[music] Tracks can have multiple artists
All checks were successful
build & deploy / test (push) Successful in 1m54s
build & deploy / build-and-deploy (push) Successful in 33s

This commit is contained in:
2026-05-28 23:32:09 -04:00
parent 17aed1191d
commit 22956c7c7f
15 changed files with 231 additions and 84 deletions

View File

@ -15,12 +15,26 @@ def clean_artist_name(name: str) -> str:
name = re.split(" w. ", name, flags=re.IGNORECASE)[0].strip()
if " featuring " in name.lower():
name = re.split(" featuring ", name, flags=re.IGNORECASE)[0].strip()
# if " & " in name.lower() and "of the wand" not in name.lower():
# name = re.split("&", name, flags=re.IGNORECASE)[0].strip()
return name
def parse_artist_names(name: str) -> list[str]:
"""Split a combined artist string (e.g. 'A & B') into individual names.
First strips feat./featuring/w. prefixes, then splits on ' & ' to
support collaboration-style credits like 'Matt Sweeney & Bonnie Prince Billy'.
"""
name = clean_artist_name(name)
if " & " in name.lower():
return [
part.strip()
for part in re.split(r"\s+&\s+", name, flags=re.IGNORECASE)
if part.strip()
]
return [name]
def get_or_create_various_artists() -> "Artist":
from music.models import Artist
@ -35,19 +49,19 @@ def deduplicate_tracks(commit=False) -> int:
from music.models import Track
duplicates = (
Track.objects.values("artist", "title")
Track.objects.values("title")
.annotate(dup_count=models.Count("id"))
.filter(dup_count__gt=1)
)
query = models.Q()
for dup in duplicates:
query |= models.Q(artist=dup["artist"], title=dup["title"])
query |= models.Q(title=dup["title"])
duplicate_tracks = Track.objects.filter(query)
for b in duplicate_tracks:
tracks = Track.objects.filter(artist=b.artist, title=b.title)
tracks = Track.objects.filter(title=b.title)
first = tracks.first()
for other in tracks.exclude(id=first.id):
print("Moving scrobbles for", other.id, " to ", first.id)
@ -74,7 +88,7 @@ def condense_albums(commit: bool = False):
for track in Track.objects.all():
albums_to_add = []
duplicates = (
Track.objects.filter(title=track.title, artist=track.artist)
Track.objects.filter(title=track.title)
.exclude(id=track.id)
.exclude(id__in=processed_ids)
)
@ -88,7 +102,7 @@ def condense_albums(commit: bool = False):
track.albums.add(dup_track.album)
# Find out if this track appears more than once
duplicates = Track.objects.filter(title=track.title, artist=track.artist)
duplicates = Track.objects.filter(title=track.title)
if duplicates.count() > 1:
logger.info(f"Track appears more than once, condensing: {track}")