From 173dd52fe19f0a6cfff72d9cf37c96cce4a24968 Mon Sep 17 00:00:00 2001
From: Deluan <deluan@navidrome.org>
Date: Tue, 20 Oct 2020 17:16:24 -0400
Subject: [PATCH] Use MBID with most occurrences

---
 persistence/album_repository.go  |  5 +++--
 persistence/artist_repository.go |  3 ++-
 persistence/helpers.go           | 32 ++++++++++++++++++++++++++++++++
 persistence/helpers_test.go      | 13 +++++++++++++
 4 files changed, 50 insertions(+), 3 deletions(-)

diff --git a/persistence/album_repository.go b/persistence/album_repository.go
index bc1bd232c..8c758ef07 100644
--- a/persistence/album_repository.go
+++ b/persistence/album_repository.go
@@ -161,8 +161,8 @@ func (r *albumRepository) refresh(ids ...string) error {
 	sel := Select(`f.album_id as id, f.album as name, f.artist, f.album_artist, f.artist_id, f.album_artist_id, 
 		f.sort_album_name, f.sort_artist_name, f.sort_album_artist_name,
 		f.order_album_name, f.order_album_artist_name, f.path,
-		f.mbz_album_id, f.mbz_album_artist_id, f.mbz_album_type, f.mbz_album_comment, f.catalog_num,
-		f.compilation, f.genre, max(f.year) as max_year, sum(f.duration) as duration, 
+		group_concat(f.mbz_album_id, ' ') as mbz_album_id, f.mbz_album_artist_id, f.mbz_album_type, f.mbz_album_comment, 
+		f.catalog_num, f.compilation, f.genre, max(f.year) as max_year, sum(f.duration) as duration, 
 		count(f.id) as song_count, a.id as current_id, 
 		group_concat(f.disc_subtitle, ' ') as disc_subtitles,
 		group_concat(f.artist, ' ') as song_artists, group_concat(f.year, ' ') as years,
@@ -211,6 +211,7 @@ func (r *albumRepository) refresh(ids ...string) error {
 			al.AlbumArtistID = al.ArtistID
 		}
 		al.MinYear = getMinYear(al.Years)
+		al.MbzAlbumID = getMbzId(r.ctx, al.MbzAlbumID, r.tableName, al.Name)
 		al.UpdatedAt = time.Now()
 		if al.CurrentId != "" {
 			toUpdate++
diff --git a/persistence/artist_repository.go b/persistence/artist_repository.go
index fe4033a5e..b39e50925 100644
--- a/persistence/artist_repository.go
+++ b/persistence/artist_repository.go
@@ -144,7 +144,7 @@ func (r *artistRepository) refresh(ids ...string) error {
 	}
 	var artists []refreshArtist
 	sel := Select("f.album_artist_id as id", "f.album_artist as name", "count(*) as album_count", "a.id as current_id",
-		"f.mbz_album_artist_id as mbz_artist_id",
+		"group_concat(f.mbz_album_artist_id , ' ') as mbz_artist_id",
 		"f.sort_album_artist_name as sort_artist_name", "f.order_album_artist_name as order_artist_name",
 		"sum(f.song_count) as song_count", "sum(f.size) as size").
 		From("album f").
@@ -164,6 +164,7 @@ func (r *artistRepository) refresh(ids ...string) error {
 		} else {
 			toInsert++
 		}
+		ar.MbzArtistID = getMbzId(r.ctx, ar.MbzArtistID, r.tableName, ar.Name)
 		err := r.Put(&ar.Artist)
 		if err != nil {
 			return err
diff --git a/persistence/helpers.go b/persistence/helpers.go
index 6bec5e989..f281a5367 100644
--- a/persistence/helpers.go
+++ b/persistence/helpers.go
@@ -1,12 +1,15 @@
 package persistence
 
 import (
+	"context"
 	"encoding/json"
 	"fmt"
 	"regexp"
 	"strings"
 
 	"github.com/Masterminds/squirrel"
+	"github.com/deluan/navidrome/consts"
+	"github.com/deluan/navidrome/log"
 	"github.com/deluan/navidrome/model"
 	"github.com/deluan/navidrome/utils"
 )
@@ -55,3 +58,32 @@ func (e existsCond) ToSql() (string, []interface{}, error) {
 	sql = fmt.Sprintf("exists (select 1 from %s where %s)", e.subTable, sql)
 	return sql, args, err
 }
+
+func getMbzId(ctx context.Context, mbzIDS, entityName, name string) string {
+	ids := strings.Fields(mbzIDS)
+	if len(ids) == 0 {
+		return ""
+	}
+	idCounts := map[string]int{}
+	for _, id := range ids {
+		if c, ok := idCounts[id]; ok {
+			idCounts[id] = c + 1
+		} else {
+			idCounts[id] = 1
+		}
+	}
+
+	var topKey string
+	var topCount int
+	for k, v := range idCounts {
+		if v > topCount {
+			topKey = k
+			topCount = v
+		}
+	}
+
+	if len(idCounts) > 1 && name != consts.VariousArtists {
+		log.Warn(ctx, "Multiple MBIDs found for "+entityName, "name", name, "mbids", idCounts)
+	}
+	return topKey
+}
diff --git a/persistence/helpers_test.go b/persistence/helpers_test.go
index 4c4a10f0c..4a9f05a46 100644
--- a/persistence/helpers_test.go
+++ b/persistence/helpers_test.go
@@ -1,6 +1,7 @@
 package persistence
 
 import (
+	"context"
 	"time"
 
 	"github.com/Masterminds/squirrel"
@@ -61,4 +62,16 @@ var _ = Describe("Helpers", func() {
 			Expect(err).To(BeNil())
 		})
 	})
+
+	Describe("getMbzId", func() {
+		It(`returns "" when no ids are passed`, func() {
+			Expect(getMbzId(context.TODO(), " ", "", "")).To(Equal(""))
+		})
+		It(`returns the only id passed`, func() {
+			Expect(getMbzId(context.TODO(), "1234 ", "", "")).To(Equal("1234"))
+		})
+		It(`returns the id with higher frequency`, func() {
+			Expect(getMbzId(context.TODO(), "1 2 3 4 1", "", "")).To(Equal("1"))
+		})
+	})
 })