#!/usr/bin/env python
#
# any-month.py
#
# What did you listen to the most on any given month? (via last.fm)
#
# Example:
#
# $ find ~/public_html/mixtape -name index.xspf -print0 | xargs -0 ./any-month.py
# Wild Orchid Children where the mexican boys go
# Crystal Castles Reckless
# Natalie Portman's Shaved Head Sophisticated Side Ponytail
# Natalie Portman's Shaved Head Iceage Babeland
# Wild Orchid Children birth of a cabin
# Wolf Parade Fancy Claps
# Crystal Castles 1991
# Portugal. The Man Church Mouth
# Feral Children The Beast=Goldmine
# Wolf Parade Grounds for Divorce
# Feral Children Zyghost
# The Sound of Animals Fighting I, The Swan
# The Sound of Animals Fighting The Ocean And The Sun
# Man Party The Party Will Come
# Man Party Sonny Boy
# UNKLE In a Broken Dream
# UNKLE 24 Frames
# Simon Viklund Rise of the Albatross
# Simon Viklund Bionic Commando Rearmed (Main Theme)
# Portugal. The Man Marching with 6
#
import collections
import datetime
import difflib
import operator
import optparse
import urllib
import xml.etree.ElementTree as ET
LFM_API_KEY = '068e57ae9465c496f8c4ceecdc4ca644'
OPTIONS = optparse.OptionParser(usage='Usage: %prog [old-playlist.xspf ...]',
description='What did you listen to the most on any given month? (via last.fm)',
epilog='The XSPFs playlists, of course, contains tracks you want to exclude.')
OPTIONS.add_option('-u', '--user',
default='quad',
help='your last.fm user (default: quad)',
metavar='USER',
action='store')
OPTIONS.add_option('-w', '--when',
default=datetime.datetime.now().strftime('%Y-%m'),
help='the year and month to report on (default: this month)',
metavar='YYYY-MM',
action='store')
class XSPF:
"""A very ghetto XSPF parser."""
NS = '{http://xspf.org/ns/0/}'
Track = collections.namedtuple('Track', 'album, creator, title')
def __init__(self, filename):
tree = ET.parse(filename)
tracks_xml = tree.find(self.NS + 'trackList').findall(self.NS + 'track')
self.tracks = [self.Track(t.find(self.NS + 'album').text,
t.find(self.NS + 'creator').text,
t.find(self.NS + 'title').text)
for t in tracks_xml]
def lfm_call(method, **kwargs):
"""Make a web service call to last.fm."""
url = "http://ws.audioscrobbler.com/2.0/?method=%s&%s&api_key=%s" % \
(method, urllib.urlencode(kwargs), LFM_API_KEY)
result = urllib.urlopen(url)
return ET.parse(result)
def get_charts(user):
"""Get all of a user's weekly charts."""
charts_resp = lfm_call('user.getWeeklyChartList', user=user)
charts_xml = charts_resp.find('weeklychartlist').findall('chart')
return [map(datetime.datetime.fromtimestamp, [int(chart.attrib['from']), int(chart.attrib['to'])])
for chart in charts_xml]
def get_weekly_tracks(user, when_from, when_to):
"""Get all of a user's weekly tracks between the specified dates."""
c_from, c_to = map(lambda d: d.strftime('%s'), [when_from, when_to])
tracks_resp = lfm_call('user.getWeeklyTrackChart', **{'user': user, 'from': c_from, 'to': c_to})
tracks_xml = tracks_resp.find('weeklytrackchart').findall('track')
return dict([((t.find('artist').text, t.find('name').text), int(t.find('playcount').text))
for t in tracks_xml])
def get_pop_tracks(user, when):
"""Get a user's popular tracks for a specified month."""
# Get all charts for the specified month.
charts = [(c_from, c_to) for c_from, c_to in get_charts(user)
if c_from.year == when.year and c_from.month == when.month]
# Get all tracks for those charts.
tracks_weekly = [get_weekly_tracks(user, c_from, c_to) for c_from, c_to in charts]
# Combine the weekly charts together.
def _(left, right):
for k, v in left.items():
right[k] = right.get(k, 0) + v
return right
tracks = reduce(_, tracks_weekly, {})
# Find the top tracks.
return sorted(tracks.iteritems(), key=operator.itemgetter(1), reverse=True)
def mask_tracks(tracks, mask):
"""Mask a selection of tracks."""
all_artists = set([a for a, t in mask])
def _(artist, title):
for maybe_artist in difflib.get_close_matches(artist, all_artists):
maybe_titles = [t for a, t in mask if a == maybe_artist]
if difflib.get_close_matches(title, maybe_titles):
return True
return [(t, pc) for t, pc in tracks if not _(*t)]
def limit_artist_occurence(tracks, max_occurence):
"""Limit the number of times an artist can occur in a track list."""
selected_artists = collections.defaultdict(int)
for index, playcount in tracks:
artist, title = index
selected_artists[artist] += 1
if (selected_artists[artist] <= max_occurence):
yield index, playcount
def prettyprint_tracklist(tracks):
"""Print a formatted track list."""
if tracks:
max_artist_length = max([len(a) for (a, t), pc in tracks])
f = "%%%us %%s" % max_artist_length
for (a, t), pc in tracks:
print f % (a, t)
def main():
# Parse the command-line arguments.
(opts, playlists) = OPTIONS.parse_args()
# Load the tracks to be masked from the playlists.
tracks_xspf = sum([XSPF(fn).tracks for fn in playlists], [])
masks = [(t.creator, t.title) for t in set(tracks_xspf)]
# Retrieve the most listened tracks for the month.
when = datetime.datetime.strptime(opts.when, '%Y-%m').date()
tracks_pop = get_pop_tracks(opts.user, when)
# Do some playlist post-processing, so as to not be too boring.
tracks = mask_tracks(tracks_pop, masks)
mix = list(limit_artist_occurence(tracks, 2))
# Only print the top 20 tracks.
num_tracks = min(20, len(mix))
prettyprint_tracklist(mix[:num_tracks])
if __name__ == '__main__':
main()
) is private.
