	})
+}
+
+

internal/repository/user_repository.go

@@ -133,26 +133,17 @@ func (r *UserRepository) TouchLogin(ctx context.Context, id string) error {
 	})
 }
 
-// Delete removes a user (soft-delete via gorm.DeletedAt), releases the unique
-// username, and drops Telegram bindings so future re-created users bind cleanly.
+// Delete 物理删除用户并级联清理其关联记录。
 func (r *UserRepository) Delete(ctx context.Context, id string) error {
-	return r.db.WithContext(ctx).Transaction(func(tx *gorm.DB) error {
-		var user model.User
-		if err := tx.Where("id = ?", id).First(&user).Error; err != nil {
-			return err
-		}
-		released := user.Username + "__deleted__" + time.Now().Format("20060102150405.000000000")
-		if len(released) > 64 {
-			sum := sha256.Sum256([]byte(user.ID + user.Username))
-			base := user.Username
-			if len(base) > 43 {
-				base = base[:43]
-			}
-			released = base + "__deleted__" + hex.EncodeToString(sum[:])[:10]
-		}
-		if err := tx.Model(&model.User{}).Where("id = ?", id).Update("username", released).Error; err != nil {
-			return err
-		}
-		return tx.Delete(&model.User{}, "id = ?", id).Error
+	return withSQLiteBusyRetry(ctx, func() error {
+		return r.db.WithContext(ctx).Transaction(func(tx *gorm.DB) error {
+			_ = tx.Unscoped().Where("user_id = ?", id).Delete(&model.RefreshToken{})
+			_ = tx.Unscoped().Where("user_id = ?", id).Delete(&model.UserPermission{})
+			_ = tx.Unscoped().Where("user_id = ?", id).Delete(&model.PlayProfile{})
+			_ = tx.Unscoped().Where("user_id = ?", id).Delete(&model.PlaybackHistory{})
+			_ = tx.Unscoped().Where("user_id = ?", id).Delete(&model.Favorite{})
+			_ = tx.Unscoped().Where("user_id = ?", id).Delete(&model.UserDevice{})
+			return tx.Unscoped().Delete(&model.User{}, "id = ?", id).Error
+		})
 	})
 }

internal/service/cloud/cloud.go

@@ -37,10 +37,11 @@ var ErrUnsupported = errors.New("unsupported cloud provider")
 
 // FileEntry is one item in a cloud directory listing.
 type FileEntry struct {
-	ID    string `json:"id"` // provider-native file id
-	Name  string `json:"name"`
-	IsDir bool   `json:"is_dir"`
-	Size  int64  `json:"size"`
+	ID       string `json:"id"` // provider-native file id
+	Name     string `json:"name"`
+	IsDir    bool   `json:"is_dir"`
+	Size     int64  `json:"size"`
+	MTime    int64  `json:"mtime,omitempty"`
 	// PickCode is 115-specific; other providers use ID directly.
 	PickCode string `json:"pick_code,omitempty"`
 }

internal/service/cloud/pan115_openapi.go

@@ -17,11 +17,20 @@ package cloud
 import (
 	"context"
 	"fmt"
+	"io"
+	"os"
+	"path/filepath"
 	"strings"
 
 	"github.com/ShukeBta/MMTL/internal/service/cloud115"
 )
 
+// OpenAPI115Provider 暴露 115 开放平台驱动接口。
+type OpenAPI115Provider interface {
+	Provider
+	OpenClient() *cloud115.OpenClient
+}
+
 // openAPI115Provider 实现 Provider 接口：List 列目录、Resolve 用 pickcode
 // 换下载直链（302 offload，无需代理）、Ping 探测根目录。
 type openAPI115Provider struct {
@@ -61,6 +70,7 @@ func (p *openAPI115Provider) List(ctx context.Context, dirID string) ([]FileEntr
 				Name:     f.FileName,
 				IsDir:    f.Category == cloud115.TypeDir,
 				Size:     f.FileSize,
+				MTime:    f.Utime,
 				PickCode: f.PickCode,
 			})
 		}
@@ -98,6 +108,39 @@ func (p *openAPI115Provider) ResolveWithUA(ctx context.Context, fileRef, ua stri
 // OpenClient 暴露底层客户端（token 刷新用）。
 func (p *openAPI115Provider) OpenClient() *cloud115.OpenClient { return p.c }
 
+// PutFileNamed 把本地元数据上传到 115 指定父目录（parentCID 为父目录 cid）。
+// io.Reader 无法携带文件名，因此走独立的 named 上传接口。将内容落为临时文件后
+// 重命名为目标文件名，再交给 115 上传（/open/upload/init 的 file_name 取真实文件名）。
+func (p *openAPI115Provider) PutFileNamed(ctx context.Context, parentCID, fileName string, r io.Reader) error {
+	tmp, err := os.CreateTemp("", "mmtl-upload-*")
+	if err != nil {
+		return fmt.Errorf("115: 创建临时文件失败：%w", err)
+	}
+	tmpPath := tmp.Name()
+	defer func() {
+		_ = tmp.Close()
+		_ = os.Remove(tmpPath)
+	}()
+	if _, err := io.Copy(tmp, r); err != nil {
+		return fmt.Errorf("115: 写入临时文件失败：%w", err)
+	}
+	if err := tmp.Close(); err != nil {
+		return fmt.Errorf("115: 关闭临时文件失败：%w", err)
+	}
+	// 重命名为目标文件名，保证上传到 115 后保留原始文件名
+	if fileName != "" && fileName != filepath.Base(tmpPath) {
+		namedPath := filepath.Join(filepath.Dir(tmpPath), fileName)
+		if err := os.Rename(tmpPath, namedPath); err == nil {
+			tmpPath = namedPath
+		}
+	}
+	_, err = p.c.Upload(ctx, tmpPath, parentCID, "", "")
+	if err != nil {
+		return err
+	}
+	return nil
+}
+
 // RefreshToken 刷新访问令牌并返回新令牌；refresh_token 失效时返回
 // cloud115.IsRefreshTokenDead(err) 为 true 的错误。
 func (p *openAPI115Provider) RefreshToken(refreshToken string) (*cloud115.TokenData, error) {

internal/service/cloud115/client.go

@@ -90,6 +90,7 @@ type RespBase struct {
 	Errno   int             `json:"errno"`
 	Message string          `json:"message"`
 	Error   string          `json:"error"`
+	Count   int64           `json:"count"`
 	Data    json.RawMessage `json:"data"`
 	Raw     json.RawMessage `json:"-"` // 原始响应体（外层附加字段用）
 }
@@ -184,7 +185,7 @@ func (c *OpenClient) doJSON(ctx context.Context, method, rawURL string, form map
 		// refresh_token 刷新后重试一次。刷新失败或重试后仍失败才返回，
 		// 避免长时间同步因 token 过期而整体失败。
 		if isTokenCode(base.Code) {
-			if access && c.tryRefreshTokenLocked() {
+			if access && c.tryRefreshTokenLocked(ctx) {
 				continue
 			}
 			if access {
@@ -258,19 +259,43 @@ func (c *OpenClient) doAuthJSONWithUA(ctx context.Context, method, rawURL string
 }
 
 // tryRefreshTokenLocked 并发安全地刷新 access_token；成功返回 true（调用方
-// 应使用内存中的新 token 重试原请求）。refresh_token 已失效时也会清空内存 token。
-func (c *OpenClient) tryRefreshTokenLocked() bool {
+// 应使用内存中的新 token 重试原请求）。
+//
+// 对"refresh_token 本身已失效/被吊销"（IsRefreshTokenDead，如 40140114/116/119/120）
+// 这类不可恢复的错误直接放弃并清空内存 token（提示需重新授权）。
+// 对其它失败（网络瞬时抖动、刷新接口可重试错误码等）做指数退避重试几次再放弃，
+// 避免同步长任务中途 token 到期时恰好撞上一个短暂的刷新失败就整体失败。
+func (c *OpenClient) tryRefreshTokenLocked(ctx context.Context) bool {
 	c.tokenMu.Lock()
 	defer c.tokenMu.Unlock()
-	token, err := c.RefreshToken(c.RefreshTokenStr)
-	if err != nil {
+	for attempt := 0; attempt < refreshAttempts; attempt++ {
+		token, err := c.RefreshToken(c.RefreshTokenStr)
+		if err == nil {
+			c.SetAuthToken(token.AccessToken, token.RefreshToken)
+			return true
+		}
 		if IsRefreshTokenDead(err) {
 			c.SetAuthToken("", "")
+			return false
+		}
+		// 可恢复失败：退避后重试。ctx 取消时立即放弃。
+		if attempt < refreshAttempts-1 {
+			select {
+			case <-ctx.Done():
+				return false
+			case <-time.After(refreshBackoff(attempt)):
+			}
 		}
-		return false
 	}
-	c.SetAuthToken(token.AccessToken, token.RefreshToken)
-	return true
+	return false
+}
+
+// refreshAttempts 是刷新 access_token 失败时的最大尝试次数（含首次）。
+const refreshAttempts = 3
+
+// refreshBackoff 返回第 attempt 次（从 0 计）刷新失败后的退避时长（指数退避）。
+func refreshBackoff(attempt int) time.Duration {
+	return time.Duration(200*(1<<attempt)) * time.Millisecond // 200ms, 400ms
 }
 
 // IsThrottleCode 判断是否为限流错误码。
@@ -280,7 +305,7 @@ func IsThrottleCode(code int) bool {
 
 func isTokenCode(code int) bool {
 	switch code {
-	case AccessTokenAuthFail, AccessAuthInvalid, AccessTokenExpiryCode, RefreshTokenInvalid:
+	case AccessTokenAuthFail, AccessAuthInvalid, AccessTokenExpiryCode, AccessTokenFormatInvalid, RefreshTokenInvalid:
 		return true
 	}
 	return false

internal/service/cloud115/cloud115_test.go

@@ -375,3 +375,100 @@ func TestThrottleCodeHandling(t *testing.T) {
 		t.Fatal("code 770004 should trigger throttle status")
 	}
 }
+
+func TestRemoteFileDetailRelativePath(t *testing.T) {
+	rootCID := "3238787832374488117" // 影视库
+
+	// 场景 1：115 目录 paths 中只有祖先目录链，不包含自身
+	d1 := &RemoteFileDetail{
+		FileId:   "3251154147730910635",
+		FileName: "出包王女",
+		Paths: []struct {
+			FileId string `json:"file_id"`
+			Name   string `json:"file_name"`
+		}{
+			{FileId: "0", Name: "根目录"},
+			{FileId: "3238787832374488117", Name: "影视库"},
+			{FileId: "3238787913223892116", Name: "动漫"},
+		},
+	}
+	if got := d1.RelativePath(rootCID); got != "动漫/出包王女" {
+		t.Errorf("d1.RelativePath = %q, want %q", got, "动漫/出包王女")
+	}
+
+	// 场景 2：祖先中间目录，自身在 paths 末尾
+	d2 := &RemoteFileDetail{
+		FileId:   "3238787913223892116",
+		FileName: "动漫",
+		Paths: []struct {
+			FileId string `json:"file_id"`
+			Name   string `json:"file_name"`
+		}{
+			{FileId: "0", Name: "根目录"},
+			{FileId: "3238787832374488117", Name: "影视库"},
+			{FileId: "3238787913223892116", Name: "动漫"},
+		},
+	}
+	if got := d2.RelativePath(rootCID); got != "动漫" {
+		t.Errorf("d2.RelativePath = %q, want %q", got, "动漫")
+	}
+
+	// 场景 3：根同步目录自身
+	d3 := &RemoteFileDetail{
+		FileId:   rootCID,
+		FileName: "影视库",
+		Paths: []struct {
+			FileId string `json:"file_id"`
+			Name   string `json:"file_name"`
+		}{
+			{FileId: "0", Name: "根目录"},
+			{FileId: rootCID, Name: "影视库"},
+		},
+	}
+	if got := d3.RelativePath(rootCID); got != "" {
+		t.Errorf("d3.RelativePath = %q, want %q", got, "")
+	}
+}
+
+// TestFsListRefreshContinue 验证 access_token 在请求中途过期（40140126）时：
+// 自动用 refresh_token 刷新得到新 token，然后对原请求重试成功（同步得以继续）。
+func TestFsListRefreshContinue(t *testing.T) {
+	var filesCalls int
+	var refreshCalls int
+	mockAPI(t, func(w http.ResponseWriter, r *http.Request) {
+		switch r.URL.Path {
+		case "/open/refreshToken":
+			refreshCalls++
+			w.Write([]byte(`{"state":true,"data":{"access_token":"at2","refresh_token":"rt2","expires_in":7200}}`))
+		case "/open/ufile/files":
+			filesCalls++
+			switch filesCalls {
+			case 1:
+				// 第一次用旧 access_token，返回过期错误，应触发刷新
+				w.Write([]byte(`{"state":false,"code":40140126,"message":"access_token 校验失败"}`))
+			default:
+				// 刷新后续请求应使用新 access_token
+				if got := r.Header.Get("Authorization"); got != "Bearer at2" {
+					t.Errorf("retried request auth = %q, want Bearer at2", got)
+				}
+				w.Write([]byte(`{"state":true,"path":[],"data":[{"fid":"200","fc":"1","fn":"a.mkv","fs":123,"pc":"pickA"}]}`))
+			}
+		default:
+			t.Errorf("unexpected path %s", r.URL.Path)
+		}
+	})
+	c := NewOpenClient("100195125", "at1", "rt1")
+	files, _, err := c.GetFsList(context.Background(), "0", 0, 100)
+	if err != nil {
+		t.Fatalf("expected sync to continue after refresh, got error: %v", err)
+	}
+	if filesCalls != 2 {
+		t.Fatalf("want 2 files calls (original + retried), got %d", filesCalls)
+	}
+	if refreshCalls == 0 {
+		t.Fatal("expected refresh_token to be used once")
+	}
+	if len(files) != 1 {
+		t.Fatalf("want 1 file, got %d", len(files))
+	}
+}

internal/service/cloud115/const.go

@@ -21,13 +21,14 @@ var (
 
 const (
 	// 业务错误码
-	AccessTokenAuthFail   = 40140126 // 访问过期，需刷新
-	AccessTokenExpiryCode = 40140125 // 访问过期，需刷新
-	AccessAuthInvalid     = 40140124 // 访问无效，需刷新
-	RefreshTokenInvalid   = 40140116 // 需重新授权
-	TokenRefreshFail      = 40140121 // 刷新失败，可重试
-	RequestMaxLimitCode   = 770004   // 访问频率过高
-	RequestRateLimitCode  = 406      // 达到访问上限
+	AccessTokenAuthFail      = 40140126 // 访问过期，需刷新
+	AccessTokenExpiryCode    = 40140125 // 访问过期，需刷新
+	AccessAuthInvalid        = 40140124 // 访问无效，需刷新
+	AccessTokenFormatInvalid = 40140123 // access_token 格式错误，需刷新
+	RefreshTokenInvalid      = 40140116 // 需重新授权
+	TokenRefreshFail         = 40140121 // 刷新失败，可重试
+	RequestMaxLimitCode      = 770004   // 访问频率过高
+	RequestRateLimitCode     = 406      // 达到访问上限
 
 	// 刷新 token 的错误码
 	RefreshTokenFormatInvalid = 40140114

internal/service/cloud115/open.go

@@ -89,6 +89,36 @@ func (c *OpenClient) GetFsList(ctx context.Context, cid string, offset, limit in
 	return files, strings.Join(pathStr, "/"), nil
 }
 
+// GetFsListFlat 递归扁平化列出 cid 下的所有文件（跨越所有子目录，不包含文件夹节点），并返回文件列表与该树下的总文件数。
+// 类似于 QMediaSync 的 115 扁平化批量拉取机制，极大地降低多层级子目录下的 API 请求次数。
+func (c *OpenClient) GetFsListFlat(ctx context.Context, cid string, offset, limit int) ([]RemoteFile, int64, error) {
+	if cid == "" {
+		cid = "0"
+	}
+	if limit <= 0 {
+		limit = 1150
+	}
+	params := map[string]string{
+		"cid":      cid,
+		"limit":    fmt.Sprint(limit),
+		"offset":   fmt.Sprint(offset),
+		"cur":      "0",
+		"show_dir": "0",
+	}
+	resp, err := c.doAuthJSON(ctx, "GET", ProAPIBase+"/open/ufile/files", params, 2)
+	if err != nil {
+		return nil, 0, err
+	}
+	if !resp.State {
+		return nil, 0, NewOpenAPIResponseError(resp.Code, resp.Errno, resp.Message, resp.Error, "115 接口调用失败")
+	}
+	files, err := openList[RemoteFile](resp.Data)
+	if err != nil {
+		return nil, 0, fmt.Errorf("115: 解析文件列表失败：%w", err)
+	}
+	return files, resp.Count, nil
+}
+
 // GetFsDetailByCid 查询文件（夹）详情。
 func (c *OpenClient) GetFsDetailByCid(ctx context.Context, fileId string) (*RemoteFileDetail, error) {
 	params := map[string]string{"file_id": fileId}
@@ -113,6 +143,48 @@ type RemoteFileDetail struct {
 	} `json:"paths"`
 }
 
+// RelativePath 计算该目录相对于根同步目录（rootCID）的相对路径。
+func (d *RemoteFileDetail) RelativePath(rootCID string) string {
+	if d == nil {
+		return ""
+	}
+	if rootCID == "" {
+		rootCID = "0"
+	}
+	if d.FileId == rootCID {
+		return ""
+	}
+	rootIdx := -1
+	for i, p := range d.Paths {
+		if p.FileId == rootCID {
+			rootIdx = i
+			break
+		}
+	}
+	var segments []string
+	start := 0
+	if rootIdx >= 0 {
+		start = rootIdx + 1
+	} else if len(d.Paths) > 0 && (d.Paths[0].FileId == "0" || d.Paths[0].FileId == "") {
+		start = 1
+	}
+	hasSelf := false
+	for i := start; i < len(d.Paths); i++ {
+		if d.Paths[i].FileId == d.FileId {
+			hasSelf = true
+		}
+		name := strings.TrimSpace(d.Paths[i].Name)
+		if name != "" {
+			segments = append(segments, name)
+		}
+	}
+	// 若 115 返回的 paths 祖先链未包含当前目录自身，则将其自身目录名 FileName 补在末尾
+	if !hasSelf && strings.TrimSpace(d.FileName) != "" && d.FileId != rootCID {
+		segments = append(segments, strings.TrimSpace(d.FileName))
+	}
+	return strings.Join(segments, "/")
+}
+
 // ─── 下载直链 ──────────────────────────────────────────────────────────────────
 
 type downloadURLData struct {

internal/service/cloud115/oss_multipart.go

@@ -0,0 +1,367 @@
+// 阿里云 OSS multipart 分片上传（用于 115 元数据上传直传）。
+// 使用 115 下发的临时 STS 凭证，将本地文件分片上传到 OSS，并经 complete 回调
+// 通知 115 完成落盘。参考 QMediaSync 的 OSSMultipartUploader 实现。
+package cloud115
+
+import (
+	"context"
+	"encoding/base64"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"io"
+	"os"
+	"sort"
+
+	"github.com/aliyun/alibabacloud-oss-go-sdk-v2/oss"
+	"github.com/aliyun/alibabacloud-oss-go-sdk-v2/oss/credentials"
+)
+
+const (
+	defaultMultipartPartSize int64 = 32 * 1024 * 1024
+	multipartPartAlign       int64 = 1024 * 1024
+	maxMultipartParts        int64 = 9999
+	maxMultipartPartSize     int64 = 5 * 1024 * 1024 * 1024
+)
+
+type ossMultipartClient interface {
+	InitiateMultipartUpload(context.Context, *oss.InitiateMultipartUploadRequest, ...func(*oss.Options)) (*oss.InitiateMultipartUploadResult, error)
+	UploadPart(context.Context, *oss.UploadPartRequest, ...func(*oss.Options)) (*oss.UploadPartResult, error)
+	ListParts(context.Context, *oss.ListPartsRequest, ...func(*oss.Options)) (*oss.ListPartsResult, error)
+	CompleteMultipartUpload(context.Context, *oss.CompleteMultipartUploadRequest, ...func(*oss.Options)) (*oss.CompleteMultipartUploadResult, error)
+	AbortMultipartUpload(context.Context, *oss.AbortMultipartUploadRequest, ...func(*oss.Options)) (*oss.AbortMultipartUploadResult, error)
+}
+
+// OSSMultipartUploader 封装 OSS multipart 上传。
+type OSSMultipartUploader struct {
+	client ossMultipartClient
+}
+
+// OSSMultipartUploadInput 是 multipart 上传输入。
+type OSSMultipartUploadInput struct {
+	Bucket        string
+	Object        string
+	Callback      string
+	CallbackVar   string
+	FilePath      string
+	FileSize      int64
+	UploadId      string
+	PartSize      int64
+	PartRetryMax  int
+	refreshClient func(context.Context) (ossMultipartClient, error)
+}
+
+// OSSMultipartUploadResult 是 multipart 上传后的结果。
+type OSSMultipartUploadResult struct {
+	CallbackResult map[string]any
+	UploadId       string
+	PartSize       int64
+	TotalParts     int
+	UploadedBytes  int64
+	UploadedParts  int
+}
+
+// CalculateMultipartPartSize 计算 OSS multipart 分片大小与分片数量。
+func CalculateMultipartPartSize(fileSize int64) (int64, int, error) {
+	if fileSize < 0 {
+		return 0, 0, fmt.Errorf("文件大小不能为负数：%d", fileSize)
+	}
+	partSize := defaultMultipartPartSize
+	minPartSize := ceilDiv(fileSize, maxMultipartParts)
+	if minPartSize > partSize {
+		partSize = roundUp(minPartSize, multipartPartAlign)
+	}
+	if partSize > maxMultipartPartSize {
+		return 0, 0, fmt.Errorf("文件过大，所需分片大小 %d 超过 OSS 上限 %d", partSize, maxMultipartPartSize)
+	}
+	totalParts := int(ceilDiv(fileSize, partSize))
+	if totalParts == 0 {
+		totalParts = 1
+	}
+	if int64(totalParts) > maxMultipartParts {
+		return 0, 0, fmt.Errorf("分片数量 %d 超过上限 %d", totalParts, maxMultipartParts)
+	}
+	return partSize, totalParts, nil
+}
+
+// NewOSSMultipartUploader 创建 OSS multipart 上传器。
+func NewOSSMultipartUploader(endpoint, accessKeyId, accessKeySecret, securityToken string) *OSSMultipartUploader {
+	return &OSSMultipartUploader{client: newOSSMultipartClient(endpoint, accessKeyId, accessKeySecret, securityToken)}
+}
+
+func newOSSMultipartClient(endpoint, accessKeyId, accessKeySecret, securityToken string) ossMultipartClient {
+	cfg := oss.LoadDefaultConfig().
+		WithCredentialsProvider(credentials.NewStaticCredentialsProvider(accessKeyId, accessKeySecret, securityToken)).
+		WithRegion("cn-shenzhen").
+		WithEndpoint(endpoint)
+	return oss.NewClient(cfg)
+}
+
+// UploadFile 上传文件并完成 OSS multipart，返回 complete callback 结果。
+func (u *OSSMultipartUploader) UploadFile(ctx context.Context, input OSSMultipartUploadInput) (map[string]any, error) {
+	result, err := u.UploadFileWithResult(ctx, input)
+	if err != nil {
+		return nil, err
+	}
+	return result.CallbackResult, nil
+}
+
+// UploadFileWithResult 上传文件并返回 multipart 结果。
+func (u *OSSMultipartUploader) UploadFileWithResult(ctx context.Context, input OSSMultipartUploadInput) (OSSMultipartUploadResult, error) {
+	if input.PartRetryMax <= 0 {
+		input.PartRetryMax = 3
+	}
+	partSize := input.PartSize
+	totalParts := 0
+	var err error
+	if partSize <= 0 {
+		partSize, totalParts, err = CalculateMultipartPartSize(input.FileSize)
+		if err != nil {
+			return OSSMultipartUploadResult{}, err
+		}
+	} else {
+		totalParts = int(ceilDiv(input.FileSize, partSize))
+	}
+
+	uploadId := input.UploadId
+	if uploadId == "" {
+		initResult, err := u.client.InitiateMultipartUpload(ctx, &oss.InitiateMultipartUploadRequest{
+			Bucket: oss.Ptr(input.Bucket),
+			Key:    oss.Ptr(input.Object),
+			RequestCommon: oss.RequestCommon{
+				Parameters: map[string]string{"sequential": "1"},
+			},
+		})
+		if err != nil {
+			return OSSMultipartUploadResult{}, fmt.Errorf("初始化 OSS multipart 失败：%w", err)
+		}
+		if initResult.UploadId == nil || *initResult.UploadId == "" {
+			return OSSMultipartUploadResult{}, fmt.Errorf("初始化 OSS multipart 返回空 upload_id")
+		}
+		uploadId = *initResult.UploadId
+	}
+
+	existingPartMap := make(map[int32]int64)
+	existingParts, err := u.ListUploadedParts(ctx, input.Bucket, input.Object, uploadId)
+	if err == nil {
+		for _, part := range existingParts {
+			existingPartMap[part.PartNumber] = part.Size
+		}
+	}
+
+	file, err := os.Open(input.FilePath)
+	if err != nil {
+		return OSSMultipartUploadResult{}, fmt.Errorf("打开待上传文件失败：%w", err)
+	}
+	defer file.Close()
+
+	var uploadedBytes int64
+	uploadedParts := 0
+	completeParts := make([]oss.UploadPart, 0, totalParts)
+	for partNumber := 1; partNumber <= totalParts; partNumber++ {
+		offset := int64(partNumber-1) * partSize
+		length := minInt64(partSize, input.FileSize-offset)
+		if length < 0 {
+			length = 0
+		}
+		if existingSize, ok := existingPartMap[int32(partNumber)]; ok && existingSize == length {
+			uploadedBytes += length
+			uploadedParts++
+		}
+		etag, err := u.uploadPartWithRetry(ctx, input, uploadId, int32(partNumber), file, offset, length)
+		if err != nil {
+			return OSSMultipartUploadResult{}, err
+		}
+		uploadedBytes += length
+		uploadedParts++
+		completeParts = append(completeParts, oss.UploadPart{
+			PartNumber: int32(partNumber),
+			ETag:       oss.Ptr(etag),
+		})
+	}
+	sort.Slice(completeParts, func(i, j int) bool {
+		return completeParts[i].PartNumber < completeParts[j].PartNumber
+	})
+
+	// 115 下发的 callback / callback_var 是 JSON 字符串，而 OSS CompleteMultipartUpload
+	// 要求 callback 参数为 Base64 编码后的 JSON，否则报 "The callback configuration is
+	// not base64 encoded"。这里把两者转为 Base64 后再提交（参考 QMediaSync 的
+	// BuildOSSCallbackHeaders）。
+	cb := input.Callback
+	cbVar := input.CallbackVar
+	if cb == "" {
+		return OSSMultipartUploadResult{}, errors.New("OSS callback 为空")
+	}
+	if !json.Valid([]byte(cb)) {
+		return OSSMultipartUploadResult{}, errors.New("解析 callback 失败：不是合法 JSON")
+	}
+	if cbVar == "" {
+		cbVar = "{}"
+	}
+	if !json.Valid([]byte(cbVar)) {
+		return OSSMultipartUploadResult{}, errors.New("解析 callback_var 失败：不是合法 JSON")
+	}
+	completeResult, err := u.client.CompleteMultipartUpload(ctx, &oss.CompleteMultipartUploadRequest{
+		Bucket:   oss.Ptr(input.Bucket),
+		Key:      oss.Ptr(input.Object),
+		UploadId: oss.Ptr(uploadId),
+		CompleteMultipartUpload: &oss.CompleteMultipartUpload{
+			Parts: completeParts,
+		},
+		Callback:    oss.Ptr(base64.StdEncoding.EncodeToString([]byte(cb))),
+		CallbackVar: oss.Ptr(base64.StdEncoding.EncodeToString([]byte(cbVar))),
+	})
+	if err != nil {
+		return OSSMultipartUploadResult{}, fmt.Errorf("完成 OSS multipart 失败：%w", err)
+	}
+	return OSSMultipartUploadResult{
+		CallbackResult: completeResult.CallbackResult,
+		UploadId:       uploadId,
+		PartSize:       partSize,
+		TotalParts:     totalParts,
+		UploadedBytes:  uploadedBytes,
+		UploadedParts:  uploadedParts,
+	}, nil
+}
+
+// ListUploadedParts 查询 OSS 已上传分片。
+func (u *OSSMultipartUploader) ListUploadedParts(ctx context.Context, bucket, object, uploadId string) ([]struct {
+	PartNumber int32
+	Size       int64
+}, error) {
+	parts := []struct {
+		PartNumber int32
+		Size       int64
+	}{}
+	result, err := u.client.ListParts(ctx, &oss.ListPartsRequest{
+		Bucket:   oss.Ptr(bucket),
+		Key:      oss.Ptr(object),
+		UploadId: oss.Ptr(uploadId),
+		MaxParts: 1000,
+	})
+	if err != nil {
+		return nil, fmt.Errorf("查询 OSS 已上传分片失败：%w", err)
+	}
+	for _, part := range result.Parts {
+		parts = append(parts, struct {
+			PartNumber int32
+			Size       int64
+		}{PartNumber: part.PartNumber, Size: part.Size})
+	}
+	return parts, nil
+}
+
+func (u *OSSMultipartUploader) uploadPartWithRetry(
+	ctx context.Context,
+	input OSSMultipartUploadInput,
+	uploadId string,
+	partNumber int32,
+	file *os.File,
+	offset, length int64,
+) (string, error) {
+	var lastErr error
+	for attempt := 0; attempt < input.PartRetryMax; attempt++ {
+		reader := io.NewSectionReader(file, offset, length)
+		result, err := u.client.UploadPart(ctx, &oss.UploadPartRequest{
+			Bucket:        oss.Ptr(input.Bucket),
+			Key:           oss.Ptr(input.Object),
+			PartNumber:    partNumber,
+			UploadId:      oss.Ptr(uploadId),
+			Body:          reader,
+			ContentLength: oss.Ptr(length),
+		})
+		if err == nil {
+			if result.ETag == nil || *result.ETag == "" {
+				return "", fmt.Errorf("OSS part %d 返回空 ETag", partNumber)
+			}
+			return *result.ETag, nil
+		}
+		lastErr = err
+		if attempt < input.PartRetryMax-1 && input.refreshClient != nil {
+			refreshed, refreshErr := input.refreshClient(ctx)
+			if refreshErr != nil {
+				lastErr = refreshErr
+				continue
+			}
+			u.client = refreshed
+		}
+	}
+	return "", fmt.Errorf("上传 OSS part %d 失败：%w", partNumber, lastErr)
+}
+
+// ParseCompleteCallbackResult 校验并解析 OSS complete 后的 115 callback 结果。
+func ParseCompleteCallbackResult(result map[string]any) (UploadCompleteResult, error) {
+	if result == nil {
+		return UploadCompleteResult{}, errors.New("OSS complete callback 结果为空")
+	}
+	if state, ok := result["state"].(bool); ok && !state {
+		return UploadCompleteResult{}, fmt.Errorf("115 callback 返回失败：%s", anyToString(result["message"]))
+	}
+	if message := anyToString(result["message"]); message != "" {
+		return UploadCompleteResult{}, fmt.Errorf("115 callback 返回错误：%s", message)
+	}
+	data, ok := result["data"].(map[string]any)
+	if !ok {
+		return UploadCompleteResult{}, errors.New("115 callback 缺少 data")
+	}
+	complete := UploadCompleteResult{
+		FileId:   anyToString(data["file_id"]),
+		PickCode: anyToString(data["pick_code"]),
+		ParentId: anyToString(data["parent_id"]),
+		Sha1:     anyToString(data["sha1"]),
+		Size:     anyToInt64(data["size"]),
+		Mtime:    anyToInt64(data["mtime"]),
+	}
+	if complete.FileId == "" || complete.PickCode == "" {
+		return UploadCompleteResult{}, errors.New("115 callback 缺少 file_id/pick_code")
+	}
+	return complete, nil
+}
+
+func ceilDiv(n, d int64) int64 {
+	if d <= 0 {
+		return 0
+	}
+	if n <= 0 {
+		return 0
+	}
+	return (n + d - 1) / d
+}
+
+func roundUp(n, align int64) int64 {
+	if align <= 0 {
+		return n
+	}
+	return ceilDiv(n, align) * align
+}
+
+func minInt64(a, b int64) int64 {
+	if a < b {
+		return a
+	}
+	return b
+}
+
+func anyToInt64(v any) int64 {
+	switch t := v.(type) {
+	case string:
+		var n int64
+		fmt.Sscanf(t, "%d", &n)
+		return n
+	case float64:
+		return int64(t)
+	case int64:
+		return t
+	case int:
+		return int64(t)
+	default:
+		return 0
+	}
+}
+
+func anyToString(v any) string {
+	if s, ok := v.(string); ok {
+		return s
+	}
+	return fmt.Sprintf("%v", v)
+}

internal/service/cloud115/queue_executor.go

@@ -26,10 +26,15 @@ var (
 	executorOnce   sync.Once
 )
 
-// GetGlobalExecutor 获取全局队列执行器单例（默认 QPS=2, QPM=120, QPH=6000，保障 115 API 调用安全不超频）。
+// GetGlobalExecutor 获取全局队列执行器单例（默认 QPS=3, QPM=200, QPH=12000，保障 115 API 调用安全不超频）。
+//
+// 历史教训：QPS 提到 8 后，下载换直链接口（/open/ufile/downurl，WAF 重点盯防对象）
+// 瞬时突发撞上 115 风控，返回阿里云 405 阻断页（HTTP 405），导致全量同步失败。
+// 因此回调到 3——这是经过实测的安全上限：宁慢勿触发风控，一旦 405 冷却 180 秒，
+// 整体吞吐反而更低。下载实际走 CDN 不受此限速影响，瓶颈仅在换链环节。
 func GetGlobalExecutor() *QueueExecutor {
 	executorOnce.Do(func() {
-		globalExecutor = NewQueueExecutor(2, 120, 6000)
+		globalExecutor = NewQueueExecutor(3, 200, 12000)
 	})
 	return globalExecutor
 }

internal/service/cloud115/sha1.go

@@ -0,0 +1,49 @@
+package cloud115
+
+import (
+	"crypto/sha1"
+	"encoding/hex"
+	"io"
+	"os"
+)
+
+// FileSHA1 计算文件完整 SHA1（小写 hex）。
+func FileSHA1(path string) (string, error) {
+	f, err := os.Open(path)
+	if err != nil {
+		return "", err
+	}
+	defer f.Close()
+	h := sha1.New()
+	if _, err := io.Copy(h, f); err != nil {
+		return "", err
+	}
+	return hex.EncodeToString(h.Sum(nil)), nil
+}
+
+// FileSHA1Partial 计算文件 [start,end]（含）字节区间的 SHA1（小写 hex）。
+// 用于 115 上传二次签名按 sign_check 指定的区间重算哈希。
+func FileSHA1Partial(path string, start, end int64) (string, error) {
+	if start < 0 {
+		start = 0
+	}
+	if end < start {
+		end = start
+	}
+	f, err := os.Open(path)
+	if err != nil {
+		return "", err
+	}
+	defer f.Close()
+	if _, err := f.Seek(start, io.SeekStart); err != nil {
+		return "", err
+	}
+	length := end - start + 1
+	h := sha1.New()
+	// io.CopyN 在文件不足 length 字节时会返回 io.EOF，导致小文件（如小于 128 KiB 的
+	// 元数据图片）无法上传。这里只拷贝实际读到的字节，文件尾对齐到区间终点即可。
+	if _, err := io.CopyN(h, f, length); err != nil && err != io.EOF {
+		return "", err
+	}
+	return hex.EncodeToString(h.Sum(nil)), nil
+}

internal/service/cloud115/upload.go

@@ -0,0 +1,355 @@
+// 115 网盘元数据上传能力：115 开放平台调度 + 阿里云 OSS 直传。
+// 参考 QMediaSync 的上传流程实现：
+//
+//	POST /open/upload/init           上传初始化/秒传调度（含二次签名）
+//	GET  /open/upload/get_token      获取 OSS 临时上传凭证（STS）
+//	OSS  multipart 分片直传 + callback 完成
+//
+// 上传目标父目录为 115 目录 ID（cid），而非路径字符串。
+package cloud115
+
+import (
+	"context"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"os"
+	"strconv"
+	"strings"
+)
+
+// 115 上传状态码。
+const (
+	UploadInitStatusNeedUpload    = 1 // 需要真实上传
+	UploadInitStatusRapidUploaded = 2 // 秒传成功
+	UploadInitStatusSignFailed    = 6 // 签名验证失败
+	UploadInitStatusNeedSign      = 7 // 需要二次签名
+	UploadInitStatusSignRejected  = 8 // 签名认证失败
+)
+
+// UploadInitRequest 是 /open/upload/init 的结构化请求。
+type UploadInitRequest struct {
+	FileName     string
+	FileSize     int64
+	ParentFileId string
+	FileSha1     string
+	Preid        string
+	PickCode     string
+	TopUpload    string
+	SignKey      string
+	SignVal      string
+}
+
+// UploadInitResult 是 /open/upload/init 的调度结果。
+type UploadInitResult struct {
+	PickCode  string
+	Status    int
+	FileId    string
+	Target    string
+	Bucket    string
+	Object    string
+	SignKey   string
+	SignCheck string
+	Callback  UploadResultCallBack
+}
+
+type uploadScheduleAPIResult struct {
+	PickCode  string          `json:"pick_code"`
+	Status    int             `json:"status"`
+	FileId    string          `json:"file_id"`
+	Target    string          `json:"target"`
+	Version   string          `json:"version"`
+	Bucket    string          `json:"bucket"`
+	Object    string          `json:"object"`
+	SignKey   string          `json:"sign_key"`
+	SignCheck string          `json:"sign_check"`
+	Callback  json.RawMessage `json:"callback"`
+}
+
+// UploadResultCallBack 是 init 返回给 OSS complete 使用的 callback 内容。
+type UploadResultCallBack struct {
+	Callback    string `json:"callback"`
+	CallbackVar string `json:"callback_var"`
+}
+
+// UploadToken 是 /open/upload/get_token 返回的 OSS STS 临时凭证。
+type UploadToken struct {
+	Endpoint         string `json:"endpoint"`
+	AccessKeySecret  string `json:"AccessKeySecret"`
+	AccessKeySecrett string `json:"AccessKeySecrett"`
+	SecurityToken    string `json:"SecurityToken"`
+	Expiration       string `json:"Expiration"`
+	AccessKeyId      string `json:"AccessKeyId"`
+}
+
+func (token *UploadToken) normalize() {
+	if token == nil {
+		return
+	}
+	if token.AccessKeySecret == "" {
+		token.AccessKeySecret = token.AccessKeySecrett
+	}
+}
+
+// UploadCompleteResult 是 OSS complete callback 成功后的远端文件定位结果。
+type UploadCompleteResult struct {
+	FileId   string
+	PickCode string
+	ParentId string
+	Sha1     string
+	Size     int64
+	Mtime    int64
+}
+
+// SignCheckRange 是 115 二次认证要求的闭区间 [start,end]。
+type SignCheckRange struct {
+	Start int64
+	End   int64
+}
+
+// UploadInit 调用 115 上传初始化/秒传调度接口。
+func (c *OpenClient) UploadInit(ctx context.Context, input UploadInitRequest) (*UploadInitResult, error) {
+	params := buildUploadInitForm(input)
+	resp, err := c.doAuthJSON(ctx, "POST", ProAPIBase+"/open/upload/init", params, 2)
+	if err != nil {
+		return nil, err
+	}
+	var raw uploadScheduleAPIResult
+	if err := json.Unmarshal(resp.Data, &raw); err != nil {
+		return nil, fmt.Errorf("115: 解析 upload/init 结果失败：%w", err)
+	}
+	callback, err := decodeUploadCallback(raw.Callback)
+	if err != nil {
+		return nil, err
+	}
+	return &UploadInitResult{
+		PickCode:  raw.PickCode,
+		Status:    raw.Status,
+		FileId:    raw.FileId,
+		Target:    raw.Target,
+		Bucket:    raw.Bucket,
+		Object:    raw.Object,
+		SignKey:   raw.SignKey,
+		SignCheck: raw.SignCheck,
+		Callback:  callback,
+	}, nil
+}
+
+func buildUploadInitForm(input UploadInitRequest) map[string]string {
+	topUpload := input.TopUpload
+	if topUpload == "" {
+		topUpload = "0"
+	}
+	params := map[string]string{
+		"file_name": input.FileName,
+		"file_size": strconv.FormatInt(input.FileSize, 10),
+		"target":    fmt.Sprintf("U_1_%s", input.ParentFileId),
+		"fileid":    input.FileSha1,
+		"preid":     input.Preid,
+		"topupload": topUpload,
+	}
+	if input.PickCode != "" {
+		params["pick_code"] = input.PickCode
+	}
+	if input.SignKey != "" && input.SignVal != "" {
+		params["sign_key"] = input.SignKey
+		params["sign_val"] = input.SignVal
+	}
+	return params
+}
+
+func decodeUploadCallback(raw json.RawMessage) (UploadResultCallBack, error) {
+	if len(raw) == 0 || string(raw) == "null" {
+		return UploadResultCallBack{}, nil
+	}
+	if raw[0] == '[' {
+		var callbacks []UploadResultCallBack
+		if err := json.Unmarshal(raw, &callbacks); err != nil {
+			return UploadResultCallBack{}, err
+		}
+		if len(callbacks) == 0 {
+			return UploadResultCallBack{}, nil
+		}
+		return callbacks[0], nil
+	}
+	var callback UploadResultCallBack
+	if err := json.Unmarshal(raw, &callback); err != nil {
+		return UploadResultCallBack{}, err
+	}
+	return callback, nil
+}
+
+func parseSignCheckRange(value string) (SignCheckRange, error) {
+	parts := strings.Split(value, "-")
+	if len(parts) != 2 {
+		return SignCheckRange{}, fmt.Errorf("sign_check 格式错误：%s", value)
+	}
+	start, err := strconv.ParseInt(strings.TrimSpace(parts[0]), 10, 64)
+	if err != nil {
+		return SignCheckRange{}, err
+	}
+	end, err := strconv.ParseInt(strings.TrimSpace(parts[1]), 10, 64)
+	if err != nil {
+		return SignCheckRange{}, err
+	}
+	if start < 0 || end < start {
+		return SignCheckRange{}, fmt.Errorf("sign_check 范围非法：%s", value)
+	}
+	return SignCheckRange{Start: start, End: end}, nil
+}
+
+// GetUploadToken 获取 115 下发的 OSS 临时上传凭证。
+func (c *OpenClient) GetUploadToken(ctx context.Context) (*UploadToken, error) {
+	resp, err := c.doAuthJSON(ctx, "GET", ProAPIBase+"/open/upload/get_token", nil, 2)
+	if err != nil {
+		return nil, err
+	}
+	var token UploadToken
+	if err := json.Unmarshal(resp.Data, &token); err != nil {
+		return nil, fmt.Errorf("115: 解析 get_token 结果失败：%w", err)
+	}
+	token.normalize()
+	return &token, nil
+}
+
+// Upload 上传单个本地文件到 115 指定父目录（cid），返回成功后的远端文件信息。
+// filePath 必须是落到磁盘的真实文件路径（调用方负责把 io.Reader 落盘为临时文件）。
+func (c *OpenClient) Upload(ctx context.Context, filePath, parentCID, signKey, signVal string) (*UploadCompleteResult, error) {
+	fileSize := fileSizeOf(filePath)
+	if fileSize < 0 {
+		return nil, fmt.Errorf("115: 无法获取文件大小：%s", filePath)
+	}
+	fileSha1, err := FileSHA1(filePath)
+	if err != nil {
+		return nil, fmt.Errorf("115: 计算文件 SHA1 失败：%w", err)
+	}
+	preSha1, err := FileSHA1Partial(filePath, 0, 128*1024-1)
+	if err != nil {
+		return nil, fmt.Errorf("115: 计算文件前 128 KiB SHA1 失败：%w", err)
+	}
+	request := UploadInitRequest{
+		FileName:     baseNameOf(filePath),
+		FileSize:     fileSize,
+		ParentFileId: parentCID,
+		FileSha1:     fileSha1,
+		Preid:        preSha1,
+		TopUpload:    "0",
+		SignKey:      signKey,
+		SignVal:      signVal,
+	}
+	initResult, err := c.UploadInit(ctx, request)
+	if err != nil {
+		return nil, fmt.Errorf("115: 上传初始化失败：%w", err)
+	}
+	status := initResult.Status
+	if status == UploadInitStatusNeedSign {
+		// 二次签名：按 sign_check 指定区间重算 sha1
+		rng, err := parseSignCheckRange(initResult.SignCheck)
+		if err != nil {
+			return nil, err
+		}
+		signValue, err := FileSHA1Partial(filePath, rng.Start, rng.End)
+		if err != nil {
+			return nil, err
+		}
+		request.SignKey = initResult.SignKey
+		request.SignVal = signValue
+		initResult, err = c.UploadInit(ctx, request)
+		if err != nil {
+			return nil, fmt.Errorf("115: 上传二次签名失败：%w", err)
+		}
+		status = initResult.Status
+	}
+	switch status {
+	case UploadInitStatusRapidUploaded:
+		// 秒传成功
+		return &UploadCompleteResult{FileId: initResult.FileId, PickCode: initResult.PickCode}, nil
+	case UploadInitStatusSignFailed:
+		return nil, fmt.Errorf("115: 签名验证后失败")
+	case UploadInitStatusSignRejected:
+		return nil, fmt.Errorf("115: 签名认证失败")
+	case UploadInitStatusNeedUpload:
+		// 真实上传：OSS multipart
+	default:
+		return &UploadCompleteResult{FileId: initResult.FileId, PickCode: initResult.PickCode}, nil
+	}
+
+	if initResult.Bucket == "" || initResult.Object == "" {
+		return nil, fmt.Errorf("115: upload/init 缺少 bucket/object 信息")
+	}
+	token, err := c.GetUploadToken(ctx)
+	if err != nil {
+		return nil, fmt.Errorf("115: 获取上传凭证失败：%w", err)
+	}
+	if token == nil || token.Endpoint == "" || token.AccessKeyId == "" || token.AccessKeySecret == "" {
+		return nil, fmt.Errorf("115: 上传凭证不完整")
+	}
+	uploader := NewOSSMultipartUploader(token.Endpoint, token.AccessKeyId, token.AccessKeySecret, token.SecurityToken)
+	result, err := uploader.UploadFile(ctx, OSSMultipartUploadInput{
+		Bucket:      initResult.Bucket,
+		Object:      initResult.Object,
+		Callback:    initResult.Callback.Callback,
+		CallbackVar: initResult.Callback.CallbackVar,
+		FilePath:    filePath,
+		FileSize:    fileSize,
+		refreshClient: func(ctx context.Context) (ossMultipartClient, error) {
+			refreshed, rerr := c.GetUploadToken(ctx)
+			if rerr != nil || refreshed == nil {
+				return nil, rerr
+			}
+			return newOSSMultipartClient(refreshed.Endpoint, refreshed.AccessKeyId, refreshed.AccessKeySecret, refreshed.SecurityToken), nil
+		},
+	})
+	if err != nil {
+		return nil, fmt.Errorf("115: OSS 上传失败：%w", err)
+	}
+	complete, err := ParseCompleteCallbackResult(result)
+	if err != nil {
+		return nil, err
+	}
+	return &complete, nil
+}
+
+// MkDir 在 115 的 parentCid 下创建目录，返回新目录 cid。
+func (c *OpenClient) MkDir(ctx context.Context, parentCID, name string) (string, error) {
+	params := map[string]string{
+		"cname": name,
+		"pid":   parentCID,
+	}
+	resp, err := c.doAuthJSON(ctx, "POST", ProAPIBase+"/open/folder/add", params, 2)
+	if err != nil {
+		return "", err
+	}
+	// /open/folder/add 结构：{ aid, cid, fid, name, pid, ... }，单一对象
+	var r struct {
+		Cid string `json:"cid"`
+	}
+	if err := json.Unmarshal(resp.Data, &r); err != nil {
+		return "", fmt.Errorf("115: 解析 folder/add 结果失败：%w", err)
+	}
+	if r.Cid == "" {
+		return "", errors.New("115: folder/add 未返回 cid")
+	}
+	return r.Cid, nil
+}
+
+func fileSizeOf(path string) int64 {
+	info, err := os.Stat(path)
+	if err != nil {
+		return -1
+	}
+	if info.IsDir() {
+		return -1
+	}
+	return info.Size()
+}
+
+func baseNameOf(path string) string {
+	s := path
+	for i := len(s) - 1; i >= 0; i-- {
+		if s[i] == '/' || s[i] == '\\' {
+			return s[i+1:]
+		}
+	}
+	return s
+}

internal/service/cloud115/upload_test.go

@@ -0,0 +1,197 @@
+package cloud115
+
+import (
+	"context"
+	"encoding/base64"
+	"io"
+	"os"
+	"path/filepath"
+	"testing"
+
+	"github.com/aliyun/alibabacloud-oss-go-sdk-v2/oss"
+)
+
+func TestFileSHA1(t *testing.T) {
+	dir := t.TempDir()
+	path := filepath.Join(dir, "a.txt")
+	if err := os.WriteFile(path, []byte("hello"), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	sum, err := FileSHA1(path)
+	if err != nil {
+		t.Fatal(err)
+	}
+	// sha1("hello") = aaf4c61ddcc5e8a2dabede0f3b482cd9aea9434d
+	if sum != "aaf4c61ddcc5e8a2dabede0f3b482cd9aea9434d" {
+		t.Errorf("unexpected sha1: %s", sum)
+	}
+}
+
+func TestFileSHA1Partial(t *testing.T) {
+	dir := t.TempDir()
+	path := filepath.Join(dir, "b.txt")
+	// 10 bytes: "0123456789"
+	if err := os.WriteFile(path, []byte("0123456789"), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	// bytes [2,4] = "234"
+	sum, err := FileSHA1Partial(path, 2, 4)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if sum != "0ec09ef9836da03f1add21e3ef607627e687e790" {
+		t.Errorf("unexpected partial sha1: %s", sum)
+	}
+}
+
+// TestFileSHA1PartialSmallerThanWindow 回归测试：经典 bug 是 io.CopyN 在文件不足
+// length 字节时返回 io.EOF。115 上传固定用 [0,128*1024-1] 窗口计算 preid，导致所有
+// 小于 128 KiB 的元数据文件（如海报/缩略图）上传必然失败。
+func TestFileSHA1PartialSmallerThanWindow(t *testing.T) {
+	dir := t.TempDir()
+	path := filepath.Join(dir, "small.bin")
+	// 6 字节小文件，不足 128 KiB 窗口
+	if err := os.WriteFile(path, []byte("abcdef"), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	sum, err := FileSHA1Partial(path, 0, 128*1024-1)
+	if err != nil {
+		t.Fatalf("compute partial sha1 for small file should not fail: %v", err)
+	}
+	// 应等于整个文件（6 字节）的 sha1
+	if sum != "1f8ac10f23c5b5bc1167bda84b833e5c057a77d2" {
+		t.Errorf("unexpected partial sha1: %s", sum)
+	}
+}
+
+func TestParseSignCheckRange(t *testing.T) {
+	rng, err := parseSignCheckRange("0-131071")
+	if err != nil {
+		t.Fatal(err)
+	}
+	if rng.Start != 0 || rng.End != 131071 {
+		t.Errorf("unexpected range: %+v", rng)
+	}
+	if _, err := parseSignCheckRange("bad"); err == nil {
+		t.Error("expected error for bad range")
+	}
+	if _, err := parseSignCheckRange("100-50"); err == nil {
+		t.Error("expected error for end<start")
+	}
+}
+
+func TestCalculateMultipartPartSize(t *testing.T) {
+	// small file: 1 MiB -> partSize 32MiB, 1 part
+	ps, parts, err := CalculateMultipartPartSize(1 << 20)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if ps != defaultMultipartPartSize {
+		t.Errorf("partSize=%d, want %d", ps, defaultMultipartPartSize)
+	}
+	if parts != 1 {
+		t.Errorf("parts=%d, want 1", parts)
+	}
+	// zero-size -> 1 part
+	_, parts, err = CalculateMultipartPartSize(0)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if parts != 1 {
+		t.Errorf("zero-size parts=%d, want 1", parts)
+	}
+	// negative -> error
+	if _, _, err := CalculateMultipartPartSize(-1); err == nil {
+		t.Error("expected error for negative size")
+	}
+}
+
+func TestBaseNameOf(t *testing.T) {
+	if got := baseNameOf("/a/b/file.nfo"); got != "file.nfo" {
+		t.Errorf("got %s", got)
+	}
+	if got := baseNameOf("a\\b\\c.jpg"); got != "c.jpg" {
+		t.Errorf("got %s", got)
+	}
+	if got := baseNameOf("top.txt"); got != "top.txt" {
+		t.Errorf("got %s", got)
+	}
+}
+
+// fakeCallbackOSSClient 捕获 CompleteMultipartUpload 收到的 callback / callback_var，
+// 用于断言已经 Base64 编码（116 要求 callback 必须是 Base64 后的 JSON，否则报
+// "The callback configuration is not base64 encoded"）。
+type fakeCallbackOSSClient struct {
+	capturedCallback    string
+	capturedCallbackVar string
+}
+
+func (c *fakeCallbackOSSClient) InitiateMultipartUpload(_ context.Context, _ *oss.InitiateMultipartUploadRequest, _ ...func(*oss.Options)) (*oss.InitiateMultipartUploadResult, error) {
+	return &oss.InitiateMultipartUploadResult{UploadId: oss.Ptr("upload-new")}, nil
+}
+func (c *fakeCallbackOSSClient) UploadPart(_ context.Context, r *oss.UploadPartRequest, _ ...func(*oss.Options)) (*oss.UploadPartResult, error) {
+	if r.Body != nil {
+		_, _ = io.Copy(io.Discard, r.Body)
+	}
+	return &oss.UploadPartResult{ETag: oss.Ptr("etag-1")}, nil
+}
+func (c *fakeCallbackOSSClient) ListParts(context.Context, *oss.ListPartsRequest, ...func(*oss.Options)) (*oss.ListPartsResult, error) {
+	return &oss.ListPartsResult{}, nil
+}
+func (c *fakeCallbackOSSClient) CompleteMultipartUpload(_ context.Context, r *oss.CompleteMultipartUploadRequest, _ ...func(*oss.Options)) (*oss.CompleteMultipartUploadResult, error) {
+	c.capturedCallback = *r.Callback
+	c.capturedCallbackVar = *r.CallbackVar
+	return &oss.CompleteMultipartUploadResult{
+		CallbackResult: map[string]any{
+			"state": true,
+			"data":  map[string]any{"file_id": "file-1", "pick_code": "pick-1"},
+		},
+	}, nil
+}
+func (c *fakeCallbackOSSClient) AbortMultipartUpload(context.Context, *oss.AbortMultipartUploadRequest, ...func(*oss.Options)) (*oss.AbortMultipartUploadResult, error) {
+	return &oss.AbortMultipartUploadResult{}, nil
+}
+
+// TestCompleteMultipartUploadCallbackBase64 回归测试：OSS CompleteMultipartUpload 的
+// callback 必须 Base64 编码，否则报 "The callback configuration is not base64 encoded"，
+// 导致大于 128 KiB 的元数据文件上传失败。
+func TestCompleteMultipartUploadCallbackBase64(t *testing.T) {
+	dir := t.TempDir()
+	path := filepath.Join(dir, "big.bin")
+	data := make([]byte, 8) // 8 字节，PartSize=8 → 1 part
+	if err := os.WriteFile(path, data, 0o644); err != nil {
+		t.Fatal(err)
+	}
+	fake := &fakeCallbackOSSClient{}
+	uploader := &OSSMultipartUploader{client: fake}
+
+	callback := `{"callbackUrl":"http://uplb.115.com/3.0/completeupload.php"}`
+	callbackVar := `{"x:pick_code":"abc"}`
+	_, err := uploader.UploadFileWithResult(context.Background(), OSSMultipartUploadInput{
+		Bucket:      "bucket-1",
+		Object:      "object-1",
+		Callback:    callback,
+		CallbackVar: callbackVar,
+		FilePath:    path,
+		FileSize:    int64(len(data)),
+		PartSize:    8,
+	})
+	if err != nil {
+		t.Fatalf("multipart 上传失败：%v", err)
+	}
+	// 捕获的 callback 必须是合法 Base64，且解码后与原 JSON 一致
+	cbBytes, err := base64.StdEncoding.DecodeString(fake.capturedCallback)
+	if err != nil {
+		t.Fatalf("callback 未 Base64 编码：%v (raw=%q)", err, fake.capturedCallback)
+	}
+	if string(cbBytes) != callback {
+		t.Errorf("callback 解码后 = %s，期望 %s", cbBytes, callback)
+	}
+	cbvBytes, err := base64.StdEncoding.DecodeString(fake.capturedCallbackVar)
+	if err != nil {
+		t.Fatalf("callback_var 未 Base64 编码：%v (raw=%q)", err, fake.capturedCallbackVar)
+	}
+	if string(cbvBytes) != callbackVar {
+		t.Errorf("callback_var 解码后 = %s，期望 %s", cbvBytes, callbackVar)
+	}
+}

internal/service/danmaku_match_test.go

@@ -92,6 +92,10 @@ func TestDanmakuFetchHashMatchLayer(t *testing.T) {
 	require.Equal(t, "xml", res.SourceType)
 	require.Contains(t, res.Raw, "弹幕Hash命中")
 	require.Empty(t, res.Candidates)
+	require.Equal(t, "测试动画", res.AnimeTitle)
+	require.Equal(t, "第1话", res.EpisodeTitle)
+	require.Equal(t, int64(25484), res.EpisodeID)
+	require.Equal(t, "hash", res.MatchMode)
 
 	// match 请求体：文件名去扩展名并 URL 转义（官方接口要求，实测验证）、
 	// hash、大小、matchMode 齐全。
@@ -274,18 +278,69 @@ func TestDanmakuSameBase(t *testing.T) {
 	require.False(t, sameDanmakuBase("", "https://api.dandanplay.net"))
 }
 
-// fetchCommentWithFallback：配置源与官方同源时不重复请求；
-// 全失败时带出最后一跳错误。
-func TestDanmakuFetchCommentWithFallback(t *testing.T) {
-	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
-		w.WriteHeader(http.StatusInternalServerError)
-	}))
-	t.Cleanup(srv.Close)
+	// fetchCommentWithFallback：配置源与官方同源时不重复请求；
+	// 全失败时带出最后一跳错误。
+	func TestDanmakuFetchCommentWithFallback(t *testing.T) {
+		srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+			w.WriteHeader(http.StatusInternalServerError)
+		}))
+		t.Cleanup(srv.Close)
 
-	svc := newDanmakuTestService(t)
-	ctx := context.Background()
-	raw, st, err := svc.fetchCommentWithFallback(ctx, srv.URL, srv.URL, "25484")
-	require.Error(t, err)
-	require.Empty(t, raw)
-	require.Equal(t, "auto", st)
-}
+		svc := newDanmakuTestService(t)
+		ctx := context.Background()
+		raw, st, err := svc.fetchCommentWithFallback(ctx, srv.URL, srv.URL, "25484")
+		require.Error(t, err)
+		require.Empty(t, raw)
+		require.Equal(t, "auto", st)
+	}
+
+	// 视频即便能命中 Hash 自动识别，当用户传入手动搜索关键词时应跳过 Hash 匹配，走关键词搜索。
+	func TestDanmakuFetchHashMatchSkippedOnManualKeyword(t *testing.T) {
+		videoPath, _ := writeDanmakuTestVideo(t, "测试动画.第01话.mkv")
+
+		// 官方服务同时提供 match 和 search：
+		// match 会返回 episodeId=25484（动画A）
+		// search 会根据关键词返回 episodeId=99999（动画B）
+		mux := http.NewServeMux()
+		var matchCalled bool
+		mux.HandleFunc("/api/v2/match", func(w http.ResponseWriter, r *http.Request) {
+			matchCalled = true
+			w.Header().Set("Content-Type", "application/json")
+			fmt.Fprint(w, `{"success":true,"isMatched":true,"matches":[{"episodeId":25484,"animeId":1001,"animeTitle":"自动识别动画A","episodeTitle":"第1话"}]}`)
+		})
+		mux.HandleFunc("/api/v2/search/episodes", func(w http.ResponseWriter, r *http.Request) {
+			w.Header().Set("Content-Type", "application/json")
+			fmt.Fprint(w, `{"hasMore":false,"animes":[{"animeId":2002,"animeTitle":"手动搜索动画B","episodes":[{"episodeId":99999,"episodeTitle":"第1话"}]}]}`)
+		})
+		mux.HandleFunc("/api/v2/comment/25484", func(w http.ResponseWriter, r *http.Request) {
+			w.Header().Set("Content-Type", "application/xml")
+			fmt.Fprint(w, `<?xml version="1.0"?><i><d p="0.5,1,16777215,user1">自动识别弹幕</d></i>`)
+		})
+		mux.HandleFunc("/api/v2/comment/99999", func(w http.ResponseWriter, r *http.Request) {
+			w.Header().Set("Content-Type", "application/xml")
+			fmt.Fprint(w, `<?xml version="1.0"?><i><d p="0.5,1,16777215,user2">手动搜索弹幕</d></i>`)
+		})
+		official := httptest.NewServer(mux)
+		t.Cleanup(official.Close)
+		overrideDanmakuOfficialBase(t, official.URL)
+
+		svc := newDanmakuTestService(t)
+		ctx := context.Background()
+		seedDanmakuVideoMedia(t, svc, "mManual", "自动识别动画A", videoPath, 32000, 1)
+
+		// 1) 默认自动识别：命中 Hash 识别
+		resAuto, err := svc.Fetch(ctx, "mManual", "", "")
+		require.NoError(t, err)
+		require.True(t, matchCalled)
+		require.Equal(t, "hash", resAuto.MatchMode)
+		require.Equal(t, int64(25484), resAuto.EpisodeID)
+		require.Contains(t, resAuto.Raw, "自动识别弹幕")
+
+		// 2) 用户传入手动搜索关键词：跳过 Hash 识别，命中搜索结果动画B
+		resManual, err := svc.Fetch(ctx, "mManual", "手动搜索动画B", "")
+		require.NoError(t, err)
+		require.Equal(t, "search", resManual.MatchMode)
+		require.Equal(t, int64(99999), resManual.EpisodeID)
+		require.Equal(t, "手动搜索动画B", resManual.AnimeTitle)
+		require.Contains(t, resManual.Raw, "手动搜索弹幕")
+	}

internal/service/danmaku_service.go

@@ -67,11 +67,17 @@ type DanmakuRenderConfig struct {
 //
 // Candidates is non-nil when multiple anime matched the search and the player
 // must ask the user which one to use (disambiguation); Raw is empty then.
+// AnimeTitle, EpisodeTitle, EpisodeID and MatchMode provide matched danmaku
+// metadata so the player UI can display which episode was loaded.
 type DanmakuFetchResult struct {
 	DanmakuRenderConfig
-	SourceType string         `json:"source_type"`
-	Raw        string         `json:"raw,omitempty"`
-	Candidates []DanmakuAnime `json:"candidates,omitempty"`
+	SourceType   string         `json:"source_type"`
+	Raw          string         `json:"raw,omitempty"`
+	Candidates   []DanmakuAnime `json:"candidates,omitempty"`
+	AnimeTitle   string         `json:"anime_title,omitempty"`
+	EpisodeTitle string         `json:"episode_title,omitempty"`
+	EpisodeID    int64          `json:"episode_id,omitempty"`
+	MatchMode    string         `json:"match_mode,omitempty"`
 }
 
 // DanmakuAnime is one search hit (an anime) with its episode list, mirroring
@@ -184,74 +190,90 @@ func (s *DanmakuService) Fetch(ctx context.Context, mediaID, keyword, episodeID
 	configured := strings.TrimRight(strings.TrimSpace(res.Source), "/")
 	official := danmakuOfficialBase
 
-	// 手动指定弹幕库：跳过识别，直接拉取该库（自定义源失败回退官方）。
-	if target := strings.TrimSpace(episodeID); target != "" {
-		raw, st, err := s.fetchCommentWithFallback(ctx, configured, official, target)
-		if err != nil {
-			s.log.Warn("danmaku comment fetch failed", zap.String("media_id", mediaID), zap.String("episode_id", target), zap.Error(err))
-			return res, err
-		}
-		res.Raw, res.SourceType = raw, st
-		return res, nil
-	}
-
-	term, media, err := s.searchTerms(ctx, mediaID)
-	if err != nil {
-		return res, err
-	}
-	manualKeyword := strings.TrimSpace(keyword) != ""
-	if kw := strings.TrimSpace(keyword); kw != "" {
-		term.name = kw
-	}
-	if strings.TrimSpace(term.name) == "" {
-		return res, nil
-	}
-
-	target := ""
-
-	// 1) hash 识别：始终走官方 /api/v2/match。
-	if media != nil && media.Path != "" {
-		if hash, ok := s.mediaHash(ctx, media); ok {
-			fileSize := media.SizeBytes
-			if strings.EqualFold(filepath.Ext(media.Path), ".strm") {
-				fileSize = 0 // strm 行的 SizeBytes 是文本大小，不是视频大小
-			}
-			matches, err := s.matchOfficial(ctx, danmakuMatchFileName(media.Path), hash, fileSize, media.DurationSec)
+		// 手动指定弹幕库：跳过识别，直接拉取该库（自定义源失败回退官方）。
+		if target := strings.TrimSpace(episodeID); target != "" {
+			raw, st, err := s.fetchCommentWithFallback(ctx, configured, official, target)
 			if err != nil {
-				s.log.Warn("danmaku hash match failed", zap.String("media_id", mediaID), zap.Error(err))
-			} else if len(matches) > 0 {
-				target = fmt.Sprintf("%d", matches[0].EpisodeID)
+				s.log.Warn("danmaku comment fetch failed", zap.String("media_id", mediaID), zap.String("episode_id", target), zap.Error(err))
+				return res, err
 			}
-		}
-	}
-
-	// 2) 按播放的文件名 + 集数搜索（keyword 手动覆盖时跳过，直接走第 3 层）。
-	if target == "" && !manualKeyword && media != nil && media.Path != "" {
-		if fileName := danmakuMatchFileName(media.Path); fileName != "" && fileName != term.name {
-			if candidates, err := s.searchCandidatesWithFallback(ctx, configured, official, fileName, term.episode); err == nil &&
-				len(candidates) == 1 && len(candidates[0].Episodes) > 0 {
-				target = fmt.Sprintf("%d", candidates[0].Episodes[0].EpisodeID)
+			res.Raw, res.SourceType = raw, st
+			if id, parseErr := strconv.ParseInt(target, 10, 64); parseErr == nil {
+				res.EpisodeID = id
 			}
-		}
-	}
-
-	// 3) 现有自动识别：标题层级（original_name → title → 文件名）+ 集数，
-	//    多结果返回候选列表交给播放器（歧义处理）。
-	if target == "" {
-		candidates, err := s.searchCandidatesWithFallback(ctx, configured, official, term.name, term.episode)
-		if err != nil {
-			s.log.Warn("danmaku search failed", zap.String("media_id", mediaID), zap.String("name", term.name), zap.String("episode", term.episode), zap.Error(err))
-			return res, err
-		}
-		if len(candidates) != 1 {
-			res.Candidates = candidates
+			res.MatchMode = "manual"
 			return res, nil
 		}
-		if len(candidates[0].Episodes) == 0 {
-			return res, errors.New("no danmaku library found for this video")
+
+		term, media, err := s.searchTerms(ctx, mediaID)
+		if err != nil {
+			return res, err
+		}
+		manualKeyword := strings.TrimSpace(keyword) != ""
+		if kw := strings.TrimSpace(keyword); kw != "" {
+			term.name = kw
+		}
+		if strings.TrimSpace(term.name) == "" {
+			return res, nil
+		}
+
+		target := ""
+
+		// 1) hash 识别：始终走官方 /api/v2/match（keyword 手动覆盖时跳过，直接走第 3 层）。
+		if target == "" && !manualKeyword && media != nil && media.Path != "" {
+			if hash, ok := s.mediaHash(ctx, media); ok {
+				fileSize := media.SizeBytes
+				if strings.EqualFold(filepath.Ext(media.Path), ".strm") {
+					fileSize = 0 // strm 行的 SizeBytes 是文本大小，不是视频大小
+				}
+				matches, err := s.matchOfficial(ctx, danmakuMatchFileName(media.Path), hash, fileSize, media.DurationSec)
+				if err != nil {
+					s.log.Warn("danmaku hash match failed", zap.String("media_id", mediaID), zap.Error(err))
+				} else if len(matches) > 0 {
+					target = fmt.Sprintf("%d", matches[0].EpisodeID)
+					res.AnimeTitle = matches[0].AnimeTitle
+					res.EpisodeTitle = matches[0].EpisodeTitle
+					res.EpisodeID = matches[0].EpisodeID
+					res.MatchMode = "hash"
+				}
+			}
+		}
+
+		// 2) 按播放的文件名 + 集数搜索（keyword 手动覆盖时跳过，直接走第 3 层）。
+		if target == "" && !manualKeyword && media != nil && media.Path != "" {
+			if fileName := danmakuMatchFileName(media.Path); fileName != "" && fileName != term.name {
+				if candidates, err := s.searchCandidatesWithFallback(ctx, configured, official, fileName, term.episode); err == nil &&
+					len(candidates) == 1 && len(candidates[0].Episodes) > 0 {
+					target = fmt.Sprintf("%d", candidates[0].Episodes[0].EpisodeID)
+					res.AnimeTitle = candidates[0].AnimeTitle
+					res.EpisodeTitle = candidates[0].Episodes[0].EpisodeTitle
+					res.EpisodeID = candidates[0].Episodes[0].EpisodeID
+					res.MatchMode = "filename"
+				}
+			}
+		}
+
+		// 3) 现有自动识别：标题层级（original_name → title → 文件名）+ 集数，
+		//    多结果返回候选列表交给播放器（歧义处理）。
+		if target == "" {
+			candidates, err := s.searchCandidatesWithFallback(ctx, configured, official, term.name, term.episode)
+			if err != nil {
+				s.log.Warn("danmaku search failed", zap.String("media_id", mediaID), zap.String("name", term.name), zap.String("episode", term.episode), zap.Error(err))
+				return res, err
+			}
+			if len(candidates) != 1 {
+				res.Candidates = candidates
+				return res, nil
+			}
+			if len(candidates[0].Episodes) == 0 {
+				return res, errors.New("no danmaku library found for this video")
+			}
+			target = fmt.Sprintf("%d", candidates[0].Episodes[0].EpisodeID)
+			res.AnimeTitle = candidates[0].AnimeTitle
+			res.EpisodeTitle = candidates[0].Episodes[0].EpisodeTitle
+			res.EpisodeID = candidates[0].Episodes[0].EpisodeID
+			res.MatchMode = "search"
 		}
-		target = fmt.Sprintf("%d", candidates[0].Episodes[0].EpisodeID)
-	}
 
 	raw, st, err := s.fetchCommentWithFallback(ctx, configured, official, target)
 	if err != nil {

internal/service/danmaku_service_test.go

@@ -142,6 +142,10 @@ func TestDanmakuFetchWithDandanplaySource(t *testing.T) {
 	require.Equal(t, "xml", res.SourceType)
 	require.Contains(t, res.Raw, "弹幕A")
 	require.Contains(t, res.Raw, `p="0.5,1,16777215,user1"`)
+	require.Equal(t, "测试动画", res.AnimeTitle)
+	require.Equal(t, "第1话", res.EpisodeTitle)
+	require.Equal(t, int64(25484), res.EpisodeID)
+	require.Equal(t, "search", res.MatchMode)
 }
 
 func TestDanmakuFetchUsesOriginalNameForSearch(t *testing.T) {
@@ -262,6 +266,8 @@ func TestDanmakuFetchWithExplicitEpisodeID(t *testing.T) {
 	require.True(t, res.Enabled)
 	require.Contains(t, res.Raw, "显式指定弹幕")
 	require.Empty(t, res.Candidates)
+	require.Equal(t, int64(99999), res.EpisodeID)
+	require.Equal(t, "manual", res.MatchMode)
 }
 func TestDetectDanmakuSourceType(t *testing.T) {
 	cases := []struct {

internal/service/database_admin.go

@@ -0,0 +1,90 @@
+package service
+
+import (
+	"context"
+	"fmt"
+	"strings"
+
+	"go.uber.org/zap"
+	"gorm.io/gorm"
+
+	"github.com/ShukeBta/MMTL/internal/config"
+	"github.com/ShukeBta/MMTL/internal/database"
+	"github.com/ShukeBta/MMTL/internal/repository"
+)
+
+// DatabaseAdminService manages database configuration, connectivity testing, and migration.
+type DatabaseAdminService struct {
+	cfg   *config.Config
+	log   *zap.Logger
+	repos *repository.Container
+	db    *gorm.DB
+}
+
+// NewDatabaseAdminService creates a new DatabaseAdminService.
+func NewDatabaseAdminService(cfg *config.Config, log *zap.Logger, repos *repository.Container, db *gorm.DB) *DatabaseAdminService {
+	if log == nil {
+		log = zap.NewNop()
+	}
+	return &DatabaseAdminService{
+		cfg:   cfg,
+		log:   log,
+		repos: repos,
+		db:    db,
+	}
+}
+
+// GetStatus returns the status of the currently active database.
+func (s *DatabaseAdminService) GetStatus(ctx context.Context) *database.DatabaseStatus {
+	return database.InspectDatabaseStatus(s.db, s.cfg)
+}
+
+// TestPostgres verifies connectivity and permissions to the specified PostgreSQL DSN.
+func (s *DatabaseAdminService) TestPostgres(ctx context.Context, dsn string) (*database.PostgresTestResult, error) {
+	return database.TestPostgres(dsn)
+}
+
+// MigrateToPostgres copies all records from the current active database to the target PostgreSQL database.
+func (s *DatabaseAdminService) MigrateToPostgres(ctx context.Context, targetDSN string) (*database.DatabaseMigrationResult, error) {
+	s.log.Info("starting user-initiated database migration to PostgreSQL", zap.String("target", database.MaskDSN(targetDSN)))
+	res, err := database.MigrateCurrentToPostgres(s.db, targetDSN, 500, s.log)
+	if err != nil {
+		s.log.Error("database migration to PostgreSQL failed", zap.Error(err))
+		return nil, err
+	}
+	s.log.Info("database migration to PostgreSQL completed successfully",
+		zap.Int64("total_rows", res.TotalRows),
+		zap.Int64("duration_ms", res.DurationMS),
+	)
+	return res, nil
+}
+
+// SaveConfig persists the database configuration to config.yaml and the database settings table.
+func (s *DatabaseAdminService) SaveConfig(ctx context.Context, dbType, dsn string) error {
+	dbType = strings.TrimSpace(dbType)
+	dsn = strings.TrimSpace(dsn)
+	if dbType == "" {
+		dbType = "postgres"
+	}
+	if dbType == "postgres" && dsn == "" {
+		return fmt.Errorf("PostgreSQL DSN 不能为空")
+	}
+
+	// 1. 保存到本地 config.yaml
+	if err := config.SaveDatabaseConfig(dbType, dsn); err != nil {
+		return fmt.Errorf("保存配置文件失败: %w", err)
+	}
+
+	// 2. 更新内存配置
+	s.cfg.Database.Type = dbType
+	s.cfg.Database.DSN = dsn
+
+	// 3. 同时更新 settings 存储库作为副本
+	if s.repos != nil && s.repos.Setting != nil {
+		_ = s.repos.Setting.Set(ctx, "database.type", dbType)
+		_ = s.repos.Setting.Set(ctx, "database.dsn", dsn)
+	}
+
+	s.log.Info("database configuration saved", zap.String("type", dbType), zap.String("dsn", database.MaskDSN(dsn)))
+	return nil
+}

internal/service/emby_compat.go

@@ -111,6 +111,7 @@ const (
 
 var (
 	embySeasonDirRE    = regexp.MustCompile(`(?i)^(season[\s._-]*\d+|s\d+|specials?|sp|ova|oad|extra|extras|第\s*[0-9一二三四五六七八九十百零两]+\s*季|特别篇|特別篇|番外|特典)$`)
+	embySeasonSuffixRE = regexp.MustCompile(`(?i)(?:[\s._-]+(?:season[\s._-]*\d+|s\d+|第\s*[0-9一二三四五六七八九十百零两]+\s*季|specials?|sp|ova|oad|extra|extras|特别篇|特別篇|番外|特典)|\s*第\s*[0-9一二三四五六七八九十百零两]+\s*季)\s*$`)
 	embyYearSuffixRE   = regexp.MustCompile(`\s*[\(（\[]\d{4}[\)）\]]\s*$`)
 	embyEpisodeTitleRE = regexp.MustCompile(`(?i)\s*[-_ ]*s\d{1,2}e\d{1,3}.*$`)
 )

internal/service/emby_media_sources.go

@@ -199,7 +199,9 @@ func (e *EmbyService) appendSubtitleStreams(ctx context.Context, streams []map[s
 	if e == nil || e.subtitle == nil || m == nil {
 		return streams
 	}
-	tracks, err := e.subtitle.Discover(ctx, m.ID)
+	// Emby 字幕只列表外挂字幕文件：云盘/strm 媒体的容器内嵌字幕不做服务端
+	// 提取，客户端直连播放直链时自行解析。
+	tracks, err := e.subtitle.DiscoverExternalOnly(ctx, m.ID)
 	if err != nil || len(tracks) == 0 {
 		return streams
 	}

internal/service/emby_playback.go

@@ -32,6 +32,9 @@ func (e *EmbyService) PlaybackInfo(ctx context.Context, mediaID, userID string)
 // so the DeliveryUrl advertised in MediaStreams lines up exactly with the
 // served track: subtitles start at 1 when no audio stream is present, otherwise
 // at 2 (after Video 0 + Audio 1).
+//
+// 只服务外挂字幕文件（DiscoverExternalOnly）：云盘/strm 媒体的容器内嵌字幕
+// 不做服务端提取，客户端直连播放直链时自行解析。
 func (e *EmbyService) ServeSubtitleStream(ctx context.Context, w io.Writer, mediaID, indexStr string, userID string) error {
 	if e == nil || e.subtitle == nil {
 		return ErrSubtitleUnavailable
@@ -44,7 +47,7 @@ func (e *EmbyService) ServeSubtitleStream(ctx context.Context, w io.Writer, medi
 	if err != nil || index < 1 {
 		return ErrSubtitleNotFound
 	}
-	tracks, err := e.subtitle.Discover(ctx, m.ID)
+	tracks, err := e.subtitle.DiscoverExternalOnly(ctx, m.ID)
 	if err != nil || len(tracks) == 0 {
 		return ErrSubtitleNotFound
 	}
@@ -74,7 +77,7 @@ func (e *EmbyService) SubtitleStreamCodec(ctx context.Context, mediaID, indexStr
 	if err != nil || index < 1 {
 		return ""
 	}
-	tracks, err := e.subtitle.Discover(ctx, m.ID)
+	tracks, err := e.subtitle.DiscoverExternalOnly(ctx, m.ID)
 	if err != nil || len(tracks) == 0 {
 		return ""
 	}

internal/service/emby_series_hierarchy_test.go

@@ -334,3 +334,105 @@ func TestEmbyCloudAnimeUsesSeriesNameFromChineseSeasonFolder(t *testing.T) {
 		t.Fatalf("cloud anime should be grouped as one series named 剑来, got %#v", items)
 	}
 }
+
+func TestEmbySeriesGroupingWithPrefixedSeasonFolders(t *testing.T) {
+	svc := newTestEmbyService(t)
+	lib := model.Library{Name: "动漫", Path: `/media/动漫`, Type: "anime", Enabled: true}
+	if err := svc.repo.Library.Create(t.Context(), &lib); err != nil {
+		t.Fatalf("create library: %v", err)
+	}
+
+	for season := 1; season <= 5; season++ {
+		for ep := 1; ep <= 3; ep++ {
+			media := model.Media{
+				Base:         model.Base{ID: fmt.Sprintf("shokugeki-s%02de%02d", season, ep)},
+				LibraryID:    lib.ID,
+				Title:        "食戟之灵",
+				OriginalName: "食戟のソーマ",
+				ScrapeStatus: "matched",
+				TMDbID:       62273,
+				BangumiID:    116461,
+				Path:         fmt.Sprintf(`/media/动漫/食戟之灵/食戟之灵 S%02d/食戟之灵 S%02dE%02d.strm`, season, season, ep),
+				SeasonNum:    season,
+				EpisodeNum:   ep,
+			}
+			if err := svc.repo.DB.Create(&media).Error; err != nil {
+				t.Fatalf("create media: %v", err)
+			}
+		}
+	}
+
+	root, err := svc.Items(t.Context(), ItemsParams{ParentID: lib.ID, Limit: 50})
+	if err != nil {
+		t.Fatalf("library items: %v", err)
+	}
+	rootItems := root["Items"].([]map[string]any)
+	if len(rootItems) != 1 {
+		t.Fatalf("expected 1 series card for 食戟之灵 across 5 seasons, got %d cards: %#v", len(rootItems), rootItems)
+	}
+	if rootItems[0]["Name"] != "食戟之灵" || rootItems[0]["Type"] != "Series" {
+		t.Fatalf("unexpected series item: %#v", rootItems[0])
+	}
+	seriesID := rootItems[0]["Id"].(string)
+
+	seasons, err := svc.Items(t.Context(), ItemsParams{ParentID: seriesID, Limit: 50})
+	if err != nil {
+		t.Fatalf("series seasons: %v", err)
+	}
+	seasonItems := seasons["Items"].([]map[string]any)
+	if len(seasonItems) != 5 {
+		t.Fatalf("expected 5 seasons, got %d: %#v", len(seasonItems), seasonItems)
+	}
+	for i, s := range seasonItems {
+		wantSeasonNum := i + 1
+		if s["Type"] != "Season" || s["IndexNumber"] != wantSeasonNum {
+			t.Errorf("season [%d] = %#v, want IndexNumber=%d", i, s, wantSeasonNum)
+		}
+	}
+
+	counts, err := svc.ItemCounts(t.Context(), "user-1")
+	if err != nil {
+		t.Fatalf("item counts: %v", err)
+	}
+	if counts["SeriesCount"] != 1 || counts["EpisodeCount"] != int64(15) {
+		t.Fatalf("counts = %#v, want 1 series and 15 episodes", counts)
+	}
+}
+
+func TestInferSeriesNameFromPath(t *testing.T) {
+	tests := []struct {
+		path string
+		want string
+	}{
+		{
+			path: `/media/动漫/食戟之灵/食戟之灵 S01/食戟之灵 S01E01.strm`,
+			want: "食戟之灵",
+		},
+		{
+			path: `/media/动漫/食戟之灵/食戟之灵 S05/食戟之灵 S05E12.strm`,
+			want: "食戟之灵",
+		},
+		{
+			path: `/media/动漫/食戟之灵/Season 02/01.mkv`,
+			want: "食戟之灵",
+		},
+		{
+			path: `/media/动漫/进击的巨人 第2季/01.mkv`,
+			want: "进击的巨人",
+		},
+		{
+			path: `cloud://openlist/国漫/剑来/第二季/04.mkv`,
+			want: "剑来",
+		},
+		{
+			path: `/media/tv/间谍过家家 (2022)/Specials/S00E01.mkv`,
+			want: "间谍过家家",
+		},
+	}
+	for _, tc := range tests {
+		got := inferSeriesNameFromPath(tc.path)
+		if got != tc.want {
+			t.Errorf("inferSeriesNameFromPath(%q) = %q, want %q", tc.path, got, tc.want)
+		}
+	}
+}

internal/service/emby_series_ids.go

@@ -31,6 +31,14 @@ func (e *EmbyService) seriesNameForMedia(m *model.Media) string {
 			return series.Title
 		}
 	}
+	if strings.EqualFold(strings.TrimSpace(m.ScrapeStatus), "matched") && strings.TrimSpace(m.Title) != "" {
+		name := strings.TrimSpace(m.Title)
+		name = embyEpisodeTitleRE.ReplaceAllString(name, "")
+		name = embyYearSuffixRE.ReplaceAllString(name, "")
+		if name != "" {
+			return name
+		}
+	}
 	if name := inferSeriesNameFromPath(m.Path); name != "" {
 		return name
 	}
@@ -53,14 +61,35 @@ func inferSeriesNameFromPath(path string) string {
 	if embySeasonDirRE.MatchString(base) {
 		dir = filepath.Dir(dir)
 		base = filepath.Base(dir)
+	} else if stripped := strings.TrimSpace(embySeasonSuffixRE.ReplaceAllString(base, "")); stripped != "" && stripped != base {
+		parentDir := filepath.Dir(dir)
+		parentBase := filepath.Base(parentDir)
+		if parentBase != "." && parentBase != string(filepath.Separator) && !isEmbyGenericContainer(parentBase) {
+			dir = parentDir
+			base = parentBase
+		} else {
+			base = stripped
+		}
 	}
 	base = strings.TrimSpace(embyYearSuffixRE.ReplaceAllString(base, ""))
-	if base == "." || base == string(filepath.Separator) {
+	if base == "." || base == string(filepath.Separator) || isEmbyGenericContainer(base) {
 		return ""
 	}
 	return base
 }
 
+func isEmbyGenericContainer(name string) bool {
+	switch strings.ToLower(strings.TrimSpace(name)) {
+	case "movie", "movies", "film", "films", "tv", "series", "show", "shows", "anime", "animation", "variety",
+		"电视剧", "剧集", "连续剧", "短剧", "国产剧", "国剧", "欧美剧", "美剧", "英剧", "日韩剧", "日剧", "韩剧", "港剧", "台剧", "港台剧",
+		"综艺", "纪录片", "儿童", "动漫", "番剧", "国漫", "日番", "韩漫", "美漫", "欧美动漫", "欧美动画", "其他动漫", "电影", "成人", "未分类",
+		"media", "downloads", "download", "videos", "video", "share", "shares":
+		return true
+	default:
+		return false
+	}
+}
+
 func stableEmbyID(prefix string, parts ...string) string {
 	h := sha256.New()
 	for _, part := range parts {

internal/service/ffmpeg_auto_install.go

@@ -1,196 +1,358 @@
 package service
 
 import (
+	"archive/tar"
 	"context"
 	"fmt"
 	"io"
 	"net/http"
 	"os"
+	"os/exec"
 	"path/filepath"
 	"runtime"
+	"strings"
 	"time"
 
+	"github.com/ulikunitz/xz"
 	"go.uber.org/zap"
 
 	"github.com/ShukeBta/MMTL/internal/config"
 )
 
-// AutoInstallFFmpeg is only called by the admin tool-install endpoint. The
-// server must not auto-download or keep ffmpeg/ffprobe running during startup.
-func AutoInstallFFmpeg(log *zap.Logger, cfg *config.Config) (ffprobePath, ffmpegPath string) {
-	// 1. 优先使用配置 / PATH / 本机常见软件目录中的现有工具。
-	if path, err := resolveLocalExecutable(cfg.App.FFprobePath, "ffprobe"); err == nil {
-		ffprobePath = path
-		cfg.App.FFprobePath = path
-		log.Info("found local ffprobe", zap.String("path", path))
-	}
-	if path, err := resolveLocalExecutable(cfg.App.FFmpegPath, "ffmpeg"); err == nil {
-		ffmpegPath = path
-		cfg.App.FFmpegPath = path
-		log.Info("found local ffmpeg", zap.String("path", path))
-	}
-	if ffprobePath != "" || ffmpegPath != "" {
-		return ffprobePath, ffmpegPath
-	}
+// ffmpegDownloadTarget 描述某个平台对应的官方构建下载源。
+type ffmpegDownloadTarget struct {
+	Label    string   // 展示名，如 "Windows x86_64"
+	Kind     string   // 压缩包类型：zip / tar.xz
+	Archives []string // 依次尝试的下载地址（主源 + 备用源）
+}
 
-	// 2. 检查默认安装位置。
-	defaultDir := getDefaultInstallDir()
-	ffprobeDefault := filepath.Join(defaultDir, "bin", "ffprobe.exe")
-	ffmpegDefault := filepath.Join(defaultDir, "bin", "ffmpeg.exe")
-
-	if _, err := os.Stat(ffprobeDefault); err == nil {
-		log.Info("在默认位置找到 ffprobe", zap.String("path", ffprobeDefault))
-		return ffprobeDefault, ffmpegDefault
-	}
-
-	// 3. 尝试自动安装。
-	log.Warn("未找到 ffmpeg/ffprobe，尝试自动安装...")
-	installed, err := tryAutoInstall(log, defaultDir)
-	if err != nil {
-		log.Error("自动安装失败，请手动安装 ffmpeg", zap.Error(err))
-		return "", ""
-	}
-
-	if installed {
-		if _, err := os.Stat(ffprobeDefault); err == nil {
-			log.Info("自动安装成功", zap.String("path", ffprobeDefault))
-			// 更新配置
-			updateConfigPaths(cfg, ffprobeDefault, ffmpegDefault)
-			return ffprobeDefault, ffmpegDefault
+// ffmpegTargetForPlatform 按当前运行环境（OS+架构）选择下载源。
+func ffmpegTargetForPlatform() (*ffmpegDownloadTarget, error) {
+	switch runtime.GOOS {
+	case "windows":
+		switch runtime.GOARCH {
+		case "amd64":
+			return &ffmpegDownloadTarget{
+				Label: "Windows x86_64",
+				Kind:  "zip",
+				Archives: []string{
+					"https://www.gyan.dev/ffmpeg/builds/ffmpeg-release-essentials.zip",
+					"https://github.com/BtbN/FFmpeg-Builds/releases/download/latest/ffmpeg-master-latest-win64-gpl.zip",
+				},
+			}, nil
+		case "386":
+			return &ffmpegDownloadTarget{
+				Label: "Windows x86",
+				Kind:  "zip",
+				Archives: []string{
+					"https://github.com/BtbN/FFmpeg-Builds/releases/download/latest/ffmpeg-master-latest-win32-gpl.zip",
+				},
+			}, nil
+		}
+	case "linux":
+		switch runtime.GOARCH {
+		case "amd64":
+			return &ffmpegDownloadTarget{
+				Label: "Linux x86_64",
+				Kind:  "tar.xz",
+				Archives: []string{
+					"https://github.com/BtbN/FFmpeg-Builds/releases/download/latest/ffmpeg-master-latest-linux64-gpl.tar.xz",
+					"https://johnvansickle.com/ffmpeg/releases/ffmpeg-release-amd64-static.tar.xz",
+				},
+			}, nil
+		case "arm64":
+			return &ffmpegDownloadTarget{
+				Label: "Linux ARM64",
+				Kind:  "tar.xz",
+				Archives: []string{
+					"https://github.com/BtbN/FFmpeg-Builds/releases/download/latest/ffmpeg-master-latest-linuxarm64-gpl.tar.xz",
+					"https://johnvansickle.com/ffmpeg/releases/ffmpeg-release-arm64-static.tar.xz",
+				},
+			}, nil
+		case "arm":
+			return &ffmpegDownloadTarget{
+				Label: "Linux ARM (32 位)",
+				Kind:  "tar.xz",
+				Archives: []string{
+					"https://github.com/BtbN/FFmpeg-Builds/releases/download/latest/ffmpeg-master-latest-linuxarmhf-gpl.tar.xz",
+				},
+			}, nil
+		case "loong64":
+			return &ffmpegDownloadTarget{
+				Label: "Linux LoongArch64",
+				Kind:  "tar.xz",
+				Archives: []string{
+					"https://github.com/BtbN/FFmpeg-Builds/releases/download/latest/ffmpeg-master-latest-linuxloongarch64-gpl.tar.xz",
+				},
+			}, nil
+		}
+	case "darwin":
+		switch runtime.GOARCH {
+		case "amd64":
+			return &ffmpegDownloadTarget{
+				Label: "macOS x86_64",
+				Kind:  "zip",
+				Archives: []string{
+					"https://github.com/BtbN/FFmpeg-Builds/releases/download/latest/ffmpeg-master-latest-osx64-gpl.zip",
+				},
+			}, nil
+		case "arm64":
+			return &ffmpegDownloadTarget{
+				Label: "macOS Apple Silicon",
+				Kind:  "zip",
+				Archives: []string{
+					"https://github.com/BtbN/FFmpeg-Builds/releases/download/latest/ffmpeg-master-latest-osxarm64-gpl.zip",
+				},
+			}, nil
 		}
 	}
-
-	return "", ""
+	return nil, fmt.Errorf("暂不支持自动下载的平台 %s/%s，请手动填写 ffmpeg/ffprobe 路径", runtime.GOOS, runtime.GOARCH)
 }
 
-// getDefaultInstallDir 返回默认安装目录
-func getDefaultInstallDir() string {
-	exePath, err := os.Executable()
+// installFFmpegTools 按平台下载并安装 ffmpeg/ffprobe 到 data/tools/ffmpeg/，
+// 返回两个可执行文件的绝对路径。progress 用于回传阶段消息（UI 展示）。
+func installFFmpegTools(ctx context.Context, log *zap.Logger, cfg *config.Config, progress func(string)) (ffmpegPath, ffprobePath string, err error) {
+	target, err := ffmpegTargetForPlatform()
 	if err != nil {
-		return "./tools/ffmpeg"
+		return "", "", err
 	}
-	exeDir := filepath.Dir(exePath)
-	return filepath.Join(exeDir, "tools", "ffmpeg")
-}
-
-// tryAutoInstall 尝试自动下载并安装 ffmpeg
-func tryAutoInstall(log *zap.Logger, installDir string) (bool, error) {
-	if runtime.GOOS == "windows" {
-		return downloadFFmpegWindows(log, installDir)
-	}
-
-	return false, fmt.Errorf("不支持的操作系统: %s", runtime.GOOS)
-}
-
-// downloadFFmpegWindows 下载 Windows 版本的 ffmpeg
-func downloadFFmpegWindows(log *zap.Logger, installDir string) (bool, error) {
-	log.Info("开始下载 ffmpeg...")
-
-	// 创建安装目录
+	installDir := filepath.Join(cfg.App.DataDir, "tools", "ffmpeg")
 	if err := os.MkdirAll(installDir, 0o750); err != nil {
-		return false, fmt.Errorf("创建安装目录失败: %w", err)
+		return "", "", fmt.Errorf("创建安装目录失败: %w", err)
 	}
 
 	tempDir, err := os.MkdirTemp("", "mmtl-ffmpeg-*")
 	if err != nil {
-		return false, fmt.Errorf("创建临时目录失败: %w", err)
+		return "", "", fmt.Errorf("创建临时目录失败: %w", err)
 	}
 	defer os.RemoveAll(tempDir)
 
-	// 下载 URL (使用 gyani.org 的静态构建)
-	arch := "win64"
-	if !is64Bit() {
-		arch = "win32"
+	progress("下载 " + target.Label + " 版本…")
+	archivePath := filepath.Join(tempDir, "ffmpeg-archive."+target.Kind)
+	if err := downloadFFmpegArchive(ctx, log, target.Archives, archivePath); err != nil {
+		return "", "", err
 	}
 
-	// 先尝试从 gyan.dev 下载（更可靠）
-	downloadURL := fmt.Sprintf("https://www.gyan.dev/ffmpeg/builds/ffmpeg-release-essentials.zip")
+	progress("解压…")
+	extractDir := filepath.Join(tempDir, "extract")
+	if err := extractFFmpegArchive(target.Kind, archivePath, extractDir); err != nil {
+		return "", "", fmt.Errorf("解压失败: %w", err)
+	}
 
-	log.Info("下载 ffmpeg", zap.String("url", downloadURL))
+	exeSuffix := ""
+	if runtime.GOOS == "windows" {
+		exeSuffix = ".exe"
+	}
+	srcFFmpeg, srcFFprobe, err := locateFFmpegBinaries(extractDir, exeSuffix)
+	if err != nil {
+		return "", "", err
+	}
 
-	// 使用 Go 下载
-	zipPath := filepath.Join(tempDir, "ffmpeg.zip")
-	if err := downloadFile(log, downloadURL, zipPath); err != nil {
-		// 尝试备用 URL
-		backupURL := fmt.Sprintf("https://github.com/BtbN/FFmpeg-Builds/releases/download/latest/ffmpeg-master-latest-%s-gpl.zip", arch)
-		log.Info("尝试备用下载地址", zap.String("url", backupURL))
-		if err2 := downloadFile(log, backupURL, zipPath); err2 != nil {
-			return false, fmt.Errorf("下载失败: %v, %v", err, err2)
+	progress("安装到 data 目录…")
+	ffmpegPath = filepath.Join(installDir, "ffmpeg"+exeSuffix)
+	ffprobePath = filepath.Join(installDir, "ffprobe"+exeSuffix)
+	if err := copyFileMode(srcFFmpeg, ffmpegPath); err != nil {
+		return "", "", fmt.Errorf("复制 ffmpeg 失败: %w", err)
+	}
+	if err := copyFileMode(srcFFprobe, ffprobePath); err != nil {
+		_ = os.Remove(ffmpegPath)
+		return "", "", fmt.Errorf("复制 ffprobe 失败: %w", err)
+	}
+
+	// 验证两个工具都能运行（失败则回滚，避免留下坏文件）。
+	for _, bin := range []string{ffmpegPath, ffprobePath} {
+		cmd := exec.CommandContext(ctx, bin, "-version") // #nosec G204 -- bin 是安装目录中刚写入的固定文件名。
+		if out, verr := cmd.Output(); verr != nil {
+			_ = os.Remove(ffmpegPath)
+			_ = os.Remove(ffprobePath)
+			return "", "", fmt.Errorf("安装后 %s 无法运行：%v", filepath.Base(bin), verr)
+		} else if log != nil {
+			log.Info("ffmpeg 工具安装验证通过", zap.String("bin", filepath.Base(bin)),
+				zap.String("version", strings.TrimSpace(strings.SplitN(string(out), "\n", 2)[0])))
 		}
 	}
-
-	// 解压
-	log.Info("解压 ffmpeg...")
-	extractDir := filepath.Join(tempDir, "extract")
-	if err := unzip(log, zipPath, extractDir); err != nil {
-		return false, fmt.Errorf("解压失败: %w", err)
-	}
-
-	packageRoot, err := findFFmpegPackageRoot(extractDir)
-	if err != nil {
-		return false, err
-	}
-	if err := copyDirContents(packageRoot, installDir); err != nil {
-		return false, fmt.Errorf("复制 ffmpeg 文件失败: %w", err)
-	}
-
-	ffmpegBin := filepath.Join(installDir, "bin", "ffmpeg.exe")
-	ffprobeBin := filepath.Join(installDir, "bin", "ffprobe.exe")
-	if _, err := os.Stat(ffmpegBin); err != nil {
-		return false, fmt.Errorf("安装后未找到 ffmpeg: %w", err)
-	}
-	if _, err := os.Stat(ffprobeBin); err != nil {
-		return false, fmt.Errorf("安装后未找到 ffprobe: %w", err)
-	}
-
-	log.Info("ffmpeg 安装完成", zap.String("dir", installDir))
-	return true, nil
+	progress("安装完成")
+	return ffmpegPath, ffprobePath, nil
 }
 
-// downloadFile 下载文件
-func downloadFile(log *zap.Logger, url, filepath string) error {
-	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Minute)
-	defer cancel()
+// downloadFFmpegArchive 按顺序尝试下载源，全部失败才返回错误。
+func downloadFFmpegArchive(ctx context.Context, log *zap.Logger, urls []string, dest string) error {
+	var lastErr error
+	for i, u := range urls {
+		if i > 0 && log != nil {
+			log.Warn("ffmpeg 主下载源不可用，切换备用源", zap.String("url", u))
+		}
+		if err := downloadFFmpegFile(ctx, log, u, dest); err != nil {
+			lastErr = err
+			if log != nil {
+				log.Warn("ffmpeg 下载失败", zap.String("url", u), zap.Error(err))
+			}
+			continue
+		}
+		return nil
+	}
+	return fmt.Errorf("所有下载源均失败：%v", lastErr)
+}
 
-	req, err := http.NewRequestWithContext(ctx, "GET", url, nil)
+// downloadFFmpegFile 下载单个归档文件（最多 10 分钟，限制大小上限）。
+func downloadFFmpegFile(ctx context.Context, log *zap.Logger, url, dest string) error {
+	ctx, cancel := context.WithTimeout(ctx, 10*time.Minute)
+	defer cancel()
+	req, err := http.NewRequestWithContext(ctx, http.MethodGet, url, nil)
 	if err != nil {
 		return err
 	}
-
+	req.Header.Set("User-Agent", "MMTL/ffmpeg-installer ("+runtime.GOOS+"/"+runtime.GOARCH+")")
 	resp, err := http.DefaultClient.Do(req)
 	if err != nil {
 		return err
 	}
 	defer resp.Body.Close()
-
 	if resp.StatusCode != http.StatusOK {
-		return fmt.Errorf("下载失败，HTTP 状态码: %d", resp.StatusCode)
+		return fmt.Errorf("下载失败，HTTP 状态码 %d", resp.StatusCode)
 	}
-
-	out, err := os.Create(filepath) // #nosec G304 -- filepath is generated by the installer under its temporary download directory.
+	out, err := os.Create(dest) // #nosec G304 -- dest 是安装器在临时目录下生成的文件。
 	if err != nil {
 		return err
 	}
 	defer out.Close()
-
-	_, err = io.Copy(out, resp.Body)
-	return err
+	n, err := io.Copy(out, io.LimitReader(resp.Body, 500<<20+1)) // 归档上限 500MB
+	if err != nil {
+		return err
+	}
+	if n > 500<<20 {
+		return fmt.Errorf("归档文件过大（>500MB）: %s", url)
+	}
+	if log != nil {
+		log.Info("ffmpeg 归档下载完成", zap.String("url", url), zap.Int64("bytes", n))
+	}
+	return nil
 }
 
-// updateConfigPaths 更新配置文件中的路径
-func updateConfigPaths(cfg *config.Config, ffprobePath, ffmpegPath string) {
-	cfg.App.FFprobePath = ffprobePath
-	cfg.App.FFmpegPath = ffmpegPath
-
-	// 保存到配置文件
-	// 这里需要调用 config 包的保存函数
-	log := zap.L().Named("config")
-	log.Info("已更新 ffmpeg 路径配置",
-		zap.String("ffprobe", ffprobePath),
-		zap.String("ffmpeg", ffmpegPath))
+// extractFFmpegArchive 按类型解压 zip 或 tar.xz。
+func extractFFmpegArchive(kind, archivePath, destDir string) error {
+	switch kind {
+	case "zip":
+		return unzip(nil, archivePath, destDir)
+	case "tar.xz":
+		return untarXZ(archivePath, destDir)
+	default:
+		return fmt.Errorf("不支持的归档类型: %s", kind)
+	}
 }
 
-// is64Bit 检查是否为 64 位系统
-func is64Bit() bool {
-	return true // 简化处理，假设为 64 位
+// untarXZ 解压 .tar.xz 归档（GNU tar + xz 流式解压，纯 Go 无外部依赖），
+// 路径安全校验与 ZIP 解压一致。
+func untarXZ(archivePath, destDir string) error {
+	if err := os.MkdirAll(destDir, 0o750); err != nil {
+		return err
+	}
+	destRoot, err := filepath.Abs(destDir)
+	if err != nil {
+		return err
+	}
+	f, err := os.Open(archivePath) // #nosec G304 -- archivePath 是安装器在临时目录下生成的文件。
+	if err != nil {
+		return err
+	}
+	defer f.Close()
+	xzReader, err := xz.NewReader(f)
+	if err != nil {
+		return err
+	}
+	tr := tar.NewReader(xzReader)
+	var totalWritten int64
+	for {
+		hdr, err := tr.Next()
+		if err == io.EOF {
+			return nil
+		}
+		if err != nil {
+			return err
+		}
+		target, err := safeZipTarget(destRoot, hdr.Name) // 与 ZIP 相同的路径穿越防护
+		if err != nil {
+			return err
+		}
+		switch hdr.Typeflag {
+		case tar.TypeDir:
+			if err := os.MkdirAll(target, 0o750); err != nil {
+				return err
+			}
+		case tar.TypeReg, tar.TypeRegA:
+			if err := os.MkdirAll(filepath.Dir(target), 0o750); err != nil {
+				return err
+			}
+			dst, err := os.OpenFile(target, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, os.FileMode(hdr.Mode).Perm())
+			if err != nil {
+				return err
+			}
+			written, copyErr := io.Copy(dst, io.LimitReader(tr, maxFFmpegZipEntryBytes+1))
+			totalWritten += written
+			closeErr := dst.Close()
+			if copyErr != nil {
+				return copyErr
+			}
+			if closeErr != nil {
+				return closeErr
+			}
+			if written > maxFFmpegZipEntryBytes || totalWritten > maxFFmpegZipTotalBytes {
+				return fmt.Errorf("tar 内容过大: %s", hdr.Name)
+			}
+		default:
+			// 符号链接/设备等一律跳过（静态构建不会依赖它们）。
+			continue
+		}
+	}
+}
+
+// locateFFmpegBinaries 在解压目录中查找 ffmpeg/ffprobe 可执行文件（兼容
+// 不同构建包的目录布局：gyan 的 bin/、BtbN/johnvansickle 的根目录等）。
+func locateFFmpegBinaries(root, exeSuffix string) (ffmpeg, ffprobe string, err error) {
+	wantFFmpeg := "ffmpeg" + strings.ToLower(exeSuffix)
+	wantFFprobe := "ffprobe" + strings.ToLower(exeSuffix)
+	err = filepath.WalkDir(root, func(path string, d os.DirEntry, err error) error {
+		if err != nil {
+			return err
+		}
+		if d.IsDir() {
+			return nil
+		}
+		switch strings.ToLower(d.Name()) {
+		case wantFFmpeg:
+			if ffmpeg == "" {
+				ffmpeg = path
+			}
+		case wantFFprobe:
+			if ffprobe == "" {
+				ffprobe = path
+			}
+		}
+		return nil
+	})
+	if err != nil {
+		return "", "", fmt.Errorf("扫描解压目录失败: %w", err)
+	}
+	if ffmpeg == "" || ffprobe == "" {
+		return "", "", fmt.Errorf("解压内容中未找到 ffmpeg/ffprobe 可执行文件")
+	}
+	return ffmpeg, ffprobe, nil
+}
+
+// copyFileMode 复制文件并赋予可执行权限。
+func copyFileMode(src, dst string) error {
+	in, err := os.Open(src) // #nosec G304 -- src 来自解压目录遍历结果。
+	if err != nil {
+		return err
+	}
+	defer in.Close()
+	out, err := os.OpenFile(dst, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, 0o755)
+	if err != nil {
+		return err
+	}
+	if _, err := io.Copy(out, in); err != nil {
+		_ = out.Close()
+		return err
+	}
+	return out.Close()
 }

internal/service/ffmpeg_install_archive.go

@@ -41,7 +41,9 @@ func unzip(log *zap.Logger, zipPath, destDir string) error {
 		}
 		info := file.FileInfo()
 		if info.Mode()&os.ModeSymlink != 0 {
-			log.Warn("跳过 ZIP 符号链接", zap.String("name", file.Name))
+			if log != nil {
+				log.Warn("跳过 ZIP 符号链接", zap.String("name", file.Name))
+			}
 			continue
 		}
 		if info.IsDir() {
@@ -110,93 +112,3 @@ func safeZipTarget(destRoot, name string) (string, error) {
 	return targetAbs, nil
 }
 
-func findFFmpegPackageRoot(root string) (string, error) {
-	var ffmpegPath string
-	var ffprobePath string
-
-	err := filepath.WalkDir(root, func(path string, d os.DirEntry, err error) error {
-		if err != nil {
-			return err
-		}
-		if d.IsDir() {
-			return nil
-		}
-
-		switch strings.ToLower(d.Name()) {
-		case "ffmpeg.exe":
-			ffmpegPath = path
-		case "ffprobe.exe":
-			ffprobePath = path
-		}
-		return nil
-	})
-	if err != nil {
-		return "", fmt.Errorf("扫描解压目录失败: %w", err)
-	}
-	if ffmpegPath == "" || ffprobePath == "" {
-		return "", fmt.Errorf("解压后未找到 ffmpeg/ffprobe 可执行文件")
-	}
-
-	return filepath.Dir(filepath.Dir(ffmpegPath)), nil
-}
-
-func copyDirContents(srcDir, dstDir string) error {
-	entries, err := os.ReadDir(srcDir)
-	if err != nil {
-		return err
-	}
-
-	for _, entry := range entries {
-		srcPath := filepath.Join(srcDir, entry.Name())
-		dstPath := filepath.Join(dstDir, entry.Name())
-		if err := copyTree(srcPath, dstPath); err != nil {
-			return err
-		}
-	}
-	return nil
-}
-
-func copyTree(srcPath, dstPath string) error {
-	info, err := os.Stat(srcPath)
-	if err != nil {
-		return err
-	}
-
-	if info.IsDir() {
-		if err := os.MkdirAll(dstPath, info.Mode()); err != nil {
-			return err
-		}
-		entries, err := os.ReadDir(srcPath)
-		if err != nil {
-			return err
-		}
-		for _, entry := range entries {
-			if err := copyTree(filepath.Join(srcPath, entry.Name()), filepath.Join(dstPath, entry.Name())); err != nil {
-				return err
-			}
-		}
-		return nil
-	}
-
-	in, err := os.Open(srcPath) // #nosec G304 -- srcPath is produced by walking the validated extracted ffmpeg package tree.
-	if err != nil {
-		return err
-	}
-	defer in.Close()
-
-	if err := os.MkdirAll(filepath.Dir(dstPath), 0o750); err != nil {
-		return err
-	}
-
-	out, err := os.Create(dstPath) // #nosec G304 -- dstPath is generated under the configured ffmpeg install directory.
-	if err != nil {
-		return err
-	}
-	defer out.Close()
-
-	if _, err := io.Copy(out, in); err != nil {
-		return err
-	}
-
-	return out.Close()
-}

internal/service/ffmpeg_status.go

@@ -1,57 +0,0 @@
-package service
-
-import (
-	"bytes"
-	"os"
-	"os/exec"
-	"runtime"
-)
-
-// CheckFFmpegStatus 检查 ffmpeg/ffprobe 状态 (供 API 使用)
-func CheckFFmpegStatus(ffprobePath, ffmpegPath string) map[string]interface{} {
-	status := map[string]interface{}{
-		"ffprobe_installed": false,
-		"ffmpeg_installed":  false,
-		"auto_installable":  runtime.GOOS == "windows",
-	}
-
-	if ffprobePath != "" {
-		if _, err := os.Stat(ffprobePath); err == nil {
-			status["ffprobe_installed"] = true
-			status["ffprobe_path"] = ffprobePath
-
-			// 获取版本
-			cmd := exec.Command(ffprobePath, "-version")
-			out, err := cmd.Output()
-			if err == nil {
-				// 提取版本信息（第一行）
-				lines := bytes.Split(out, []byte("\n"))
-				if len(lines) > 0 {
-					version := string(bytes.TrimSpace(lines[0]))
-					status["ffprobe_version"] = version
-					status["ffprobe_security"] = EvaluateFFmpegSecurity(version)
-				}
-			}
-		}
-	}
-
-	if ffmpegPath != "" {
-		if _, err := os.Stat(ffmpegPath); err == nil {
-			status["ffmpeg_installed"] = true
-			status["ffmpeg_path"] = ffmpegPath
-
-			cmd := exec.Command(ffmpegPath, "-version")
-			out, err := cmd.Output()
-			if err == nil {
-				lines := bytes.Split(out, []byte("\n"))
-				if len(lines) > 0 {
-					version := string(bytes.TrimSpace(lines[0]))
-					status["ffmpeg_version"] = version
-					status["ffmpeg_security"] = EvaluateFFmpegSecurity(version)
-				}
-			}
-		}
-	}
-
-	return status
-}

internal/service/ffmpeg_tools.go

@@ -0,0 +1,211 @@
+// Package service — ffmpeg/ffprobe 自动下载安装。
+//
+// FFmpegToolsService 负责「一键下载」：点击后按当前运行平台（OS+架构）选择
+// 官方构建包（Windows: gyan.dev / BtbN；Linux: BtbN / johnvansickle；
+// macOS: BtbN），下载解压 ffmpeg/ffprobe 到 data 目录（data/tools/ffmpeg/），
+// 并把绝对路径写入设置（ffmpeg.path / ffprobe.path），系统随即使用安装的
+// 工具，无需手动填写路径。
+package service
+
+import (
+	"context"
+	"errors"
+	"os"
+	"os/exec"
+	"path/filepath"
+	"runtime"
+	"strings"
+	"sync"
+	"time"
+
+	"go.uber.org/zap"
+
+	"github.com/ShukeBta/MMTL/internal/config"
+	"github.com/ShukeBta/MMTL/internal/repository"
+)
+
+// FFmpegToolsService 管理 ffmpeg/ffprobe 的自动下载安装状态（单飞：同一时间
+// 只允许一个安装任务）。
+type FFmpegToolsService struct {
+	cfg  *config.Config
+	log  *zap.Logger
+	repo *repository.Container
+
+	mu      sync.Mutex
+	running bool
+	msg     string    // 最近阶段/结果消息
+	errMsg  string    // 最近一次失败原因
+	started time.Time // 最近一次安装开始时间
+	done    time.Time // 最近一次安装结束时间
+}
+
+// NewFFmpegToolsService 构造工具安装服务。
+func NewFFmpegToolsService(cfg *config.Config, log *zap.Logger, repo *repository.Container) *FFmpegToolsService {
+	return &FFmpegToolsService{cfg: cfg, log: log, repo: repo}
+}
+
+// ffmpegInstallDir 返回 data 目录下的安装位置。
+func (s *FFmpegToolsService) ffmpegInstallDir() string {
+	return filepath.Join(s.cfg.App.DataDir, "tools", "ffmpeg")
+}
+
+// installedBinaries 检查安装目录中是否已存在 ffmpeg/ffprobe 可执行文件。
+func (s *FFmpegToolsService) installedBinaries() (ffmpeg, ffprobe string) {
+	exe := ""
+	if runtime.GOOS == "windows" {
+		exe = ".exe"
+	}
+	ffmpeg = filepath.Join(s.ffmpegInstallDir(), "ffmpeg"+exe)
+	ffprobe = filepath.Join(s.ffmpegInstallDir(), "ffprobe"+exe)
+	if _, err := os.Stat(ffmpeg); err != nil {
+		return "", ""
+	}
+	if _, err := os.Stat(ffprobe); err != nil {
+		return "", ""
+	}
+	return ffmpeg, ffprobe
+}
+
+// ffToolVersion 取工具第一行版本信息。
+func ffToolVersion(path string) string {
+	if path == "" {
+		return ""
+	}
+	out, err := exec.Command(path, "-version").Output() // #nosec G204 -- path 来自配置/安装目录中的已知工具。
+	if err != nil {
+		return ""
+	}
+	line := strings.SplitN(strings.TrimSpace(string(out)), "\n", 2)
+	if len(line) == 0 || strings.TrimSpace(line[0]) == "" {
+		return ""
+	}
+	return strings.TrimSpace(line[0])
+}
+
+// ffToolInfo 是单个工具的安装状态（返回给前端展示）。
+type ffToolInfo struct {
+	Installed bool   `json:"installed"`
+	Path      string `json:"path,omitempty"`
+	Version   string `json:"version,omitempty"`
+}
+
+// Status 返回当前安装状态（供 GET /api/admin/tools/ffmpeg/status 使用）。
+func (s *FFmpegToolsService) Status(ctx context.Context) map[string]any {
+	s.mu.Lock()
+	running, msg, errMsg, started, done := s.running, s.msg, s.errMsg, s.started, s.done
+	s.mu.Unlock()
+
+	startedAt, doneAt := "", ""
+	if !started.IsZero() {
+		startedAt = started.Format(time.RFC3339)
+	}
+	if !done.IsZero() {
+		doneAt = done.Format(time.RFC3339)
+	}
+
+	out := map[string]any{
+		"installing":  running,
+		"message":     msg,
+		"error":       errMsg,
+		"started_at":  startedAt,
+		"finished_at": doneAt,
+		"install_dir": s.ffmpegInstallDir(),
+	}
+	target, targetErr := ffmpegTargetForPlatform()
+	if targetErr != nil {
+		out["target"] = map[string]any{"label": targetErr.Error()}
+	} else {
+		out["target"] = map[string]any{
+			"os":    runtime.GOOS,
+			"arch":  runtime.GOARCH,
+			"label": target.Label,
+		}
+	}
+	// 报告「系统当前实际会使用」的工具：优先已生效配置（安装完成会把设置指到
+	// data 目录），其次 PATH / 常见目录。
+	ffmpegPath, ferr := resolveLocalExecutable(s.cfg.App.FFmpegPath, "ffmpeg")
+	ffprobePath, perr := resolveLocalExecutable(s.cfg.App.FFprobePath, "ffprobe")
+	out["ffmpeg"] = ffToolInfo{Installed: ferr == nil, Path: ffmpegPath, Version: ffToolVersion(ffmpegPath)}
+	out["ffprobe"] = ffToolInfo{Installed: perr == nil, Path: ffprobePath, Version: ffToolVersion(ffprobePath)}
+	return out
+}
+
+// StartInstall 启动后台安装（幂等）。正在安装时返回错误；data 目录已有完整
+// 工具时直接应用路径设置并返回（无需重新下载）。
+func (s *FFmpegToolsService) StartInstall(ctx context.Context) error {
+	s.mu.Lock()
+	if s.running {
+		s.mu.Unlock()
+		return errors.New("工具正在安装中，请稍候")
+	}
+	if ffmpeg, ffprobe := s.installedBinaries(); ffmpeg != "" && ffprobe != "" {
+		s.mu.Unlock()
+		s.setMessage("检测到已安装，直接应用配置")
+		if err := s.applyInstalledPaths(ctx, ffmpeg, ffprobe); err != nil {
+			return err
+		}
+		return nil
+	}
+	s.running = true
+	s.errMsg = ""
+	s.started = time.Now()
+	s.mu.Unlock()
+
+	s.setMessage("准备下载…")
+	go s.runInstall()
+	return nil
+}
+
+// runInstall 在后台执行下载、解压、验证与配置落盘。
+func (s *FFmpegToolsService) runInstall() {
+	defer func() {
+		s.mu.Lock()
+		s.running = false
+		s.done = time.Now()
+		s.mu.Unlock()
+	}()
+	ctx, cancel := context.WithTimeout(context.Background(), 15*time.Minute)
+	defer cancel()
+
+	ffmpegPath, ffprobePath, err := installFFmpegTools(ctx, s.log, s.cfg, s.setMessage)
+	if err != nil {
+		s.mu.Lock()
+		s.errMsg = err.Error()
+		s.msg = "安装失败"
+		s.mu.Unlock()
+		s.log.Error("ffmpeg 工具安装失败", zap.Error(err))
+		return
+	}
+	if err := s.applyInstalledPaths(ctx, ffmpegPath, ffprobePath); err != nil {
+		s.mu.Lock()
+		s.errMsg = "安装完成，但写入设置失败：" + err.Error()
+		s.msg = "安装完成，设置写入失败"
+		s.mu.Unlock()
+		s.log.Error("写入 ffmpeg 工具路径设置失败", zap.Error(err))
+		return
+	}
+	s.setMessage("安装完成")
+	s.log.Info("ffmpeg 工具安装完成",
+		zap.String("ffmpeg", ffmpegPath), zap.String("ffprobe", ffprobePath))
+}
+
+// applyInstalledPaths 把安装后的路径写入设置表并热应用到运行配置。
+func (s *FFmpegToolsService) applyInstalledPaths(ctx context.Context, ffmpeg, ffprobe string) error {
+	if s.repo != nil && s.repo.Setting != nil {
+		if err := s.repo.Setting.Set(ctx, "ffmpeg.path", ffmpeg); err != nil {
+			return err
+		}
+		if err := s.repo.Setting.Set(ctx, "ffprobe.path", ffprobe); err != nil {
+			return err
+		}
+	}
+	ApplyRuntimeSetting(s.cfg, "ffmpeg.path", ffmpeg)
+	ApplyRuntimeSetting(s.cfg, "ffprobe.path", ffprobe)
+	return nil
+}
+
+func (s *FFmpegToolsService) setMessage(msg string) {
+	s.mu.Lock()
+	s.msg = msg
+	s.mu.Unlock()
+}

internal/service/ffmpeg_tools_test.go

@@ -0,0 +1,81 @@
+package service
+
+import (
+	"context"
+	"runtime"
+	"strings"
+	"testing"
+
+	"go.uber.org/zap"
+
+	"github.com/ShukeBta/MMTL/internal/config"
+)
+
+func TestFFmpegTargetForPlatform(t *testing.T) {
+	target, err := ffmpegTargetForPlatform()
+	switch runtime.GOOS {
+	case "windows", "linux", "darwin":
+		if err != nil {
+			t.Fatalf("supported platform %s/%s should resolve a target: %v", runtime.GOOS, runtime.GOARCH, err)
+		}
+		if target == nil || target.Label == "" || len(target.Archives) == 0 {
+			t.Fatalf("target incomplete: %#v", target)
+		}
+		if target.Kind != "zip" && target.Kind != "tar.xz" {
+			t.Fatalf("unexpected archive kind: %s", target.Kind)
+		}
+		for _, u := range target.Archives {
+			if !strings.HasPrefix(u, "https://") {
+				t.Fatalf("archive url not https: %s", u)
+			}
+		}
+	default:
+		if err == nil {
+			t.Fatalf("unsupported platform %s/%s should fail", runtime.GOOS, runtime.GOARCH)
+		}
+	}
+}
+
+func TestSafeZipTargetRejectsTraversal(t *testing.T) {
+	root := t.TempDir()
+	for _, name := range []string{"../evil", "..\\evil", "/etc/passwd", "a/../../evil"} {
+		if _, err := safeZipTarget(root, name); err == nil {
+			t.Fatalf("expected traversal rejection for %q", name)
+		}
+	}
+	if _, err := safeZipTarget(root, "bin/ffmpeg.exe"); err != nil {
+		t.Fatalf("valid relative path should pass: %v", err)
+	}
+}
+
+func TestFFmpegToolsStatusNoPanic(t *testing.T) {
+	svc := NewFFmpegToolsService(&config.Config{}, zap.NewNop(), nil)
+	st := svc.Status(context.Background())
+	for _, key := range []string{"installing", "message", "error", "install_dir", "target", "ffmpeg", "ffprobe"} {
+		if _, ok := st[key]; !ok {
+			t.Fatalf("status missing key %q: %#v", key, st)
+		}
+	}
+	ffmpeg, ok := st["ffmpeg"].(ffToolInfo)
+	if !ok {
+		t.Fatalf("ffmpeg field not ffToolInfo: %T", st["ffmpeg"])
+	}
+	if ffmpeg.Installed {
+		t.Fatalf("empty config should not report installed ffmpeg: %#v", st)
+	}
+}
+
+func TestStartInstallRejectsConcurrent(t *testing.T) {
+	svc := NewFFmpegToolsService(&config.Config{App: config.AppConfig{DataDir: t.TempDir()}}, zap.NewNop(), nil)
+	// 不真实运行：直接占用 running 标记模拟进行中的安装。
+	svc.mu.Lock()
+	svc.running = true
+	svc.mu.Unlock()
+	if err := svc.StartInstall(context.Background()); err == nil {
+		t.Fatalf("second install while running should be rejected")
+	}
+	svc.mu.Lock()
+	svc.running = false
+	svc.mu.Unlock()
+}
+

internal/service/local_metadata_read.go

@@ -49,7 +49,7 @@ func findMovieNFO(mediaPath, libraryRoot string) (*nfoDocument, string, error) {
 			continue
 		}
 		seen[key] = struct{}{}
-		if doc, _, err := readNFO(path); err == nil {
+		if doc, _, err := decodeNFOFile(path); err == nil && doc != nil {
 			return doc, path, nil
 		} else if err != nil && !errors.Is(err, os.ErrNotExist) {
 			return nil, "", err
@@ -62,7 +62,7 @@ func findMovieNFO(mediaPath, libraryRoot string) (*nfoDocument, string, error) {
 			for _, match := range matches {
 				baseKey := strings.ToLower(strings.ReplaceAll(strings.TrimSuffix(filepath.Base(match), filepath.Ext(match)), "-", ""))
 				if strings.Contains(baseKey, codeKey) || strings.Contains(codeKey, baseKey) {
-					if doc, _, err := readNFO(match); err == nil {
+					if doc, _, err := decodeNFOFile(match); err == nil && doc != nil {
 						return doc, match, nil
 					} else if err != nil && !errors.Is(err, os.ErrNotExist) {
 						return nil, "", err
@@ -71,7 +71,7 @@ func findMovieNFO(mediaPath, libraryRoot string) (*nfoDocument, string, error) {
 			}
 		}
 		if len(matches) == 1 {
-			if doc, _, err := readNFO(matches[0]); err == nil {
+			if doc, _, err := decodeNFOFile(matches[0]); err == nil && doc != nil {
 				return doc, matches[0], nil
 			} else if err != nil && !errors.Is(err, os.ErrNotExist) {
 				return nil, "", err
@@ -84,21 +84,33 @@ func findMovieNFO(mediaPath, libraryRoot string) (*nfoDocument, string, error) {
 func readSeriesMetadata(mediaPath, libraryRoot string) (*LocalMetadata, error) {
 	var meta *LocalMetadata
 	showBaseDir := ""
-	if showDoc, showPath, err := findShowNFO(mediaPath, libraryRoot); err == nil && showDoc != nil {
+	if showDoc, showPath, partial, err := findShowNFO(mediaPath, libraryRoot); err == nil && showDoc != nil {
 		showBaseDir = filepath.Dir(showPath)
 		meta = metadataFromDoc(showDoc, showBaseDir, true)
+		// A truncated show NFO still yields a usable title/poster (decodePartialNFO
+		// only surfaces docs with recoverable fields). Keep HasNFO so the recovered
+		// title participates in series grouping; a fully partial episode match
+		// below can still demote it.
+		meta.HasNFO = !partial || meta.HasNFO
 	} else if err != nil && !errors.Is(err, os.ErrNotExist) {
-		return nil, err
+		// A damaged show NFO must not discard the whole series: readLocalScanMetadata
+		// treats this as a hard failure and leaves every episode without local
+		// metadata. When the show level fails we still merge the episode NFO and
+		// local artwork below instead of returning early.
+		meta = nil
 	}
 
-	if episodeDoc, episodePath, err := readNFO(nfoPath(mediaPath)); err == nil {
+	if episodeDoc, episodePath, episodePartial, err := readEpisodeNFO(nfoPath(mediaPath)); err == nil && episodeDoc != nil {
 		episodeMeta := metadataFromDoc(episodeDoc, filepath.Dir(episodePath), true)
 		if meta == nil {
 			meta = &LocalMetadata{}
 		}
 		mergeEpisodeMetadata(meta, episodeMeta, episodeDoc)
-	} else if err != nil && !errors.Is(err, os.ErrNotExist) {
-		return nil, err
+		// Only a fully valid episode NFO can confirm the match; a truncated one
+		// keeps whatever the show NFO supplied but must not force scrape_status.
+		if episodePartial {
+			meta.HasNFO = false
+		}
 	}
 	if meta == nil {
 		meta = metadataFromArtwork(mediaPath, showBaseDir)
@@ -108,19 +120,49 @@ func readSeriesMetadata(mediaPath, libraryRoot string) (*LocalMetadata, error) {
 	return meta, nil
 }
 
-func readNFO(path string) (*nfoDocument, string, error) {
+// readEpisodeNFO reads the sidecar NFO next to an episode file and reports both
+// the recovered document and whether it came from a truncated/malformed file.
+func readEpisodeNFO(path string) (*nfoDocument, string, bool, error) {
 	body, err := os.ReadFile(path) // #nosec G304 -- path is a discovered NFO sidecar under the configured library root.
 	if err != nil {
-		return nil, "", err
+		return nil, "", false, err
 	}
-	var doc nfoDocument
-	if err := xml.Unmarshal(body, &doc); err != nil {
-		return nil, "", err
+	doc, partial, err := decodePartialNFO(body)
+	if err != nil {
+		return nil, "", false, err
 	}
-	return &doc, path, nil
+	return doc, path, partial, nil
 }
 
-func findShowNFO(mediaPath, libraryRoot string) (*nfoDocument, string, error) {
+func decodePartialNFO(body []byte) (*nfoDocument, bool, error) {
+	var doc nfoDocument
+	err := xml.Unmarshal(body, &doc)
+	if err == nil {
+		return &doc, false, nil
+	}
+	if !isLikelyTruncatedXMLError(err) {
+		// A real parse error unrelated to truncation: don't trust partial fields.
+		return nil, false, err
+	}
+	// Unmarshal still fills the elements it closed before hitting the cut; treat
+	// those as partial show metadata instead of dropping everything.
+	if doc.Title == "" && doc.OriginalTitle == "" && len(doc.Thumbs) == 0 &&
+		doc.Premiered == "" && doc.Plot == "" {
+		return nil, false, err
+	}
+	return &doc, true, nil
+}
+
+func isLikelyTruncatedXMLError(err error) bool {
+	for _, msg := range []string{"unexpected EOF", "EOF"} {
+		if strings.Contains(strings.ToLower(err.Error()), strings.ToLower(msg)) {
+			return true
+		}
+	}
+	return false
+}
+
+func findShowNFO(mediaPath, libraryRoot string) (*nfoDocument, string, bool, error) {
 	dir := filepath.Dir(mediaPath)
 	root := filepath.Clean(libraryRoot)
 	for {
@@ -133,19 +175,33 @@ func findShowNFO(mediaPath, libraryRoot string) (*nfoDocument, string, error) {
 		names = append(names, base+".nfo")
 		for _, name := range names {
 			path := filepath.Join(dir, name)
-			if doc, _, err := readNFO(path); err == nil {
-				return doc, path, nil
-			} else if err != nil && !errors.Is(err, os.ErrNotExist) {
-				return nil, "", err
+			doc, partial, err := decodeNFOFile(path)
+			if doc != nil {
+				return doc, path, partial, nil
+			}
+			if err != nil && !errors.Is(err, os.ErrNotExist) {
+				return nil, "", false, err
 			}
 		}
 		if samePath(dir, root) {
-			return nil, "", os.ErrNotExist
+			return nil, "", false, os.ErrNotExist
 		}
 		parent := filepath.Dir(dir)
 		if parent == dir {
-			return nil, "", os.ErrNotExist
+			return nil, "", false, os.ErrNotExist
 		}
 		dir = parent
 	}
 }
+
+func decodeNFOFile(path string) (*nfoDocument, bool, error) {
+	body, err := os.ReadFile(path) // #nosec G304 -- path is a discovered NFO sidecar under the configured library root.
+	if err != nil {
+		return nil, false, err
+	}
+	doc, partial, err := decodePartialNFO(body)
+	if err != nil {
+		return nil, false, err
+	}
+	return doc, partial, nil
+}

internal/service/local_metadata_test.go

@@ -267,3 +267,111 @@ func TestReadLocalMetadataWithoutNFOStillFindsArtwork(t *testing.T) {
 		t.Fatalf("unexpected artwork metadata: %+v", got)
 	}
 }
+
+// TestReadLocalMetadataRecoversFromTruncatedShowNFO mirrors a real-world issue:
+// some anime tvshow.nfo files are truncated mid-URL (unexpected EOF). Before the
+// fix this discarded the whole series, leaving episodes pending with per-episode
+// titles and no poster. The recoverable fields (title/year) and the matching
+// episode NFO + local artwork must still be applied.
+func TestReadLocalMetadataRecoversFromTruncatedShowNFO(t *testing.T) {
+	root := t.TempDir()
+	showDir := filepath.Join(root, "夏日重现 (2022)")
+	seasonDir := filepath.Join(showDir, "Season 1")
+	if err := os.MkdirAll(seasonDir, 0o755); err != nil {
+		t.Fatal(err)
+	}
+	mediaPath := filepath.Join(seasonDir, "S01E01.mkv")
+	if err := os.WriteFile(mediaPath, []byte("x"), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	// tvshow.nfo cut off inside a <thumb> URL, like the broken real files.
+	truncated := `<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
+<tvshow>
+  <title>夏日重现</title>
+  <originaltitle>サマータイムレンダ</originaltitle>
+  <year>2022</year>
+  <plot>听闻自己青梅竹马死讯。</plot>
+  <thumb aspect="poster">https://image.tmdb.org/t/p/original/2koyWLm6iVn5OTEExTjKzVms5Iz.jpg</thumb>
+  <fanart>
+    <thumb>https://image.tmdb.org/t/p/original/p2eZlGwd8OjkWpwD2hSoBiIlHBZ.jpg</thu`
+	if err := os.WriteFile(filepath.Join(showDir, "tvshow.nfo"), []byte(truncated), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	// The episode sidecar NFO is complete and should still be merged.
+	if err := os.WriteFile(nfoPath(mediaPath), []byte(`<episodedetails><title>再见了夏日</title><season>1</season><episode>1</episode></episodedetails>`), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	// Local artwork next to the show folder.
+	poster := filepath.Join(showDir, "poster.jpg")
+	if err := os.WriteFile(poster, []byte("jpg"), 0o644); err != nil {
+		t.Fatal(err)
+	}
+
+	got, err := ReadLocalMetadata(mediaPath, root, true)
+	if err != nil {
+		t.Fatalf("ReadLocalMetadata returned error on truncated show NFO: %v", err)
+	}
+	if got == nil {
+		t.Fatal("metadata is nil; truncated show NFO discarded the series")
+	}
+	// The recovered show title takes precedence over the episode title.
+	if got.Title != "夏日重现" {
+		t.Fatalf("Title = %q, want recovered show title 夏日重现", got.Title)
+	}
+	if got.Year != 2022 {
+		t.Fatalf("Year = %d, want 2022", got.Year)
+	}
+	if got.EpisodeTitle != "再见了夏日" || got.SeasonNum != 1 || got.EpisodeNum != 1 {
+		t.Fatalf("episode metadata not preserved: %+v", got)
+	}
+	// Prior to the fix the episode metadata was dropped entirely; the poster comes
+	// from the local poster.jpg next to the show.
+	if got.PosterURL != poster {
+		t.Fatalf("PosterURL = %q, want local poster %q", got.PosterURL, poster)
+	}
+	// A truncated show NFO must not be treated as an authoritative match by
+	// itself; since the episode NFO is valid we still mark it matched so the
+	// recovered series title participates in grouping.
+	if !got.HasNFO {
+		t.Fatalf("HasNFO = false, want true (episode NFO is valid)")
+	}
+}
+
+// TestReadLocalMetadataKeepsArtworkOnlyWhenNoUsableNFO verifies that when a show
+// NFO is truncated AND yields no recoverable fields, we still fall back to local
+// artwork instead of returning an error.
+func TestReadLocalMetadataArtworkFallbackOnGarbageShowNFO(t *testing.T) {
+	root := t.TempDir()
+	showDir := filepath.Join(root, "Some Show")
+	seasonDir := filepath.Join(showDir, "Season 1")
+	if err := os.MkdirAll(seasonDir, 0o755); err != nil {
+		t.Fatal(err)
+	}
+	mediaPath := filepath.Join(seasonDir, "S01E01.mkv")
+	if err := os.WriteFile(mediaPath, []byte("x"), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	// Severely truncated: no recoverable title/fields at all.
+	if err := os.WriteFile(filepath.Join(showDir, "tvshow.nfo"), []byte(`<tvshow><title>半截`), 0o644); err != nil {
+		t.Fatal(err)
+	}
+	// Episode-level backdrop sits next to the episode file.
+	backdrop := filepath.Join(seasonDir, "S01E01-backdrop.jpg")
+	if err := os.WriteFile(backdrop, []byte("jpg"), 0o644); err != nil {
+		t.Fatal(err)
+	}
+
+	got, err := ReadLocalMetadata(mediaPath, root, true)
+	if err != nil {
+		t.Fatalf("ReadLocalMetadata should not error when NFO is unusable: %v", err)
+	}
+	if got == nil {
+		t.Fatal("metadata is nil; artwork fallback missing")
+	}
+	if got.HasNFO {
+		t.Fatalf("HasNFO = true, want false for unusable NFO")
+	}
+	if got.BackdropURL != backdrop {
+		t.Fatalf("expected episode artwork fallback, got %+v", got)
+	}
+}

internal/service/manual_scrape_providers.go

@@ -32,18 +32,21 @@ func (s *ScraperService) manualTMDbCandidates(ctx context.Context, query string,
 	for _, typ := range manualTMDbSearchTypes(mediaType) {
 		switch typ {
 		case "movie":
-			if matches, err := s.tmdb.SearchMovieCandidates(ctx, query, year); err == nil {
+			if matches, err := s.tmdb.SearchMovieCandidates(ctx, query, year); err == nil && len(matches) > 0 {
 				for _, match := range matches {
 					out = append(out, manualTMDbCandidate{MediaType: "movie", Match: match})
 				}
 			}
 		case "tv":
-			if matches, err := s.tmdb.SearchTVCandidates(ctx, query, year); err == nil {
+			if matches, err := s.tmdb.SearchTVCandidates(ctx, query, year); err == nil && len(matches) > 0 {
 				for _, match := range matches {
 					out = append(out, manualTMDbCandidate{MediaType: "tv", Match: match})
 				}
 			}
 		}
+		if len(out) > 0 {
+			break
+		}
 	}
 	return out
 }
@@ -70,15 +73,15 @@ func manualTMDbIDSearchTypes(mediaType string) []string {
 
 func manualTMDbSearchTypes(mediaType string) []string {
 	if strings.TrimSpace(mediaType) == "" {
-		return []string{"movie", "tv"}
+		return []string{"tv", "movie"}
 	}
 	switch normalizeMediaType(mediaType, "", "") {
 	case "tv", "anime", "variety":
 		return []string{"tv", "movie"}
 	case "movie", "adult":
-		return []string{"movie"}
-	default:
 		return []string{"movie", "tv"}
+	default:
+		return []string{"tv", "movie"}
 	}
 }
 

internal/service/manual_scrape_test.go

@@ -179,9 +179,9 @@ func TestManualSearchFallsBackToMovieFolderForGenericQuery(t *testing.T) {
 	if len(results) != 1 || results[0].TMDbID != 27205 {
 		t.Fatalf("manual search results=%#v, want folder fallback candidate; queries=%v", results, queries)
 	}
-	if len(queries) < 2 || queries[0] != "00000" || queries[1] != "inception" {
-		t.Fatalf("manual search queries=%v, want explicit query then folder fallback", queries)
-	}
+		if len(queries) < 2 || queries[0] != "00000" || queries[len(queries)-1] != "inception" {
+			t.Fatalf("manual search queries=%v, want explicit query then folder fallback", queries)
+		}
 }
 
 func TestManualSearchReturnsMovieFallbackForTVTypedTMDbSearch(t *testing.T) {

internal/service/media_library.go

@@ -4,14 +4,88 @@ import (
 	"context"
 
 	"github.com/ShukeBta/MMTL/internal/model"
+	"github.com/ShukeBta/MMTL/internal/repository"
 	"gorm.io/gorm"
 )
 
+type LibraryPreviewItem struct {
+	model.Library
+	Total int64        `json:"total"`
+	Cards []SeriesCard `json:"cards"`
+}
+
 // ListLibraries returns every library configured on the server.
 func (s *MediaService) ListLibraries(ctx context.Context) ([]model.Library, error) {
 	return s.repo.Library.List(ctx)
 }
 
+// ListLibrariesWithPreview returns libraries populated with item counts and latest preview cards.
+func (s *MediaService) ListLibrariesWithPreview(ctx context.Context, libraries []model.Library, visibility MediaVisibility, cardLimit int) ([]LibraryPreviewItem, error) {
+	if cardLimit <= 0 {
+		cardLimit = 10
+	}
+	out := make([]LibraryPreviewItem, len(libraries))
+	if len(libraries) == 0 {
+		return out, nil
+	}
+
+	libIDs := make([]string, 0, len(libraries))
+	for i, lib := range libraries {
+		out[i] = LibraryPreviewItem{
+			Library: lib,
+			Total:   0,
+			Cards:   []SeriesCard{},
+		}
+		libIDs = append(libIDs, lib.ID)
+	}
+
+	visibility = ExpandMediaVisibilityForMergedCloudLibraries(ctx, s.repo, visibility)
+	filter := repository.MediaQueryFilter{
+		IncludeNSFW:       visibility.IncludeNSFW,
+		AllowedLibraryIDs: visibility.AllowedLibraryIDs,
+		HiddenLibraryIDs:  visibility.HiddenLibraryIDs,
+	}
+
+	counts, err := s.repo.Media.CountByLibraries(ctx, libIDs, filter)
+	if err != nil {
+		return nil, err
+	}
+
+	for i := range out {
+		if total, ok := counts[out[i].ID]; ok {
+			out[i].Total = total
+		}
+	}
+
+	fetchCount := cardLimit * 4
+	if fetchCount < 60 {
+		fetchCount = 60
+	} else if fetchCount > 200 {
+		fetchCount = 200
+	}
+
+	for i := range out {
+		if out[i].Total == 0 {
+			continue
+		}
+		items, _, err := s.repo.Media.ListByLibrariesFiltered(ctx, []string{out[i].ID}, 0, fetchCount, filter)
+		if err != nil {
+			continue
+		}
+		s.attachLibraryMetadata(ctx, items)
+		cards := groupMediaSeriesCards(items)
+		if len(cards) > cardLimit {
+			cards = cards[:cardLimit]
+		}
+		if cards == nil {
+			cards = []SeriesCard{}
+		}
+		out[i].Cards = cards
+	}
+
+	return out, nil
+}
+
 // DeleteLibrary removes a library and its media rows. The on-disk files are
 // left untouched.
 func (s *MediaService) DeleteLibrary(ctx context.Context, id string) error {
@@ -43,10 +117,10 @@ func (s *MediaService) DeleteLibrary(ctx context.Context, id string) error {
 		if err := tx.Unscoped().Where("library_id = ?", id).Delete(&model.Media{}).Error; err != nil {
 			return err
 		}
-		if err := hardDeleteLibraryRoots(ctx, tx, id); err != nil {
-			return err
-		}
-		return tx.Delete(&model.Library{}, "id = ?", id).Error
+			if err := hardDeleteLibraryRoots(ctx, tx, id); err != nil {
+				return err
+			}
+			return tx.Unscoped().Delete(&model.Library{}, "id = ?", id).Error
 	})
 	if err == nil {
 		s.invalidateMediaCache(ctx)

internal/service/media_library_roots.go

@@ -4,6 +4,7 @@ import (
 	"context"
 	"errors"
 	"fmt"
+	"os"
 	"path/filepath"
 	"strings"
 
@@ -51,7 +52,14 @@ func (s *MediaService) CreateLibraryWithRootsAndCover(ctx context.Context, name,
 		s.invalidateMediaCache(ctx)
 		return lib, nil
 	}
-	lib := &model.Library{Name: strings.TrimSpace(name), Path: roots[0].Path, Type: kind, CoverURL: strings.TrimSpace(coverURL), Enabled: true}
+	lib := &model.Library{
+		Name:            strings.TrimSpace(name),
+		Path:            roots[0].Path,
+		Type:            kind,
+		CoverURL:        strings.TrimSpace(coverURL),
+		Enabled:         true,
+		CarouselEnabled: false,
+	}
 	if err := s.repo.Library.CreateWithRoots(ctx, lib, roots); err != nil {
 		return nil, err
 	}
@@ -59,11 +67,72 @@ func (s *MediaService) CreateLibraryWithRootsAndCover(ctx context.Context, name,
 	return lib, nil
 }
 
+// CreateLibrariesPerSubfolder 为 parent 目录下的每个直接子目录各建一个媒体库，
+// 媒体库名取子目录名，路径指向该子目录。kind 为空时按子目录名推断类型。
+func (s *MediaService) CreateLibrariesPerSubfolder(ctx context.Context, parent, kind, coverURL string) ([]model.Library, error) {
+	parent = strings.TrimSpace(parent)
+	if parent == "" {
+		return nil, errors.New("parent path required")
+	}
+	dir, err := resolveAccessibleLibraryPath(parent)
+	if err != nil {
+		return nil, err
+	}
+	entries, err := os.ReadDir(dir)
+	if err != nil {
+		return nil, fmt.Errorf("read directory failed: %w", err)
+	}
+	subdirs := make([]string, 0, len(entries))
+	for _, entry := range entries {
+		if !entry.IsDir() {
+			continue
+		}
+		if strings.HasPrefix(entry.Name(), ".") {
+			continue
+		}
+		subdirs = append(subdirs, filepath.Join(dir, entry.Name()))
+	}
+	if len(subdirs) == 0 {
+		return nil, errors.New("no subfolders found")
+	}
+	created := make([]model.Library, 0, len(subdirs))
+	for _, subdir := range subdirs {
+		name := filepath.Base(subdir)
+		lib, err := s.CreateLibraryWithRootsAndCover(ctx, name, kind, coverURL, []LibraryRootInput{{Path: subdir}})
+		if err != nil {
+			return nil, fmt.Errorf("create library for %s: %w", subdir, err)
+		}
+		created = append(created, *lib)
+	}
+	return created, nil
+}
+
 func (s *MediaService) UpdateLibraryCover(ctx context.Context, libraryID, coverURL string) error {
 	return s.repo.DB.WithContext(ctx).Model(&model.Library{}).Where("id = ?", libraryID).
 		Update("cover_url", strings.TrimSpace(coverURL)).Error
 }
 
+// UpdateLibraryFields updates sort_order / carousel_enabled on a library.
+func (s *MediaService) UpdateLibraryFields(ctx context.Context, libraryID string, sortOrder *int, carouselEnabled *bool) error {
+	updates := map[string]any{}
+	if sortOrder != nil {
+		updates["sort_order"] = *sortOrder
+	}
+	if carouselEnabled != nil {
+		updates["carousel_enabled"] = *carouselEnabled
+	}
+	if len(updates) == 0 || strings.TrimSpace(libraryID) == "" {
+		return nil
+	}
+	return s.repo.DB.WithContext(ctx).Model(&model.Library{}).
+		Where("id = ?", libraryID).Updates(updates).Error
+}
+
+// ReorderLibraries persists a full media-library ordering.
+func (s *MediaService) ReorderLibraries(ctx context.Context, ids []string) error {
+	return s.repo.Library.SetSortOrder(ctx, ids)
+}
+
 func (s *MediaService) findLogicalLibrary(ctx context.Context, name, kind string) (*model.Library, error) {
 	if s == nil || s.repo == nil || s.repo.Library == nil {
 		return nil, nil

internal/service/media_library_test.go

@@ -0,0 +1,94 @@
+package service
+
+import (
+	"fmt"
+	"testing"
+	"time"
+
+	"go.uber.org/zap"
+
+	"github.com/ShukeBta/MMTL/internal/config"
+	"github.com/ShukeBta/MMTL/internal/model"
+	"github.com/ShukeBta/MMTL/internal/repository"
+)
+
+func TestListLibrariesWithPreview(t *testing.T) {
+	db := newServiceTestDB(t, &model.Library{}, &model.Media{})
+	repos := repository.New(db)
+
+	lib1 := model.Library{Name: "电影", Path: "/media/movies", Type: "movie", Enabled: true}
+	if err := repos.Library.Create(t.Context(), &lib1); err != nil {
+		t.Fatal(err)
+	}
+	lib2 := model.Library{Name: "动漫", Path: "/media/anime", Type: "anime", Enabled: true}
+	if err := repos.Library.Create(t.Context(), &lib2); err != nil {
+		t.Fatal(err)
+	}
+
+	now := time.Date(2026, 7, 2, 12, 0, 0, 0, time.UTC)
+	var rows []model.Media
+
+	// Add 5 movies to lib1
+	for i := 1; i <= 5; i++ {
+		rows = append(rows, model.Media{
+			Base:      model.Base{ID: fmt.Sprintf("movie-%02d", i), CreatedAt: now.Add(time.Duration(i) * time.Hour)},
+			LibraryID: lib1.ID,
+			Title:     fmt.Sprintf("电影%d", i),
+			Path:      fmt.Sprintf("/media/movies/电影%d/movie%d.mp4", i, i),
+			PosterURL: fmt.Sprintf("/api/media/movie-%02d/poster", i),
+		})
+	}
+
+	// Add 12 episodes of 1 anime to lib2
+	for i := 1; i <= 12; i++ {
+		rows = append(rows, model.Media{
+			Base:       model.Base{ID: fmt.Sprintf("anime-ep-%02d", i), CreatedAt: now.Add(time.Duration(i) * time.Minute)},
+			LibraryID:  lib2.ID,
+			Title:      fmt.Sprintf("某动漫 第%d集", i),
+			Path:       fmt.Sprintf("/media/anime/某动漫/Season 01/某动漫.S01E%02d.mp4", i),
+			SeasonNum:  1,
+			EpisodeNum: i,
+			PosterURL:  "/api/media/anime-01/poster",
+		})
+	}
+
+	if err := repos.DB.Create(&rows).Error; err != nil {
+		t.Fatal(err)
+	}
+
+	svc := NewMediaService(&config.Config{}, zap.NewNop(), repos)
+
+	previews, err := svc.ListLibrariesWithPreview(t.Context(), []model.Library{lib1, lib2}, MediaVisibility{IncludeNSFW: true}, 10)
+	if err != nil {
+		t.Fatalf("ListLibrariesWithPreview failed: %v", err)
+	}
+
+	if len(previews) != 2 {
+		t.Fatalf("got %d previews, want 2", len(previews))
+	}
+
+	// Verify lib1 (movies)
+	if previews[0].ID != lib1.ID {
+		t.Errorf("preview[0].ID = %q, want %q", previews[0].ID, lib1.ID)
+	}
+	if previews[0].Total != 5 {
+		t.Errorf("preview[0].Total = %d, want 5", previews[0].Total)
+	}
+	if len(previews[0].Cards) != 5 {
+		t.Errorf("preview[0].Cards count = %d, want 5", len(previews[0].Cards))
+	}
+
+	// Verify lib2 (anime)
+	if previews[1].ID != lib2.ID {
+		t.Errorf("preview[1].ID = %q, want %q", previews[1].ID, lib2.ID)
+	}
+	if previews[1].Total != 12 {
+		t.Errorf("preview[1].Total = %d, want 12", previews[1].Total)
+	}
+	// 12 episodes should be grouped into 1 SeriesCard with Count = 12
+	if len(previews[1].Cards) != 1 {
+		t.Errorf("preview[1].Cards count = %d, want 1", len(previews[1].Cards))
+	} else if previews[1].Cards[0].Count != 12 {
+		t.Errorf("preview[1].Cards[0].Count = %d, want 12", previews[1].Cards[0].Count)
+	}
+}

internal/service/media_recycle.go

@@ -10,25 +10,10 @@ import (
 
 const maxRecycleBinRecords = 200
 
-// SoftDelete moves a media row to the recycle bin (gorm soft delete).
-// The on-disk file is kept; admins can purge it later.
+// SoftDelete 物理删除媒体记录（统一硬删除以降低 SQLite 存储与索引压力）。
 func (s *MediaService) SoftDelete(ctx context.Context, id string) error {
-	media, err := s.repo.Media.FindByID(ctx, id)
-	if err != nil {
-		return err
-	}
-	if media != nil && isCloudMediaPath(media.Path) {
-		err := s.repo.DB.WithContext(ctx).Unscoped().Where("id = ?", id).Delete(&model.Media{}).Error
-		if err == nil {
-			s.invalidateMediaCache(ctx)
-		}
-		return err
-	}
-	err = s.repo.DB.WithContext(ctx).Where("id = ?", id).Delete(&model.Media{}).Error
+	err := s.repo.DB.WithContext(ctx).Unscoped().Where("id = ?", id).Delete(&model.Media{}).Error
 	if err == nil {
-		if pruneErr := pruneRecycleBinRows(ctx, s.repo.DB, maxRecycleBinRecords); pruneErr != nil {
-			return pruneErr
-		}
 		s.invalidateMediaCache(ctx)
 	}
 	return err

internal/service/media_series.go

@@ -2,6 +2,8 @@ package service
 
 import (
 	"context"
+	"errors"
+	"path/filepath"
 	"regexp"
 	"sort"
 	"strings"
@@ -75,6 +77,74 @@ func (s *MediaService) ListLibrarySeriesEpisodes(ctx context.Context, libraryID,
 	return out, nil
 }
 
+func (s *MediaService) ListMediaEpisodes(ctx context.Context, mediaID string, visibility MediaVisibility) ([]model.Media, error) {
+	target, err := s.repo.Media.FindByID(ctx, mediaID)
+	if err != nil {
+		return nil, err
+	}
+	if target == nil {
+		return nil, errors.New("media not found")
+	}
+	if !visibility.Allows(target) {
+		return nil, errors.New("media not found")
+	}
+	if target.LibraryID == "" {
+		return []model.Media{*target}, nil
+	}
+	rows, _, err := s.listAllMediaVisible(ctx, target.LibraryID, visibility)
+	if err != nil {
+		return nil, err
+	}
+	if len(rows) == 0 {
+		return []model.Media{*target}, nil
+	}
+
+	resolver := newMediaSeriesKeyResolver(rows)
+	targetKey := resolver.key(*target)
+
+	out := make([]model.Media, 0)
+	if targetKey != "" {
+		for _, row := range rows {
+			if resolver.key(row) == targetKey {
+				out = append(out, row)
+			}
+		}
+	}
+
+	// 如果没有聚合到多集，尝试同父目录匹配
+	if len(out) <= 1 && target.Path != "" {
+		targetDir := filepath.Dir(strings.ReplaceAll(target.Path, "\\", "/"))
+		dirMatches := make([]model.Media, 0)
+		for _, row := range rows {
+			if row.Path != "" && filepath.Dir(strings.ReplaceAll(row.Path, "\\", "/")) == targetDir {
+				dirMatches = append(dirMatches, row)
+			}
+		}
+		if len(dirMatches) > 1 {
+			out = dirMatches
+		}
+	}
+
+	if len(out) == 0 {
+		out = []model.Media{*target}
+	}
+
+	sort.SliceStable(out, func(i, j int) bool {
+		if out[i].SeasonNum != out[j].SeasonNum {
+			return out[i].SeasonNum < out[j].SeasonNum
+		}
+		if out[i].EpisodeNum != out[j].EpisodeNum {
+			return out[i].EpisodeNum < out[j].EpisodeNum
+		}
+		if out[i].Path != out[j].Path {
+			return out[i].Path < out[j].Path
+		}
+		return out[i].CreatedAt.Before(out[j].CreatedAt)
+	})
+
+	return out, nil
+}
+
 func (s *MediaService) listAllMediaVisible(ctx context.Context, libraryID string, visibility MediaVisibility) ([]model.Media, int64, error) {
 	const pageSize = 2000
 	var all []model.Media

internal/service/organizer_directory_versions.go

@@ -229,9 +229,9 @@ func (o *OrganizerService) replaceVersions(ctx context.Context, src string, exis
 			o.log.Warn("organize replace remove existing failed",
 				zap.String("path", e), zap.Error(err))
 		}
-		if o.repo != nil && o.repo.DB != nil {
-			_ = o.repo.DB.WithContext(ctx).Where("path = ?", e).Delete(&model.Media{}).Error
-		}
+			if o.repo != nil && o.repo.DB != nil {
+				_ = o.repo.DB.WithContext(ctx).Unscoped().Where("path = ?", e).Delete(&model.Media{}).Error
+			}
 	}
 	// Move staged file + sidecars into the final path.
 	if err := os.Rename(stage, dst); err != nil {

internal/service/organizer_reclassify_media_rows.go

@@ -97,7 +97,7 @@ func (o *OrganizerService) deleteMediaRowForPath(ctx context.Context, path strin
 	if o == nil || o.repo == nil || o.repo.DB == nil {
 		return
 	}
-	_ = o.repo.DB.WithContext(ctx).Where("path = ?", path).Delete(&model.Media{}).Error
+	_ = o.repo.DB.WithContext(ctx).Unscoped().Where("path = ?", path).Delete(&model.Media{}).Error
 }
 
 func (o *OrganizerService) mediaPathExists(ctx context.Context, path string) bool {

internal/service/playback.go

@@ -196,18 +196,18 @@ func (p *PlaybackService) AddToPlaylist(ctx context.Context, playlistID, mediaID
 	return p.repo.DB.Create(item).Error
 }
 
-// RemoveFromPlaylist removes a media item from a playlist (idempotent).
+// RemoveFromPlaylist 物理删除播放列表项（幂等）。
 func (p *PlaybackService) RemoveFromPlaylist(ctx context.Context, playlistID, mediaID string) error {
-	return p.repo.DB.
+	return p.repo.DB.WithContext(ctx).Unscoped().
 		Where("playlist_id = ? AND media_id = ?", playlistID, mediaID).
 		Delete(&model.PlaylistItem{}).Error
 }
 
-// DeletePlaylist removes a playlist and all of its items.
+// DeletePlaylist 物理删除播放列表及其全部条目。
 func (p *PlaybackService) DeletePlaylist(ctx context.Context, playlistID string) error {
-	if err := p.repo.DB.Where("playlist_id = ?", playlistID).
+	if err := p.repo.DB.WithContext(ctx).Unscoped().Where("playlist_id = ?", playlistID).
 		Delete(&model.PlaylistItem{}).Error; err != nil {
 		return err
 	}
-	return p.repo.DB.Where("id = ?", playlistID).Delete(&model.Playlist{}).Error
+	return p.repo.DB.WithContext(ctx).Unscoped().Where("id = ?", playlistID).Delete(&model.Playlist{}).Error
 }

internal/service/scanner_prune.go

@@ -11,13 +11,12 @@ import (
 	"github.com/ShukeBta/MMTL/internal/model"
 )
 
-// RemovePath deletes the media row for a path that has disappeared from disk
-// (incremental delete used by the watcher on Remove/Rename events).
+// RemovePath 物理删除磁盘上已不存在的媒体记录。
 func (s *ScannerService) RemovePath(ctx context.Context, path string) (int64, error) {
 	if _, err := os.Stat(path); err == nil {
 		return 0, nil // still exists; nothing to remove
 	}
-	res := s.repo.DB.WithContext(ctx).
+	res := s.repo.DB.WithContext(ctx).Unscoped().
 		Where("path = ?", path).
 		Delete(&model.Media{})
 	if res.Error == nil && res.RowsAffected > 0 {
@@ -55,7 +54,7 @@ func (s *ScannerService) pruneMissingMedia(ctx context.Context, libraryID string
 		}
 		stale = append(stale, row.ID)
 	}
-	return s.deleteMediaByIDs(ctx, stale, false)
+	return s.deleteMediaByIDs(ctx, stale, true)
 }
 
 func (s *ScannerService) pruneMissingMediaForRoot(ctx context.Context, libraryID, rootID, rootPath string, seen map[string]struct{}) (int64, error) {
@@ -92,7 +91,7 @@ func (s *ScannerService) pruneMissingMediaForRoot(ctx context.Context, libraryID
 		}
 		stale = append(stale, row.ID)
 	}
-	return s.deleteMediaByIDs(ctx, stale, false)
+	return s.deleteMediaByIDs(ctx, stale, true)
 }
 
 func pathBelongsToRoot(pathValue, rootPath string) bool {

internal/service/scraper_queue.go

@@ -0,0 +1,354 @@
+package service
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"strings"
+	"sync"
+	"time"
+
+	"go.uber.org/zap"
+
+	"github.com/ShukeBta/MMTL/internal/model"
+)
+
+type ScrapeQueueCounts struct {
+	Pending  int64 `json:"pending"`
+	Running  int64 `json:"running"`
+	Done     int64 `json:"done"`
+	Failed   int64 `json:"failed"`
+	Canceled int64 `json:"canceled"`
+}
+
+type ScrapeQueueSnapshot struct {
+	Counts   ScrapeQueueCounts  `json:"counts"`
+	Tasks    []model.ScrapeTask `json:"tasks"`
+	Total    int64              `json:"total"`
+	Page     int                `json:"page"`
+	PageSize int                `json:"page_size"`
+}
+
+// Start 启动刮削任务队列的后台消费者。
+func (s *ScraperService) Start(ctx context.Context) {
+	if s == nil {
+		return
+	}
+	go s.queueWorker(ctx)
+}
+
+func (s *ScraperService) queueWorker(ctx context.Context) {
+	const claimBatch = 4
+	sem := make(chan struct{}, 2) // 最大并发刮削数：2
+
+	for {
+		select {
+		case <-ctx.Done():
+			return
+		default:
+		}
+
+		tasks, err := s.repo.ScrapeTask.ClaimPending(ctx, claimBatch)
+		if err != nil {
+			if s.log != nil {
+				s.log.Warn("claim pending scrape task failed", zap.Error(err))
+			}
+			sleepContext(ctx, 3*time.Second)
+			continue
+		}
+		if len(tasks) == 0 {
+			sleepContext(ctx, 2*time.Second)
+			continue
+		}
+
+		var wg sync.WaitGroup
+		for i := range tasks {
+			wg.Add(1)
+			go func(t *model.ScrapeTask) {
+				defer wg.Done()
+				select {
+				case <-ctx.Done():
+					return
+				case sem <- struct{}{}:
+				}
+				defer func() { <-sem }()
+
+				s.processScrapeTask(ctx, t)
+			}(&tasks[i])
+		}
+		wg.Wait()
+	}
+}
+
+func (s *ScraperService) processScrapeTask(ctx context.Context, task *model.ScrapeTask) {
+	media, err := s.repo.Media.FindByID(ctx, task.MediaID)
+	if err != nil || media == nil {
+		now := time.Now()
+		task.Status = model.ScrapeTaskFailed
+		task.Error = "媒体项已不存在或被删除"
+		task.FinishedAt = &now
+		_ = s.repo.ScrapeTask.Update(ctx, task)
+		return
+	}
+
+	epArtwork := task.EpisodeImages
+	options := ScrapeOptions{
+		EpisodeArtwork: &epArtwork,
+		IncludeMatched: task.RefreshMatched,
+		RetryNoMatch:   true,
+	}
+
+	enrichErr := s.EnrichOneWithOptions(ctx, media, options)
+	now := time.Now()
+	task.FinishedAt = &now
+
+	refreshed, _ := s.repo.Media.FindByID(ctx, media.ID)
+	if refreshed != nil && refreshed.ScrapeStatus == "matched" {
+		task.Status = model.ScrapeTaskDone
+		task.Error = ""
+		task.MatchedTitle = refreshed.Title
+		task.MatchedYear = refreshed.Year
+		task.PosterURL = refreshed.PosterURL
+		task.BackdropURL = refreshed.BackdropURL
+		if refreshed.TMDbID > 0 {
+			task.Provider = "tmdb"
+		} else if strings.TrimSpace(refreshed.DoubanID) != "" {
+			task.Provider = "douban"
+		} else if refreshed.BangumiID > 0 {
+			task.Provider = "bangumi"
+		} else if strings.TrimSpace(refreshed.TheTVDBID) != "" {
+			task.Provider = "thetvdb"
+		} else {
+			task.Provider = "metatube"
+		}
+	} else {
+		task.Status = model.ScrapeTaskFailed
+		if enrichErr != nil {
+			task.Error = enrichErr.Error()
+		} else if refreshed != nil && refreshed.ScrapeStatus == "no_match" {
+			task.Error = "未搜索到匹配的元数据"
+		} else {
+			task.Error = "刮削未完成匹配"
+		}
+	}
+
+	_ = s.repo.ScrapeTask.Update(ctx, task)
+
+	if s.hub != nil {
+		s.hub.Publish("scraper_queue", map[string]any{
+			"task_id": task.ID,
+			"status":  task.Status,
+			"title":   task.MediaTitle,
+		})
+	}
+}
+
+func (s *ScraperService) mediaKind(m *model.Media, lib *model.Library) string {
+	if m == nil {
+		return ""
+	}
+	if lib != nil && lib.Type != "" {
+		return lib.Type
+	}
+	if mediaIsEpisodic(m, lib) {
+		return "tv"
+	}
+	return "movie"
+}
+
+// EnqueueMedia 把单个媒体项放入刮削队列。
+func (s *ScraperService) EnqueueMedia(ctx context.Context, mediaID string, options ScrapeOptions) (*model.ScrapeTask, error) {
+	if s == nil || s.repo == nil {
+		return nil, errors.New("scraper service not initialized")
+	}
+	media, err := s.repo.Media.FindByID(ctx, mediaID)
+	if err != nil || media == nil {
+		return nil, errors.New("media not found")
+	}
+
+	if active, _ := s.repo.ScrapeTask.FindActiveByMediaID(ctx, mediaID); active != nil {
+		return active, nil
+	}
+
+	libName := ""
+	var lib *model.Library
+	if strings.TrimSpace(media.LibraryID) != "" {
+		lib, _ = s.repo.Library.FindByID(ctx, media.LibraryID)
+		if lib != nil {
+			libName = lib.Name
+		}
+	}
+
+	task := &model.ScrapeTask{
+		MediaID:        media.ID,
+		LibraryID:      media.LibraryID,
+		LibraryName:    libName,
+		MediaTitle:     media.Title,
+		MediaPath:      media.Path,
+		MediaType:      s.mediaKind(media, lib),
+		Status:         model.ScrapeTaskPending,
+		EpisodeImages:  options.episodeArtworkEnabled(),
+		RefreshMatched: options.IncludeMatched || options.RefreshWeakMatched,
+	}
+
+	if err := s.repo.ScrapeTask.Create(ctx, task); err != nil {
+		return nil, err
+	}
+	return task, nil
+}
+
+// EnqueueLibrary 把指定媒体库内的所有候选媒体批量推入刮削队列。
+func (s *ScraperService) EnqueueLibrary(ctx context.Context, libraryID string, options ScrapeOptions) (int, error) {
+	if s == nil || s.repo == nil {
+		return 0, errors.New("scraper service not initialized")
+	}
+
+	lib, err := s.repo.Library.FindByID(ctx, libraryID)
+	if err != nil || lib == nil {
+		return 0, errors.New("library not found")
+	}
+
+	rows, err := s.scrapeCandidateRows(ctx, libraryID, options)
+	if err != nil {
+		return 0, err
+	}
+	if len(rows) == 0 {
+		return 0, nil
+	}
+
+	tasks := make([]model.ScrapeTask, 0, len(rows))
+	for _, m := range rows {
+		tasks = append(tasks, model.ScrapeTask{
+			MediaID:        m.ID,
+			LibraryID:      lib.ID,
+			LibraryName:    lib.Name,
+			MediaTitle:     m.Title,
+			MediaPath:      m.Path,
+			MediaType:      s.mediaKind(&m, lib),
+			Status:         model.ScrapeTaskPending,
+			EpisodeImages:  options.episodeArtworkEnabled(),
+			RefreshMatched: options.IncludeMatched || options.RefreshWeakMatched,
+		})
+	}
+
+	if err := s.repo.ScrapeTask.CreateBatch(ctx, tasks); err != nil {
+		return 0, err
+	}
+	return len(tasks), nil
+}
+
+// EnqueueAll 把所有已启用媒体库的媒体推入刮削队列。
+func (s *ScraperService) EnqueueAll(ctx context.Context, options ScrapeOptions) (int, error) {
+	libs, err := s.repo.Library.List(ctx)
+	if err != nil {
+		return 0, err
+	}
+	total := 0
+	for _, lib := range libs {
+		if !lib.Enabled {
+			continue
+		}
+		n, err := s.EnqueueLibrary(ctx, lib.ID, options)
+		if err != nil {
+			if s.log != nil {
+				s.log.Warn("enqueue library for scrape failed", zap.String("library", lib.ID), zap.Error(err))
+			}
+			continue
+		}
+		total += n
+	}
+	return total, nil
+}
+
+func (s *ScraperService) ScrapeQueueSnapshot(ctx context.Context, status string, page, pageSize int) (*ScrapeQueueSnapshot, error) {
+	tasks, total, err := s.repo.ScrapeTask.List(ctx, status, page, pageSize)
+	if err != nil {
+		return nil, err
+	}
+	countsMap, err := s.repo.ScrapeTask.CountByStatus(ctx)
+	if err != nil {
+		return nil, err
+	}
+	snap := &ScrapeQueueSnapshot{
+		Counts: ScrapeQueueCounts{
+			Pending:  countsMap[model.ScrapeTaskPending],
+			Running:  countsMap[model.ScrapeTaskRunning],
+			Done:     countsMap[model.ScrapeTaskDone],
+			Failed:   countsMap[model.ScrapeTaskFailed],
+			Canceled: countsMap[model.ScrapeTaskCanceled],
+		},
+		Tasks:    tasks,
+		Total:    total,
+		Page:     page,
+		PageSize: pageSize,
+	}
+	return snap, nil
+}
+
+func (s *ScraperService) CancelScrapeTask(ctx context.Context, id string) error {
+	task, err := s.repo.ScrapeTask.FindByID(ctx, id)
+	if err != nil || task == nil {
+		return errors.New("刮削任务不存在")
+	}
+	if task.Status != model.ScrapeTaskPending && task.Status != model.ScrapeTaskRunning {
+		return errors.New("任务已完成或已终止，无法取消")
+	}
+	now := time.Now()
+	task.Status = model.ScrapeTaskCanceled
+	task.Error = "已取消"
+	task.FinishedAt = &now
+	return s.repo.ScrapeTask.Update(ctx, task)
+}
+
+func (s *ScraperService) RetryScrapeTask(ctx context.Context, id string) error {
+	task, err := s.repo.ScrapeTask.FindByID(ctx, id)
+	if err != nil || task == nil {
+		return errors.New("刮削任务不存在")
+	}
+	if task.Status != model.ScrapeTaskFailed && task.Status != model.ScrapeTaskCanceled {
+		return errors.New("只有失败或已取消的任务可以重试")
+	}
+	task.Status = model.ScrapeTaskPending
+	task.Error = ""
+	task.RetryCount = 0
+	task.StartedAt = nil
+	task.FinishedAt = nil
+	return s.repo.ScrapeTask.Update(ctx, task)
+}
+
+func (s *ScraperService) DeleteScrapeTask(ctx context.Context, id string) error {
+	return s.repo.ScrapeTask.Delete(ctx, id)
+}
+
+func (s *ScraperService) BatchActionScrapeTasks(ctx context.Context, action string, ids []string) (int64, error) {
+	switch action {
+	case "delete":
+		return s.repo.ScrapeTask.DeleteBatch(ctx, ids)
+	case "retry":
+		return s.repo.ScrapeTask.RetryBatch(ctx, ids)
+	case "cancel":
+		return s.repo.ScrapeTask.CancelBatch(ctx, ids)
+	default:
+		return 0, fmt.Errorf("不支持的操作: %s", action)
+	}
+}
+
+func (s *ScraperService) ClearDoneScrapeTasks(ctx context.Context) (int64, error) {
+	return s.repo.ScrapeTask.ClearDone(ctx)
+}
+
+func (s *ScraperService) ClearFinishedScrapeTasks(ctx context.Context) (int64, error) {
+	return s.repo.ScrapeTask.ClearFinished(ctx)
+}
+
+func (s *ScraperService) ClearCanceledScrapeTasks(ctx context.Context) (int64, error) {
+	return s.repo.ScrapeTask.ClearCanceled(ctx)
+}
+
+func (s *ScraperService) RetryAllFailedScrapeTasks(ctx context.Context) (int64, error) {
+	return s.repo.ScrapeTask.RetryAllFailed(ctx)
+}
+
+func (s *ScraperService) CancelPendingScrapeTasks(ctx context.Context) (int64, error) {
+	return s.repo.ScrapeTask.CancelPending(ctx)
+}

internal/service/service.go

@@ -58,11 +58,13 @@ type Container struct {
 	Device           *DeviceService
 	Cache            *RuntimeCacheService
 	Sessions         *SessionTrackerService
-	RecognitionWords *RecognitionWordsService
-	Danmaku          *DanmakuService
-	Strm             *StrmService
+		RecognitionWords *RecognitionWordsService
+		Danmaku          *DanmakuService
+		Strm             *StrmService
+		Database         *DatabaseAdminService
+		FFTools          *FFmpegToolsService
 
-	stopCtx    context.Context
+		stopCtx    context.Context
 	stopCancel context.CancelFunc
 
 	// ReloadHTTPServer 由 cmd/server 注入。HTTPS 相关设置保存后，handler
@@ -102,7 +104,12 @@ func (c *Container) Boot() {
 		c.Strm.Start(c.stopCtx)
 	}
 
-	// Mgo 保号规则巡检：默认关闭，由管理员通过 Telegram Bot 命令开启。
+		// 启动刮削队列后台消费者
+		if c.Scraper != nil {
+			c.Scraper.Start(c.stopCtx)
+		}
+
+		// Mgo 保号规则巡检：默认关闭，由管理员通过 Telegram Bot 命令开启。
 	// 每天触发一次评估；规则里的窗口可随机，不固定。
 	if c.Device != nil {
 		go c.runInactivitySweeper(c.stopCtx)

internal/service/service_builder.go

@@ -111,6 +111,8 @@ func (b *serviceContainerBuilder) initContentServices() {
 	b.c.Profile = NewProfileService(b.log, b.repos)
 	b.c.Audit = NewAuditService(b.log, b.repos)
 	b.c.Strm = NewStrmService(b.cfg, b.log, b.repos, b.c.Crypto)
+	// ffmpeg/ffprobe 一键下载安装（data/tools/ffmpeg/）。
+	b.c.FFTools = NewFFmpegToolsService(b.cfg, b.log, b.repos)
 	// 弹幕 hash 识别需要把 strm 指向解析成可拉取的直链/本地路径。
 	b.c.Danmaku.SetStrmResolver(b.c.Strm.ResolvePlay)
 }
@@ -118,6 +120,7 @@ func (b *serviceContainerBuilder) initContentServices() {
 func (b *serviceContainerBuilder) initAccessAndStorageServices() {
 	b.c.PlayProfiles = NewPlayProfileService(b.log, b.repos)
 	b.c.Permissions = NewPermissionService(b.log, b.repos)
+	b.c.Database = NewDatabaseAdminService(b.cfg, b.log, b.repos, b.repos.DB)
 	b.c.Emby.SetRuntimeCache(b.c.Cache)
 	b.c.Emby.SetSubtitleService(b.c.Subtitle)
 	b.c.Scheduler = NewSchedulerService(

internal/service/stream_normalize_test.go

@@ -22,11 +22,18 @@ func TestNormalizeCloudPlayTarget(t *testing.T) {
 	if parsed.IsAbs() || parsed.Host != "" {
 		t.Fatalf("normalized target should be relative, got %q", got)
 	}
-	if parsed.Query().Get("ref") != ref {
-		t.Fatalf("ref round-trip failed: %q", parsed.Query().Get("ref"))
-	}
+		if parsed.Query().Get("ref") != ref {
+			t.Fatalf("ref round-trip failed: %q", parsed.Query().Get("ref"))
+		}
 
-	// 非云盘播放 URL 保持原样（WebDAV/直链等）。
+		strmStale := "http://bwg.linkmy.fun:1314/api/strm/play/cloud115/video.mkv?acct=abc&pickcode=123"
+		gotStrm := normalizeCloudPlayTarget(strmStale)
+		wantStrm := "/api/strm/play/cloud115/video.mkv?acct=abc&pickcode=123"
+		if gotStrm != wantStrm {
+			t.Fatalf("normalizeCloudPlayTarget(strm) = %q, want %q", gotStrm, wantStrm)
+		}
+
+		// 非云盘播放 URL 保持原样（WebDAV/直链等）。
 	passthrough := "https://dav.example.com/media/file.mkv"
 	if got := normalizeCloudPlayTarget(passthrough); got != passthrough {
 		t.Fatalf("non-cloud target should pass through, got %q", got)

internal/service/stream_redirect.go

@@ -15,11 +15,22 @@ import (
 // /api/cloud/play 路径，由 absoluteInternalRedirect 基于「当前请求」补全
 // host，从而对历史脏数据免疫。
 func normalizeCloudPlayTarget(raw string) string {
-	typ, ref, ok := parseCloudMediaPlaybackURL(raw)
-	if !ok {
+	raw = strings.TrimSpace(raw)
+	if raw == "" {
 		return raw
 	}
-	return BuildRelativeCloudPlayURL(typ, ref)
+	if typ, ref, ok := parseCloudMediaPlaybackURL(raw); ok {
+		return BuildRelativeCloudPlayURL(typ, ref)
+	}
+	if u, err := url.Parse(raw); err == nil {
+		path := strings.ToLower(u.Path)
+		if strings.HasPrefix(path, "/api/strm/play/") || strings.HasPrefix(path, "/api/cloud/play/") || strings.HasPrefix(path, "/api/stream/") {
+			u.Scheme = ""
+			u.Host = ""
+			return u.String()
+		}
+	}
+	return raw
 }
 
 // BuildRelativeCloudPlayURL 构造相对的云盘播放 API 路径。

internal/service/strm_play.go

@@ -116,6 +116,64 @@ func (s *StrmService) resolveLocalPlay(ctx context.Context, rawPath string) (*St
 	return nil, errors.New("文件不在任何本地同步目录内")
 }
 
+// ResolvePlayTarget 解析媒体行固化的播放目标（STRMURL 或 .strm 文件内容）为
+// 可播放结果，供弹幕 hash、内嵌字幕提取等「先解析直链再读取远端」的场景复用。
+// 支持：
+//   - /api/strm/play/{provider}/video{ext}?acct=..&pickcode=.. （常规格式，含账号）
+//   - /api/cloud/play/{type}?ref=.. （旧格式，无账号 → 取该类型第一个启用账号）
+//   - 绝对 http(s) 链接（直接透传）
+//   - 其余协议（webdav:// 等）返回错误，由调用方决定是否静默跳过
+func (s *StrmService) ResolvePlayTarget(ctx context.Context, raw string) (*StrmPlayResult, error) {
+	raw = strings.TrimSpace(raw)
+	if raw == "" {
+		return nil, errors.New("空播放目标")
+	}
+	u, err := url.Parse(raw)
+	if err != nil {
+		return nil, fmt.Errorf("解析播放目标失败: %w", err)
+	}
+	lowerPath := strings.ToLower(u.Path)
+	switch {
+	case strings.HasPrefix(lowerPath, "/api/strm/play/"):
+		segs := strings.Split(strings.TrimPrefix(u.Path, "/api/strm/play/"), "/")
+		if len(segs) < 1 || strings.TrimSpace(segs[0]) == "" {
+			return nil, errors.New("无效的 strm 播放地址")
+		}
+		return s.ResolvePlay(ctx, segs[0], u.Query())
+	case strings.HasPrefix(lowerPath, "/api/cloud/play/"):
+		typ := strings.TrimSpace(strings.TrimPrefix(u.Path, "/api/cloud/play/"))
+		acct, err := s.firstEnabledAccountOf(ctx, typ)
+		if err != nil || acct == nil {
+			return nil, errors.New("没有可用的网盘账号，无法解析直链")
+		}
+		q := u.Query()
+		q.Set("acct", acct.ID)
+		return s.ResolvePlay(ctx, typ, q)
+	case u.Scheme == "http" || u.Scheme == "https":
+		return &StrmPlayResult{RedirectURL: raw}, nil
+	default:
+		return nil, fmt.Errorf("不支持的播放目标协议: %s", u.Scheme)
+	}
+}
+
+// firstEnabledAccountOf 返回指定提供方第一个凭据可用的启用账号。
+func (s *StrmService) firstEnabledAccountOf(ctx context.Context, provider string) (*model.StrmAccount, error) {
+	accounts, err := s.repo.StrmAccount.List(ctx)
+	if err != nil {
+		return nil, err
+	}
+	for i := range accounts {
+		a := &accounts[i]
+		if !a.Enabled || a.Provider != provider {
+			continue
+		}
+		if _, err := s.providerFor(ctx, a); err == nil {
+			return a, nil
+		}
+	}
+	return nil, nil
+}
+
 // ProxyDirect 反向代理渲染直链内容（保留 Range 请求头以支持拖动播放）。
 func (s *StrmService) ProxyDirect(ctx context.Context, w http.ResponseWriter, r *http.Request, link *cloud.DirectLink) error {
 	if link == nil || link.URL == "" {

internal/service/strm_queue.go

@@ -13,6 +13,7 @@ import (
 	"os"
 	"path/filepath"
 	"strings"
+	"sync"
 	"time"
 
 	"go.uber.org/zap"
@@ -27,7 +28,12 @@ const (
 )
 
 // downloadWorker 下载队列 worker：认领 → 解析直链 → 下载 → 落盘。
+//
+// 采用「批量认领 + 全局并发限流」：一次认领数个任务，用 StrmService 上的全局信号量
+// 限制整个进程「同时换直链+下载」的并发数（与 115 换链风控匹配，见 strmDownloadSemCap），
+// 同时让下载充分并行。换链走全局令牌桶（QPS=3）兜底，下载走 CDN 不限速。
 func (s *StrmService) downloadWorker(ctx context.Context) {
+	const claimBatch = 12 // 每次批量认领的任务数
 	for {
 		select {
 		case <-ctx.Done():
@@ -42,7 +48,7 @@ func (s *StrmService) downloadWorker(ctx context.Context) {
 			sleepContext(ctx, left)
 			continue
 		}
-		tasks, err := s.repo.StrmDownload.ClaimPendingDownload(ctx, 1)
+		tasks, err := s.repo.StrmDownload.ClaimPendingDownload(ctx, claimBatch)
 		if err != nil {
 			s.log.Warn("claim strm download task failed", zap.Error(err))
 			sleepContext(ctx, 3*time.Second)
@@ -52,9 +58,21 @@ func (s *StrmService) downloadWorker(ctx context.Context) {
 			sleepContext(ctx, 2*time.Second)
 			continue
 		}
+		// 并发处理本批任务：每个任务先获取全局下载槽位，槽位内部执行换链+下载。
+		// 信号量与令牌桶双重限速，确保任意时刻并发换链请求不超过安全阈值。
+		var wg sync.WaitGroup
 		for i := range tasks {
-			s.processDownloadTask(ctx, &tasks[i])
+			wg.Add(1)
+			go func(i int) {
+				defer wg.Done()
+				if !s.acquireDownloadSlot(ctx) {
+					return
+				}
+				defer s.releaseDownloadSlot()
+				s.processDownloadTask(ctx, &tasks[i])
+			}(i)
 		}
+		wg.Wait()
 	}
 }
 
@@ -139,7 +157,7 @@ func (s *StrmService) processUploadTask(ctx context.Context, task *model.StrmUpl
 		}
 	}
 	if task.Provider == model.StrmProvider115 {
-		finish(model.StrmTaskFailed, "115 网盘暂不支持元数据上传")
+		s.processUpload115(ctx, task)
 		return
 	}
 	acct, err := s.repo.StrmAccount.FindByID(ctx, task.AccountID)
@@ -178,6 +196,48 @@ func (s *StrmService) processUploadTask(ctx context.Context, task *model.StrmUpl
 	finish(model.StrmTaskDone, "")
 }
 
+// processUpload115 115 元数据上传：task.RemotePath 存的是父目录 cid，FileName 为远端文件名。
+func (s *StrmService) processUpload115(ctx context.Context, task *model.StrmUploadTask) {
+	finish := func(status, message string) {
+		now := time.Now()
+		task.Status = status
+		task.Error = message
+		task.FinishedAt = &now
+		if err := s.repo.StrmUpload.Update(context.Background(), task); err != nil {
+			s.log.Warn("update strm upload task failed", zap.Error(err))
+		}
+	}
+	acct, err := s.repo.StrmAccount.FindByID(ctx, task.AccountID)
+	if err != nil || acct == nil {
+		finish(model.StrmTaskFailed, "网盘账号不存在")
+		return
+	}
+	provider, err := s.providerFor(ctx, acct)
+	if err != nil {
+		s.uploadTaskFailWithRetry(task, err.Error())
+		return
+	}
+	named, ok := provider.(interface {
+		PutFileNamed(ctx context.Context, parentCID, fileName string, r io.Reader) error
+	})
+	if !ok {
+		finish(model.StrmTaskFailed, "该网盘不支持元数据上传")
+		return
+	}
+	f, err := os.Open(task.LocalPath)
+	if err != nil {
+		s.uploadTaskFailWithRetry(task, "打开本地文件失败："+err.Error())
+		return
+	}
+	if err := named.PutFileNamed(ctx, task.RemotePath, task.FileName, f); err != nil {
+		_ = f.Close()
+		s.uploadTaskFailWithRetry(task, "上传失败："+err.Error())
+		return
+	}
+	_ = f.Close()
+	finish(model.StrmTaskDone, "")
+}
+
 // downloadTaskFailWithRetry 下载失败任务按退避重试，超过上限标记 failed。
 func (s *StrmService) downloadTaskFailWithRetry(task *model.StrmDownloadTask, message string) {
 	if !retryTask(&task.RetryCount, &task.Status, &task.Error, &task.NextTryAt, &task.FinishedAt, message) {
@@ -491,6 +551,44 @@ func (s *StrmService) RetryUploadTask(ctx context.Context, id string) error {
 
 // ─── 下载队列批量操作（handler 使用） ─────────────────────────────────────────
 
+// DeleteDownloadTask 删除一个下载任务记录。
+func (s *StrmService) DeleteDownloadTask(ctx context.Context, id string) error {
+	return s.repo.StrmDownload.Delete(ctx, id)
+}
+
+// DeleteUploadTask 删除一个上传任务记录。
+func (s *StrmService) DeleteUploadTask(ctx context.Context, id string) error {
+	return s.repo.StrmUpload.Delete(ctx, id)
+}
+
+// BatchActionDownloadTasks 对选中的下载任务执行批量操作（delete / retry / cancel）。
+func (s *StrmService) BatchActionDownloadTasks(ctx context.Context, action string, ids []string) (int64, error) {
+	switch action {
+	case "delete":
+		return s.repo.StrmDownload.DeleteBatch(ctx, ids)
+	case "retry":
+		return s.repo.StrmDownload.RetryBatch(ctx, ids)
+	case "cancel":
+		return s.repo.StrmDownload.CancelBatch(ctx, ids)
+	default:
+		return 0, fmt.Errorf("不支持的批量操作: %s", action)
+	}
+}
+
+// BatchActionUploadTasks 对选中的上传任务执行批量操作（delete / retry / cancel）。
+func (s *StrmService) BatchActionUploadTasks(ctx context.Context, action string, ids []string) (int64, error) {
+	switch action {
+	case "delete":
+		return s.repo.StrmUpload.DeleteBatch(ctx, ids)
+	case "retry":
+		return s.repo.StrmUpload.RetryBatch(ctx, ids)
+	case "cancel":
+		return s.repo.StrmUpload.CancelBatch(ctx, ids)
+	default:
+		return 0, fmt.Errorf("不支持的批量操作: %s", action)
+	}
+}
+
 // ClearDoneDownloadTasks 清空全部已完成下载记录，返回删除数量。
 func (s *StrmService) ClearDoneDownloadTasks(ctx context.Context) (int64, error) {
 	return s.repo.StrmDownload.ClearDone(ctx)
@@ -501,6 +599,16 @@ func (s *StrmService) ClearFinishedDownloadTasks(ctx context.Context) (int64, er
 	return s.repo.StrmDownload.ClearFinished(ctx)
 }
 
+// ClearCanceledDownloadTasks 清空全部已取消的下载记录，返回删除数量。
+func (s *StrmService) ClearCanceledDownloadTasks(ctx context.Context) (int64, error) {
+	return s.repo.StrmDownload.ClearCanceled(ctx)
+}
+
+// ClearCanceledUploadTasks 清空全部已取消的上传记录，返回删除数量。
+func (s *StrmService) ClearCanceledUploadTasks(ctx context.Context) (int64, error) {
+	return s.repo.StrmUpload.ClearCanceled(ctx)
+}
+
 // RetryAllFailedDownloadTasks 批量重试所有失败下载任务，返回重新入队数量。
 func (s *StrmService) RetryAllFailedDownloadTasks(ctx context.Context) (int64, error) {
 	return s.repo.StrmDownload.RetryAllFailed(ctx)
@@ -511,6 +619,11 @@ func (s *StrmService) CancelPendingDownloadTasks(ctx context.Context) (int64, er
 	return s.repo.StrmDownload.CancelPending(ctx)
 }
 
+// CancelPendingUploadTasks 批量取消所有排队上传任务，返回取消数量。
+func (s *StrmService) CancelPendingUploadTasks(ctx context.Context) (int64, error) {
+	return s.repo.StrmUpload.CancelPending(ctx)
+}
+
 func sleepContext(ctx context.Context, d time.Duration) {
 	select {
 	case <-ctx.Done():

internal/service/strm_service.go

@@ -90,11 +90,48 @@ type StrmService struct {
 	running       map[string]context.CancelFunc // sync path id -> cancel
 	oauthSessions map[string]*strm115AuthSession
 	wafUntil      time.Time // 115 风控/限流熔断截止时间（由 mu 保护）
+
+	downloadSem chan struct{} // 全局下载并发信号量：限制整个进程同时进行「换直链+下载」的并发数
+	downloadSemOnce sync.Once
 }
 
 // strmWAFCooldown 检测到 115 风控/限流后下载队列的全局冷却时长。
 const strmWAFCooldown = 3 * time.Minute
 
+// strmDownloadSemCap 全局同时进行「换直链+下载」的并发上限。
+//
+// 115 对换直链接口（/open/ufile/downurl）风控极严：过去把全局 QPS 提到 8 或让多
+// worker 高并发换链，会瞬时撞上 WAF 返回 405 阻断页并触发 180 秒冷却，反而更慢。
+// 因此用信号量把整个进程同时换直链的并发数压到 3，与令牌桶限速共同兜底：
+// 宁可下载稍慢，也绝不触发风控。下载本身走 CDN 不限速。
+const strmDownloadSemCap = 3
+
+// ensureDownloadSem 惰性初始化全局共享的下载并发信号量。
+func (s *StrmService) ensureDownloadSem() {
+	s.downloadSemOnce.Do(func() {
+		s.downloadSem = make(chan struct{}, strmDownloadSemCap)
+	})
+}
+
+// acquireDownloadSlot 获取一个下载并发槽位（等待/取消安全）。
+func (s *StrmService) acquireDownloadSlot(ctx context.Context) bool {
+	s.ensureDownloadSem()
+	select {
+	case s.downloadSem <- struct{}{}:
+		return true
+	case <-ctx.Done():
+		return false
+	}
+}
+
+// releaseDownloadSlot 释放一个下载并发槽位。
+func (s *StrmService) releaseDownloadSlot() {
+	if s.downloadSem == nil {
+		return
+	}
+	<-s.downloadSem
+}
+
 // NewStrmService constructs the STRM service.
 func NewStrmService(cfg *config.Config, log *zap.Logger, repos *repository.Container, crypto *CryptoService) *StrmService {
 	return &StrmService{
@@ -113,6 +150,7 @@ func NewStrmService(cfg *config.Config, log *zap.Logger, repos *repository.Conta
 // Start 启动下载/上传队列 worker、定时同步巡检、115 token 刷新与队列清理。
 func (s *StrmService) Start(ctx context.Context) {
 	s.sync115RelayKey(ctx)
+	s.recoverInterruptedSyncs(ctx)
 	downloadThreads := s.strmIntSetting(ctx, StrmSettingDownloadThreads, 3)
 	if downloadThreads < 1 {
 		downloadThreads = 1
@@ -141,6 +179,21 @@ func (s *StrmService) Start(ctx context.Context) {
 		zap.Int("upload_threads", uploadThreads))
 }
 
+// recoverInterruptedSyncs 在服务启动时自愈重置因服务重启遗留的 running 状态。
+func (s *StrmService) recoverInterruptedSyncs(ctx context.Context) {
+	paths, err := s.repo.StrmSyncPath.List(ctx)
+	if err == nil {
+		for i := range paths {
+			p := &paths[i]
+			if p.LastSyncStatus == model.StrmSyncRecordRunning {
+				p.LastSyncStatus = model.StrmSyncRecordCanceled
+				p.LastSyncMessage = "服务重启，已重置同步状态"
+				_ = s.repo.StrmSyncPath.Update(ctx, p)
+			}
+		}
+	}
+}
+
 func (s *StrmService) Stop() {
 	s.stopOnce.Do(func() { close(s.stopCh) })
 }
@@ -396,6 +449,38 @@ func (s *StrmService) ListSyncRecords(ctx context.Context, pathID string, limit
 	return s.repo.StrmSyncRecord.List(ctx, pathID, limit)
 }
 
+// DeleteSyncRecord 删除单条同步记录。
+func (s *StrmService) DeleteSyncRecord(ctx context.Context, id string) error {
+	if err := s.repo.StrmSyncRecord.Delete(ctx, id); err != nil {
+		return err
+	}
+	return nil
+}
+
+// ClearSyncRecords 清空某同步目录（pathID 为空则全部）的同步记录，返回删除条数。
+func (s *StrmService) ClearSyncRecords(ctx context.Context, pathID string) (int64, error) {
+	if pathID != "" {
+		return s.repo.StrmSyncRecord.DeleteBySyncPathID(ctx, pathID)
+	}
+	var total int64
+	// 全量清空：分页拉取物理删除所有记录
+	for {
+		rows, err := s.repo.StrmSyncRecord.List(ctx, "", 200)
+		if err != nil {
+			return total, err
+		}
+		if len(rows) == 0 {
+			return total, nil
+		}
+		for _, rec := range rows {
+			if err := s.repo.StrmSyncRecord.Delete(ctx, rec.ID); err != nil {
+				return total, err
+			}
+		}
+		total += int64(len(rows))
+	}
+}
+
 // CreateSyncPath 校验并创建同步目录。
 func (s *StrmService) CreateSyncPath(ctx context.Context, p *model.StrmSyncPath) (*model.StrmSyncPath, error) {
 	if err := s.validateSyncPath(ctx, p); err != nil {
@@ -404,6 +489,9 @@ func (s *StrmService) CreateSyncPath(ctx context.Context, p *model.StrmSyncPath)
 	if strings.TrimSpace(p.Name) == "" {
 		p.Name = "同步目录 " + time.Now().Format("01-02 15:04")
 	}
+	if p.SyncMode == "" {
+		p.SyncMode = model.StrmSyncTypeIncremental
+	}
 	if p.EnableCron && strings.TrimSpace(p.Cron) == "" {
 		return nil, errors.New("启用定时同步需要填写 cron 表达式")
 	}
@@ -430,6 +518,12 @@ func (s *StrmService) UpdateSyncPath(ctx context.Context, id string, p *model.St
 	p.LastSyncAt = existing.LastSyncAt
 	p.LastSyncStatus = existing.LastSyncStatus
 	p.LastSyncMessage = existing.LastSyncMessage
+	if p.SyncMode == "" {
+		p.SyncMode = existing.SyncMode
+		if p.SyncMode == "" {
+			p.SyncMode = model.StrmSyncTypeIncremental
+		}
+	}
 	if p.EnableCron && strings.TrimSpace(p.Cron) == "" {
 		return nil, errors.New("启用定时同步需要填写 cron 表达式")
 	}

internal/service/strm_sync.go

@@ -20,6 +20,7 @@ import (
 
 	"github.com/ShukeBta/MMTL/internal/model"
 	"github.com/ShukeBta/MMTL/internal/service/cloud"
+	"github.com/ShukeBta/MMTL/internal/service/cloud115"
 )
 
 // strmSyncState 是一次同步执行的上下文。
@@ -31,16 +32,29 @@ type strmSyncState struct {
 	provider cloud.Provider // local 提供方为 nil
 	cfg      *strmPathConfig
 	rec      *model.StrmSyncRecord
+	syncType string
 
-	mu         sync.Mutex
-	processed  int              // 已处理文件计数（用于定期落库进度）
-	seenVideo  map[string]bool  // "v:"+去掉扩展名的相对路径 → 远端存在该视频
-	seenMeta   map[string]bool  // "m:"+相对路径 → 远端存在该元数据
-	remoteMeta map[string]int64 // 远端元数据大小（上传比对用）
+	mu                  sync.Mutex
+	processed           int              // 已处理文件计数（用于定期落库进度）
+	lastProgressFlush   time.Time        // 上次进度落库时间
+	seenVideo           map[string]bool  // "v:"+去掉扩展名的相对路径 → 远端存在该视频
+	seenMeta            map[string]bool  // "m:"+相对路径 → 远端存在该元数据
+	remoteMeta          map[string]int64 // 远端元数据大小（上传比对用）
+	seenMetaTarget      map[string]cloud.FileEntry
+	seenVideoTarget     map[string]cloud.FileEntry
+	activeDownloadPaths map[string]bool // 本地已在排队/进行的下载任务路径（内存去重）
+	activeUploadPaths   map[string]bool // 本地已在排队/进行的上传任务路径（内存去重）
+	pendingDownloads    []*model.StrmDownloadTask
+	pendingUploads      []*model.StrmUploadTask
+	dirCache            sync.Map          // dirID (string) -> relativePath (string)
+	dirPathToID         map[string]string // relativePath (string) -> dirID（115 上传父目录寻址用，walk 后构建）
+
+	scanIncomplete atomic.Bool // 远端目录树/文件列表本次扫描不完整 → 禁止增量 prune 误删本地文件
 }
 
 // StartSync 启动一次同步（异步执行，同一目录同时只允许一个任务）。
-func (s *StrmService) StartSync(ctx context.Context, pathID string) error {
+// syncType 支持 "incremental"（默认增量）和 "full"（全量同步）。
+func (s *StrmService) StartSync(ctx context.Context, pathID string, syncType ...string) error {
 	p, err := s.repo.StrmSyncPath.FindByID(ctx, pathID)
 	if err != nil || p == nil {
 		return errNotFoundOr(err, "同步目录不存在")
@@ -64,9 +78,20 @@ func (s *StrmService) StartSync(ctx context.Context, pathID string) error {
 	s.running[pathID] = cancel
 	s.mu.Unlock()
 
+	mode := model.StrmSyncTypeIncremental
+	if len(syncType) > 0 && syncType[0] != "" {
+		mode = syncType[0]
+	} else if p.SyncMode != "" {
+		mode = p.SyncMode
+	}
+	if mode != model.StrmSyncTypeFull {
+		mode = model.StrmSyncTypeIncremental
+	}
+
 	now := time.Now()
 	rec := &model.StrmSyncRecord{
 		SyncPathID: pathID,
+		SyncType:   mode,
 		Status:     model.StrmSyncRecordRunning,
 		StartedAt:  &now,
 	}
@@ -84,15 +109,27 @@ func (s *StrmService) StartSync(ctx context.Context, pathID string) error {
 	return nil
 }
 
-// CancelSync 取消正在进行的同步。
+// CancelSync 取消正在进行的同步（若为僵尸运行状态则直接自愈重置）。
 func (s *StrmService) CancelSync(ctx context.Context, pathID string) error {
 	s.mu.Lock()
-	defer s.mu.Unlock()
 	cancel, exists := s.running[pathID]
-	if !exists {
-		return errors.New("该目录当前没有进行中的同步")
+	if exists {
+		delete(s.running, pathID)
+	}
+	s.mu.Unlock()
+
+	if exists && cancel != nil {
+		cancel()
+	}
+
+	// 无论内存中是否活跃，确保同步目录状态正确重置为已取消
+	if p, err := s.repo.StrmSyncPath.FindByID(ctx, pathID); err == nil && p != nil {
+		if p.LastSyncStatus == model.StrmSyncRecordRunning {
+			p.LastSyncStatus = model.StrmSyncRecordCanceled
+			p.LastSyncMessage = "已取消"
+			_ = s.repo.StrmSyncPath.Update(ctx, p)
+		}
 	}
-	cancel()
 	return nil
 }
 
@@ -140,14 +177,17 @@ func (s *StrmService) runSync(ctx context.Context, p *model.StrmSyncPath, rec *m
 		return
 	}
 	st := &strmSyncState{
-		s:          s,
-		ctx:        ctx,
-		p:          p,
-		cfg:        cfg,
-		rec:        rec,
-		seenVideo:  map[string]bool{},
-		seenMeta:   map[string]bool{},
-		remoteMeta: map[string]int64{},
+		s:               s,
+		ctx:             ctx,
+		p:               p,
+		cfg:             cfg,
+		rec:             rec,
+		syncType:        rec.SyncType,
+		seenVideo:       map[string]bool{},
+		seenMeta:        map[string]bool{},
+		remoteMeta:      map[string]int64{},
+		seenMetaTarget:  map[string]cloud.FileEntry{},
+		seenVideoTarget: map[string]cloud.FileEntry{},
 	}
 	if p.Provider != model.StrmProviderLocal {
 		acct, err := s.repo.StrmAccount.FindByID(ctx, p.AccountID)
@@ -191,36 +231,77 @@ func (s *StrmService) finishSync(p *model.StrmSyncPath, rec *model.StrmSyncRecor
 	p.LastSyncStatus = status
 	p.LastSyncMessage = message
 	if status != model.StrmSyncRecordFailed && message == "" {
-		p.LastSyncMessage = fmt.Sprintf("完成：新增/更新 %d 个 strm，下载 %d 个元数据，清理 %d 个文件",
-			rec.NewStrm, rec.NewMeta, rec.Pruned)
+		syncTypeLabel := "增量"
+		if rec.SyncType == model.StrmSyncTypeFull {
+			syncTypeLabel = "全量"
+		}
+		p.LastSyncMessage = fmt.Sprintf("[%s] 完成：新增/更新 %d 个 strm，跳过 %d 个，下载 %d 个元数据，上传 %d 个元数据，清理 %d 个文件",
+			syncTypeLabel, rec.NewStrm, rec.Skipped, rec.NewMeta, rec.Uploaded, rec.Pruned)
 	}
 	if err := s.repo.StrmSyncPath.Update(context.Background(), p); err != nil {
 		s.log.Warn("update strm sync path failed", zap.Error(err))
 	}
 	s.log.Info("strm sync finished",
-		zap.String("path_id", p.ID), zap.String("status", status),
-		zap.Int64("new_strm", rec.NewStrm), zap.Int64("new_meta", rec.NewMeta),
-		zap.Int64("pruned", rec.Pruned), zap.String("message", message))
+		zap.String("path_id", p.ID), zap.String("sync_type", rec.SyncType), zap.String("status", status),
+		zap.Int64("new_strm", rec.NewStrm), zap.Int64("skipped", rec.Skipped), zap.Int64("new_meta", rec.NewMeta),
+		zap.Int64("uploaded", rec.Uploaded), zap.Int64("pruned", rec.Pruned), zap.String("message", message))
 }
 
 func (st *strmSyncState) run() error {
 	if err := ensureLocalDir(st.p.LocalPath); err != nil {
 		return fmt.Errorf("创建输出目录失败：%w", err)
 	}
+	if st.cfg.DownloadMeta {
+		if active, err := st.s.repo.StrmDownload.GetActiveLocalPathMap(st.ctx, st.p.ID); err == nil {
+			st.activeDownloadPaths = active
+		} else {
+			st.activeDownloadPaths = map[string]bool{}
+		}
+	}
+	if st.cfg.UploadMeta {
+		if active, err := st.s.repo.StrmUpload.GetActiveLocalPathMap(st.ctx, st.p.ID); err == nil {
+			st.activeUploadPaths = active
+		} else {
+			st.activeUploadPaths = map[string]bool{}
+		}
+	}
+
 	if st.provider != nil {
-		if err := st.walkRemote(); err != nil {
-			return err
+		if open115, ok := st.provider.(cloud.OpenAPI115Provider); ok && st.p.Provider == model.StrmProvider115 {
+			if err := st.walk115Flat(open115.OpenClient()); err != nil {
+				return err
+			}
+		} else {
+			if err := st.walkRemote(); err != nil {
+				return err
+			}
 		}
 	} else {
 		if err := st.walkLocalSource(); err != nil {
 			return err
 		}
 	}
+	st.flushPendingDownloads()
 	st.flushProgress()
-	if st.cfg.UploadMeta && st.provider != nil && st.p.Provider != model.StrmProvider115 {
+	if st.cfg.UploadMeta && st.provider != nil {
+		// 115 上传需要父目录 cid，先用 dirCache 构建「路径 → cid」反向索引
+		if st.p.Provider == model.StrmProvider115 {
+			reversed := map[string]string{}
+			st.dirCache.Range(func(key, value any) bool {
+				path, ok := value.(string)
+				if ok && path != "" {
+					if id, ok2 := key.(string); ok2 {
+						reversed[path] = id
+					}
+				}
+				return true
+			})
+			st.dirPathToID = reversed
+		}
 		if err := st.scanLocalMetaForUpload(); err != nil {
 			return err
 		}
+		st.flushPendingUploads()
 	}
 	if err := st.pruneLocal(); err != nil {
 		return err
@@ -239,6 +320,7 @@ const strmScanWorkers = 8
 // 多个 worker 并行执行 List（受全局 115 令牌桶限流约束），子目录动态
 // 入队；任一目录失败则取消其余 worker 并返回错误（与旧串行版语义一致）。
 func (st *strmSyncState) walkRemote() error {
+	defer st.flushPendingDownloads()
 	root := strings.TrimSpace(st.p.RemotePath)
 	if root == "" {
 		root = "/"
@@ -374,6 +456,286 @@ func (st *strmSyncState) isMetaExt(ext string) bool {
 	return false
 }
 
+// cleanDirRel 对 115 扁平化拉取的目录相对路径逐段套用目录级文件名清洗，
+// 确保与 walkRemote / joinLocalRel（sanitizeRelativePath）使用同一套清洗规则。
+// 若不清洗，目录名中的冒号等非法字符会直达 rel，而 seenVideo/seenMeta 的 key
+// 与磁盘实际路径不一致，导致 pruneLocal 误删已下载的 strm / 元数据。
+// 空 rel（根目录）原样返回。
+func cleanDirRel(rel string) string {
+	if rel == "" {
+		return ""
+	}
+	parts := strings.Split(rel, "/")
+	out := make([]string, 0, len(parts))
+	for _, part := range parts {
+		if part == "" {
+			continue
+		}
+		clean := cleanEntryName(part, true)
+		if clean != "" && clean != "." && clean != ".." {
+			out = append(out, clean)
+		}
+	}
+	return strings.Join(out, "/")
+}
+
+// walk115Flat 使用 115 开放平台扁平化分页批量拉取机制与目录拓扑缓存（参考 QMediaSync）。
+// 极大地降低 API 请求次数并支持毫秒级/秒级增量同步。
+func (st *strmSyncState) walk115Flat(open115 *cloud115.OpenClient) error {
+	defer st.flushPendingDownloads()
+	ctx, cancel := context.WithCancel(st.ctx)
+	defer cancel()
+	rootCID := strings.TrimSpace(st.p.RemotePath)
+	if rootCID == "" {
+		rootCID = "0"
+	}
+
+	// 1. 目录拓扑缓存处理
+	st.dirCache.Store(rootCID, "")
+	if st.syncType == model.StrmSyncTypeFull {
+		// 全量同步：清空本路径的历史目录缓存
+		if err := st.s.repo.StrmDirCache.DeleteBySyncPathID(ctx, st.p.ID); err != nil {
+			st.s.log.Warn("delete strm dir cache failed", zap.Error(err))
+		}
+	} else {
+		// 增量同步：预加载历史目录缓存（过滤历史一对多塌陷冲突的脏数据以自愈刷新）
+		cached, err := st.s.repo.StrmDirCache.ListBySyncPathID(ctx, st.p.ID)
+		if err == nil {
+			pathCounts := make(map[string]int, len(cached))
+			for _, item := range cached {
+				pathCounts[item.Path]++
+			}
+			for _, item := range cached {
+				// 若同一个 path 对应了多个不同 dir_id，说明包含历史层级塌陷的脏数据，不预加载，让后续步骤重新向 115 获取精确路径
+				if pathCounts[item.Path] > 1 {
+					continue
+				}
+				st.dirCache.Store(item.DirID, cleanDirRel(item.Path))
+			}
+		}
+	}
+
+	// 2. 探测文件总数
+	const pageSize = 1150
+	firstBatch, totalCount, err := open115.GetFsListFlat(ctx, rootCID, 0, pageSize)
+	if err != nil {
+		return fmt.Errorf("115: 获取文件列表失败：%w", err)
+	}
+
+	st.updateSyncMessage(fmt.Sprintf("正在拉取远端文件列表 (共 %d 个文件)...", totalCount))
+
+	allFiles := make([]cloud115.RemoteFile, 0, totalCount)
+	allFiles = append(allFiles, firstBatch...)
+
+	// 3. 并发分页拉取剩余文件
+	if totalCount > int64(len(firstBatch)) {
+		totalPages := int((totalCount + pageSize - 1) / pageSize)
+		type pageTask struct {
+			offset int
+		}
+		pageTasks := make([]pageTask, 0, totalPages-1)
+		for page := 1; page < totalPages; page++ {
+			pageTasks = append(pageTasks, pageTask{offset: page * pageSize})
+		}
+
+		var (
+			filesMu  sync.Mutex
+			wg       sync.WaitGroup
+			taskCh   = make(chan pageTask, len(pageTasks))
+			errMu    sync.Mutex
+			fetchErr error
+		)
+
+		for _, t := range pageTasks {
+			taskCh <- t
+		}
+		close(taskCh)
+
+		workers := 8
+		if len(pageTasks) < workers {
+			workers = len(pageTasks)
+		}
+
+		for i := 0; i < workers; i++ {
+			wg.Add(1)
+			go func() {
+				defer wg.Done()
+				for t := range taskCh {
+					if ctx.Err() != nil {
+						return
+					}
+					files, _, err := open115.GetFsListFlat(ctx, rootCID, t.offset, pageSize)
+					if err != nil {
+						errMu.Lock()
+						if fetchErr == nil {
+							fetchErr = err
+						}
+						errMu.Unlock()
+						return
+					}
+					filesMu.Lock()
+					allFiles = append(allFiles, files...)
+					filesMu.Unlock()
+				}
+			}()
+		}
+		wg.Wait()
+		if fetchErr != nil {
+			return fmt.Errorf("115: 分页拉取失败：%w", fetchErr)
+		}
+	}
+
+	if ctx.Err() != nil {
+		return ctx.Err()
+	}
+
+	// 4. 收集所有未在缓存中的父目录 ID (file.Pid)
+	missingPids := make(map[string]struct{})
+	for _, f := range allFiles {
+		pid := f.Pid
+		if pid == "" || pid == rootCID {
+			continue
+		}
+		if _, ok := st.dirCache.Load(pid); !ok {
+			missingPids[pid] = struct{}{}
+		}
+	}
+
+	// 并发补全未知目录详情与祖先链
+	if len(missingPids) > 0 {
+		pidList := make([]string, 0, len(missingPids))
+		for pid := range missingPids {
+			pidList = append(pidList, pid)
+		}
+
+		pidCh := make(chan string, len(pidList))
+		for _, pid := range pidList {
+			pidCh <- pid
+		}
+		close(pidCh)
+
+		var (
+			pwg        sync.WaitGroup
+			dirWorkers = 8
+			doneDirs   atomic.Int64
+			totalDirs  = len(pidList)
+			errMu      sync.Mutex
+			firstErr   error
+		)
+		if len(pidList) < dirWorkers {
+			dirWorkers = len(pidList)
+		}
+
+		st.updateSyncMessage(fmt.Sprintf("正在解析目录树 (0/%d)...", totalDirs))
+
+		for i := 0; i < dirWorkers; i++ {
+			pwg.Add(1)
+			go func() {
+				defer pwg.Done()
+				for pid := range pidCh {
+					if ctx.Err() != nil {
+						return
+					}
+					if _, loaded := st.dirCache.Load(pid); loaded {
+						if n := doneDirs.Add(1); n%20 == 0 || n == int64(totalDirs) {
+							st.updateSyncMessage(fmt.Sprintf("正在解析目录树 (%d/%d)...", n, totalDirs))
+						}
+						continue
+					}
+					detail, err := open115.GetFsDetailByCid(ctx, pid)
+					if err != nil {
+						// 目录详情解析失败会导致下游文件 rel 无法还原真实父路径，
+						// seen key 与磁盘路径对不上：增量 prune 会误删本地文件、上传会
+						// 误传本地未变文件、下载会重复下载。这里不是降级容错，而是
+						// 直接中止整个同步——宁可本次同步失败，也不带着损坏的相对路径
+						// 继续执行造成大规模误删/误传/重下（参考用户反馈"云盘没动却重下重传"）。
+						errMu.Lock()
+						if firstErr == nil {
+							firstErr = fmt.Errorf("115: 解析目录树失败（file_id=%s）：%w", pid, err)
+						}
+						errMu.Unlock()
+						st.scanIncomplete.Store(true)
+						cancel()
+						return
+					} else if detail != nil {
+						// 解析相对路径
+						relPath := cleanDirRel(detail.RelativePath(rootCID))
+						st.dirCache.Store(pid, relPath)
+						_ = st.s.repo.StrmDirCache.Set(ctx, st.p.ID, pid, relPath)
+
+						// 顺便解析并缓存 detail.Paths 中包含的中间各层级目录
+						for _, ancestor := range detail.Paths {
+							if ancestor.FileId == "0" || ancestor.FileId == rootCID {
+								continue
+							}
+							if _, loaded := st.dirCache.Load(ancestor.FileId); !loaded {
+								subDetail := &cloud115.RemoteFileDetail{
+									FileId:   ancestor.FileId,
+									FileName: ancestor.Name,
+									Paths:    nil,
+								}
+								for _, p := range detail.Paths {
+									subDetail.Paths = append(subDetail.Paths, p)
+									if p.FileId == ancestor.FileId {
+										break
+									}
+								}
+								ancestorRel := cleanDirRel(subDetail.RelativePath(rootCID))
+								st.dirCache.Store(ancestor.FileId, ancestorRel)
+								_ = st.s.repo.StrmDirCache.Set(ctx, st.p.ID, ancestor.FileId, ancestorRel)
+							}
+						}
+					}
+					if n := doneDirs.Add(1); n%10 == 0 || n == int64(totalDirs) {
+						st.updateSyncMessage(fmt.Sprintf("正在解析目录树 (%d/%d)...", n, totalDirs))
+					}
+				}
+			}()
+		}
+		pwg.Wait()
+		if firstErr != nil {
+			// 目录树解析失败会导致 rel 塌缩，若继续处理会让大量本地文件
+			// 被错误判定为"云端不存在"而重复下载/上传，并可能误删本地文件。
+			// 中止本次同步，避免在损坏的相对路径上执行任何写操作。
+			return firstErr
+		}
+	}
+
+	st.updateSyncMessage(fmt.Sprintf("正在生成 STRM 与同步文件 (共 %d 个)...", len(allFiles)))
+
+	// 5. 分类处理所有文件
+	for _, f := range allFiles {
+		if ctx.Err() != nil {
+			return ctx.Err()
+		}
+		cleanName := cleanEntryName(f.FileName, false)
+		var rel string
+		if f.Pid == "" || f.Pid == rootCID {
+			rel = cleanName
+		} else {
+			if parentVal, ok := st.dirCache.Load(f.Pid); ok && parentVal.(string) != "" {
+				rel = cleanDirRel(parentVal.(string)) + "/" + cleanName
+			} else {
+				// 父目录不在目录缓存，无法还原真实相对路径。若继续用塌缩后的
+				// 根路径处理，该文件会被错误判定，导致重复下载/上传或误删本地文件。
+				// 目录树不完整时宁可中止本次同步，也不带着损坏的 rel 继续执行。
+				return fmt.Errorf("115: 文件 %s 的父目录未解析成功，目录树不完整，中止同步以防误删/误传", cleanName)
+			}
+		}
+		entry := cloud.FileEntry{
+			ID:       f.FileId,
+			Name:     f.FileName,
+			IsDir:    false,
+			Size:     f.FileSize,
+			MTime:    f.Utime,
+			PickCode: f.PickCode,
+		}
+		st.processRemoteFile(entry, rel)
+	}
+
+	return nil
+}
+
 // handleVideo 生成/更新 .strm 文件。
 func (st *strmSyncState) handleVideo(entry cloud.FileEntry, rel, ext string) {
 	relSansExt := rel[:len(rel)-len(ext)]
@@ -391,6 +753,30 @@ func (st *strmSyncState) handleVideo(entry cloud.FileEntry, rel, ext string) {
 		st.s.log.Warn("strm target path out of root", zap.String("rel", targetRel), zap.Error(err))
 		return
 	}
+
+	st.mu.Lock()
+	if st.seenVideoTarget == nil {
+		st.seenVideoTarget = map[string]cloud.FileEntry{}
+	}
+	if _, exists := st.seenVideoTarget[target]; exists {
+		st.mu.Unlock()
+		st.touchProgress()
+		return
+	}
+	st.seenVideoTarget[target] = entry
+	st.mu.Unlock()
+
+	// 增量同步模式快速检查：本地 strm 文件存在、非空且修改时间与远端 mtime 一致，直接跳过无需读磁盘
+	if st.syncType == model.StrmSyncTypeIncremental && entry.MTime > 0 {
+		if info, err := os.Stat(target); err == nil && info.Size() > 0 && info.ModTime().Unix() == entry.MTime {
+			st.mu.Lock()
+			st.rec.Skipped++
+			st.mu.Unlock()
+			st.touchProgress()
+			return
+		}
+	}
+
 	content, err := st.strmContent(entry, rel, ext)
 	if err != nil {
 		// 并发 worker 下 rec.Message 无锁写会有数据竞争，这里仅记录日志；
@@ -403,6 +789,11 @@ func (st *strmSyncState) handleVideo(entry cloud.FileEntry, rel, ext string) {
 		existing = string(data)
 	}
 	if existing == content {
+		// 对齐本地 strm 修改时间为远端 mtime，便于后续秒级比对
+		if entry.MTime > 0 {
+			mTime := time.Unix(entry.MTime, 0)
+			_ = os.Chtimes(target, mTime, mTime)
+		}
 		st.mu.Lock()
 		st.rec.Skipped++
 		st.mu.Unlock()
@@ -423,6 +814,10 @@ func (st *strmSyncState) handleVideo(entry cloud.FileEntry, rel, ext string) {
 		st.s.log.Warn("rename strm failed", zap.String("file", target), zap.Error(err))
 		return
 	}
+	if entry.MTime > 0 {
+		mTime := time.Unix(entry.MTime, 0)
+		_ = os.Chtimes(target, mTime, mTime)
+	}
 	st.mu.Lock()
 	st.rec.NewStrm++
 	st.mu.Unlock()
@@ -482,6 +877,36 @@ func (st *strmSyncState) recordRemoteMeta(entry cloud.FileEntry, rel string) {
 	st.mu.Unlock()
 }
 
+func (st *strmSyncState) flushPendingDownloads() {
+	st.mu.Lock()
+	if len(st.pendingDownloads) == 0 {
+		st.mu.Unlock()
+		return
+	}
+	batch := st.pendingDownloads
+	st.pendingDownloads = nil
+	st.mu.Unlock()
+
+	if err := st.s.repo.StrmDownload.CreateInBatches(st.ctx, batch, 100); err != nil {
+		st.s.log.Warn("batch enqueue strm download tasks failed", zap.Error(err))
+	}
+}
+
+func (st *strmSyncState) flushPendingUploads() {
+	st.mu.Lock()
+	if len(st.pendingUploads) == 0 {
+		st.mu.Unlock()
+		return
+	}
+	batch := st.pendingUploads
+	st.pendingUploads = nil
+	st.mu.Unlock()
+
+	if err := st.s.repo.StrmUpload.CreateInBatches(st.ctx, batch, 100); err != nil {
+		st.s.log.Warn("batch enqueue strm upload tasks failed", zap.Error(err))
+	}
+}
+
 // handleMeta 元数据入下载队列（本地已存在且大小一致则跳过）。
 func (st *strmSyncState) handleMeta(entry cloud.FileEntry, rel, ext string) {
 	st.recordRemoteMeta(entry, rel)
@@ -490,14 +915,40 @@ func (st *strmSyncState) handleMeta(entry cloud.FileEntry, rel, ext string) {
 	if err != nil {
 		return
 	}
+
+	st.mu.Lock()
+	if st.seenMetaTarget == nil {
+		st.seenMetaTarget = map[string]cloud.FileEntry{}
+	}
+	if _, exists := st.seenMetaTarget[target]; exists {
+		// 该本地目标路径在当前批次中已被处理（存在同名/重名冲突），直接忽略重复项，避免多份不同大小的文件在本地交替覆盖导致增量死循环
+		st.mu.Unlock()
+		st.touchProgress()
+		return
+	}
+	st.seenMetaTarget[target] = entry
+	st.mu.Unlock()
+
 	if info, err := os.Stat(target); err == nil && info.Size() == entry.Size {
 		st.touchProgress()
 		return
 	}
-	if st.taskExists("download", st.p.ID, target) {
+	st.mu.Lock()
+	if st.activeDownloadPaths == nil {
+		if active, err := st.s.repo.StrmDownload.GetActiveLocalPathMap(st.ctx, st.p.ID); err == nil {
+			st.activeDownloadPaths = active
+		} else {
+			st.activeDownloadPaths = map[string]bool{}
+		}
+	}
+	if st.activeDownloadPaths[target] {
+		st.mu.Unlock()
 		st.touchProgress()
 		return
 	}
+	st.activeDownloadPaths[target] = true
+	st.mu.Unlock()
+
 	task := &model.StrmDownloadTask{
 		SyncPathID: st.p.ID,
 		AccountID:  st.p.AccountID,
@@ -516,13 +967,16 @@ func (st *strmSyncState) handleMeta(entry cloud.FileEntry, rel, ext string) {
 	if st.p.Provider != model.StrmProvider115 {
 		task.RemoteRef = entry.ID
 	}
-	if err := st.s.repo.StrmDownload.Create(st.ctx, task); err != nil {
-		st.s.log.Warn("enqueue strm download task failed", zap.Error(err))
-		return
-	}
+
 	st.mu.Lock()
+	st.pendingDownloads = append(st.pendingDownloads, task)
+	shouldFlush := len(st.pendingDownloads) >= 100
 	st.rec.NewMeta++
 	st.mu.Unlock()
+
+	if shouldFlush {
+		st.flushPendingDownloads()
+	}
 	st.touchProgress()
 }
 
@@ -580,10 +1034,22 @@ func (st *strmSyncState) walkLocalSource() error {
 		if err != nil {
 			return nil
 		}
+		mTime := info.ModTime()
+		if st.syncType == model.StrmSyncTypeIncremental {
+			if tInfo, err := os.Stat(target); err == nil && tInfo.Size() > 0 && tInfo.ModTime().Unix() == mTime.Unix() {
+				st.mu.Lock()
+				st.rec.Skipped++
+				st.mu.Unlock()
+				st.touchProgress()
+				return nil
+			}
+		}
 		if data, err := os.ReadFile(target); err == nil && string(data) == content {
+			_ = os.Chtimes(target, mTime, mTime)
 			st.mu.Lock()
 			st.rec.Skipped++
 			st.mu.Unlock()
+			st.touchProgress()
 			return nil
 		}
 		if err := os.MkdirAll(filepath.Dir(target), 0o755); err != nil {
@@ -592,18 +1058,28 @@ func (st *strmSyncState) walkLocalSource() error {
 		tmp := target + ".tmp"
 		if err := os.WriteFile(tmp, []byte(content), 0o644); err == nil {
 			_ = os.Rename(tmp, target)
+			_ = os.Chtimes(target, mTime, mTime)
 		} else {
 			_ = os.Remove(tmp)
 		}
 		st.mu.Lock()
 		st.rec.NewStrm++
 		st.mu.Unlock()
+		st.touchProgress()
 		return nil
 	})
 }
 
 // scanLocalMetaForUpload 扫描本地元数据，与远端比对后入上传队列。
 func (st *strmSyncState) scanLocalMetaForUpload() error {
+	defer st.flushPendingUploads()
+	if st.activeUploadPaths == nil {
+		if active, err := st.s.repo.StrmUpload.GetActiveLocalPathMap(st.ctx, st.p.ID); err == nil {
+			st.activeUploadPaths = active
+		} else {
+			st.activeUploadPaths = map[string]bool{}
+		}
+	}
 	localRoot := filepath.Clean(st.p.LocalPath)
 	return filepath.WalkDir(localRoot, func(path string, d os.DirEntry, err error) error {
 		if err != nil {
@@ -640,26 +1116,35 @@ func (st *strmSyncState) scanLocalMetaForUpload() error {
 			// 网盘端已存在该元数据文件，跳过上传
 			return nil
 		}
-		if st.taskExists("upload", st.p.ID, path) {
+		st.mu.Lock()
+		if st.activeUploadPaths != nil && st.activeUploadPaths[path] {
+			st.mu.Unlock()
 			return nil
 		}
+		if st.activeUploadPaths != nil {
+			st.activeUploadPaths[path] = true
+		}
+		st.mu.Unlock()
+
 		task := &model.StrmUploadTask{
 			SyncPathID: st.p.ID,
 			AccountID:  st.p.AccountID,
 			Provider:   st.p.Provider,
 			FileName:   filepath.Base(rel),
 			LocalPath:  path,
-			RemotePath: st.remoteUploadPath(rel),
+			RemotePath: st.uploadRemoteTarget(rel),
 			Size:       info.Size(),
 			Status:     model.StrmTaskPending,
 		}
-		if err := st.s.repo.StrmUpload.Create(st.ctx, task); err != nil {
-			st.s.log.Warn("enqueue strm upload task failed", zap.Error(err))
-			return nil
-		}
 		st.mu.Lock()
+		st.pendingUploads = append(st.pendingUploads, task)
+		shouldFlush := len(st.pendingUploads) >= 100
 		st.rec.Uploaded++
 		st.mu.Unlock()
+
+		if shouldFlush {
+			st.flushPendingUploads()
+		}
 		return nil
 	})
 }
@@ -673,6 +1158,31 @@ func (st *strmSyncState) remoteUploadPath(rel string) string {
 	return root + "/" + rel
 }
 
+// uploadRemoteTarget 返回上传任务的目标远端描述。
+//   - 115：返回父目录 cid（供 PutFileNamed 定位），基于 dirPathToID 把父目录相对路径映射到 cid。
+//   - 网盘桥接（clouddrive2/openlist）：返回完整远端路径。
+func (st *strmSyncState) uploadRemoteTarget(rel string) string {
+	if st.p.Provider == model.StrmProvider115 {
+		dir := rel
+		if idx := strings.LastIndexByte(dir, '/'); idx >= 0 {
+			dir = dir[:idx]
+		} else {
+			dir = ""
+		}
+		if dir == "" {
+			// 文件在同步根目录下，父目录即 115 同步根目录 ID
+			return st.p.RemotePath
+		}
+		if cid, ok := st.dirPathToID[dir]; ok && cid != "" {
+			return cid
+		}
+		// 父目录未在缓存中（父目录可能本次未扫描到），降级为用户配置的同步根 cid，
+		// 由上传端尽力处理（可能失败记日志，不影响下载）。
+		return st.p.RemotePath
+	}
+	return st.remoteUploadPath(rel)
+}
+
 // taskExists 检查是否已有同目录、同目标的进行中/已完成任务（避免重复入队）。
 func (st *strmSyncState) taskExists(kind, syncPathID, localPath string) bool {
 	ctx := st.ctx
@@ -688,6 +1198,14 @@ func (st *strmSyncState) taskExists(kind, syncPathID, localPath string) bool {
 
 // pruneLocal 清理本地多余 .strm 与元数据（远端已不存在），可选删除空目录。
 func (st *strmSyncState) pruneLocal() error {
+	// 增量同步保护：本次远端扫描不完整（目录详情解析失败 / 文件父路径降级）时，
+	// seenVideo/seenMeta 覆盖不全，按"远端不存在"清理会误删刚下载或已存在的本地文件，
+	// 进而触发"下次增量重新下载"的循环。此时跳过清理，仅做进度落库。
+	if st.syncType == model.StrmSyncTypeIncremental && st.scanIncomplete.Load() {
+		st.s.log.Warn("strm 增量同步跳过清理：本次远端扫描不完整，prune 已禁用",
+			zap.String("path_id", st.p.ID))
+		return nil
+	}
 	localRoot := filepath.Clean(st.p.LocalPath)
 	var dirs []string
 	err := filepath.WalkDir(localRoot, func(path string, d os.DirEntry, err error) error {
@@ -748,12 +1266,16 @@ func (st *strmSyncState) pruneLocal() error {
 	return nil
 }
 
-// touchProgress 每处理若干个文件落库一次进度。
+// touchProgress 进度计数并限流防抖落库（避免高频写 SQLite 导致锁竞争）。
 func (st *strmSyncState) touchProgress() {
 	st.mu.Lock()
 	st.rec.Total++
 	st.processed++
-	flush := st.processed%100 == 0
+	now := time.Now()
+	flush := st.processed%100 == 0 || (st.processed%20 == 0 && now.Sub(st.lastProgressFlush) >= 2*time.Second)
+	if flush {
+		st.lastProgressFlush = now
+	}
 	st.mu.Unlock()
 	if flush {
 		st.flushProgress()
@@ -769,6 +1291,18 @@ func (st *strmSyncState) flushProgress() {
 	}
 }
 
+// updateSyncMessage 实时更新同步阶段提示信息，让前端界面清晰了解当前进度。
+func (st *strmSyncState) updateSyncMessage(msg string) {
+	st.mu.Lock()
+	st.rec.Message = msg
+	st.p.LastSyncMessage = msg
+	rec := *st.rec
+	p := *st.p
+	st.mu.Unlock()
+	_ = st.s.repo.StrmSyncRecord.Update(st.ctx, &rec)
+	_ = st.s.repo.StrmSyncPath.Update(st.ctx, &p)
+}
+
 // ─── 定时同步巡检 ──────────────────────────────────────────────────────────────
 
 func (s *StrmService) cronLoop(ctx context.Context) {

internal/service/strm_sync_test.go

@@ -2,10 +2,13 @@ package service
 
 import (
 	"context"
+	"net/http"
+	"net/http/httptest"
 	"net/url"
 	"os"
 	"path/filepath"
 	"strings"
+	"sync"
 	"testing"
 	"time"
 
@@ -18,6 +21,7 @@ import (
 	"github.com/ShukeBta/MMTL/internal/model"
 	"github.com/ShukeBta/MMTL/internal/repository"
 	"github.com/ShukeBta/MMTL/internal/service/cloud"
+	"github.com/ShukeBta/MMTL/internal/service/cloud115"
 )
 
 // testStrmService 构建带内存库的 StrmService。
@@ -35,7 +39,7 @@ func testStrmService(t *testing.T) *StrmService {
 		t.Cleanup(func() { _ = sqlDB.Close() })
 	}
 	if err := db.AutoMigrate(&model.StrmAccount{}, &model.StrmSyncPath{}, &model.StrmSyncRecord{},
-		&model.StrmDownloadTask{}, &model.StrmUploadTask{}, &model.Setting{}); err != nil {
+		&model.StrmDownloadTask{}, &model.StrmUploadTask{}, &model.StrmDirCache{}, &model.Setting{}); err != nil {
 		t.Fatal(err)
 	}
 	repos := repository.New(db)
@@ -159,6 +163,56 @@ func TestLocalStrmSync(t *testing.T) {
 	}
 }
 
+// TestStrmFullAndIncrementalSync 测试增量同步与全量同步模式切换及记录
+func TestStrmFullAndIncrementalSync(t *testing.T) {
+	svc := testStrmService(t)
+	src := t.TempDir()
+	out := t.TempDir()
+
+	writeFile(t, filepath.Join(src, "电影", "星际穿越.mkv"), "fake-video-data")
+
+	p := syncPathRecord(t, svc, model.StrmProviderLocal, src, out, true)
+
+	// 1. 默认触发增量同步
+	if err := svc.StartSync(context.Background(), p.ID, model.StrmSyncTypeIncremental); err != nil {
+		t.Fatal(err)
+	}
+	record := waitSyncDone(t, svc, p.ID, 10*time.Second)
+	if record.Status != model.StrmSyncRecordDone {
+		t.Fatalf("sync status = %s, message = %s", record.Status, record.Message)
+	}
+	if record.SyncType != model.StrmSyncTypeIncremental {
+		t.Fatalf("expected sync_type = incremental, got %s", record.SyncType)
+	}
+	if record.NewStrm != 1 {
+		t.Fatalf("expected 1 new strm, got %d", record.NewStrm)
+	}
+
+	// 2. 再次执行增量同步，应当跳过
+	if err := svc.StartSync(context.Background(), p.ID, model.StrmSyncTypeIncremental); err != nil {
+		t.Fatal(err)
+	}
+	record = waitSyncDone(t, svc, p.ID, 10*time.Second)
+	if record.SyncType != model.StrmSyncTypeIncremental {
+		t.Fatalf("expected sync_type = incremental, got %s", record.SyncType)
+	}
+	if record.Skipped != 1 {
+		t.Fatalf("expected 1 skipped, got %d", record.Skipped)
+	}
+
+	// 3. 执行全量同步
+	if err := svc.StartSync(context.Background(), p.ID, model.StrmSyncTypeFull); err != nil {
+		t.Fatal(err)
+	}
+	record = waitSyncDone(t, svc, p.ID, 10*time.Second)
+	if record.SyncType != model.StrmSyncTypeFull {
+		t.Fatalf("expected sync_type = full, got %s", record.SyncType)
+	}
+	if record.Status != model.StrmSyncRecordDone {
+		t.Fatalf("full sync failed: status = %s, message = %s", record.Status, record.Message)
+	}
+}
+
 // TestStrmCronMatches cron 表达式匹配。
 func TestStrmCronMatches(t *testing.T) {
 	cases := []struct {
@@ -446,3 +500,211 @@ func TestWalkRemoteConcurrent(t *testing.T) {
 		t.Errorf("生成的 .strm 数量 = %d，期望 5", strmCount)
 	}
 }
+
+// TestStrmBatchEnqueueAndConcurrentClaim 测试大规模批量入库及多协程并发认领无死锁
+func TestStrmBatchEnqueueAndConcurrentClaim(t *testing.T) {
+	svc := testStrmService(t)
+	ctx := context.Background()
+
+	// 1. 批量插入 200 个下载任务
+	tasks := make([]*model.StrmDownloadTask, 0, 200)
+	for i := 0; i < 200; i++ {
+		tasks = append(tasks, &model.StrmDownloadTask{
+			SyncPathID: "test-sync-path",
+			AccountID:  "test-acct",
+			Provider:   model.StrmProvider115,
+			FileName:   filepath.Base(string(rune('a'+i%26))) + ".nfo",
+			LocalPath:  filepath.Join(t.TempDir(), string(rune('a'+i%26)), "test.nfo"),
+			Status:     model.StrmTaskPending,
+		})
+	}
+	if err := svc.repo.StrmDownload.CreateInBatches(ctx, tasks, 50); err != nil {
+		t.Fatalf("CreateInBatches failed: %v", err)
+	}
+
+	// 2. 验证 ActiveLocalPathMap
+	activeMap, err := svc.repo.StrmDownload.GetActiveLocalPathMap(ctx, "test-sync-path")
+	if err != nil {
+		t.Fatalf("GetActiveLocalPathMap failed: %v", err)
+	}
+	if len(activeMap) == 0 {
+		t.Fatal("expected active local path map to have entries")
+	}
+
+	// 3. 模拟 6 个 worker 并发 ClaimPendingDownload
+	claimedCount := 0
+	var claimMu sync.Mutex
+	var wg sync.WaitGroup
+	for w := 0; w < 6; w++ {
+		wg.Add(1)
+		go func() {
+			defer wg.Done()
+			for {
+				batch, err := svc.repo.StrmDownload.ClaimPendingDownload(ctx, 10)
+				if err != nil {
+					t.Errorf("concurrent ClaimPendingDownload failed: %v", err)
+					return
+				}
+				if len(batch) == 0 {
+					return
+				}
+				claimMu.Lock()
+				claimedCount += len(batch)
+				claimMu.Unlock()
+			}
+		}()
+	}
+	wg.Wait()
+
+	if claimedCount != 200 {
+		t.Fatalf("expected all 200 tasks claimed, got %d", claimedCount)
+	}
+}
+
+// TestStrmDuplicateFileConflictResolution 测试远端存在多个同名不同大小文件时，本地确定性仲裁，避免增量死循环
+func TestStrmDuplicateFileConflictResolution(t *testing.T) {
+	svc := testStrmService(t)
+	localDir := t.TempDir()
+
+	p := &model.StrmSyncPath{
+		Base:         model.Base{ID: "dup-test-path"},
+		Provider:     model.StrmProvider115,
+		RemotePath:   "root",
+		LocalPath:    localDir,
+		DownloadMeta: true,
+	}
+
+	st := &strmSyncState{
+		s:               svc,
+		ctx:             context.Background(),
+		p:               p,
+		cfg:             &strmPathConfig{DownloadMeta: true, MetaExt: []string{"nfo"}},
+		rec:             &model.StrmSyncRecord{},
+		seenMeta:        map[string]bool{},
+		remoteMeta:      map[string]int64{},
+		seenMetaTarget:  map[string]cloud.FileEntry{},
+		seenVideoTarget: map[string]cloud.FileEntry{},
+	}
+
+	// 模拟远端同目录下存在两个同名不同大小的 nfo 文件 (115 历史重复上传)
+	// entry1: 较早文件 (MTime: 1000, Size: 100)
+	entry1 := cloud.FileEntry{ID: "f1", Name: "test.nfo", Size: 100, MTime: 1000, PickCode: "p1"}
+	// entry2: 较新文件 (MTime: 2000, Size: 200)
+	entry2 := cloud.FileEntry{ID: "f2", Name: "test.nfo", Size: 200, MTime: 2000, PickCode: "p2"}
+
+	// 第一次全量处理：两者都在列表中
+	st.handleMeta(entry1, "test.nfo", ".nfo")
+	st.handleMeta(entry2, "test.nfo", ".nfo")
+	st.flushPendingDownloads()
+
+	// 验证仲裁结果：最终只产生 1 个下载任务，且使用的是首个匹配项 (Size 100/p1)
+	tasks, _, err := svc.repo.StrmDownload.List(context.Background(), "", 1, 10)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if len(tasks) != 1 {
+		t.Fatalf("expected 1 download task after conflict resolution, got %d", len(tasks))
+	}
+	if tasks[0].Size != 100 || tasks[0].RemoteRef != "p1" {
+		t.Fatalf("expected task with size 100/p1, got size=%d ref=%s", tasks[0].Size, tasks[0].RemoteRef)
+	}
+
+	// 模拟该任务下载落盘完成
+	writeFile(t, filepath.Join(localDir, "test.nfo"), strings.Repeat("x", 100))
+
+	// 第二次增量同步：两者再次依次扫描
+	st2 := &strmSyncState{
+		s:               svc,
+		ctx:             context.Background(),
+		p:               p,
+		cfg:             &strmPathConfig{DownloadMeta: true, MetaExt: []string{"nfo"}},
+		rec:             &model.StrmSyncRecord{},
+		seenMeta:        map[string]bool{},
+		remoteMeta:      map[string]int64{},
+		seenMetaTarget:  map[string]cloud.FileEntry{},
+		seenVideoTarget: map[string]cloud.FileEntry{},
+	}
+	st2.handleMeta(entry1, "test.nfo", ".nfo")
+	st2.handleMeta(entry2, "test.nfo", ".nfo")
+	st2.flushPendingDownloads()
+
+	// 验证：不会新增任何下载任务，NewMeta 为 0，增量跳过
+	if st2.rec.NewMeta != 0 {
+		t.Fatalf("expected 0 new meta on incremental sync, got %d", st2.rec.NewMeta)
+	}
+
+}
+
+// TestWalk115FlatAbortsOnDirResolveFailure 回归测试：115 开放平台 token 失效/目录详情
+// 解析失败时，同步必须中止而不是带着塌缩的 rel 继续处理，否则会导致本地大量元数据
+// 被误判为"云端不存在"而重复下载/上传，甚至误删本地文件（用户反馈"云盘没动却重下重传"）。
+func TestWalk115FlatAbortsOnDirResolveFailure(t *testing.T) {
+	svc := testStrmService(t)
+	localDir := t.TempDir()
+
+	acct := &model.StrmAccount{
+		Name:     "fake115",
+		Provider: "cloud115",
+		Config:   "{}",
+		Enabled:  true,
+	}
+	if err := svc.repo.StrmAccount.Create(context.Background(), acct); err != nil {
+		t.Fatal(err)
+	}
+	p := &model.StrmSyncPath{
+		Base:       model.Base{ID: "abort-path"},
+		AccountID:  acct.ID,
+		Provider:   model.StrmProvider115,
+		RemotePath: "0",
+		LocalPath:  localDir,
+	}
+
+	// 115 mock：文件列表返回一个视频（父目录 999 不在缓存，需要 get_info），
+	// get_info 恒返回 access_token 格式错误（40140123）→ 目录树解析失败。
+	api := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		switch r.URL.Path {
+		case "/open/ufile/files":
+			w.Write([]byte(`{"state":true,"count":1,"data":[{"fid":"100","pid":"999","fc":1,"fn":"movie.mkv","pc":"pc1","upt":1700000000,"fs":1024}]}`))
+		case "/open/folder/get_info":
+			w.Write([]byte(`{"state":false,"code":40140123,"message":"access_token 格式错误"}`))
+		default:
+			t.Errorf("unexpected path %s", r.URL.Path)
+		}
+	}))
+	defer api.Close()
+
+	oldPro := cloud115.ProAPIBase
+	cloud115.ProAPIBase = api.URL
+	defer func() { cloud115.ProAPIBase = oldPro }()
+
+	oc := cloud115.NewOpenClient("app", "at", "rt")
+	st := &strmSyncState{
+		s:          svc,
+		ctx:        context.Background(),
+		p:          p,
+		provider:   cloud.NewOpenAPI115("app", "at", "rt"),
+		cfg:        &strmPathConfig{VideoExt: []string{"mkv"}, MetaExt: []string{"nfo"}, AddPath: 1, DownloadMeta: false},
+		rec:        &model.StrmSyncRecord{},
+		syncType:   model.StrmSyncTypeFull,
+		dirCache:   sync.Map{},
+		seenVideo:  map[string]bool{},
+		seenMeta:   map[string]bool{},
+		remoteMeta: map[string]int64{},
+	}
+	err := st.walk115Flat(oc)
+	if err == nil {
+		t.Fatal("expected walk115Flat to abort on dir-resolve failure, got nil error")
+	}
+
+	// 中止后不允许产生任何部分写入（本地不允许生成 .strm 文件）。
+	var strmCount int
+	_ = filepath.WalkDir(localDir, func(path string, d os.DirEntry, err error) error {
+		if err == nil && !d.IsDir() && strings.HasSuffix(d.Name(), ".strm") {
+			strmCount++
+		}
+		return nil
+	})
+	if strmCount != 0 {
+		t.Fatalf("expected no .strm written after abort, got %d", strmCount)
+	}
+}

internal/service/subtitle.go

@@ -1,8 +1,8 @@
 // Package service — subtitle handling.
 //
-// SubtitleService finds external subtitle files next to a media file AND
-// embedded text subtitle tracks inside the media container, exposing both as
-// WebVTT so the browser <track> element can load them directly.
+// SubtitleService finds external subtitle files next to a media file and
+// exposes them as WebVTT so the browser <track> element can load them
+// directly, or as the original bytes for Emby/Jellyfin clients.
 //
 // External-subtitle discovery rules (matching the legacy Python defaults):
 //
@@ -12,25 +12,17 @@
 //     ?lang=zh / ?lang=en.
 //
 // Supported extensions: .srt, .ass, .ssa, .vtt.
-//
-// Embedded subtitles are probed with ffprobe and exposed as
-// path "embedded:<stream-index>"; the browser endpoint extracts the stream
-// via ffmpeg into a cached .vtt file.
 package service
 
 import (
 	"context"
-	"encoding/json"
 	"errors"
 	"fmt"
 	"io"
 	"os"
-	"os/exec"
 	"path/filepath"
 	"regexp"
-	"strconv"
 	"strings"
-	"time"
 
 	"go.uber.org/zap"
 
@@ -71,6 +63,16 @@ var extToCodec = map[string]string{
 // relative; the caller should prepend /api/subtitles/<media_id>?path=...
 // when serializing for the frontend.
 func (s *SubtitleService) Discover(ctx context.Context, mediaID string) ([]SubtitleTrack, error) {
+	return s.discover(ctx, mediaID)
+}
+
+// DiscoverExternalOnly 只返回媒体旁边的外挂字幕文件，不含容器内嵌字幕轨。
+// Emby 字幕接口（/Videos/:id/Subtitles/...）用。
+func (s *SubtitleService) DiscoverExternalOnly(ctx context.Context, mediaID string) ([]SubtitleTrack, error) {
+	return s.discover(ctx, mediaID)
+}
+
+func (s *SubtitleService) discover(ctx context.Context, mediaID string) ([]SubtitleTrack, error) {
 	m, err := s.repo.Media.FindByID(ctx, mediaID)
 	if err != nil {
 		return nil, err
@@ -118,94 +120,9 @@ func (s *SubtitleService) Discover(ctx context.Context, mediaID string) ([]Subti
 			})
 		}
 	}
-
-	// 容器内嵌文本字幕轨（MKV/MP4 等封装内的字幕流）：本地真实文件才可
-	// 探测提取；cloud:// 与 .strm 媒体跳过。探测失败静默忽略（无 ffprobe
-	// 或没有字幕流都属正常）。
-	if embedded, ok := s.discoverEmbeddedTracks(ctx, m.Path); ok {
-		tracks = append(tracks, embedded...)
-	}
 	return tracks, nil
 }
 
-// embeddedCodecOK 只暴露可提取为 WebVTT 的文本字幕编解码器；位图字幕
-// （PGS/DVDSUB/DVBSUB）浏览器无法渲染，跳过。
-func embeddedCodecOK(codec string) bool {
-	switch strings.ToLower(codec) {
-	case "subrip", "srt", "mov_text", "text", "webvtt", "ass", "ssa", "ttml", "sami":
-		return true
-	default:
-		return false
-	}
-}
-
-// ffprobeSubtitleStream 是 ffprobe -show_streams 输出的字幕流字段。
-type ffprobeSubtitleStream struct {
-	Index int               `json:"index"`
-	Codec string            `json:"codec_name"`
-	Tags  map[string]string `json:"tags"`
-}
-
-type ffprobeSubtitleContainer struct {
-	Streams []ffprobeSubtitleStream `json:"streams"`
-}
-
-// discoverEmbeddedTracks 用 ffprobe 探测媒体容器内的文本字幕轨。
-// 返回 (tracks, ok)：ok=false 表示该媒体不适用（非本地文件/ffprobe 不可用）。
-func (s *SubtitleService) discoverEmbeddedTracks(ctx context.Context, mediaPath string) ([]SubtitleTrack, bool) {
-	if strings.HasPrefix(strings.ToLower(strings.TrimSpace(mediaPath)), "cloud://") ||
-		strings.HasSuffix(strings.ToLower(strings.TrimSpace(mediaPath)), ".strm") {
-		return nil, false
-	}
-	bin, err := resolveLocalExecutable(s.cfg.App.FFprobePath, "ffprobe")
-	if err != nil {
-		return nil, false
-	}
-	if _, err := os.Stat(mediaPath); err != nil {
-		return nil, false
-	}
-
-	probeCtx, cancel := context.WithTimeout(ctx, 15*time.Second)
-	defer cancel()
-	cmd := exec.CommandContext(probeCtx, bin, // #nosec G204 -- bin resolved by resolveLocalExecutable; args are fixed probes.
-		"-v", "error",
-		"-select_streams", "s",
-		"-show_entries", "stream=index,codec_name:stream_tags=language,title",
-		"-of", "json",
-		"--", mediaPath,
-	)
-	out, err := cmd.Output()
-	if err != nil {
-		return nil, false
-	}
-	var container ffprobeSubtitleContainer
-	if err := json.Unmarshal(out, &container); err != nil {
-		return nil, false
-	}
-
-	tracks := make([]SubtitleTrack, 0, len(container.Streams))
-	for _, stream := range container.Streams {
-		if !embeddedCodecOK(stream.Codec) {
-			continue
-		}
-		lang := strings.ToLower(strings.TrimSpace(stream.Tags["language"]))
-		if lang == "" {
-			lang = "und"
-		}
-		label := stream.Tags["title"]
-		if label == "" {
-			label = lang
-		}
-		tracks = append(tracks, SubtitleTrack{
-			Lang:  lang,
-			Label: "内置字幕 · " + label,
-			Path:  "embedded:" + strconv.Itoa(stream.Index),
-			Codec: stream.Codec,
-		})
-	}
-	return tracks, true
-}
-
 // langTag matches the .zh / .zh-cn / .chs language sub-extensions.
 var langTag = regexp.MustCompile(`(?i)\.([a-z]{2,3}(?:[-_][a-z]{2,4})?)$`)
 
@@ -222,18 +139,13 @@ func detectLang(name, base string) string {
 }
 
 // Serve writes the subtitle file as WebVTT (.vtt). SRT/SSA files are
-// converted minimally on the fly; embedded container tracks (path
-// "embedded:<index>") are extracted via ffmpeg into a cached .vtt.
-// Returns ErrSubtitleNotFound when the path is rejected (path traversal /
-// not in the media directory).
+// converted minimally on the fly. Returns ErrSubtitleNotFound when the
+// path is rejected (path traversal / not in the media directory).
 func (s *SubtitleService) Serve(ctx context.Context, mediaID, sub string, w io.Writer) error {
 	m, err := s.repo.Media.FindByID(ctx, mediaID)
 	if err != nil || m == nil {
 		return errors.New("media not found")
 	}
-	if strings.HasPrefix(sub, "embedded:") {
-		return s.ServeEmbeddedToVTT(ctx, m.Path, sub, w)
-	}
 	abs, err := filepath.Abs(sub)
 	if err != nil {
 		return err
@@ -266,81 +178,6 @@ func (s *SubtitleService) Serve(ctx context.Context, mediaID, sub string, w io.W
 	return err
 }
 
-// embeddedSubtitleCachePath 内嵌字幕提取后的 WebVTT 缓存路径
-// （按媒体路径哈希 + 轨道号定位，跨媒体互不干扰）。
-func (s *SubtitleService) embeddedSubtitleCachePath(mediaPath string, idx int) string {
-	hash := fmt.Sprintf("%x", fnvHash(mediaPath))
-	return filepath.Join(s.cfg.Cache.CacheDir, "subs", hash, fmt.Sprintf("s%d.vtt", idx))
-}
-
-// ServeEmbeddedToVTT 把容器内第 idx 个字幕轨提取为 WebVTT 输出。
-// 提取结果缓存在 cache 目录，媒体文件更新（mtime 变化）后自动重新提取。
-func (s *SubtitleService) ServeEmbeddedToVTT(ctx context.Context, mediaPath, streamRef string, w io.Writer) error {
-	idx, err := strconv.Atoi(strings.TrimPrefix(streamRef, "embedded:"))
-	if err != nil || idx < 0 {
-		return errors.New("invalid embedded subtitle index")
-	}
-	ffmpegBin, err := resolveLocalExecutable(s.cfg.App.FFmpegPath, "ffmpeg")
-	if err != nil {
-		return fmt.Errorf("ffmpeg 不可用，无法提取内嵌字幕：%w", err)
-	}
-	info, err := os.Stat(mediaPath)
-	if err != nil {
-		return errors.New("media file not found")
-	}
-
-	cachePath := s.embeddedSubtitleCachePath(mediaPath, idx)
-
-	if cached, statErr := os.Stat(cachePath); statErr == nil && !info.ModTime().After(cached.ModTime()) {
-		f, openErr := os.Open(cachePath) // #nosec G304 -- cachePath is generated under the cache dir.
-		if openErr == nil {
-			defer f.Close()
-			_, copyErr := io.Copy(w, f)
-			return copyErr
-		}
-	}
-
-	// 缓存未命中或媒体已更新：ffmpeg 提取到临时文件后原子改名。
-	if err := os.MkdirAll(filepath.Dir(cachePath), 0o750); err != nil {
-		return err
-	}
-	tmp := cachePath + ".tmp"
-	extractCtx, cancel := context.WithTimeout(ctx, 60*time.Second)
-	defer cancel()
-	cmd := exec.CommandContext(extractCtx, ffmpegBin, // #nosec G204 -- bin resolved by resolveLocalExecutable; args fixed extraction.
-		"-v", "error", "-y",
-		"-i", mediaPath,
-		"-map", "0:s:"+strconv.Itoa(idx),
-		"-f", "webvtt",
-		tmp,
-	)
-	if out, runErr := cmd.CombinedOutput(); runErr != nil {
-		_ = os.Remove(tmp)
-		return fmt.Errorf("提取内嵌字幕失败（轨道 %d，可能为位图字幕或轨道无效）：%s", idx, strings.TrimSpace(string(out)))
-	}
-	if err := os.Rename(tmp, cachePath); err != nil {
-		_ = os.Remove(tmp)
-		return err
-	}
-	f, err := os.Open(cachePath) // #nosec G304 -- cachePath is generated under the cache dir.
-	if err != nil {
-		return err
-	}
-	defer f.Close()
-	_, err = io.Copy(w, f)
-	return err
-}
-
-// fnvHash 简单 32 位 FNV-1a 哈希，用于生成稳定的缓存子目录名。
-func fnvHash(s string) uint32 {
-	var h uint32 = 2166136261
-	for i := 0; i < len(s); i++ {
-		h ^= uint32(s[i])
-		h *= 16777619
-	}
-	return h
-}
-
 // ServeRaw writes the subtitle file in its original format without any
 // WebVTT conversion. Emby/Jellyfin clients advertise the source codec (ASS,
 // subrip, etc.) in MediaStreams, then fetch the subtitle bytes via the
@@ -368,4 +205,4 @@ func (s *SubtitleService) ServeRaw(ctx context.Context, mediaID, sub string, w i
 	defer f.Close()
 	_, err = io.Copy(w, f)
 	return err
-}
+}

internal/service/tmdb_search.go

@@ -57,7 +57,7 @@ func (t *TMDbProvider) searchMovieCandidates(ctx context.Context, query string,
 
 	apiKey := t.resolveAPIKey(ctx)
 	if apiKey == "" {
-		return nil, nil
+		return nil, errors.New("TMDb API Key 未配置，请先在「系统设置 → API配置」中填写")
 	}
 	base := t.resolveBaseURL(ctx)
 
@@ -65,7 +65,7 @@ func (t *TMDbProvider) searchMovieCandidates(ctx context.Context, query string,
 	q.Set("api_key", apiKey)
 	q.Set("query", query)
 	q.Set("language", language)
-	q.Set("include_adult", "false")
+	q.Set("include_adult", "true")
 	if year > 0 {
 		q.Set("year", fmt.Sprintf("%d", year))
 	}
@@ -79,6 +79,12 @@ func (t *TMDbProvider) searchMovieCandidates(ctx context.Context, query string,
 	if err := t.getJSON(ctx, u, &p); err != nil {
 		return nil, err
 	}
+	if len(p.Results) == 0 && year > 0 {
+		// 移除年份限制重试一次，避免年份微小差异（如 2019 vs 2020）导致无搜索结果
+		q.Del("year")
+		u = base + "/search/movie?" + q.Encode()
+		_ = t.getJSON(ctx, u, &p)
+	}
 	if len(p.Results) == 0 {
 		return nil, nil
 	}
@@ -137,7 +143,7 @@ func (t *TMDbProvider) searchTVCandidates(ctx context.Context, query string, yea
 
 	apiKey := t.resolveAPIKey(ctx)
 	if apiKey == "" {
-		return nil, nil
+		return nil, errors.New("TMDb API Key 未配置，请先在「系统设置 → API配置」中填写")
 	}
 	base := t.resolveBaseURL(ctx)
 
@@ -145,7 +151,7 @@ func (t *TMDbProvider) searchTVCandidates(ctx context.Context, query string, yea
 	q.Set("api_key", apiKey)
 	q.Set("query", query)
 	q.Set("language", language)
-	q.Set("include_adult", "false")
+	q.Set("include_adult", "true")
 	if year > 0 {
 		q.Set("first_air_date_year", fmt.Sprintf("%d", year))
 	}
@@ -159,6 +165,12 @@ func (t *TMDbProvider) searchTVCandidates(ctx context.Context, query string, yea
 	if err := t.getJSON(ctx, u, &p); err != nil {
 		return nil, err
 	}
+	if len(p.Results) == 0 && year > 0 {
+		// 移除年份限制重试一次，避免年份微小差异导致无搜索结果
+		q.Del("first_air_date_year")
+		u = base + "/search/tv?" + q.Encode()
+		_ = t.getJSON(ctx, u, &p)
+	}
 	if len(p.Results) == 0 {
 		return nil, nil
 	}

internal/service/visibility.go

@@ -123,12 +123,20 @@ func LibraryVisibleForUser(ctx context.Context, repo *repository.Container, lib
 		return false
 	}
 	if repo != nil && repo.DB != nil {
-		var count int64
+		var totalCount int64
 		_ = repo.DB.WithContext(ctx).Model(&model.Media{}).
-			Where("library_id = ? AND nsfw = ?", lib.ID, true).
-			Count(&count).Error
-		if count > 0 {
-			return false
+			Where("library_id = ?", lib.ID).
+			Count(&totalCount).Error
+		if totalCount > 0 {
+			var nsfwCount int64
+			_ = repo.DB.WithContext(ctx).Model(&model.Media{}).
+				Where("library_id = ? AND nsfw = ?", lib.ID, true).
+				Count(&nsfwCount).Error
+			// 仅当整库媒体全部为成人内容（纯成人库）时才隐藏整库；
+			// 含有普通内容的混合媒体库保持库本身可见，具体 NSFW 条目在媒体列表内过滤。
+			if nsfwCount == totalCount {
+				return false
+			}
 		}
 	}
 	return true

web/embed.go

@@ -0,0 +1,23 @@
+// Package webui embeds the built React SPA so a single binary can serve the
+// UI without requiring a separate web/dist directory on disk. The embed
+// happens at compile time, so `web/dist` must exist before `go build` runs
+// (the CI pipeline builds it via `npm run build` first).
+package webui
+
+import (
+	"embed"
+	"io/fs"
+)
+
+//go:embed all:dist
+var distFS embed.FS
+
+// DistFS returns the embedded SPA build artifacts rooted at the directory
+// containing index.html (i.e. web/dist).
+func DistFS() fs.FS {
+	sub, err := fs.Sub(distFS, "dist")
+	if err != nil {
+		panic(err)
+	}
+	return sub
+}