feat(uploads): match release groups with MusicBrainz and AcoustID

services/api/internal/uploads/release_groups.go

@@ -0,0 +1,144 @@
+package uploads
+
+import (
+	"context"
+	"crypto/sha256"
+	"encoding/hex"
+	"path"
+	"regexp"
+	"sort"
+	"strconv"
+	"strings"
+
+	"github.com/google/uuid"
+	"pipu/api/internal/library"
+	"pipu/api/internal/library/meta"
+	"pipu/api/internal/platform/apperr"
+)
+
+type ReleaseGroup struct {
+	ID       string      `json:"id"`
+	Artist   string      `json:"artist"`
+	Album    string      `json:"album"`
+	FileIDs  []uuid.UUID `json:"file_ids"`
+	HasCover bool        `json:"has_cover"`
+	State    string      `json:"state"`
+}
+type releaseGroupDraft struct {
+	Key, Artist, Album string
+	Files              []uuid.UUID
+}
+
+var discDirectory = regexp.MustCompile(`(?i)^(?:cd|disc|disk|диск)[ ._-]*0*\d+(?:\s*[-(].*)?$`)
+
+func releaseFolder(f File) string {
+	name := f.OriginalName
+	if f.ClientRelPath != nil {
+		name = *f.ClientRelPath
+	}
+	dir := path.Dir(strings.ReplaceAll(name, `\`, "/"))
+	for dir != "." && dir != "/" && discDirectory.MatchString(path.Base(dir)) {
+		dir = path.Dir(dir)
+	}
+	if dir == "." || dir == "/" {
+		return ""
+	}
+	return strings.ToLower(dir)
+}
+func releaseBase(f File) string {
+	m := f.Meta
+	year := ""
+	if m.Year != nil {
+		year = strconv.Itoa(*m.Year)
+	}
+	if m.Album != "" && m.Album != library.UnknownAlbum {
+		if m.AlbumArtist != "" && m.AlbumArtist != meta.UnknownArtist {
+			return "tags\x00" + meta.Key(m.AlbumArtist) + "\x00" + meta.Key(m.Album) + "\x00" + year
+		}
+		if folder := releaseFolder(f); folder != "" {
+			return "album-folder\x00" + folder + "\x00" + meta.Key(m.Album) + "\x00" + year
+		}
+		return "tags\x00" + meta.Key(m.Artist) + "\x00" + meta.Key(m.Album) + "\x00" + year
+	}
+	return "folder\x00" + releaseFolder(f)
+}
+func groupReleaseFiles(files []File) []releaseGroupDraft {
+	files = releaseFiles(files)
+	known := map[string]map[string]bool{}
+	for _, f := range files {
+		base := releaseBase(f)
+		if known[base] == nil {
+			known[base] = map[string]bool{}
+		}
+		if id := tagID(f.Tags, "musicbrainz_albumid", "musicbrainz album id", "musicbrainz release id", "musicbrainz_releaseid"); id != "" {
+			known[base][id] = true
+		}
+	}
+	groups := map[string]*releaseGroupDraft{}
+	for _, f := range files {
+		base := releaseBase(f)
+		id := tagID(f.Tags, "musicbrainz_albumid", "musicbrainz album id", "musicbrainz release id", "musicbrainz_releaseid")
+		if id == "" && len(known[base]) == 1 {
+			for value := range known[base] {
+				id = value
+			}
+		}
+		key := base
+		if id != "" {
+			key = "mbid\x00" + id
+		}
+		sum := sha256.Sum256([]byte(key))
+		key = hex.EncodeToString(sum[:])
+		if groups[key] == nil {
+			artist := f.Meta.AlbumArtist
+			if artist == "" {
+				artist = f.Meta.Artist
+			}
+			groups[key] = &releaseGroupDraft{Key: key, Artist: artist, Album: f.Meta.Album, Files: []uuid.UUID{}}
+		}
+		groups[key].Files = append(groups[key].Files, f.ID)
+	}
+	out := []releaseGroupDraft{}
+	for _, g := range groups {
+		sort.Slice(g.Files, func(i, j int) bool { return g.Files[i].String() < g.Files[j].String() })
+		out = append(out, *g)
+	}
+	sort.Slice(out, func(i, j int) bool { return out[i].Files[0].String() < out[j].Files[0].String() })
+	return out
+}
+
+// releaseScope resolves only a group belonging to this batch. A mixed batch requires explicit selection.
+func (s *Service) releaseScope(ctx context.Context, batch uuid.UUID, requested string) ([]File, []ReleaseGroup, string, error) {
+	if err := s.store.EnsureReleaseGroups(ctx, batch); err != nil {
+		return nil, nil, "", notFound(err)
+	}
+	groups, err := s.store.ReleaseGroups(ctx, batch)
+	if err != nil {
+		return nil, nil, "", err
+	}
+	if requested == "" && len(groups) == 1 {
+		requested = groups[0].ID
+	}
+	if requested != "" {
+		found := false
+		for _, g := range groups {
+			found = found || g.ID == requested
+		}
+		if !found {
+			return nil, groups, "", apperr.NotFound("uploads.release_group_not_found", "release group does not belong to batch")
+		}
+		files, err := s.store.AllInReleaseGroup(ctx, batch, requested)
+		return files, groups, requested, err
+	}
+	if len(groups) > 1 {
+		return nil, groups, "", nil
+	}
+	files, err := s.store.AllInBatch(ctx, batch)
+	return files, groups, "", err
+}
+func firstGroup(values []string) string {
+	if len(values) > 0 {
+		return values[0]
+	}
+	return ""
+}