From 3d3679c21c062bfd3089aa81f7d8f53cf3804fe8 Mon Sep 17 00:00:00 2001 From: ShukeBta <272197458+ShukeBta@users.noreply.github.com> Date: Thu, 25 Jun 2026 02:32:55 +0800 Subject: [PATCH] fix: prevent episode metadata from splitting series --- internal/service/local_metadata.go | 7 +- internal/service/local_metadata_test.go | 33 +++++++++ internal/service/media_series.go | 6 +- .../service/organizer_directory_metadata.go | 3 + .../service/organizer_scrape_safety_test.go | 67 +++++++++++++++++++ internal/service/scraper_query.go | 33 +++++++++ internal/service/scraper_query_test.go | 36 ++++++++++ web/src/utils/groupSeries.ts | 14 +++- 8 files changed, 192 insertions(+), 7 deletions(-) diff --git a/internal/service/local_metadata.go b/internal/service/local_metadata.go index ae205ff..10b0abb 100644 --- a/internal/service/local_metadata.go +++ b/internal/service/local_metadata.go @@ -419,13 +419,10 @@ func mergeArtworkMetadata(meta *LocalMetadata, mediaPath, showBaseDir string) { func mergeEpisodeMetadata(dst, episode *LocalMetadata, doc *nfoDocument) { showTitle := cleanXMLText(doc.ShowTitle) // 整剧标题: 优先 (MoviePilot 在单集 NFO 里也会写整剧名); - // 其次保留 dst 已有(来自 tvshow.nfo);最后才退而用单集名占位。 + // 其次保留 dst 已有(来自 tvshow.nfo)。不要把单集 当整剧标题, + // 否则会把"第 11 集/第几期"整理成整剧目录并污染后续 TMDb 查询。 if showTitle != "" { dst.Title = showTitle - } else if dst.Title == "" { - if episodeTitle := cleanXMLText(doc.Title); episodeTitle != "" { - dst.Title = episodeTitle - } } // 注意: 不要把单集名 / 单集 originaltitle 写进 OriginalName(整剧原名,分组键)。 if episodeTitle := firstText(episode.EpisodeTitle, doc.Title); episodeTitle != "" && !strings.EqualFold(episodeTitle, showTitle) { diff --git a/internal/service/local_metadata_test.go b/internal/service/local_metadata_test.go index c4636e2..cd3b2c7 100644 --- a/internal/service/local_metadata_test.go +++ b/internal/service/local_metadata_test.go @@ -79,6 +79,39 @@ func TestReadLocalEpisodeMetadataMergesShowAndEpisode(t *testing.T) { } } +func TestReadLocalEpisodeMetadataWithoutShowTitleDoesNotUseEpisodeTitleAsSeries(t *testing.T) { + root := t.TempDir() + showDir := filepath.Join(root, "哈哈哈哈哈") + seasonDir := filepath.Join(showDir, "Season 06") + if err := os.MkdirAll(seasonDir, 0o755); err != nil { + t.Fatal(err) + } + mediaPath := filepath.Join(seasonDir, "哈哈哈哈哈 - S06E11.mkv") + if err := os.WriteFile(mediaPath, []byte("x"), 0o644); err != nil { + t.Fatal(err) + } + if err := os.WriteFile(nfoPath(mediaPath), []byte(`<episodedetails><title>第 11 集6114375419`), 0o644); err != nil { + t.Fatal(err) + } + + got, err := ReadLocalMetadata(mediaPath, root, true) + if err != nil { + t.Fatal(err) + } + if got == nil { + t.Fatal("metadata is nil") + } + if got.Title != "" { + t.Fatalf("episode title must not become series title, got %q", got.Title) + } + if got.EpisodeTitle != "第 11 集" || got.SeasonNum != 6 || got.EpisodeNum != 11 { + t.Fatalf("episode metadata not preserved: %+v", got) + } + if got.TMDbID != 0 { + t.Fatalf("episode-level tmdb id must not become series tmdb id, got %d", got.TMDbID) + } +} + func TestReadLocalEpisodeMetadataIgnoresNoneNumericFields(t *testing.T) { root := t.TempDir() showDir := filepath.Join(root, "链锯人 总集篇 (2025)") diff --git a/internal/service/media_series.go b/internal/service/media_series.go index d181db1..5ba2155 100644 --- a/internal/service/media_series.go +++ b/internal/service/media_series.go @@ -221,7 +221,11 @@ func seriesTitleFromMediaPath(path string) string { if dirIndex < 0 { return "" } - return normalizeSeriesPathTitle(parts[dirIndex]) + title := normalizeSeriesPathTitle(parts[dirIndex]) + if unsafeAutomaticEpisodeQuery(title) { + return "" + } + return title } func seriesDisplayTitle(media model.Media) string { diff --git a/internal/service/organizer_directory_metadata.go b/internal/service/organizer_directory_metadata.go index 2c3a341..0d5144c 100644 --- a/internal/service/organizer_directory_metadata.go +++ b/internal/service/organizer_directory_metadata.go @@ -138,6 +138,9 @@ func organizeMetadataMatchTrusted(query string, sourceYear int, match *Match) bo if match == nil || strings.TrimSpace(match.Title) == "" { return false } + if unsafeAutomaticEpisodeQuery(query) { + return false + } if sourceYear > 0 && match.Year > 0 { diff := sourceYear - match.Year if diff < 0 { diff --git a/internal/service/organizer_scrape_safety_test.go b/internal/service/organizer_scrape_safety_test.go index 0f35854..18b8d90 100644 --- a/internal/service/organizer_scrape_safety_test.go +++ b/internal/service/organizer_scrape_safety_test.go @@ -75,6 +75,73 @@ func TestOrganizeDirectoryRejectsWrongYearScraperRename(t *testing.T) { } } +func TestOrganizeDirectoryDoesNotUseEpisodeNFOTitleAsSeriesTitle(t *testing.T) { + var queries []string + upstream := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) { + queries = append(queries, r.URL.Query().Get("query")) + w.Header().Set("Content-Type", "application/json") + if r.URL.Path != "/search/tv" { + http.NotFound(w, r) + return + } + if r.URL.Query().Get("query") != "哈哈哈哈哈" { + _ = json.NewEncoder(w).Encode(map[string]any{"results": []any{}}) + return + } + _ = json.NewEncoder(w).Encode(map[string]any{ + "results": []map[string]any{{ + "id": 112732, + "name": "哈哈哈哈哈", + "first_air_date": "2026-01-01", + }}, + }) + })) + defer upstream.Close() + + repos := newOrganizerTestRepo(t) + cfg := &config.Config{} + cfg.Secrets.TMDbAPIKey = "test-key" + cfg.Secrets.TMDbAPIProxy = upstream.URL + scraper := NewScraperService(cfg, zap.NewNop(), repos, NewTMDbProvider(cfg, zap.NewNop(), nil), nil, nil, nil, NewHub(zap.NewNop())) + + root := t.TempDir() + src := filepath.Join(root, "downloads") + dest := filepath.Join(root, "media") + sourceFile := filepath.Join(src, "哈哈哈哈哈", "Season 06", "哈哈哈哈哈 - S06E11.mkv") + writeOrgFile(t, sourceFile, "episode") + if err := os.WriteFile(nfoPath(sourceFile), []byte(`第 11 集6117129825`), 0o644); err != nil { + t.Fatal(err) + } + + organizer := NewOrganizerService(cfg, 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 != 1 { + t.Fatalf("organized = %d, want 1; items=%#v errors=%#v queries=%v", res.Organized, res.Items, res.Errors, queries) + } + rejected := filepath.Join(dest, "电视剧", "第 11 集", "Season 06", "第 11 集 - S06E11.mkv") + if _, err := os.Stat(rejected); err == nil { + t.Fatalf("episode title should not be used as series folder: %q", rejected) + } + want := filepath.Join(dest, "电视剧", "哈哈哈哈哈", "Season 06", "哈哈哈哈哈 - S06E11.mkv") + if _, err := os.Stat(want); err != nil { + t.Fatalf("expected organized file at %q: %v; items=%#v queries=%v", want, err, res.Items, queries) + } + for _, query := range queries { + if unsafeAutomaticEpisodeQuery(query) { + t.Fatalf("organizer queried unsafe episode title %q; all queries=%v", query, queries) + } + } +} + func TestOrganizeDirectoryDedupsByExternalIDBeforeRename(t *testing.T) { scraper, repos, closeServer := newTestScraper(t) defer closeServer() diff --git a/internal/service/scraper_query.go b/internal/service/scraper_query.go index 995daf5..96123af 100644 --- a/internal/service/scraper_query.go +++ b/internal/service/scraper_query.go @@ -55,6 +55,13 @@ var releaseBoundaryTokenSet = map[string]struct{}{ "x264": {}, "x265": {}, "h264": {}, "h265": {}, "hevc": {}, "avc": {}, } +var ( + episodeOnlyQueryRE = regexp.MustCompile(`(?i)^\s*(?:e(?:p(?:isode)?)?\s*\d{1,3}|episode\s*\d{1,3}|第\s*[0-9一二三四五六七八九十百零两]+\s*[集期话話](?:\s*[上下])?)\s*$`) + episodeTitleQueryRE = regexp.MustCompile(`^\s*第\s*[0-9一二三四五六七八九十百零两]+\s*[集期话話](?:\s*[上下])?\s*[:：].+`) + genericEpisodeWordsRE = regexp.MustCompile(`^\s*第\s*[集期话話]\s*$`) + episodeReleaseTitleTagRE = regexp.MustCompile(`(?i)(?:^|[\s._-])s\d{1,2}e\d{1,3}(?:[\s._-]|$)`) +) + // bracketedTag matches "[anything]", "(anything)" or "{anything}" segments. var bracketedTag = regexp.MustCompile(`[\[\(\{][^\]\)\}]*[\]\)\}]`) var multiWordNoise = []*regexp.Regexp{ @@ -150,6 +157,9 @@ func scrapeQueryCandidates(m *model.Media, lib *model.Library) []string { cleaned = strings.TrimSpace(raw) } for _, candidate := range titleCandidates(cleaned) { + if unsafeAutomaticEpisodeQuery(candidate) { + continue + } key := strings.ToLower(candidate) if _, ok := seen[key]; ok || candidate == "" { continue @@ -394,3 +404,26 @@ func librarySupportsSeasons(lib *model.Library) bool { return false } } + +func unsafeAutomaticEpisodeQuery(query string) bool { + query = strings.TrimSpace(query) + if query == "" { + return true + } + if episodeOnlyQueryRE.MatchString(query) || genericEpisodeWordsRE.MatchString(query) { + return true + } + if episodeTitleQueryRE.MatchString(query) { + return true + } + _, episode := ParseEpisode(query) + if episode > 0 && !looksLikeSeriesReleaseTitle(query) { + return true + } + return false +} + +func looksLikeSeriesReleaseTitle(query string) bool { + cleaned, _ := CleanQuery(query) + return strings.TrimSpace(cleaned) != "" && episodeReleaseTitleTagRE.MatchString(query) +} diff --git a/internal/service/scraper_query_test.go b/internal/service/scraper_query_test.go index 5ded697..0b0f70b 100644 --- a/internal/service/scraper_query_test.go +++ b/internal/service/scraper_query_test.go @@ -272,6 +272,42 @@ func TestScrapeQueryCandidatesUseSeriesLibraryRootWhenMountedAtShowFolder(t *tes } } +func TestScrapeQueryCandidatesSkipEpisodeOnlyTitles(t *testing.T) { + lib := &model.Library{ + Path: `F:\media\电视剧\欧美剧`, + Type: "tv", + } + media := &model.Media{ + Title: "第1期上：最狠开局！五哈团命悬一线好刺激", + Path: `F:\media\电视剧\欧美剧\第1期上：最狠开局！五哈团命悬一线好刺激 (2026)\Season 6\第1期上：最狠开局！五哈团命悬一线好刺激 - S06E01 - 第 1 集.mkv`, + SeasonNum: 6, + EpisodeNum: 1, + } + + got := scrapeQueryCandidates(media, lib) + for _, candidate := range got { + if unsafeAutomaticEpisodeQuery(candidate) { + t.Fatalf("query candidates kept unsafe episode title %q: %#v", candidate, got) + } + } +} + +func TestOrganizeMetadataRejectsEpisodeOnlyQuery(t *testing.T) { + match := &Match{Title: "错误节目", Year: 2026, TMDbID: 284725} + for _, query := range []string{"第 11 集", "第1期上：最狠开局！五哈团命悬一线好刺激"} { + if organizeMetadataMatchTrusted(query, 2026, match) { + t.Fatalf("episode-only query %q must not be trusted for automatic match", query) + } + } +} + +func TestSeriesTitleFromMediaPathIgnoresEpisodeOnlyFolder(t *testing.T) { + got := seriesTitleFromMediaPath(`F:\media\电视剧\欧美剧\第 11 集 (2026)\Season 6\第 11 集 - S06E11.mkv`) + if got != "" { + t.Fatalf("series title from episode-only folder = %q, want empty", got) + } +} + func TestMediaIsEpisodicUsesEpisodePatternInPath(t *testing.T) { lib := &model.Library{ Path: `/media/movies`, diff --git a/web/src/utils/groupSeries.ts b/web/src/utils/groupSeries.ts index dbc8ff8..c31880b 100644 --- a/web/src/utils/groupSeries.ts +++ b/web/src/utils/groupSeries.ts @@ -122,7 +122,8 @@ const SERIES_SPECIAL_CJK_RE = function normalizePathSeriesTitle(value?: string): string { const title = normalizeTitle(value) const stripped = stripSeriesSpecialSuffix(title) - return stripped || title + const normalized = stripped || title + return unsafeEpisodeTitle(normalized) ? '' : normalized } function stripSeriesSpecialSuffix(title: string): string { @@ -133,6 +134,17 @@ function stripSeriesSpecialSuffix(title: string): string { return title } +const EPISODE_ONLY_TITLE_RE = + /^(?:e(?:p(?:isode)?)?\s*\d{1,3}|episode\s*\d{1,3}|第\s*[0-9一二三四五六七八九十百零两]+\s*[集期话話](?:\s*[上下])?|第\s*[集期话話])$/i + +const EPISODE_TITLE_RE = + /^第\s*[0-9一二三四五六七八九十百零两]+\s*[集期话話](?:\s*[上下])?\s*[:：].+/ + +function unsafeEpisodeTitle(title: string): boolean { + const value = title.trim() + return EPISODE_ONLY_TITLE_RE.test(value) || EPISODE_TITLE_RE.test(value) +} + export function seriesTitleFromPath(path?: string): string { if (!path) return '' const parts = path.split(/[\\/]+/).filter(Boolean)