Fetches the recent tracks of a last.fm user within a date range and converts it to a list of Spotify URLs (to be copied directly into the Spotify program) sorted descending by play count

lastfm_scrobbles_to_spotify.py

#!/usr/bin/env python2.7

import requests
import datetime
import calendar
import time
import json
from pprint import pprint


def timestamp(dt):
    return calendar.timegm(dt.utctimetuple())


def spotify_id(artist, name):
    payload = {'q': artist + " - " + name}
    data = requests\
        .get("http://ws.spotify.com/search/1/track.json", params=payload)\
        .json()
    for track in data['tracks']:
        if not 'karaoke' in track['name'].lower():
            return track['href']
    raise Exception("Failed to get Spotify ID")


def recent_tracks(name, start, end):
    num = 1
    page = 0
    while page < num:
        print("{}/{}: recent tracks".format(page + 1, num))

        # HTTP API call
        payload = {'method': 'user.getrecenttracks',
                   'user': name,
                   'api_key': lastfm_key,
                   'format': 'json',
                   'from': timestamp(start),
                   'to': timestamp(end),
                   'page': page}
        response = requests\
            .get("http://ws.audioscrobbler.com/2.0/", params=payload)\
            .json()

        # yield each track
        if 'recenttracks' in response:
            page += 1
            data = response['recenttracks']
            if '@attr' in data:
                num = int(data['@attr']['totalPages'])
                for track in data['track']:
                    try:
                        # test existence of these keys
                        track['artist']['#text']
                        track['name']
                        track['url']
                        yield track
                    except: pass
                time.sleep(0.5)
            else:
                return # no tracks found
        else:
            print("Got invalid result {0}".format(response))
            time.sleep(5)


def top_tracks(name, start, end):
    tracks = {}
    for track in recent_tracks(name, start, end):
        url = track['url']
        if url in tracks:
            tracks[url]['count'] += 1
        else:
            stored = {'count': 1}
            stored.update(track)
            tracks[url] = stored
    return sorted(tracks.values(), key=lambda r: -r['count'])


lastfm_key = 'CHANGE THIS'
user = 'CHANGE THIS'
start = datetime.datetime(2013, 1, 1)
length = datetime.timedelta(weeks=52)
end = start + length

for track in top_tracks(user, start=start, end=end):
    count = track['count']
    artist = track['artist']['#text']
    title = track['name']
    spotify_url = ""
    try:
        spotify_url = spotify_id(artist, title)\
            .replace("spotify:track:", "http://open.spotify.com/track/")
    except: pass
    print("{: <55} [{}] {} - {}"\
        .format(spotify_url, count, artist, title)).encode('utf-8')