Deduplicate organize by multi-source metadata

internal/model/model.go

@@ -91,6 +91,8 @@ type Media struct {
 	ScrapeStatus string  `gorm:"size:16;default:pending" json:"scrape_status"`
 	TMDbID       int     `json:"tmdb_id"`
 	BangumiID    int     `json:"bangumi_id"`
+	DoubanID     string  `gorm:"column:douban_id;size:32" json:"douban_id,omitempty"`
+	TheTVDBID    string  `gorm:"column:thetvdb_id;size:64" json:"thetvdb_id,omitempty"`
 	Languages    string  `gorm:"size:64"  json:"languages,omitempty"` // 逗号分隔的 ISO 639-1 代码，如 "zh,en"
 	Countries    string  `gorm:"size:128" json:"countries,omitempty"` // 逗号分隔的 ISO 3166-1，如 "CN,US"
 	Genres       string  `gorm:"size:255" json:"genres,omitempty"`    // 逗号分隔的类型名，如 "Action,Animation"
@@ -150,6 +152,8 @@ type Series struct {
 	Year        int     `json:"year"`
 	TMDbID      int     `json:"tmdb_id"`
 	BangumiID   int     `json:"bangumi_id"`
+	DoubanID    string  `gorm:"column:douban_id;size:32" json:"douban_id,omitempty"`
+	TheTVDBID   string  `gorm:"column:thetvdb_id;size:64" json:"thetvdb_id,omitempty"`
 }
 
 // PlaybackHistory 记录当前播放位置以支持续播。

internal/repository/repository.go

@@ -385,6 +385,12 @@ func (r *MediaRepository) upsert(ctx context.Context, m *model.Media) error {
 		if m.BangumiID > 0 {
 			setIfChanged(updates, "bangumi_id", existing.BangumiID, m.BangumiID)
 		}
+		if m.DoubanID != "" {
+			setIfChanged(updates, "douban_id", existing.DoubanID, m.DoubanID)
+		}
+		if m.TheTVDBID != "" {
+			setIfChanged(updates, "thetvdb_id", existing.TheTVDBID, m.TheTVDBID)
+		}
 		if m.Languages != "" {
 			setIfChanged(updates, "languages", existing.Languages, m.Languages)
 		}

internal/service/cloud_metadata.go

@@ -206,6 +206,12 @@ func mergeCloudMetadata(dst, src *LocalMetadata) *LocalMetadata {
 	if src.TMDbID > 0 {
 		dst.TMDbID = src.TMDbID
 	}
+	if src.DoubanID != "" {
+		dst.DoubanID = src.DoubanID
+	}
+	if src.TheTVDBID != "" {
+		dst.TheTVDBID = src.TheTVDBID
+	}
 	if src.SeasonNum > 0 {
 		dst.SeasonNum = src.SeasonNum
 	}

internal/service/douban.go

@@ -115,6 +115,23 @@ func (d *DoubanProvider) Search(ctx context.Context, query string) (*DoubanMatch
 	}, nil
 }
 
+func (d *DoubanProvider) SearchMatch(ctx context.Context, query string) (*Match, error) {
+	got, err := d.Search(ctx, query)
+	if err != nil || got == nil {
+		return nil, err
+	}
+	match := &Match{
+		DoubanID:  got.DoubanID,
+		Title:     got.Title,
+		PosterURL: got.Img,
+		Rating:    got.Rating,
+	}
+	if len(got.Year) >= 4 {
+		_, _ = fmt.Sscanf(got.Year[:4], "%d", &match.Year)
+	}
+	return match, nil
+}
+
 func (d *DoubanProvider) setHeaders(req *http.Request) {
 	req.Header.Set("User-Agent", userAgents[secureRandomIntn(len(userAgents))])
 	req.Header.Set("Referer", "https://movie.douban.com/")

internal/service/duplicate.go

@@ -1,10 +1,13 @@
 // Package service — duplicate-file finder.
 //
-// DuplicateService computes a sparse-sample SHA-256 (head + middle + tail,
-// 1 MiB each, plus the file size to break collisions) for every media
-// file and groups identical hashes into "duplicate sets". The first row
-// (preferring scraped + larger files) is kept as the primary; the rest
-// get is_duplicate = true and duplicate_of pointing at the primary.
+// DuplicateService finds duplicate media by two signals:
+//
+//   - external identity: same TMDb / Bangumi / Douban / TheTVDB id and, for
+//     episodes, same season+episode;
+//   - sparse file hash: same head + middle + tail SHA-256 and same size.
+//
+// The first row (preferring scraped + larger files) is kept as the primary;
+// the rest get is_duplicate = true and duplicate_of pointing at the primary.
 //
 // Why sparse: a full hash on a 50 GB Blu-ray remux takes minutes; the
 // 3-window 3 MiB sample is enough to differentiate real-world copies
@@ -121,34 +124,19 @@ func (d *DuplicateService) Detect(ctx context.Context, libraryID string) (*Repor
 		groups[r.FileHash] = append(groups[r.FileHash], r)
 	}
 
+	markedIDs := map[string]struct{}{}
 	for hash, group := range groups {
 		if len(group) < 2 {
 			continue
 		}
-		primary := pickPrimary(group)
-		dupes := make([]model.Media, 0, len(group)-1)
-		for _, m := range group {
-			if m.ID == primary.ID {
-				continue
-			}
-			dupes = append(dupes, m)
-			if err := d.repo.DB.WithContext(ctx).
-				Model(&model.Media{}).
-				Where("id = ?", m.ID).
-				Updates(map[string]any{
-					"is_duplicate": true,
-					"duplicate_of": primary.ID,
-				}).Error; err != nil {
-				d.log.Warn("dup mark failed", zap.Error(err))
-				continue
-			}
-			rep.ItemsMarked++
+		d.markDuplicateGroup(ctx, rep, hash, group, markedIDs)
+	}
+
+	for key, group := range groupByExternalIdentity(rows) {
+		if len(group) < 2 {
+			continue
 		}
-		rep.Groups = append(rep.Groups, Group{
-			Hash:       hash,
-			Primary:    primary,
-			Duplicates: dupes,
-		})
+		d.markDuplicateGroup(ctx, rep, key, group, markedIDs)
 	}
 	rep.GroupsFound = len(rep.Groups)
 	if d.hub != nil {
@@ -161,6 +149,72 @@ func (d *DuplicateService) Detect(ctx context.Context, libraryID string) (*Repor
 	return rep, nil
 }
 
+func (d *DuplicateService) markDuplicateGroup(ctx context.Context, rep *Report, key string, group []model.Media, markedIDs map[string]struct{}) {
+	primary := pickPrimary(group)
+	dupes := make([]model.Media, 0, len(group)-1)
+	for _, m := range group {
+		if m.ID == primary.ID || m.DuplicateOf == primary.ID {
+			continue
+		}
+		if _, ok := markedIDs[m.ID]; ok {
+			continue
+		}
+		dupes = append(dupes, m)
+		if err := d.repo.DB.WithContext(ctx).
+			Model(&model.Media{}).
+			Where("id = ?", m.ID).
+			Updates(map[string]any{
+				"is_duplicate": true,
+				"duplicate_of": primary.ID,
+			}).Error; err != nil {
+			d.log.Warn("dup mark failed", zap.Error(err))
+			continue
+		}
+		markedIDs[m.ID] = struct{}{}
+		rep.ItemsMarked++
+	}
+	if len(dupes) == 0 {
+		return
+	}
+	rep.Groups = append(rep.Groups, Group{
+		Hash:       key,
+		Primary:    primary,
+		Duplicates: dupes,
+	})
+}
+
+func groupByExternalIdentity(rows []model.Media) map[string][]model.Media {
+	groups := map[string][]model.Media{}
+	for _, row := range rows {
+		key := mediaExternalIdentityKey(row)
+		if key == "" {
+			continue
+		}
+		groups[key] = append(groups[key], row)
+	}
+	return groups
+}
+
+func mediaExternalIdentityKey(row model.Media) string {
+	var key string
+	switch {
+	case row.TMDbID > 0:
+		key = fmt.Sprintf("tmdb:%d", row.TMDbID)
+	case row.BangumiID > 0:
+		key = fmt.Sprintf("bangumi:%d", row.BangumiID)
+	case row.DoubanID != "":
+		key = "douban:" + row.DoubanID
+	case row.TheTVDBID != "":
+		key = "thetvdb:" + row.TheTVDBID
+	default:
+		return ""
+	}
+	if row.SeasonNum > 0 || row.EpisodeNum > 0 {
+		key += fmt.Sprintf(":s%d:e%d", row.SeasonNum, row.EpisodeNum)
+	}
+	return key
+}
+
 // Current returns duplicate groups already marked in the database. It keeps
 // the UI useful after a prior scan and avoids requiring POST on page load.
 func (d *DuplicateService) Current(ctx context.Context, libraryID string) (*Report, error) {

internal/service/duplicate_test.go

@@ -0,0 +1,71 @@
+package service
+
+import (
+	"path/filepath"
+	"testing"
+
+	"go.uber.org/zap"
+
+	"github.com/ShukeBta/MediaStationGo/internal/model"
+)
+
+func TestDuplicateDetectMarksExternalIdentityDuplicates(t *testing.T) {
+	repos := newOrganizerTestRepo(t)
+	root := t.TempDir()
+	firstPath := filepath.Join(root, "show-a.mkv")
+	secondPath := filepath.Join(root, "show-b.mkv")
+	writeOrgFile(t, firstPath, "first-release")
+	writeOrgFile(t, secondPath, "second-release")
+
+	lib := model.Library{Name: "剧集", Path: root, Type: "tv", Enabled: true}
+	if err := repos.Library.Create(t.Context(), &lib); err != nil {
+		t.Fatal(err)
+	}
+	first := model.Media{
+		LibraryID:    lib.ID,
+		Title:        "间谍过家家",
+		Path:         firstPath,
+		SizeBytes:    13,
+		SeasonNum:    1,
+		EpisodeNum:   1,
+		TMDbID:       12345,
+		ScrapeStatus: "matched",
+	}
+	second := model.Media{
+		LibraryID:    lib.ID,
+		Title:        "Spy Family",
+		Path:         secondPath,
+		SizeBytes:    14,
+		SeasonNum:    1,
+		EpisodeNum:   1,
+		TMDbID:       12345,
+		ScrapeStatus: "matched",
+	}
+	if err := repos.DB.Create(&first).Error; err != nil {
+		t.Fatal(err)
+	}
+	if err := repos.DB.Create(&second).Error; err != nil {
+		t.Fatal(err)
+	}
+
+	report, err := NewDuplicateService(zap.NewNop(), repos, nil).Detect(t.Context(), lib.ID)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if report.ItemsMarked != 1 || report.GroupsFound != 1 {
+		t.Fatalf("report = %#v, want one external identity duplicate", report)
+	}
+	var rows []model.Media
+	if err := repos.DB.Find(&rows).Error; err != nil {
+		t.Fatal(err)
+	}
+	marked := 0
+	for _, row := range rows {
+		if row.IsDuplicate && row.DuplicateOf != "" {
+			marked++
+		}
+	}
+	if marked != 1 {
+		t.Fatalf("marked duplicate rows = %d, want 1; rows=%#v", marked, rows)
+	}
+}

internal/service/local_metadata.go

@@ -25,6 +25,8 @@ type LocalMetadata struct {
 	PosterURL    string
 	BackdropURL  string
 	TMDbID       int
+	DoubanID     string
+	TheTVDBID    string
 	SeasonNum    int
 	EpisodeNum   int
 	Genres       string
@@ -257,6 +259,8 @@ func metadataFromDoc(doc *nfoDocument, baseDir string, seriesLike bool) *LocalMe
 		PosterURL:    firstRemoteURL(baseDir, nfoPosterValues(doc)...),
 		BackdropURL:  firstRemoteURL(baseDir, nfoBackdropValues(doc)...),
 		TMDbID:       doc.TMDbID,
+		DoubanID:     externalIDFromUniqueIDs(doc.UniqueIDs, "douban"),
+		TheTVDBID:    externalIDFromUniqueIDs(doc.UniqueIDs, "thetvdb", "tvdb"),
 		SeasonNum:    doc.Season,
 		EpisodeNum:   doc.Episode,
 		Genres:       joinNFOValues(adultAwareGenres(doc)),
@@ -391,6 +395,12 @@ func mergeEpisodeMetadata(dst, episode *LocalMetadata, doc *nfoDocument) {
 	if episode.TMDbID > 0 {
 		dst.TMDbID = episode.TMDbID
 	}
+	if episode.DoubanID != "" {
+		dst.DoubanID = episode.DoubanID
+	}
+	if episode.TheTVDBID != "" {
+		dst.TheTVDBID = episode.TheTVDBID
+	}
 	if episode.SeasonNum > 0 {
 		dst.SeasonNum = episode.SeasonNum
 	}
@@ -409,13 +419,24 @@ func mergeEpisodeMetadata(dst, episode *LocalMetadata, doc *nfoDocument) {
 }
 
 func tmdbIDFromUniqueIDs(ids []nfoUniqueID) int {
+	value := externalIDFromUniqueIDs(ids, "tmdb")
+	if value == "" {
+		return 0
+	}
+	v, _ := strconv.Atoi(value)
+	return v
+}
+
+func externalIDFromUniqueIDs(ids []nfoUniqueID, types ...string) string {
 	for _, id := range ids {
-		if strings.EqualFold(strings.TrimSpace(id.Type), "tmdb") {
-			v, _ := strconv.Atoi(strings.TrimSpace(id.Value))
-			return v
+		idType := strings.TrimSpace(id.Type)
+		for _, typ := range types {
+			if strings.EqualFold(idType, typ) {
+				return strings.TrimSpace(id.Value)
+			}
 		}
 	}
-	return 0
+	return ""
 }
 
 func firstRemoteURL(baseDir string, values ...string) string {

internal/service/organizer_directory.go

@@ -264,7 +264,9 @@ func (o *OrganizerService) organizeSourceFile(ctx context.Context, src, sourceRo
 	if layout.MediaType == "" {
 		layout.MediaType = o.inferMediaTypeForSourceFile(src, title, season, episode)
 	}
+	var metadataMatch *Match
 	if match := o.lookupOrganizeMetadata(ctx, src, sourceRoot, layout.MediaType, title, year, season, episode, metadataCache); match != nil {
+		metadataMatch = match
 		if matchedTitle := sanitizeFilename(strings.TrimSpace(match.Title)); matchedTitle != "" {
 			title = matchedTitle
 			parsedTitle = strings.TrimSpace(match.Title)
@@ -321,9 +323,10 @@ func (o *OrganizerService) organizeSourceFile(ctx context.Context, src, sourceRo
 
 	// 去重候选：合并「目的地媒体库已扫描入库的同一媒体（按标题/年份/季集匹配，
 	// 不受目录大小写或布局影响）」与「目标文件夹内已存在的同名视频文件」。
+	externalExisting := o.existingByExternalIdentity(ctx, destRoot, metadataMatch, season, episode)
 	identityExisting := o.existingByIdentity(ctx, destRoot, parsedTitle, year, season, episode)
 	folderExisting := o.existingByFolder(destDir, episodeTag)
-	existing := mergeExistingVersionPaths(identityExisting, folderExisting)
+	existing := mergeExistingVersionPaths(externalExisting, identityExisting, folderExisting)
 	if len(existing) > 0 {
 		srcArea := o.resolutionArea(ctx, src)
 		bestArea := 0
@@ -357,7 +360,7 @@ func (o *OrganizerService) organizeSourceFile(ctx context.Context, src, sourceRo
 		}
 		// 去重：目的地已存在同一媒体且不低于来源分辨率，跳过不再整理过去。
 		reason := organizeSkipTargetExists
-		if len(identityExisting) > 0 || o.allExistingPathsInDB(ctx, existing) {
+		if len(externalExisting) > 0 || len(identityExisting) > 0 || o.allExistingPathsInDB(ctx, existing) {
 			reason = organizeSkipDuplicateLibrary
 		}
 		o.log.Debug("organize skip duplicate",
@@ -501,7 +504,9 @@ func (o *OrganizerService) lookupOrganizeMetadata(ctx context.Context, src, sour
 					zap.String("title", match.Title),
 					zap.Int("year", match.Year),
 					zap.Int("tmdb_id", match.TMDbID),
-					zap.Int("bangumi_id", match.BangumiID))
+					zap.Int("bangumi_id", match.BangumiID),
+					zap.String("douban_id", match.DoubanID),
+					zap.String("thetvdb_id", match.TheTVDBID))
 			}
 			return match
 		}
@@ -522,6 +527,8 @@ func organizeMatchFromLocalMetadata(local *LocalMetadata) *Match {
 		Year:         local.Year,
 		Rating:       local.Rating,
 		TMDbID:       local.TMDbID,
+		DoubanID:     local.DoubanID,
+		TheTVDBID:    local.TheTVDBID,
 		NSFW:         local.NSFW,
 	}
 	if local.Genres != "" {
@@ -980,6 +987,50 @@ func (o *OrganizerService) existingByIdentity(ctx context.Context, destRoot, tit
 	return out
 }
 
+func (o *OrganizerService) existingByExternalIdentity(ctx context.Context, destRoot string, match *Match, season, episode int) []string {
+	if o.repo == nil || o.repo.DB == nil || match == nil {
+		return nil
+	}
+	var conds []string
+	var args []any
+	if match.TMDbID > 0 {
+		conds = append(conds, "tm_db_id = ?")
+		args = append(args, match.TMDbID)
+	}
+	if match.BangumiID > 0 {
+		conds = append(conds, "bangumi_id = ?")
+		args = append(args, match.BangumiID)
+	}
+	if strings.TrimSpace(match.DoubanID) != "" {
+		conds = append(conds, "douban_id = ?")
+		args = append(args, strings.TrimSpace(match.DoubanID))
+	}
+	if strings.TrimSpace(match.TheTVDBID) != "" {
+		conds = append(conds, "thetvdb_id = ?")
+		args = append(args, strings.TrimSpace(match.TheTVDBID))
+	}
+	if len(conds) == 0 {
+		return nil
+	}
+	q := o.repo.DB.WithContext(ctx).Model(&model.Media{}).
+		Where("deleted_at IS NULL").
+		Where("("+strings.Join(conds, " OR ")+")", args...)
+	if season > 0 || episode > 0 {
+		q = q.Where("season_num = ? AND episode_num = ?", season, episode)
+	}
+	var rows []model.Media
+	if err := q.Find(&rows).Error; err != nil {
+		return nil
+	}
+	var out []string
+	for _, row := range rows {
+		if row.Path != "" && pathWithin(row.Path, destRoot) {
+			out = append(out, row.Path)
+		}
+	}
+	return out
+}
+
 // existingByFolder returns video files already present in destDir that
 // represent the same media. For an episode (episodeTag != "") it matches files
 // carrying the same SxxExx tag; for a movie it matches every video file in the

internal/service/organizer_scrape_test.go

@@ -1,6 +1,9 @@
 package service
 
 import (
+	"encoding/json"
+	"net/http"
+	"net/http/httptest"
 	"os"
 	"path/filepath"
 	"testing"
@@ -102,6 +105,107 @@ func TestOrganizeDirectoryUsesScraperMatchBeforeRename(t *testing.T) {
 	}
 }
 
+func TestOrganizeDirectoryDedupsByExternalIDBeforeRename(t *testing.T) {
+	scraper, repos, closeServer := newTestScraper(t)
+	defer closeServer()
+
+	root := t.TempDir()
+	src := filepath.Join(root, "downloads")
+	dest := filepath.Join(root, "media")
+	sourceFile := filepath.Join(src, "Spy.x.Family.S01E01.2022.2160p.mkv")
+	writeOrgFile(t, sourceFile, "episode")
+
+	existingPath := filepath.Join(dest, "电视剧", "旧错误名", "Season 01", "旧错误名 - S01E01.mkv")
+	writeOrgFile(t, existingPath, "existing")
+	lib := model.Library{Name: "剧集", Path: filepath.Join(dest, "电视剧"), Type: "tv", Enabled: true}
+	if err := repos.Library.Create(t.Context(), &lib); err != nil {
+		t.Fatal(err)
+	}
+	if err := repos.DB.Create(&model.Media{
+		LibraryID:    lib.ID,
+		Title:        "旧错误名",
+		Path:         existingPath,
+		SeasonNum:    1,
+		EpisodeNum:   1,
+		TMDbID:       12345,
+		ScrapeStatus: "matched",
+	}).Error; err != nil {
+		t.Fatal(err)
+	}
+
+	organizer := NewOrganizerService(&config.Config{}, zap.NewNop(), repos)
+	organizer.SetScraper(scraper)
+	res, err := organizer.OrganizeDirectory(t.Context(), OrganizeOptions{
+		SourcePath:   src,
+		DestPath:     dest,
+		TransferMode: TransferCopy,
+		MediaType:    "tv",
+	})
+	if err != nil {
+		t.Fatalf("organize directory: %v", err)
+	}
+	if res.Organized != 0 || res.Skipped != 1 {
+		t.Fatalf("organize result = organized %d skipped %d, want 0/1; items=%#v errors=%#v", res.Organized, res.Skipped, res.Items, res.Errors)
+	}
+	if len(res.Items) != 1 || res.Items[0].Reason != organizeSkipDuplicateLibrary {
+		t.Fatalf("source should be skipped as external-id duplicate: %#v", res.Items)
+	}
+	if _, err := os.Stat(sourceFile); err != nil {
+		t.Fatalf("duplicate source should remain untouched: %v", err)
+	}
+}
+
+func TestOrganizeDirectoryUsesBangumiForAnimeRename(t *testing.T) {
+	upstream := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if r.URL.Path != "/search/subject/frieren" {
+			http.NotFound(w, r)
+			return
+		}
+		w.Header().Set("Content-Type", "application/json")
+		_ = json.NewEncoder(w).Encode(map[string]any{
+			"results": 1,
+			"list": []map[string]any{{
+				"id":       889,
+				"name":     "Frieren",
+				"name_cn":  "葬送的芙莉莲",
+				"air_date": "2023-09-29",
+			}},
+		})
+	}))
+	defer upstream.Close()
+
+	repos := newOrganizerTestRepo(t)
+	cfg := &config.Config{}
+	bangumi := NewBangumiProvider(cfg, zap.NewNop())
+	bangumi.base = upstream.URL
+	scraper := NewScraperService(cfg, zap.NewNop(), repos, nil, bangumi, nil, nil, NewHub(zap.NewNop()))
+
+	root := t.TempDir()
+	src := filepath.Join(root, "downloads")
+	dest := filepath.Join(root, "media")
+	sourceFile := filepath.Join(src, "Frieren.S01E01.1080p.mkv")
+	writeOrgFile(t, sourceFile, "episode")
+
+	organizer := NewOrganizerService(cfg, zap.NewNop(), repos)
+	organizer.SetScraper(scraper)
+	res, err := organizer.OrganizeDirectory(t.Context(), OrganizeOptions{
+		SourcePath:   src,
+		DestPath:     dest,
+		TransferMode: TransferCopy,
+		MediaType:    "anime",
+	})
+	if err != nil {
+		t.Fatalf("organize directory: %v", err)
+	}
+	want := filepath.Join(dest, "电视剧", "葬送的芙莉莲", "Season 01", "葬送的芙莉莲 - S01E01.mkv")
+	if res.Organized != 1 {
+		t.Fatalf("organized = %d, want 1; items=%#v errors=%#v", res.Organized, res.Items, res.Errors)
+	}
+	if _, err := os.Stat(want); err != nil {
+		t.Fatalf("organized file should use Bangumi metadata path %q: %v", want, err)
+	}
+}
+
 func TestOrganizeScanAndScrapeRetriesNoMatchRows(t *testing.T) {
 	scraper, repos, closeServer := newTestScraper(t)
 	defer closeServer()

internal/service/scanner.go

@@ -1734,6 +1734,12 @@ func applyLocalMetadata(m *model.Media, local *LocalMetadata) {
 	if local.TMDbID > 0 {
 		m.TMDbID = local.TMDbID
 	}
+	if local.DoubanID != "" {
+		m.DoubanID = local.DoubanID
+	}
+	if local.TheTVDBID != "" {
+		m.TheTVDBID = local.TheTVDBID
+	}
 	if local.SeasonNum > 0 {
 		m.SeasonNum = local.SeasonNum
 	}
@@ -1768,6 +1774,8 @@ func localHasDescriptiveMetadata(local *LocalMetadata) bool {
 		local.Overview != "" ||
 		local.Rating > 0 ||
 		local.TMDbID > 0 ||
+		local.DoubanID != "" ||
+		local.TheTVDBID != "" ||
 		local.Genres != "" ||
 		local.Countries != "" ||
 		local.Languages != ""

internal/service/scraper.go

@@ -34,6 +34,7 @@ type ScraperService struct {
 	tmdb    *TMDbProvider
 	bangumi *BangumiProvider
 	thetvdb *TheTVDBProvider
+	douban  *DoubanProvider
 	fanart  *FanartProvider
 	adult   *AdultProvider
 	hub     *Hub
@@ -61,6 +62,10 @@ func NewScraperService(
 	}
 }
 
+func (s *ScraperService) SetDouban(douban *DoubanProvider) {
+	s.douban = douban
+}
+
 // yearPattern extracts a 4-digit year (1900-2099).
 var yearPattern = regexp.MustCompile(`(?:^|[^\d])(19\d{2}|20\d{2})(?:[^\d]|$)`)
 
@@ -309,6 +314,12 @@ func (s *ScraperService) applyProviderMatch(ctx context.Context, m *model.Media,
 	if match.BangumiID > 0 {
 		updates["bangumi_id"] = match.BangumiID
 	}
+	if match.DoubanID != "" {
+		updates["douban_id"] = match.DoubanID
+	}
+	if match.TheTVDBID != "" {
+		updates["thetvdb_id"] = match.TheTVDBID
+	}
 	if match.NSFW {
 		updates["nsfw"] = true
 	}
@@ -368,6 +379,8 @@ func (s *ScraperService) applyProviderMatch(ctx context.Context, m *model.Media,
 		"title":      match.Title,
 		"tmdb_id":    match.TMDbID,
 		"bangumi_id": match.BangumiID,
+		"douban_id":  match.DoubanID,
+		"thetvdb_id": match.TheTVDBID,
 		"source":     map[bool]string{true: "adult"}[match.NSFW],
 	})
 	return nil
@@ -408,6 +421,12 @@ func (s *ScraperService) applyLocalMetadataMatch(ctx context.Context, m *model.M
 	if next.BangumiID > 0 {
 		updates["bangumi_id"] = next.BangumiID
 	}
+	if next.DoubanID != "" {
+		updates["douban_id"] = next.DoubanID
+	}
+	if next.TheTVDBID != "" {
+		updates["thetvdb_id"] = next.TheTVDBID
+	}
 	if next.SeasonNum > 0 {
 		updates["season_num"] = next.SeasonNum
 	}
@@ -431,10 +450,11 @@ func (s *ScraperService) applyLocalMetadataMatch(ctx context.Context, m *model.M
 		return err
 	}
 	s.hub.Publish("scrape", map[string]any{
-		"media_id": m.ID,
-		"title":    next.Title,
-		"tmdb_id":  next.TMDbID,
-		"source":   "local_nfo",
+		"media_id":  m.ID,
+		"title":     next.Title,
+		"tmdb_id":   next.TMDbID,
+		"douban_id": next.DoubanID,
+		"source":    "local_nfo",
 	})
 	return nil
 }
@@ -558,7 +578,7 @@ func librarySupportsSeasons(lib *model.Library) bool {
 // 库类型决定首选 provider：
 //
 //	anime  -> Bangumi  -> TMDb /search/tv  -> TMDb /search/movie
-//	tv     -> TheTVDB  -> TMDb /search/tv  -> TMDb /search/movie
+//	tv     -> TMDb /search/tv -> TMDb /search/movie -> TheTVDB
 //	movie  -> TMDb /search/movie
 //	(空)    -> TMDb /search/movie
 //
@@ -578,14 +598,6 @@ func (s *ScraperService) lookup(ctx context.Context, lib *model.Library, query s
 				s.log.Debug("bangumi search failed", zap.String("query", query), zap.Error(err))
 			}
 		}
-	case "tv", "variety", "show", "shows":
-		if s.thetvdb != nil && s.thetvdb.Enabled() {
-			if m, err := s.thetvdb.SearchSeries(ctx, query); err == nil && m != nil {
-				return m
-			} else if err != nil {
-				s.log.Debug("thetvdb search failed", zap.String("query", query), zap.Error(err))
-			}
-		}
 	}
 	if s.tmdb != nil && s.tmdb.Enabled() {
 		// anime / tv 先用 TMDb /search/tv（剧名通常是 TV 类目）。
@@ -602,6 +614,20 @@ func (s *ScraperService) lookup(ctx context.Context, lib *model.Library, query s
 			s.log.Debug("tmdb movie search failed", zap.String("query", query), zap.Error(err))
 		}
 	}
+	if (kind == "anime" || kind == "tv" || kind == "variety" || kind == "show" || kind == "shows") && s.thetvdb != nil && s.thetvdb.Enabled() {
+		if m, err := s.thetvdb.SearchSeries(ctx, query); err == nil && m != nil {
+			return m
+		} else if err != nil {
+			s.log.Debug("thetvdb search failed", zap.String("query", query), zap.Error(err))
+		}
+	}
+	if s.douban != nil && s.douban.Enabled() {
+		if m, err := s.douban.SearchMatch(ctx, query); err == nil && m != nil {
+			return m
+		} else if err != nil {
+			s.log.Debug("douban search failed", zap.String("query", query), zap.Error(err))
+		}
+	}
 	return nil
 }
 
@@ -724,6 +750,9 @@ func (s *ScraperService) AnyEnabled() bool {
 	if s.adult != nil && s.adult.Enabled() {
 		return true
 	}
+	if s.douban != nil && s.douban.Enabled() {
+		return true
+	}
 	return false
 }
 

internal/service/service.go

@@ -96,9 +96,11 @@ func New(cfg *config.Config, log *zap.Logger, repos *repository.Container) *Cont
 	tmdb := NewTMDbProvider(cfg, log, apiConfig)
 	bangumi := NewBangumiProvider(cfg, log)
 	thetvdb := NewTheTVDBProvider(cfg, log)
+	douban := NewDoubanProvider(cfg, log)
 	fanart := NewFanartProvider(cfg, log)
 	adult := NewAdultProvider(log, apiConfig)
 	scraper := NewScraperService(cfg, log, repos, tmdb, bangumi, thetvdb, fanart, hub, adult)
+	scraper.SetDouban(douban)
 	organizer := NewOrganizerService(cfg, log, repos)
 	organizer.SetProbe(probe)
 	organizer.SetScraper(scraper)
@@ -124,7 +126,6 @@ func New(cfg *config.Config, log *zap.Logger, repos *repository.Container) *Cont
 	emby.SetCloudProbe(storageCfg, probe)
 	downloadClients := NewDownloadClientService(log, repos)
 	assistant := NewAssistantService(log, repos, ai)
-	douban := NewDoubanProvider(cfg, log)
 	scheduler := NewSchedulerService(log, repos, scanner, transcoder, organizer, storageCfg, hub, cfg.Cache.CacheDir)
 	scheduler.SetTaskTracker(tasks)
 

internal/service/thetvdb.go

@@ -128,6 +128,7 @@ func (t *TheTVDBProvider) SearchSeries(ctx context.Context, query string) (*Matc
 	}
 	r := p.Data[0]
 	m := &Match{
+		TheTVDBID: r.ID,
 		Title:     r.Name,
 		Overview:  r.Overview,
 		PosterURL: r.Image,

internal/service/tmdb.go

@@ -126,6 +126,8 @@ func (t *TMDbProvider) resolveBaseURL(ctx context.Context) string {
 type Match struct {
 	TMDbID       int      `json:"tmdb_id"`
 	BangumiID    int      `json:"bangumi_id"`
+	DoubanID     string   `json:"douban_id,omitempty"`
+	TheTVDBID    string   `json:"thetvdb_id,omitempty"`
 	Title        string   `json:"title"`
 	OriginalName string   `json:"original_name,omitempty"`
 	Overview     string   `json:"overview"`