Merge pull request #2999 from metabrainz/ansh/fix-artist-metadata

fix: Release Group Artist Credits are duplicated
metabrainz · Nov 19, 2024 · 44f7929 · 44f7929
2 parents b1f879e + 15b0ff1
commit 44f7929
Showing 1 changed file with 6 additions and 4 deletions.
diff --git a/mbid_mapping/mapping/mb_artist_metadata_cache.py b/mbid_mapping/mapping/mb_artist_metadata_cache.py
@@ -85,7 +85,7 @@ def create_json_data(self, row):
                     "name": release_group_name,
                     "mbid": release_group_mbid,
                     "artist_credit_name": artist_credit_name,
-                    "artists": release_group_artists
+                    "artists": sorted(release_group_artists, key=lambda x: x.pop("position", float('inf')))
                 }
                 if year is not None:
                     date = str(year)
@@ -199,11 +199,12 @@ def get_metadata_cache_query(self, with_values=False):
                                  , array_agg(DISTINCT rgst.name ORDER BY rgst.name)
                                      FILTER (WHERE rgst.name IS NOT NULL)
                                    AS secondary_types
-                                 , jsonb_agg(jsonb_build_object(
+                                 , jsonb_agg(DISTINCT jsonb_build_object(
                                         'artist_mbid', a2.gid::TEXT,
                                         'artist_credit_name', acn2.name,
-                                        'join_phrase', acn2.join_phrase
-                                   ) ORDER BY acn2.position) AS release_group_artists
+                                        'join_phrase', acn2.join_phrase,
+                                        'position', acn2.position
+                                   )) AS release_group_artists
                               FROM musicbrainz.artist a
                               JOIN musicbrainz.artist_credit_name acn
                                 ON a.id = acn.artist
@@ -427,3 +428,4 @@ def create_mb_artist_metadata_cache(use_lb_conn: bool):
 def incremental_update_mb_artist_metadata_cache(use_lb_conn: bool):
     """ Update the MB metadata cache incrementally """
     incremental_update_metadata_cache(MusicBrainzArtistMetadataCache, MB_ARTIST_METADATA_CACHE_TIMESTAMP_KEY, use_lb_conn)
+