def parse_itunes(fname):
itunes_fname = SAVE_ITUNES_DB_FNAME
if itunes_fname==None:
itunes_fname = 'itunes.db'
parser = xml.sax.make_parser()
handler = ITunesHandler()
parser.setContentHandler(handler)
parser.parse(fname)
parsed_artists = [] #strings of artist names
parsed_artists_lower = [] #lowercase strings of all artist name strings added to parsed_artists
parsed_tracks = []
for track in handler.tracks:
parsed_tracks.append(track)
if type(track.artist) != type(None): # otherwise blank artist field causes error
if unicode.lower(track.artist) not in parsed_artists_lower:
parsed_artists.append(track.artist)
parsed_artists_lower.append(unicode.lower(track.artist))
for i in range(len(parsed_artists)):
parsed_artists[i] = Itunes_Artist(parsed_artists[i]) #convert strings into Itunes_Artist objects
count = 0
total_count = len(parsed_artists)
report_at = [10,20,30,40,50,60,70,80,90,99,1000] #1000 included just in case
for artist in parsed_artists: #itunes_artist objects
for track in parsed_tracks: #itunes_track objects
if type(track.artist) != type(None) and type(artist.name) != type(None) and type(track.track) != type(None):
if unicode.lower(track.artist) == unicode.lower(artist.name) and unicode.lower(track.track) not in artist.track_names_lower:
artist.track_names_lower.append(string.lower(track.track))
artist.tracks.append(track)
elif unicode.lower(track.artist) == unicode.lower(artist.name) and unicode.lower(track.track) in artist.track_names_lower:
artist.duplicate_track_names.append(unicode.lower(track.track))
count += 1
progress = int((float(count) / total_count) * 100)
if progress >= report_at[0]:
print 'Progress: %3d' % (progress),'%'
report_at.remove(report_at[0])
print 'Processing duplicate tracks...'
for artist in parsed_artists:
[artist.duplicate_tracks.append(track) for track in artist.tracks if string.lower(track.track) in artist.duplicate_track_names]
print 'Creating library file...'
library = Itunes_Library(parsed_artists)
print 'Complete! Library being saved as \'%s\''%(itunes_fname)
datamgmt.save(library,itunes_fname)
def process_info(fname='incompleteartists.db',v=False): #get playcount info for all artists, returns a new list of artists
"""Gets all playcount info for all artist objects. v=verbose"""
global artists
import random
artists = datamgmt.load(fname)
fname = 'artists.db'
random.shuffle(artists)
newartists =[]
artist_count = len(artists)
print 'Getting info for each track by %d artists. This could take a very long time (approximately 1 minute per hundred tracks).' % artist_count
for i in range(artist_count):
print '%d/%d' % (i,artist_count)
try:
processed_artist = process_artist(artists[i],v)
except:
try:
processed_artist = process_artist(artists[i],v)
except:
print 'Errors processing',unicode(artists[i])
newartists.append(processed_artist)
artists = newartists
print 'calculating ratios...'
calculate_ratios()
try:
datamgmt.save(artists,fname)
print 'saved successfully'
except:
try:
NETWORK.disable_caching()
datamgmt.save(artists,fname)
print 'data saved successfully'
NETWORK.enable_caching()
except:
global errors
errors = newartists
print 'Error during save. Artists stored as var: errors'
def get_lfm_info(itunes_library):
""" Gets info from last.fm for each Itunes_Artist object in the artists field of an itunes_library object. Saves a list of Hybrid_Track files"""
tracks = [] #list of all tracks of all artists
#create list of all tracks
print 'creating list of all tracks'
for artist in itunes_library.artists:
for track in artist.tracks:
tracks.append(track)
print 'list created, %d tracks to parse'%(len(tracks))
report_at = [5,10,20,30,40,50,60,70,80,90,1000]
total_count = len(tracks)
#lists for fail/success tracks
fails = []
no_fail=[]
#make each Itunes_Track in tracks into a Hybrid_Track
for count in range(total_count):
try:
try:
tracks[count] = Hybrid_Track(tracks[count])
no_fail.append(tracks[count])
except pylast.WSError:
print 'error with', unicode(tracks[count])
fails.append(tracks[count])
except:
print 'ascii error on track',count #this except statement SHOULD be unneccessary now, with the track printed in unicode
fails.append(tracks[count])
# print 'xml error on:',tracks[count]
progress = int((float(count)/total_count)*100)
if progress >= report_at[0]:
print 'Progress %3d' % (progress),'%'
report_at.remove(report_at[0])
print 'Progress: 100 %'
print 'Total failed tracks: %d' % (len(fails))
#TEST-CODE:
checksum = len(no_fail)+len(fails)
if checksum != len(tracks):
print 'ERROR HAPPEN FFFFFF'
tracks = no_fail
#save list of Itunes_Artist objects containing Hybrid_Track objects within their tracks field.
#10/12/09: currently saves list of Hybrid_Track objects
try:
artists = []
artistnames=[]
for track in tracks:
if track.get_artist().get_name() not in artistnames:
artists.append(track.get_artist())
artistnames.append(track.get_artist().get_name())
for artist in artists:
artist.tracks = []
for track in tracks:
if track.get_artist() == artist:
artist.tracks.append(track)
artist.trackcount=len(artist.tracks)
except:
return tracks
print 'saving!'
datamgmt.save(artists,'incompleteartists.db')
datamgmt.save(fails,'failedtracks.db')
if debug==True:
return artists,fails
