优化

internal/service/adult_scraper_parse.go

@@ -23,6 +23,7 @@ func parseAdultDetailHTML(body, code, source, detailURL string) *Match {
 	}
 
 	match := &Match{
+		Provider:     strings.ToLower(strings.TrimSpace(source)),
 		OriginalName: code,
 		MediaType:    "adult",
 		NSFW:         true,

internal/service/metatube.go

@@ -310,6 +310,7 @@ func (p *MetaTubeProvider) convertSearchResultToMatch(cfg MetaTubeConfig, query
 	}
 
 	return &Match{
+		Provider:     "metatube",
 		MediaType:    "adult",
 		Title:        formattedTitle,
 		OriginalName: code,
@@ -383,6 +384,7 @@ func (p *MetaTubeProvider) convertMovieInfoToMatch(cfg MetaTubeConfig, movie *Me
 	}
 
 	return &Match{
+		Provider:     "metatube",
 		MediaType:    "adult",
 		Title:        formattedTitle,
 		OriginalName: code,

internal/service/metatube_test.go

@@ -62,6 +62,9 @@ func TestMetaTubeProviderSearch(t *testing.T) {
 		t.Fatalf("expected 1 match, got %d", len(matches))
 	}
 	m := matches[0]
+	if m.Provider != "metatube" {
+		t.Errorf("expected provider metatube, got %q", m.Provider)
+	}
 	if m.OriginalName != "IPX-235" {
 		t.Errorf("expected original name IPX-235, got %s", m.OriginalName)
 	}

internal/service/scraper.go

@@ -93,7 +93,11 @@ func (s *ScraperService) EnrichOneWithOptions(ctx context.Context, m *model.Medi
 	}
 	if match == nil {
 		if local != nil && !local.PathHint {
-			return s.applyLocalMetadataMatch(ctx, m, local)
+			err := s.applyLocalMetadataMatch(ctx, m, local)
+			if err == nil {
+				options.recordProvider("local")
+			}
+			return err
 		}
 		if err := s.repo.DB.WithContext(ctx).Model(&model.Media{}).Where("id = ?", m.ID).
 			Update("scrape_status", "no_match").Error; err != nil {
@@ -227,6 +231,11 @@ func (s *ScraperService) applyProviderMatchWithOptions(ctx context.Context, m *m
 		"thetvdb_id": match.TheTVDBID,
 		"source":     map[bool]string{true: "adult"}[match.NSFW],
 	})
+	provider := strings.ToLower(strings.TrimSpace(match.Provider))
+	if provider == "" {
+		provider = "unknown"
+	}
+	options.recordProvider(provider)
 	return nil
 }
 

internal/service/scraper_artwork.go

@@ -2,6 +2,7 @@ package service
 
 import (
 	"context"
+	"net/url"
 	"strings"
 	"time"
 
@@ -20,10 +21,34 @@ func (s *ScraperService) prepareScrapedArtworkURL(ctx context.Context, mediaID,
 	if s == nil || s.images == nil || !isHTTPish(candidate) {
 		return candidate, ""
 	}
-	fetchCtx, cancel := context.WithTimeout(context.WithoutCancel(ctx), 15*time.Second)
+	originalSource := metaTubeOriginalArtworkURL(candidate)
+	timeout := 15 * time.Second
+	if originalSource != "" {
+		// MetaTube face processing can become very slow under a batch scrape.
+		// Fail over quickly so the same source can be processed locally.
+		timeout = 8 * time.Second
+	}
+	fetchCtx, cancel := context.WithTimeout(context.WithoutCancel(ctx), timeout)
 	err := s.images.PrefetchRemote(fetchCtx, candidate)
 	cancel()
 	if err != nil {
+		if originalSource != "" {
+			fallbackCtx, fallbackCancel := context.WithTimeout(context.WithoutCancel(ctx), 15*time.Second)
+			fallbackErr := s.images.PrefetchRemote(fallbackCtx, originalSource)
+			fallbackCancel()
+			if fallbackErr == nil {
+				s.log.Warn("MetaTube artwork processing failed; using local face-aware crop",
+					zap.String("media_id", mediaID),
+					zap.String("field", field),
+					zap.String("candidate", candidate),
+					zap.String("source", originalSource),
+					zap.Error(err))
+				if current != "" && isHTTPish(current) {
+					return originalSource, current
+				}
+				return originalSource, ""
+			}
+		}
 		if current != "" {
 			s.log.Warn("scrape artwork prefetch failed; keeping existing artwork",
 				zap.String("media_id", mediaID),
@@ -46,6 +71,18 @@ func (s *ScraperService) prepareScrapedArtworkURL(ctx context.Context, mediaID,
 	return candidate, ""
 }
 
+func metaTubeOriginalArtworkURL(raw string) string {
+	parsed, err := url.Parse(strings.TrimSpace(raw))
+	if err != nil || !strings.Contains(parsed.Path, "/v1/images/primary/") {
+		return ""
+	}
+	source := strings.TrimSpace(parsed.Query().Get("url"))
+	if !isHTTPish(source) {
+		return ""
+	}
+	return source
+}
+
 func (s *ScraperService) removeCachedScrapedArtwork(urls ...string) {
 	if s == nil || s.images == nil {
 		return

internal/service/scraper_artwork_test.go

@@ -3,8 +3,11 @@ package service
 import (
 	"bytes"
 	"errors"
+	"image"
+	"image/color"
 	"io"
 	"net/http"
+	"net/url"
 	"os"
 	"path/filepath"
 	"strings"
@@ -67,10 +70,14 @@ func TestApplyProviderMatchInvalidatesMediaCache(t *testing.T) {
 	if err := repos.DB.Create(&media).Error; err != nil {
 		t.Fatal(err)
 	}
-	match := &Match{Title: "Matched", PosterURL: "https://image.tmdb.org/t/p/w500/poster.jpg", BackdropURL: "https://image.tmdb.org/t/p/w1280/backdrop.jpg"}
+	match := &Match{Provider: "metatube", Title: "Matched", PosterURL: "https://image.tmdb.org/t/p/w500/poster.jpg", BackdropURL: "https://image.tmdb.org/t/p/w1280/backdrop.jpg"}
-	if err := scraper.applyProviderMatch(t.Context(), &media, &lib, match); err != nil {
+	usedProvider := ""
+	if err := scraper.applyProviderMatchWithOptions(t.Context(), &media, &lib, match, ScrapeOptions{resultProvider: &usedProvider}); err != nil {
 		t.Fatal(err)
 	}
+	if usedProvider != "metatube" {
+		t.Fatalf("recorded provider = %q, want metatube", usedProvider)
+	}
 
 	var stale map[string]string
 	if cache.GetJSON(t.Context(), "media:list:stale", &stale) {
@@ -145,6 +152,92 @@ func TestApplyProviderMatchKeepsExistingArtworkWhenNewPrefetchFails(t *testing.T
 	}
 }
 
+func TestApplyProviderMatchFallsBackToLocalFaceCropWhenMetaTubeImageFails(t *testing.T) {
+	scraper, repos, closeServer := newTestScraper(t)
+	defer closeServer()
+
+	wideCover := createTestImage(
+		900,
+		600,
+		color.RGBA{R: 255, A: 255},
+		color.RGBA{B: 255, A: 255},
+	)
+	sourceURL := "https://images.example.com/wide-cover.jpg"
+	images := NewImageProxy(&config.Config{Cache: config.CacheConfig{CacheDir: filepath.Join(t.TempDir(), "cache")}}, zap.NewNop())
+	images.client = &http.Client{Transport: imageRoundTripFunc(func(req *http.Request) (*http.Response, error) {
+		if req.URL.Host == "metatube.example.com" {
+			return &http.Response{
+				StatusCode: http.StatusGatewayTimeout,
+				Status:     "504 Gateway Timeout",
+				Header:     make(http.Header),
+				Body:       io.NopCloser(strings.NewReader("timeout")),
+				Request:    req,
+			}, nil
+		}
+		return &http.Response{
+			StatusCode: http.StatusOK,
+			Status:     "200 OK",
+			Header:     http.Header{"Content-Type": []string{"image/jpeg"}},
+			Body:       io.NopCloser(bytes.NewReader(wideCover)),
+			Request:    req,
+		}, nil
+	})}
+	scraper.SetImageProxy(images)
+
+	libPath := t.TempDir()
+	lib := model.Library{Name: "Adult", Path: libPath, Type: "adult", Enabled: true}
+	if err := repos.Library.Create(t.Context(), &lib); err != nil {
+		t.Fatal(err)
+	}
+	media := model.Media{
+		LibraryID:    lib.ID,
+		Title:        "Raw",
+		Path:         filepath.Join(libPath, "IPX-235.mp4"),
+		PosterURL:    filepath.Join(libPath, "old-poster.jpg"),
+		NSFW:         true,
+		ScrapeStatus: "matched",
+	}
+	if err := repos.DB.Create(&media).Error; err != nil {
+		t.Fatal(err)
+	}
+	query := url.Values{
+		"auto":    {"true"},
+		"pos":     {"1"},
+		"quality": {"90"},
+		"ratio":   {"-1"},
+		"url":     {sourceURL},
+	}
+	match := &Match{
+		Title:     "IPX-235 Matched",
+		MediaType: "adult",
+		NSFW:      true,
+		PosterURL: "https://metatube.example.com/v1/images/primary/AVE/123?" + query.Encode(),
+	}
+	if err := scraper.applyProviderMatch(t.Context(), &media, &lib, match); err != nil {
+		t.Fatal(err)
+	}
+
+	var stored model.Media
+	if err := repos.DB.First(&stored, "id = ?", media.ID).Error; err != nil {
+		t.Fatal(err)
+	}
+	if filepath.Base(stored.PosterURL) != "IPX-235-poster.jpg" {
+		t.Fatalf("poster = %q, want local face-cropped sidecar", stored.PosterURL)
+	}
+	data, err := os.ReadFile(stored.PosterURL)
+	if err != nil {
+		t.Fatal(err)
+	}
+	cropped, _, err := image.Decode(bytes.NewReader(data))
+	if err != nil {
+		t.Fatal(err)
+	}
+	ratio := float64(cropped.Bounds().Dx()) / float64(cropped.Bounds().Dy())
+	if ratio < 0.65 || ratio > 0.68 {
+		t.Fatalf("local fallback poster ratio = %.3f", ratio)
+	}
+}
+
 func TestApplyProviderMatchReplacesArtworkAndRemovesOldCache(t *testing.T) {
 	scraper, repos, closeServer := newTestScraper(t)
 	defer closeServer()

internal/service/scraper_library.go

@@ -27,11 +27,13 @@ func (s *ScraperService) lookup(ctx context.Context, lib *model.Library, media *
 	}
 	if s.tmdb != nil && s.tmdb.Enabled() {
 		if match := s.lookupAutomaticTMDb(ctx, kind, query, year); match != nil {
+			match.Provider = "tmdb"
 			return match
 		}
 	}
 	if s.douban != nil && s.douban.Enabled() {
 		if m, err := s.douban.SearchMatch(ctx, query); err == nil && m != nil && metadataMatchCompatibleWithType(kind, m) {
+			m.Provider = "douban"
 			return m
 		} else if err != nil {
 			s.log.Debug("douban search failed", zap.String("query", query), zap.Error(err))
@@ -39,6 +41,7 @@ func (s *ScraperService) lookup(ctx context.Context, lib *model.Library, media *
 	}
 	if s.bangumi != nil && s.bangumi.Enabled() {
 		if m, err := s.bangumi.Search(ctx, query); err == nil && m != nil && metadataMatchCompatibleWithType(kind, m) {
+			m.Provider = "bangumi"
 			return m
 		} else if err != nil {
 			s.log.Debug("bangumi search failed", zap.String("query", query), zap.Error(err))
@@ -46,6 +49,7 @@ func (s *ScraperService) lookup(ctx context.Context, lib *model.Library, media *
 	}
 	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 && metadataMatchCompatibleWithType(kind, m) {
+			m.Provider = "thetvdb"
 			return m
 		} else if err != nil {
 			s.log.Debug("thetvdb search failed", zap.String("query", query), zap.Error(err))

internal/service/scraper_lookup.go

@@ -23,6 +23,7 @@ func (s *ScraperService) matchFromMediaExternalIDs(ctx context.Context, m *model
 		}
 		if match := s.manualTMDbMatchByID(ctx, m.TMDbID, normalizeMediaType(mediaType, m.Title, "")); match != nil {
 			if s.mediaExternalIDMatchTrusted(m, lib, match, "tmdb") {
+				match.Provider = "tmdb"
 				preferExistingLocalizedEpisodeTitle(m, lib, match)
 				return match
 			}
@@ -31,6 +32,7 @@ func (s *ScraperService) matchFromMediaExternalIDs(ctx context.Context, m *model
 	if strings.TrimSpace(m.DoubanID) != "" && s.douban != nil && s.douban.Enabled() {
 		if match, err := s.douban.GetMatchByID(ctx, strings.TrimSpace(m.DoubanID)); err == nil && match != nil {
 			if s.mediaExternalIDMatchTrusted(m, lib, match, "douban") {
+				match.Provider = "douban"
 				preferExistingLocalizedEpisodeTitle(m, lib, match)
 				return match
 			}
@@ -41,6 +43,7 @@ func (s *ScraperService) matchFromMediaExternalIDs(ctx context.Context, m *model
 	if m.BangumiID > 0 && s.bangumi != nil && s.bangumi.Enabled() {
 		if match, err := s.bangumi.GetSubject(ctx, m.BangumiID); err == nil && match != nil {
 			if s.mediaExternalIDMatchTrusted(m, lib, match, "bangumi") {
+				match.Provider = "bangumi"
 				preferExistingLocalizedEpisodeTitle(m, lib, match)
 				return match
 			}
@@ -51,6 +54,7 @@ func (s *ScraperService) matchFromMediaExternalIDs(ctx context.Context, m *model
 	if strings.TrimSpace(m.TheTVDBID) != "" && s.thetvdb != nil && s.thetvdb.Enabled() {
 		if match, err := s.thetvdb.GetSeriesMatchByID(ctx, strings.TrimSpace(m.TheTVDBID)); err == nil && match != nil {
 			if s.mediaExternalIDMatchTrusted(m, lib, match, "thetvdb") {
+				match.Provider = "thetvdb"
 				preferExistingLocalizedEpisodeTitle(m, lib, match)
 				return match
 			}

internal/service/scraper_options.go

@@ -8,6 +8,7 @@ type ScrapeOptions struct {
 	DeferEpisodeDetails bool
 	ForceRematch        bool
 	RebuildIdentity     bool
+	resultProvider      *string
 }
 
 func (o ScrapeOptions) episodeArtworkEnabled() bool {
@@ -18,3 +19,9 @@ func skipEpisodeArtworkOptions(retryNoMatch bool) ScrapeOptions {
 	episodeArtwork := false
 	return ScrapeOptions{RetryNoMatch: retryNoMatch, EpisodeArtwork: &episodeArtwork}
 }
+
+func (o ScrapeOptions) recordProvider(provider string) {
+	if o.resultProvider != nil {
+		*o.resultProvider = provider
+	}
+}

internal/service/scraper_queue.go

@@ -115,10 +115,12 @@ func (s *ScraperService) processScrapeTask(ctx context.Context, task *model.Scra
 	}
 
 	epArtwork := task.EpisodeImages
+	usedProvider := ""
 	options := ScrapeOptions{
 		EpisodeArtwork: &epArtwork,
 		IncludeMatched: task.RefreshMatched,
 		RetryNoMatch:   true,
+		resultProvider: &usedProvider,
 	}
 
 	enrichErr := s.EnrichOneWithOptions(ctx, media, options)
@@ -133,16 +135,9 @@ func (s *ScraperService) processScrapeTask(ctx context.Context, task *model.Scra
 		task.MatchedYear = refreshed.Year
 		task.PosterURL = refreshed.PosterURL
 		task.BackdropURL = refreshed.BackdropURL
-		if refreshed.TMDbID > 0 {
+		task.Provider = strings.ToLower(strings.TrimSpace(usedProvider))
-			task.Provider = "tmdb"
+		if task.Provider == "" {
-		} else if strings.TrimSpace(refreshed.DoubanID) != "" {
+			task.Provider = "unknown"
-			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

internal/service/tmdb_types.go

@@ -4,6 +4,7 @@ package service
 // across providers; provider-specific IDs sit side-by-side so the scraper
 // orchestrator can write them all into a single update.
 type Match struct {
+	Provider      string   `json:"provider,omitempty"`
 	TMDbID        int      `json:"tmdb_id"`
 	BangumiID     int      `json:"bangumi_id"`
 	DoubanID      string   `json:"douban_id,omitempty"`