Python script to get the top 100 songs scrobbled on Last.fm in all the metros in U.S and associated stats and track info.

laststats.py

#Fetches the top 100 songs scrobbled in all the metros in U.S and associated stats and track info.

import urllib
import urllib2
from lxml import etree
import os
try:
    import cpickle as pickle
except:
    import pickle

lastfm_apikey = '' # use your own api key here
lastfm_url = 'http://ws.audioscrobbler.com/2.0'

method_metros = 'geo.getMetros'
method_toptracks = 'geo.gettoptracks'
method_trackinfo = 'track.getInfo'

def call_lastfm(method, **kwargs):
    data = {}
    data['method'] = method
    data['api_key'] = lastfm_apikey
    for key in kwargs:
        data[key] = kwargs[key]
    values = urllib.urlencode(data)
    url = lastfm_url + '?' + values
    print url
    response = urllib2.urlopen(url);
    return etree.fromstring(response.read())

def get_data():
    with open('laststats.csv', 'w') as f:
        us = 'united states'
        f.write('city,artist,track,listeners,playcount')
        metros = call_lastfm(method_metros, country=us)
        musicd = None
        pfile = 'musicd.p'
        if os.path.isfile(pfile):
            musicd = pickle.load(open(pfile, 'rb'))
        else:
            musicd = {}
        for metro in metros.iter('metro'):
            city = metro.find('name').text
            toptracks = call_lastfm(method_toptracks, country=us, location=city, limit=100)
            for track in toptracks.iter('track'):
                artist = track.find('artist').find('name').text.encode('utf-8')
                name = track.find('name').text.encode('utf-8')
                key = '%s-%s' % (artist, name)
                listeners = None
                playcount = None
                if key in musicd:
                    listeners, playcount = musicd[key]
                else:
                    info = call_lastfm(method_trackinfo, artist=artist, track=name)
                    listeners = info.find('track').find('listeners').text
                    playcount = info.find('track').find('playcount').text
                    musicd[key] = (listeners, playcount)
                f.write('\n%s,%s,%s,%s,%s' % (city, artist, name, listeners, playcount))
        pickle.dump(musicd, open(pfile, 'wb'))

if __name__ == '__main__':
    get_data();