def get_by_mbid(cls, mbid):
""" Fetches the artist and releases from MB if not in the database. """
if mbid in cls.blacklisted:
raise cls.Blacklisted()
try:
return cls.objects.get(mbid=mbid)
except cls.DoesNotExist:
pass
artist_data = mb.get_artist(mbid)
if artist_data is None:
return None
if not artist_data:
raise cls.Unknown
artist = Artist(
mbid=mbid, name=artist_data['name'], sort_name=artist_data['sort-name'],
disambiguation=artist_data.get('disambiguation', ''))
try:
artist.save()
except IntegrityError:
# The artist was added while we were querying MB.
return cls.objects.get(mbid=mbid)
# Add a few release groups immediately.
# Sleep 1s to comply with the MB web service.
sleep(1)
LIMIT = 100
release_groups = mb.get_release_groups(mbid, limit=LIMIT, offset=0)
if release_groups:
with transaction.commit_on_success():
for rg_data in release_groups:
# Ignoring releases without a release date or a type.
if rg_data.get('first-release-date') and rg_data.get('type'):
release_group = ReleaseGroup(
artist=artist,
mbid=rg_data['id'],
name=rg_data['title'],
type=rg_data['type'],
date=str_to_date(rg_data['first-release-date']),
is_deleted=False)
release_group.save()
if release_groups is None or len(release_groups) == LIMIT:
# Add the remaining release groups
Job.add_release_groups(artist)
return artist
def calendar(request):
date_str = request.GET.get('date', None)
today = int(date.today().strftime('%Y%m%d'))
if date_str:
if not re.match("^\d{4}(-\d{2})?(-\d{2})?$", date_str):
return redirect('/calendar')
date_int = str_to_date(date_str)
else:
date_int = today
try:
offset = int(request.GET.get('offset', 0))
except ValueError:
return HttpResponseNotFound()
PER_PAGE = 10
limit = PER_PAGE + 1
releases = list(ReleaseGroup.get_calendar(date_int, limit, offset))
if len(releases) == limit:
if releases[0].date == releases[-1].date:
next_date = date_to_str(date_int)
next_offset = offset + PER_PAGE
releases = releases[:-1]
else:
if offset:
i = min(i for i in xrange(limit) if releases[i].date != releases[0].date)
next_date = date_to_str(releases[i].date)
next_offset = 0
releases = releases[:i]
else:
next_date = date_to_str(releases[-1].date)
next_offset = 0
releases = [r for r in releases if r.date != releases[-1].date]
else:
next_date = None
next_offset = 0
for i, release in enumerate(releases):
if i > 0 and release.date == releases[i - 1].date:
release.date_first = None
else:
release.date_first = release.date_str
release.date_str = None
return render(request, 'calendar.html', {
'releases': releases,
'next_date': next_date,
'next_offset': next_offset})
def add_release_groups(mbid):
logging.info('[JOB] Fetching release groups for artist %s' % mbid)
try:
artist = Artist.objects.get(mbid=mbid)
except Artist.DoesNotExist:
logging.warning('[ERR] Cannot find by mbid, skipping' % mbid)
return True
LIMIT = 100
offset = 0
while True:
tools.sleep()
logging.info('[JOB] Fetching release groups at offset %d' % offset)
release_groups = mb.get_release_groups(mbid,
limit=LIMIT,
offset=offset)
if release_groups:
with transaction.commit_on_success():
for rg_data in release_groups:
# Ignoring releases without a release date or a type.
if rg_data.get('first-release-date') and rg_data.get(
'type'):
q = ReleaseGroup.objects.filter(artist=artist,
mbid=rg_data['id'])
if q.exists():
continue
release_group = ReleaseGroup(
artist=artist,
mbid=rg_data['id'],
name=rg_data['title'],
type=rg_data['type'],
date=str_to_date(rg_data['first-release-date']),
is_deleted=False)
release_group.save()
if release_groups is None:
logging.warning('[ERR] MusicBrainz error, retrying')
continue
if len(release_groups) < LIMIT:
break
offset += LIMIT
return True
def add_release_groups(mbid):
logging.info('[JOB] Fetching release groups for artist %s' % mbid)
try:
artist = Artist.objects.get(mbid=mbid)
except Artist.DoesNotExist:
logging.warning('[ERR] Cannot find by mbid, skipping' % mbid)
return True
LIMIT = 100
offset = 0
while True:
tools.sleep()
logging.info('[JOB] Fetching release groups at offset %d' % offset)
release_groups = mb.get_release_groups(mbid, limit=LIMIT, offset=offset)
if release_groups:
with transaction.commit_on_success():
for rg_data in release_groups:
# Ignoring releases without a release date or a type.
if rg_data.get('first-release-date') and rg_data.get('type'):
q = ReleaseGroup.objects.filter(
artist=artist, mbid=rg_data['id'])
if q.exists():
continue
release_group = ReleaseGroup(
artist=artist,
mbid=rg_data['id'],
name=rg_data['title'],
type=rg_data['type'],
date=str_to_date(rg_data['first-release-date']),
is_deleted=False)
release_group.save()
if release_groups is None:
logging.warning('[ERR] MusicBrainz error, retrying')
continue
if len(release_groups) < LIMIT:
break
offset += LIMIT
return True
def check():
logging.info('Start checking artists')
checked_artists = 0
checked_release_groups = 0
day = datetime.datetime.utcnow().day
artist = None
while True:
# Get the next artist.
artists = Artist.objects.order_by('mbid')
if artist:
artists = artists.filter(mbid__gt=artist.mbid)
try:
artist = artists[0]
except IndexError:
break # last artist
checked_artists += 1
# Artist names don't change that often. Update artists at most once
# a month, unless we are debugging.
if DEBUG or day == 1:
jobs.process()
tools.sleep()
logging.info('Updating artist %s' % artist.mbid)
artist_data = mb.get_artist(artist.mbid)
if not artist_data:
# TODO: musicbrainz/network error or deleted?
logging.warning('Could not fetch artist data')
elif artist_data['id'] != artist.mbid:
# Requested and returned mbids are different if the artist has been merged.
logging.info('Merging into artist %s' % artist_data['id'])
try:
new_artist = Artist.get_by_mbid(artist_data['id'])
except (Artist.Blacklisted, Artist.Unknown):
continue
if not new_artist:
continue
cursor = connection.cursor()
cursor.execute(
"""
UPDATE OR REPLACE "app_userartist"
SET "artist_id" = %s
WHERE "artist_id" = %s
""", [new_artist.id, artist.id])
# Mark release groups as deleted.
n = artist.releasegroup_set.update(is_deleted=True)
logging.info('Deleted %s release groups' % n)
continue
else:
# Update artist info if changed.
updated = False
if artist.name != artist_data['name']:
artist.name = artist_data['name']
updated = True
if artist.sort_name != artist_data['sort-name']:
artist.sort_name = artist_data['sort-name']
updated = True
if artist.disambiguation != artist_data.get('disambiguation', ''):
artist.disambiguation = artist_data.get('disambiguation', '')
updated = True
if updated:
logging.info('Artist changed, updating')
artist.save()
else:
logging.info('Checking artist %s' % artist.mbid)
current = {rg.mbid: rg for rg in ReleaseGroup.objects.filter(artist=artist)}
# Get release groups
LIMIT = 100
offset = 0
while True:
jobs.process()
tools.sleep()
release_groups = mb.get_release_groups(artist.mbid, LIMIT, offset)
if release_groups is None:
logging.warning('Could not fetch release groups, retrying')
continue
logging.info('Fetched %s release groups' % len(release_groups))
with transaction.commit_on_success():
for rg_data in release_groups:
mbid = rg_data['id']
# Ignore releases without a release date or a type.
if not rg_data.get('first-release-date') or not rg_data.get('type'):
if mbid in current:
release_group = current[mbid]
if not release_group.is_deleted:
release_group.is_deleted = True
release_group.save()
logging.info('Deleted release group %s' % mbid)
continue
checked_release_groups += 1
release_date = str_to_date(rg_data['first-release-date'])
if mbid in current:
release_group = current[mbid]
updated = False
if release_group.is_deleted:
release_group.is_deleted = False
updated = True
# Work-around MBS-4285.
if release_group.name != rg_data['title'] and rg_data['title']:
release_group.name = rg_data['title']
updated = True
if release_group.type != rg_data['type']:
release_group.type = rg_data['type']
updated = True
if release_group.date != release_date:
release_group.date = release_date
updated = True
if updated:
release_group.save()
logging.info('Updated release group %s' % mbid)
del current[mbid]
else:
release_group = ReleaseGroup(
artist=artist,
mbid=rg_data['id'],
name=rg_data['title'],
type=rg_data['type'],
date=release_date,
is_deleted=False)
release_group.save()
logging.info('Created release group %s' % mbid)
# Notify users
cursor = connection.cursor()
cursor.execute(
"""
INSERT INTO "app_notification" ("user_id", "release_group_id")
SELECT "app_userartist"."user_id", "app_releasegroup"."id"
FROM "app_userartist"
JOIN "app_artist" ON "app_artist"."id" = "app_userartist"."artist_id"
JOIN "app_releasegroup" ON "app_releasegroup"."artist_id" = "app_artist"."id"
WHERE "app_releasegroup"."id" = %s
""", [release_group.id])
logging.info('Notified %d users' % cursor.rowcount)
if len(release_groups) < LIMIT: break
offset += LIMIT
with transaction.commit_on_success():
for mbid in current:
release_group = current[mbid]
if not release_group.is_deleted:
release_group.is_deleted = True
release_group.save()
logging.info('Deleted release group %s' % mbid)
logging.info('Checked %d artists and %d release groups' % (checked_artists, checked_release_groups))
def check():
logging.info('Start checking artists')
sent_notifications = 0
checked_artists = 0
checked_release_groups = 0
day = datetime.datetime.utcnow().day
artist = None
while True:
# Get the next artist.
artists = Artist.objects.order_by('mbid')
if artist:
artists = artists.filter(mbid__gt=artist.mbid)
try:
artist = artists[0]
except IndexError:
break # last artist
checked_artists += 1
# Artist names don't change that often. Update artists at most 3 times
# a month, unless we are debugging.
if DEBUG or day in (1, 11, 21):
jobs.process()
tools.sleep()
logging.info('Updating artist %s' % artist.mbid)
artist_data = mb.get_artist(artist.mbid)
if not artist_data:
# TODO: musicbrainz/network error or deleted?
logging.warning('Could not fetch artist data')
elif artist_data['id'] != artist.mbid:
# Requested and returned mbids are different if the artist has been merged.
logging.info('Merging into artist %s' % artist_data['id'])
try:
new_artist = Artist.get_by_mbid(artist_data['id'])
except (Artist.Blacklisted, Artist.Unknown):
continue
if not new_artist:
continue
cursor = connection.cursor()
cursor.execute(
"""
UPDATE OR REPLACE "app_userartist"
SET "artist_id" = %s
WHERE "artist_id" = %s
""", [new_artist.id, artist.id])
# Delete the artist and its release groups.
# Use SQL, delete() is buggy, see Django bug #16426.
# TODO: possible FK constraint fail in app_star.
cursor.execute(
"""
DELETE FROM "app_releasegroup"
WHERE "artist_id" = %s
""", [artist.id])
logging.info('Deleted release groups')
cursor.execute(
"""
DELETE FROM "app_artist"
WHERE "id" = %s
""", [artist.id])
logging.info('Deleted the artist')
continue
else:
# Update artist info if changed.
updated = False
if artist.name != artist_data['name']:
artist.name = artist_data['name']
updated = True
if artist.sort_name != artist_data['sort-name']:
artist.sort_name = artist_data['sort-name']
updated = True
if artist.disambiguation != artist_data.get('disambiguation', ''):
artist.disambiguation = artist_data.get('disambiguation', '')
updated = True
if updated:
logging.info('Artist changed, updating')
artist.save()
else:
logging.info('Checking artist %s' % artist.mbid)
current = {rg.mbid: rg for rg in ReleaseGroup.objects.filter(artist=artist)}
# Get release groups
LIMIT = 100
offset = 0
while True:
sent_notifications += notifications.send()
release_groups = mb.get_release_groups(artist.mbid, LIMIT, offset)
if release_groups is None:
logging.warning('Could not fetch release groups, retrying')
continue
logging.info('Fetched %s release groups' % len(release_groups))
with transaction.commit_on_success():
for rg_data in release_groups:
mbid = rg_data['id']
# Ignore releases without a release date or a type.
release_date = str_to_date(rg_data.get('first-release-date'))
if not release_date or not rg_data.get('type'):
if mbid in current:
release_group = current[mbid]
if not release_group.is_deleted:
release_group.is_deleted = True
release_group.save()
logging.info('Deleted release group %s' % mbid)
continue
checked_release_groups += 1
if mbid in current:
release_group = current[mbid]
updated = False
if release_group.is_deleted:
release_group.is_deleted = False
updated = True
# Work-around MBS-4285.
if release_group.name != rg_data['title'] and rg_data['title']:
release_group.name = rg_data['title']
updated = True
if release_group.type != rg_data['type']:
release_group.type = rg_data['type']
updated = True
if release_group.date != release_date:
release_group.date = release_date
updated = True
if updated:
release_group.save()
logging.info('Updated release group %s' % mbid)
del current[mbid]
elif rg_data['title']:
release_group = ReleaseGroup(
artist=artist,
mbid=rg_data['id'],
name=rg_data['title'],
type=rg_data['type'],
date=release_date,
is_deleted=False)
release_group.save()
logging.info('Created release group %s' % mbid)
# Notify users
cursor = connection.cursor()
cursor.execute(
"""
INSERT INTO "app_notification" ("user_id", "release_group_id")
SELECT "app_userartist"."user_id", "app_releasegroup"."id"
FROM "app_userartist"
JOIN "app_artist" ON "app_artist"."id" = "app_userartist"."artist_id"
JOIN "app_releasegroup" ON "app_releasegroup"."artist_id" = "app_artist"."id"
WHERE "app_releasegroup"."id" = %s
""", [release_group.id])
logging.info('Will notify %d users' % cursor.rowcount)
if len(release_groups) < LIMIT: break
offset += LIMIT
with transaction.commit_on_success():
for mbid in current:
release_group = current[mbid]
if not release_group.is_deleted:
release_group.is_deleted = True
release_group.save()
logging.info('Deleted release group %s' % mbid)
return (checked_artists, checked_release_groups, sent_notifications)