listenbrainz: implement duplicate listen check on import

internal/backends/listenbrainz/listenbrainz.go

@@ -24,12 +24,14 @@ import (
 	"go.uploadedlobster.com/scotty/internal/config"
 	"go.uploadedlobster.com/scotty/internal/i18n"
 	"go.uploadedlobster.com/scotty/internal/models"
+	"go.uploadedlobster.com/scotty/internal/similarity"
 	"go.uploadedlobster.com/scotty/internal/version"
 )
 
 type ListenBrainzApiBackend struct {
 	client          Client
 	username        string
+	checkDuplicates bool
 	existingMbids   map[string]bool
 }
 
@@ -44,6 +46,10 @@ func (b *ListenBrainzApiBackend) Options() []models.BackendOption {
 		Name:  "token",
 		Label: i18n.Tr("Access token"),
 		Type:  models.Secret,
+	}, {
+		Name:  "check-duplicate-listens",
+		Label: i18n.Tr("Check for duplicate listens on import (slower)"),
+		Type:  models.Bool,
 	}}
 }
 
@@ -51,6 +57,7 @@ func (b *ListenBrainzApiBackend) FromConfig(config *config.ServiceConfig) models
 	b.client = NewClient(config.GetString("token"))
 	b.client.MaxResults = MaxItemsPerGet
 	b.username = config.GetString("username")
+	b.checkDuplicates = config.GetBool("check-duplicate-listens", false)
 	return b
 }
 
@@ -117,6 +124,7 @@ func (b *ListenBrainzApiBackend) ExportListens(oldestTimestamp time.Time, result
 
 func (b *ListenBrainzApiBackend) ImportListens(export models.ListensResult, importResult models.ImportResult, progress chan models.Progress) (models.ImportResult, error) {
 	total := len(export.Items)
+	p := models.Progress{}.FromImportResult(importResult)
 	for i := 0; i < total; i += MaxListensPerRequest {
 		listens := export.Items[i:min(i+MaxListensPerRequest, total)]
 		count := len(listens)
@@ -130,6 +138,21 @@ func (b *ListenBrainzApiBackend) ImportListens(export models.ListensResult, impo
 		}
 
 		for _, l := range listens {
+			if b.checkDuplicates {
+				isDupe, err := b.checkDuplicateListen(l)
+				p.Elapsed += 1
+				progress <- p
+				if err != nil {
+					return importResult, err
+				} else if isDupe {
+					count -= 1
+					msg := fmt.Sprintf("Ignored duplicate listen %v: \"%v\" by %v (%v)",
+						l.ListenedAt, l.TrackName, l.ArtistName(), l.RecordingMbid)
+					importResult.ImportErrors = append(importResult.ImportErrors, msg)
+					continue
+				}
+			}
+
 			l.FillAdditionalInfo()
 			listen := Listen{
 				ListenedAt: l.ListenedAt.Unix(),
@@ -142,17 +165,22 @@ func (b *ListenBrainzApiBackend) ImportListens(export models.ListensResult, impo
 			}
 			listen.TrackMetadata.AdditionalInfo["submission_client"] = version.AppName
 			listen.TrackMetadata.AdditionalInfo["submission_client_version"] = version.AppVersion
+
 			submission.Payload = append(submission.Payload, listen)
 		}
 
+		if len(submission.Payload) > 0 {
 			_, err := b.client.SubmitListens(submission)
 			if err != nil {
 				return importResult, err
 			}
+		}
 
+		if count > 0 {
 			importResult.UpdateTimestamp(listens[count-1].ListenedAt)
+		}
 		importResult.ImportCount += count
-		progress <- models.Progress{}.FromImportResult(importResult)
+		progress <- p.FromImportResult(importResult)
 	}
 
 	return importResult, nil
@@ -258,6 +286,33 @@ func (b *ListenBrainzApiBackend) ImportLoves(export models.LovesResult, importRe
 	return importResult, nil
 }
 
+var defaultDuration = time.Duration(3 * time.Minute)
+
+const trackSimilarityThreshold = 0.9
+
+func (b *ListenBrainzApiBackend) checkDuplicateListen(listen models.Listen) (bool, error) {
+	// Find listens
+	duration := listen.Duration
+	if duration == 0 {
+		duration = defaultDuration
+	}
+	minTime := listen.ListenedAt.Add(-duration)
+	maxTime := listen.ListenedAt.Add(duration)
+	candidates, err := b.client.GetListens(b.username, maxTime, minTime)
+	if err != nil {
+		return false, err
+	}
+
+	for _, c := range candidates.Payload.Listens {
+		sim := similarity.CompareTracks(listen.Track, c.TrackMetadata.AsTrack())
+		if sim >= trackSimilarityThreshold {
+			return true, nil
+		}
+	}
+
+	return false, nil
+}
+
 func (lbListen Listen) AsListen() models.Listen {
 	listen := models.Listen{
 		ListenedAt: time.Unix(lbListen.ListenedAt, 0),