def merge_hints(query_track, consensus_track, part, do_quick_match):
# If we did a quick match, read tags, as it may have much better tags.
track_title = query_track.name
if do_quick_match:
track_title = improve_from_tag(track_title, part, 'title')
merged_track = Media.Track(
index=int(query_track.index) if query_track.index is not None else -1,
album=toBytes(consensus_track.album),
artist=toBytes(consensus_track.artist),
title=toBytes(track_title),
disc=toBytes(consensus_track.disc),
album_thumb_url=toBytes(consensus_track.album_thumb_url),
artist_thumb_url=toBytes(consensus_track.artist_thumb_url),
year=toBytes(consensus_track.year),
album_guid=toBytes(consensus_track.album_guid),
artist_guid=toBytes(consensus_track.artist_guid))
merged_track.parts.append(part)
#if DEBUG:
# merged_track.name = toBytes(merged_track.name + ' [MERGED GN MISS]')
return merged_track
def merge_hints(query_track, consensus_track, part, do_quick_match):
# If we did a quick match, read tags, as it may have much better tags.
track_title = query_track.name
if do_quick_match:
track_title = improve_from_tag(track_title, part, 'title')
merged_track = Media.Track(
index=int(query_track.index) if query_track.index is not None else -1,
album=toBytes(consensus_track.album),
artist=toBytes(consensus_track.artist),
title=toBytes(track_title),
disc=toBytes(consensus_track.disc),
album_thumb_url=toBytes(consensus_track.album_thumb_url),
artist_thumb_url=toBytes(consensus_track.artist_thumb_url),
year=toBytes(consensus_track.year),
album_guid=toBytes(consensus_track.album_guid),
artist_guid=toBytes(consensus_track.artist_guid))
merged_track.parts.append(part)
#if DEBUG:
# merged_track.name = toBytes(merged_track.name + ' [MERGED GN MISS]')
return merged_track
def merge_hints(query_track, consensus_track, part, do_quick_match):
# If we did a quick match, read tags, as it may have much better tags.
track_title = query_track.name
if do_quick_match:
track_title = improve_from_tag(track_title, part, 'title')
# We don't want to use consensus disc numbers, since tags are more reliable. It's common for bonus discs, etc. to get "split".
try: disc = improve_from_tag('1', part, 'discnumber').split('/')[0].split('of')[0].strip()
except: disc = '1'
merged_track = Media.Track(
index=int(query_track.index) if (query_track.index is not None and str(query_track.index).isdigit()) else -1,
album=toBytes(consensus_track.album),
artist=toBytes(consensus_track.artist),
title=toBytes(track_title),
disc=disc,
album_thumb_url=toBytes(consensus_track.album_thumb_url),
artist_thumb_url=toBytes(consensus_track.artist_thumb_url),
year=toBytes(consensus_track.year),
album_guid=toBytes(consensus_track.album_guid),
artist_guid=toBytes(consensus_track.artist_guid))
merged_track.parts.append(part)
return merged_track
def merge_hints(query_track, consensus_track, part, do_quick_match):
# If we did a quick match, read tags, as it may have much better tags.
track_title = query_track.name
if do_quick_match:
track_title = improve_from_tag(track_title, part, 'title')
# We don't want to use consensus disc numbers, since tags are more reliable. It's common for bonus discs, etc. to get "split".
try:
disc = improve_from_tag(
'1', part, 'discnumber').split('/')[0].split('of')[0].strip()
except:
disc = '1'
merged_track = Media.Track(
index=int(query_track.index) if
(query_track.index is not None
and str(query_track.index).isdigit()) else -1,
album=toBytes(consensus_track.album),
artist=toBytes(consensus_track.artist),
title=toBytes(track_title),
disc=disc,
album_thumb_url=toBytes(consensus_track.album_thumb_url),
artist_thumb_url=toBytes(consensus_track.artist_thumb_url),
year=toBytes(consensus_track.year),
album_guid=toBytes(consensus_track.album_guid),
artist_guid=toBytes(consensus_track.artist_guid))
merged_track.parts.append(part)
return merged_track
def improve_from_tag(existing, file, tag):
tags = None
try:
tags = mutagen.File(file, easy=True)
except:
Log('There was an exception thrown reading tags.')
if tags and tag in tags:
existing = tags[tag][0]
return toBytes(existing)
def lookup(query_list, result_list, language=None, fingerprint=False, mixed=False, multiple=False, do_quick_match=False):
# This shouldn't happen, but be safe.
if len(query_list) == 0:
return (0, 0, 0)
# See if input looks like a sane album
(sane_input_tracks, unique_input_albums, input_discs) = compute_input_sanity(query_list)
# Build up the query with the contents of the query list.
args = ''
parts = {}
Log('Running Gracenote match on %d tracks with fingerprinting: %d and mixedContent: %d and multiple: %d' % (len(query_list), fingerprint, mixed, multiple))
for i, track in enumerate(query_list):
# We need to pass at least a path and an identifier for each track that we know about.
args += '&tracks[%d].path=%s' % (i, quote(track.parts[0], ''))
args += '&tracks[%d].userData=%d' % (i, i)
# Keep track of the identifier -> part mapping so we can reassemble later.
parts[i] = track.parts[0]
if track.name:
args += '&tracks[%d].title=%s' % (i, quote(toBytes(track.title or track.name), ''))
if track.artist and track.artist != 'Various Artists':
args += '&tracks[%d].artist=%s' % (i, quote(toBytes(track.artist), ''))
if track.album_artist:
args += '&tracks[%d].albumArtist=%s' % (i, quote(toBytes(track.album_artist), ''))
elif track.artist and track.artist != 'Various Artists':
args += '&tracks[%d].albumArtist=%s' % (i, quote(toBytes(track.artist), ''))
if track.album and track.album != '[Unknown Album]':
args += '&tracks[%d].album=%s' % (i, quote(toBytes(track.album), ''))
if track.index:
args += '&tracks[%d].index=%s' % (i, track.index)
if track.disc:
args += '&tracks[%d].parentIndex=%s' % (i, track.disc)
Log(' - %s/%s - %s/%s - %s' % (toBytes(track.artist), toBytes(track.album), toBytes(track.disc), toBytes(track.index), toBytes(track.name)))
url = 'http://127.0.0.1:32400/services/gracenote/search?fingerprint=%d&mixedContent=%d&multiple=%d%s&lang=%s' % (fingerprint, mixed, multiple, args, language)
try:
res = minidom.parse(urlopen(url))
except Exception, e:
Log('Error parsing Gracenote response: ' + str(e))
return (0, 0, 0)
def improve_from_tag(existing, file, tag):
tags = mutagen.File(file, easy=True)
if tags and tag in tags:
existing = tags[tag][0]
return toBytes(existing)
def Scan(path, files, media_list, subdirs, language=None, root=None, respect_tags=False):
# Scan for audio files.
AudioFiles.Scan(path, files, media_list, subdirs, root)
root_str = root or ''
loc_str = os.path.join(root_str, path)
Log('Scanning: ' + loc_str)
Log('Files: ' + str(files))
Log('Subdirs: ' + str(subdirs))
# Look at the files and determine whether we can do a quick match (minimal tag parsing).
do_quick_match = True
mixed = False
# Make sure we're looking at a leaf directory (no audio files below here).
if len(subdirs) > 0:
Log('Found directories below this one; won\'t attempt quick matching.')
do_quick_match = False
if files:
# Make sure we're not sitting in the section root.
parent_path = os.path.split(files[0])[0]
if parent_path == root:
Log('File(s) are in section root; doing expensive matching with mixed content.')
do_quick_match = False
mixed = True
# Make sure we have reliable track indices for all files and there are no dupes.
tracks = {}
for f in files:
try:
index = re.search(r'^([0-9]{1,2})[^0-9].*', os.path.split(f)[-1]).groups(0)[0]
except:
do_quick_match = False
Log('Couldn\'t find track indices in all filenames; doing expensive matching.')
break
if tracks.get(index):
do_quick_match = False
mixed = True
Log('Found duplicate track index: %s; doing expensive matching with mixed content.' % index)
break
else:
tracks[index] = True
# Read the first track's tags to check for milti-disc and VA.
if do_quick_match:
disc = album_artist = None
try:
(artist, album, title, track, disc, album_artist, compil) = AudioFiles.getInfoFromTag(files[0], language)
except:
Log('Exception reading tags from first file; doing expensive matching.')
do_quick_match = False
# Make sure we are on the first disc.
if disc is not None and disc > 1:
Log('Skipping quick match because of non-first disc.')
do_quick_match = False
# We want to read all the tags for VA albums to pick up track artists.
if album_artist is not None and album_artist == 'Various Artists':
Log('Skipping quick match for Various Artists album.')
do_quick_match = False
artist = None
album = None
if do_quick_match:
Log('Doing quick match')
# See if we have some consensus on artist/album by reading a few tags.
for i in range(3):
if i < len(files):
this_artist = this_album = tags = None
try: tags = mutagen.File(files[i], easy=True)
except: Log('There was an exception thrown reading tags.')
if tags:
# See if there's an album artist tag.
album_artist_tags = [t for t in ['albumartist', 'TPE2', 'performer'] if t in tags]
album_artist_tag = album_artist_tags[0] if len(album_artist_tags) else None
this_artist = tags[album_artist_tag][0] if album_artist_tag else tags['artist'][0] if 'artist' in tags else None
this_album = tags['album'][0] if 'album' in tags else None
if artist and artist != this_artist:
Log('Found different artists in tags (%s vs. %s); doing expensive matching.' % (artist, this_artist))
do_quick_match = False
break
if album and album != this_album:
Log('Found different albums in tags (%s vs. %s); doing expensive matching.' % (artist, this_artist))
do_quick_match = False
break
artist = this_artist
album = this_album
if not artist or not album:
Log('Couldn\'t determine unique artist or album from tags; doing expensive matching.')
do_quick_match = False
query_list = []
result_list = []
fingerprint = False
# Directory looks clean, let's build a query list directly from info gleaned from file names.
if do_quick_match:
Log('Building query list for quickmatch with artist: %s, album: %s' % (artist, album))
# Determine if the artist and/or album appears in all filenames, since we'll want to strip these out for clean titles.
strip_artist = True if len([f for f in files if artist.lower() in Unicodize(os.path.basename(f), language).lower()]) == len(files) else False
strip_album = True if len([f for f in files if album.lower() in Unicodize(os.path.basename(f), language).lower()]) == len(files) else False
for f in files:
try:
filename = os.path.splitext(os.path.split(f)[1])[0]
(head, index, title) = re.split(r'^([0-9]{1,2})', filename)
# Replace underscores and dots with spaces.
title = re.sub(r'[_\. ]+', ' ', title)
# Things in parens seem to confuse Gracenote, so let's strip them out.
title = re.sub(r' ?\(.*\)', '', title)
# Remove artist name from title if it appears in all of them.
if strip_artist and len(files) > 2:
title = re.sub(r'(?i)' + artist, '', title)
# Remove album title from title if it appears in all of them.
if strip_album and len(files) > 2:
title = re.sub(r'(?i)' + album, '', title)
# Remove any remaining index-, artist-, and album-related cruft from the head of the track title.
title = re.sub(r'^[\W\-]+', '', title).strip()
# Last chance for artist or album prefix.
if not strip_artist and Unicodize(title, language).lower().find(artist.lower()) == 0:
title = title[len(artist):]
if not strip_album and Unicodize(title, language).lower().find(album.lower()) == 0:
title = title[len(album):]
t = Media.Track(artist=toBytes(artist), album=toBytes(album), title=toBytes(title), index=int(index))
t.parts.append(f)
Log(' - Adding: %s - %s' % (index, title))
query_list.append(t)
except Exception as e:
Log('Error preparing tracks for quick matching: ' + str(e))
# Otherwise, let's do old school directory crawling and tag reading.
else:
AudioFiles.Process(path, files, media_list, subdirs, root)
query_list = list(media_list)
# Try as-is first (ask for everything at once).
discs = [query_list]
final_match = run_queries(discs, result_list, language, fingerprint, mixed, do_quick_match)
# If the match was still shitty, and it looks like we have multiple discs, try splitting.
if final_match < 75:
discs = group_tracks_by_disc(query_list)
if len(discs) > 1:
Log('Result still looked bad, we will try splitting into separate per-disc queries.')
other_result_list = []
other_match = run_queries(discs, other_result_list, language, fingerprint, mixed, do_quick_match)
if other_match > final_match:
Log('The split result was best, we will use it.')
result_list = other_result_list
final_match = other_match
# If we have a crappy match, don't use it.
if final_match < 50.0:
Log('That was terrible, let us not use it.')
result_list = []
# Finalize the results.
used_tags = False
del media_list[:]
if len(result_list) > 0:
# Gracenote results.
for result in result_list:
media_list.append(result)
else:
# We bailed during the GN lookup, fall back to tags.
used_tags = True
AudioFiles.Process(path, files, media_list, subdirs, root)
# If we wanted to respect tags, then make sure we used tags.
if not used_tags and respect_tags:
# Let's grab tag results, and then set GUIDs we found.
tag_media_list = []
AudioFiles.Process(path, files, tag_media_list, subdirs, root)
# Now suck GN data out.
path_map = {}
for track in media_list:
path_map[track.parts[0]] = track
for track in tag_media_list:
if track.parts[0] in path_map:
gn_track = path_map[track.parts[0]]
track.guid = gn_track.guid
track.album_guid = gn_track.album_guid
track.artist_guid = gn_track.artist_guid
track.album_thumb_url = gn_track.album_thumb_url
track.artist_thumb_url = gn_track.artist_thumb_url
# If the tags failed, fill in key data from Gracenote.
if track.album == '[Unknown Album]':
track.album = gn_track.album
if track.artist == '[Unknown Artist]':
track.artist = gn_track.artist
media_list[:] = tag_media_list
average_album_ratio = total_album_ratio / len(query_list)
# If we've got really excellent track matches on a good number of tracks, then it's likely
# that the GN match is just calling the artist different (VA vs artist, etc.) Prefer the name
# in the tag if we have one and it's consistent.
#
track_min_ratio = 0.88
if average_album_ratio > 0.90:
track_min_ratio = 0.75
if average_album_ratio > 0.98 and number_of_matched_tracks == len(query_list):
track_min_ratio = 0.50
Log('Track average lev ratio %f, album lev ratio %f, required track ratio: %f' % (average_track_ratio, average_album_ratio, track_min_ratio))
if len(query_list) >= 4 and average_track_ratio > track_min_ratio:
if number_of_artists == 1:
Log('Using override artist of %s' % toBytes(query_list[0].artist))
artist_override = query_list[0].artist
elif len(query_list) < 4 or average_track_ratio < 0.75 or ratio < 0.20:
return (0, 0, 0)
# Check for Various Artists albums which come back matching to an artist, or movie name.
number_of_album_artists = len(set([q.album_artist for q in query_list if q.album_artist]))
if number_of_artists > 1 and number_of_album_artists == 1 and query_list[0].album_artist and LevenshteinRatio(query_list[0].album_artist, 'Various Artists') > 0.9:
Log('Using override artist of Various Artists')
artist_override = 'Various Artists'
# Restore track artists from tags if necessary.
for i, query_track in enumerate(query_list):
if str(i) in matched_tracks:
track = matched_tracks[str(i)]
if query_track.artist and not track.getAttribute('originalTitle'):
def lookup(query_list,
result_list,
language=None,
fingerprint=False,
mixed=False,
multiple=False,
do_quick_match=False):
# This shouldn't happen, but be safe.
if len(query_list) == 0:
return (0, 0, 0)
# See if input looks like a sane album
(sane_input_tracks, unique_input_albums,
input_discs) = compute_input_sanity(query_list)
# Build up the query with the contents of the query list.
args = ''
parts = {}
Log('Running Gracenote match on %d tracks with fingerprinting: %d and mixedContent: %d and multiple: %d'
% (len(query_list), fingerprint, mixed, multiple))
for i, track in enumerate(query_list):
# We need to pass at least a path and an identifier for each track that we know about.
args += '&tracks[%d].path=%s' % (i, quote(track.parts[0], ''))
args += '&tracks[%d].userData=%d' % (i, i)
# Keep track of the identifier -> part mapping so we can reassemble later.
parts[i] = track.parts[0]
if track.name:
args += '&tracks[%d].title=%s' % (
i, quote(toBytes(track.title or track.name), ''))
if track.artist and track.artist != 'Various Artists':
args += '&tracks[%d].artist=%s' % (
i, quote(toBytes(track.artist), ''))
if track.album_artist:
args += '&tracks[%d].albumArtist=%s' % (
i, quote(toBytes(track.album_artist), ''))
elif track.artist and track.artist != 'Various Artists':
args += '&tracks[%d].albumArtist=%s' % (
i, quote(toBytes(track.artist), ''))
if track.album and track.album != '[Unknown Album]':
args += '&tracks[%d].album=%s' % (i, quote(toBytes(track.album),
''))
if track.index:
args += '&tracks[%d].index=%s' % (i, track.index)
if track.disc:
args += '&tracks[%d].parentIndex=%s' % (i, track.disc)
Log(' - %s/%s - %s/%s - %s' %
(toBytes(track.artist), toBytes(track.album), toBytes(
track.disc), toBytes(track.index), toBytes(track.name)))
url = 'http://127.0.0.1:32400/services/gracenote/search?fingerprint=%d&mixedContent=%d&multiple=%d%s&lang=%s' % (
fingerprint, mixed, multiple, args, language)
try:
res = minidom.parse(urlopen(url))
except Exception, e:
Log('Error parsing Gracenote response: ' + str(e))
return (0, 0, 0)
def Scan(path,
files,
media_list,
subdirs,
language=None,
root=None,
respect_tags=False):
# Scan for audio files.
AudioFiles.Scan(path, files, media_list, subdirs, root)
root_str = root or ''
loc_str = os.path.join(root_str, path)
Log('Scanning: ' + loc_str)
Log('Files: ' + str(files))
Log('Subdirs: ' + str(subdirs))
# Look at the files and determine whether we can do a quick match (minimal tag parsing).
do_quick_match = True
mixed = False
# Make sure we're looking at a leaf directory (no audio files below here).
if len(subdirs) > 0:
Log('Found directories below this one; won\'t attempt quick matching.')
do_quick_match = False
if files:
# Make sure we're not sitting in the section root.
parent_path = os.path.split(files[0])[0]
if parent_path == root:
Log('File(s) are in section root; doing expensive matching with mixed content.'
)
do_quick_match = False
mixed = True
# Make sure we have reliable track indices for all files and there are no dupes.
tracks = {}
for f in files:
try:
index = re.search(r'^([0-9]{1,2})[^0-9].*',
os.path.split(f)[-1]).groups(0)[0]
except:
do_quick_match = False
Log('Couldn\'t find track indices in all filenames; doing expensive matching.'
)
break
if tracks.get(index):
do_quick_match = False
mixed = True
Log('Found duplicate track index: %s; doing expensive matching with mixed content.'
% index)
break
else:
tracks[index] = True
# Read the first track's tags to check for milti-disc and VA.
if do_quick_match:
disc = album_artist = None
try:
(artist, album, title, track, disc, album_artist,
compil) = AudioFiles.getInfoFromTag(files[0], language)
except:
Log('Exception reading tags from first file; doing expensive matching.'
)
do_quick_match = False
# Make sure we are on the first disc.
if disc is not None and disc > 1:
Log('Skipping quick match because of non-first disc.')
do_quick_match = False
# We want to read all the tags for VA albums to pick up track artists.
if album_artist is not None and album_artist == 'Various Artists':
Log('Skipping quick match for Various Artists album.')
do_quick_match = False
artist = None
album = None
if do_quick_match:
Log('Doing quick match')
# See if we have some consensus on artist/album by reading a few tags.
for i in range(3):
if i < len(files):
this_artist = this_album = tags = None
try:
tags = mutagen.File(files[i], easy=True)
except:
Log('There was an exception thrown reading tags.')
if tags:
# See if there's an album artist tag.
album_artist_tags = [
t for t in ['albumartist', 'TPE2', 'performer']
if t in tags
]
album_artist_tag = album_artist_tags[0] if len(
album_artist_tags) else None
this_artist = tags[album_artist_tag][
0] if album_artist_tag else tags['artist'][
0] if 'artist' in tags else None
this_album = tags['album'][
0] if 'album' in tags else None
if artist and artist != this_artist:
Log('Found different artists in tags (%s vs. %s); doing expensive matching.'
% (artist, this_artist))
do_quick_match = False
break
if album and album != this_album:
Log('Found different albums in tags (%s vs. %s); doing expensive matching.'
% (album, this_album))
do_quick_match = False
break
artist = this_artist
album = this_album
if not artist or not album:
Log('Couldn\'t determine unique artist or album from tags; doing expensive matching.'
)
do_quick_match = False
query_list = []
result_list = []
fingerprint = False
# Directory looks clean, let's build a query list directly from info gleaned from file names.
if do_quick_match:
Log('Building query list for quickmatch with artist: %s, album: %s'
% (artist, album))
# Determine if the artist and/or album appears in all filenames, since we'll want to strip these out for clean titles.
strip_artist = True if len([
f for f in files if artist.lower() in Unicodize(
os.path.basename(f), language).lower()
]) == len(files) else False
strip_album = True if len([
f for f in files if album.lower() in Unicodize(
os.path.basename(f), language).lower()
]) == len(files) else False
for f in files:
try:
filename = os.path.splitext(os.path.split(f)[1])[0]
(head, index, title) = re.split(r'^([0-9]{1,2})', filename)
# Replace underscores and dots with spaces.
title = re.sub(r'[_\. ]+', ' ', title)
# Things in parens seem to confuse Gracenote, so let's strip them out.
title = re.sub(r' ?\(.*\)', '', title)
# Remove artist name from title if it appears in all of them.
if strip_artist and len(files) > 2:
title = re.sub(r'(?i)' + artist, '', title)
# Remove album title from title if it appears in all of them.
if strip_album and len(files) > 2:
title = re.sub(r'(?i)' + album, '', title)
# Remove any remaining index-, artist-, and album-related cruft from the head of the track title.
title = re.sub(r'^[\W\-]+', '', title).strip()
# Last chance for artist or album prefix.
if not strip_artist and Unicodize(
title, language).lower().find(artist.lower()) == 0:
title = title[len(artist):]
if not strip_album and Unicodize(
title, language).lower().find(album.lower()) == 0:
title = title[len(album):]
t = Media.Track(artist=toBytes(artist),
album=toBytes(album),
title=toBytes(title),
index=int(index))
t.parts.append(f)
Log(' - Adding: %s - %s' % (index, title))
query_list.append(t)
except Exception as e:
Log('Error preparing tracks for quick matching: ' + str(e))
# Otherwise, let's do old school directory crawling and tag reading.
else:
AudioFiles.Process(path, files, media_list, subdirs, root)
query_list = list(media_list)
# Try as-is first (ask for everything at once).
discs = [query_list]
final_match = run_queries(discs, result_list, language, fingerprint,
mixed, do_quick_match)
# If the match was still shitty, and it looks like we have multiple discs, try splitting.
if final_match < 75:
discs = group_tracks_by_disc(query_list)
if len(discs) > 1:
Log('Result still looked bad, we will try splitting into separate per-disc queries.'
)
other_result_list = []
other_match = run_queries(discs, other_result_list, language,
fingerprint, mixed, do_quick_match)
if other_match > final_match:
Log('The split result was best, we will use it.')
result_list = other_result_list
final_match = other_match
# If we have a crappy match, don't use it.
if final_match < 50.0:
Log('That was terrible, let us not use it.')
result_list = []
# Finalize the results.
used_tags = False
del media_list[:]
if len(result_list) > 0:
# Gracenote results.
for result in result_list:
media_list.append(result)
else:
# We bailed during the GN lookup, fall back to tags.
used_tags = True
AudioFiles.Process(path, files, media_list, subdirs, root)
# If we wanted to respect tags, then make sure we used tags.
if not used_tags and respect_tags:
# Let's grab tag results, and then set GUIDs we found.
tag_media_list = []
AudioFiles.Process(path, files, tag_media_list, subdirs, root)
# Now suck GN data out.
path_map = {}
for track in media_list:
path_map[track.parts[0]] = track
for track in tag_media_list:
if track.parts[0] in path_map:
gn_track = path_map[track.parts[0]]
track.guid = gn_track.guid
track.album_guid = gn_track.album_guid
track.artist_guid = gn_track.artist_guid
track.album_thumb_url = gn_track.album_thumb_url
track.artist_thumb_url = gn_track.artist_thumb_url
# If the tags failed, fill in key data from Gracenote.
if track.album == '[Unknown Album]':
track.album = gn_track.album
if track.artist == '[Unknown Artist]':
track.artist = gn_track.artist
media_list[:] = tag_media_list
# that the GN match is just calling the artist different (VA vs artist, etc.) Prefer the name
# in the tag if we have one and it's consistent.
#
track_min_ratio = 0.88
if average_album_ratio > 0.90:
track_min_ratio = 0.75
if average_album_ratio > 0.98 and number_of_matched_tracks == len(
query_list):
track_min_ratio = 0.50
Log('Track average lev ratio %f, album lev ratio %f, required track ratio: %f'
% (average_track_ratio, average_album_ratio, track_min_ratio))
if len(query_list) >= 4 and average_track_ratio > track_min_ratio:
if number_of_artists == 1:
Log('Using override artist of %s' %
toBytes(query_list[0].artist))
artist_override = query_list[0].artist
elif len(query_list
) < 4 or average_track_ratio < 0.75 or ratio < 0.20:
return (0, 0, 0)
# Check for Various Artists albums which come back matching to an artist, or movie name.
number_of_album_artists = len(
set([q.album_artist for q in query_list if q.album_artist]))
if number_of_artists > 1 and number_of_album_artists == 1 and query_list[
0].album_artist and LevenshteinRatio(query_list[0].album_artist,
'Various Artists') > 0.9:
Log('Using override artist of Various Artists')
artist_override = 'Various Artists'
# Restore track artists from tags if necessary.
def improve_from_tag(existing, file, tag):
tags = mutagen.File(file, easy=True)
if tags and tag in tags:
existing = tags[tag][0]
return toBytes(existing)
def Scan(path, files, mediaList, subdirs, exts, root=None):
files_to_whack = []
plexignore_files = []
plexignore_dirs = []
use_unicode = os.path.supports_unicode_filenames
# Build a list of things to ignore based on a .plexignore file in this dir.
if root and Utils.ContainsFile(files, '.plexignore'):
ParsePlexIgnore(os.path.join(root,path,'.plexignore'), plexignore_files, plexignore_dirs)
# Also look for a .plexignore in the 'root' for this source.
if root and files and root != os.path.dirname(files[0]):
if Utils.ContainsFile(os.listdir(root), '.plexignore'):
ParsePlexIgnore(os.path.join(root,'.plexignore'), plexignore_files, plexignore_dirs)
for f in files:
# Only use unicode if it's supported, which it is on Windows and OS X,
# but not Linux. This allows things to work with non-ASCII characters
# without having to go through a bunch of work to ensure the Linux
# filesystem is UTF-8 "clean".
#
if use_unicode:
try: filename = unicode(f.decode('utf-8'))
except: files_to_whack.append(f)
else:
filename = f
(basename, ext) = os.path.splitext(f)
basename = os.path.basename(basename)
# If extension is wrong, don't include.
if not ext.lower()[1:] in exts:
files_to_whack.append(f)
# Broken symlinks and zero byte files need not apply.
if os.path.exists(filename) == False or os.path.getsize(filename) == 0:
files_to_whack.append(f)
# Remove unreadable files.
if not os.access(filename, os.R_OK):
# If access() claims the file is unreadable, try to read a byte just to be sure.
try:
read_file = open(f,'rb')
read_file.read(1)
read_file.close()
except:
files_to_whack.append(f)
# Remove hidden files.
if len(basename) == 0 or basename[0] == '.':
files_to_whack.append(f)
# Remove .plexignore file regex matches.
for rx in plexignore_files:
if re.match(rx, os.path.basename(f), re.IGNORECASE):
files_to_whack.append(f)
# Remove files that look DRM'd.
try:
mp4_file = open(f, 'rb')
codec = mp4.MP4Info(mp4.Atoms(mp4_file), mp4_file).codec
if codec in ['drms', 'enca', 'encv']:
Utils.Log('Skipping file %s because it looks DRM-protected (has codec: %s)' % (toBytes(f), codec))
files_to_whack.append(f)
except:
pass
# See what directories to ignore.
ignore_dirs_total = IGNORE_DIRS
if len(path) == 0:
ignore_dirs_total += ROOT_IGNORE_DIRS
dirs_to_whack = []
for dir in subdirs:
# See which directories to get rid of.
baseDir = os.path.basename(dir)
for rx in ignore_dirs_total:
if re.match(rx, baseDir, re.IGNORECASE):
dirs_to_whack.append(dir)
break
# Add glob matches from .plexignore before whacking.
for pattern in plexignore_dirs:
for match in glob.glob(pattern):
if os.path.isdir(match):
dirs_to_whack.append(os.path.dirname(match))
else:
files_to_whack.append(match)
# Whack files.
files_to_whack = list(set(files_to_whack))
for f in files_to_whack:
if f in files:
files.remove(f)
# Remove the directories.
dirs_to_whack = list(set(dirs_to_whack))
for f in dirs_to_whack:
if f in subdirs:
subdirs.remove(f)
# that the GN match is just calling the artist different (VA vs artist, etc.) Prefer the name
# in the tag if we have one and it's consistent.
#
track_min_ratio = 0.88
if average_album_ratio > 0.90:
track_min_ratio = 0.75
if average_album_ratio > 0.98 and number_of_matched_tracks == len(
query_list):
track_min_ratio = 0.50
Log('Track average lev ratio %f, album lev ratio %f, required track ratio: %f'
% (average_track_ratio, average_album_ratio, track_min_ratio))
if len(query_list) >= 4 and average_track_ratio > track_min_ratio:
if number_of_artists == 1:
Log('Using override artist of %s' %
toBytes(query_list[0].artist))
artist_override = query_list[0].artist
elif len(query_list
) < 4 or average_track_ratio < 0.75 or ratio < 0.20:
return (0, 0, 0)
# Check for Various Artists albums which come back matching to an artist, or movie name.
number_of_album_artists = len(
set([q.album_artist for q in query_list if q.album_artist]))
if number_of_artists > 1 and number_of_album_artists == 1 and query_list[
0].album_artist and LevenshteinRatio(query_list[0].album_artist,
'Various Artists') > 0.9:
Log('Using override artist of Various Artists')
artist_override = 'Various Artists'
# Restore track artists from tags if necessary.
average_album_ratio = total_album_ratio / len(query_list)
# If we've got really excellent track matches on a good number of tracks, then it's likely
# that the GN match is just calling the artist different (VA vs artist, etc.) Prefer the name
# in the tag if we have one and it's consistent.
#
track_min_ratio = 0.88
if average_album_ratio > 0.90:
track_min_ratio = 0.75
if average_album_ratio > 0.98 and number_of_matched_tracks == len(query_list):
track_min_ratio = 0.50
Log('Track average lev ratio %f, album lev ratio %f, required track ratio: %f' % (average_track_ratio, average_album_ratio, track_min_ratio))
if len(query_list) >= 4 and average_track_ratio > track_min_ratio:
if number_of_artists == 1:
Log('Using override artist of %s' % toBytes(query_list[0].artist))
artist_override = query_list[0].artist
elif len(query_list) < 4 or average_track_ratio < 0.75 or ratio < 0.20:
return (0, 0, 0)
# Check for Various Artists albums which come back matching to an artist, or movie name.
number_of_album_artists = len(set([q.album_artist for q in query_list if q.album_artist]))
if number_of_artists > 1 and number_of_album_artists == 1 and query_list[0].album_artist and LevenshteinRatio(query_list[0].album_artist, 'Various Artists') > 0.9:
Log('Using override artist of Various Artists')
artist_override = 'Various Artists'
# Restore track artists from tags if necessary.
for i, query_track in enumerate(query_list):
if str(i) in matched_tracks:
track = matched_tracks[str(i)]
if query_track.artist and not track.getAttribute('originalTitle'):
