Python script to delete duplicate tracks from Google Play Music library.

README.md

Python script to delete duplicate tracks from Google Play Music library. Please retry until "No duplicate songs" message is displayed. Use this script at your own risk. gmusicapi is required.

gpm_cleaner.py

#!/usr/bin/env python

from gmusicapi import Mobileclient
from getpass import getpass

client = Mobileclient()
client.login( raw_input( "Username: " ), getpass() )

print "Getting all songs ..."
all_songs = client.get_all_songs()
new_songs = {}
old_songs = {}

for song in all_songs:
    song_id = song.get('id')
    timestamp = song.get('recentTimestamp')
    
    key = "%s: %d-%02d %s" % ( song.get('album'), song.get('discNumber'), song.get('trackNumber'), song.get('title') )
    
    if key in new_songs:
        if new_songs[key]['timestamp'] < timestamp:
            old_songs[key] = new_songs[key]
            new_songs[key] = { 'id': song_id, 'timestamp': timestamp }
        else:
            old_songs[key] = { 'id': song_id, 'timestamp': timestamp }
    
    new_songs[key] = { 'id': song_id, 'timestamp': timestamp }

if len( old_songs ):
    print "Duplicate songs"
    
    old_song_ids = []
    
    for key in sorted( old_songs.keys() ):
        old_song_ids.append( old_songs[key]['id'] )
        print "    " + key.encode('utf-8')
    
    if raw_input( "Delete duplicate songs? (y, n): ") is 'y':
        print "Deleting songs ..."
        client.delete_songs( old_song_ids )
else:
    print "No duplicate songs"

Below is the code to remove duplicates from all playlists (rather than songs in the library). Don't forget the , Mobileclient.FROM_MAC_ADDRESS from TheRealWag's comment if it's appropriate.

#!/usr/bin/env python

from gmusicapi import Mobileclient
from getpass import getpass

client = Mobileclient()
client.login( raw_input( "Username: " ), getpass() )

print "Getting all playlists ..."
all_plists = client.get_all_user_playlist_contents()

all_dup_entries = []

for plist in all_plists:
    unique_track_list = []

    duplen = len( all_dup_entries )

    for entry in plist.get('tracks'):
        track_id = entry.get('trackId')
        entry_id = entry.get('id')

        if track_id in unique_track_list:
            all_dup_entries.append(entry_id)
        else:
            unique_track_list.append(track_id)

    duplen = len( all_dup_entries ) - duplen

    if duplen:
        print "\n " + plist.get('name').encode('utf-8') + " - " + str(duplen) + " duplicates"

    del unique_track_list
    del duplen

if len( all_dup_entries ):
    print "Duplicate playlist entries exist."

    if raw_input( "Delete duplicate entries? (y, n): ") is 'y':
        print "Deleting entries ..."
        client.remove_entries_from_playlist( all_dup_entries )
else:
    print "No duplicate entries"

Can you tell me what i'm doing wrong? I get this error:

python2 gm-cleaner.py
Username: XXXXXXXX
Password:
Getting all songs ...
Traceback (most recent call last):
File "gm-cleaner.py", line 18, in
key = "%s: %d-%02d %s" % ( song.get('album'), song.get('discNumber'), song.get('trackNumber'), song.get('title') )
TypeError: %d format: a number is required, not NoneType