parses outside lands lineup and generates a json of artists along with Spotify, Echo Nest and Rdio IDs

parse.py

import sys
from bs4 import BeautifulSoup
import json
import pyen
en = pyen.Pyen()

def get_fid(artist, idspace):
    if 'foreign_ids' in artist and len(artist['foreign_ids']) > 0:
        for fids in artist['foreign_ids']:
            if fids['catalog'] == idspace:
                return fids['foreign_id']
    return None

def en_artist_lookup(name):
    response = en.get('artist/search', name=name,
        bucket=['id:spotify', 'id:rdio-US'])
    artists = response['artists']
    if len(artists) > 0:
        artist = artists[0]
        print artist
        enid = artist['id']
        spid = get_fid(artist, 'spotify')
        rdio = get_fid(artist, 'rdio-US')
        if rdio:
            rdio = rdio.split(':')[2]

        print name, '/', artist['name'], enid, spid, rdio
        ids = {
            'echonest':enid,
            'spotify':spid,
            'rdio':rdio
        }
        return ids
    return  {}
    
    
if __name__ == '__main__':
    lineup = []
    f = open(sys.argv[1])
    html_doc = f.read()
    soup = BeautifulSoup(html_doc)
    for a in soup.find_all('a', class_='band'):
        name = a.text.strip()
        ids = en_artist_lookup(name)
        artist = {
            'artist' : name,
            'link' : a['href'],
            'ids' : ids
        }
        lineup.append(artist)

    out = open('lineup_2014.json', 'w')
    print >> out, json.dumps(lineup, indent=4)
    out.close()