Short script to pull attendance information for the bonnaroo lineup.

bonnaroo_attendance.py

from bs4 import BeautifulSoup

import csv
import requests


def get_linup():
    """Returns a request response object of the lineup html"""
    return requests.get('http://lineup.bonnaroo.com')


def get_artist_list(lineup):
    """Create an artist-id list to be used to search for their numbers"""
    soup = BeautifulSoup(lineup.text)
    artist_id_list = []
    for i in soup.find_all('a', {'class': 'band'}):
        artist_id_list.append(i['id'])

    return artist_id_list


def get_people_added(artist_id_list):
    """
    Use artist-id to look up how many people are attending the show and
    write results to a csv file for future processing
    """

    d = []

    for i in artist_id_list:
        url = 'http://lineup.bonnaroo.com/band/' + i + '/fpup?version=fb'
        r = requests.get(url)
        soup = BeautifulSoup(r.text)
        artist_info = soup.find("div", {"class": "fl-pop-similar"})
        count = artist_info.p.text.split(' ')[0]
        artist = artist_info.p.a.text
        print count, artist
        d.append({'artist': artist.encode('utf-8'), 'count': count})

    with open('bonnaroo.csv', 'w') as csvfile:
        fieldnames = ['artist', 'count']
        writer = csv.DictWriter(csvfile, fieldnames=fieldnames)
        writer.writeheader()
        for i in d:
            writer.writerow(i)


if __name__ == '__main__':
    lineup = get_linup()
    artist_list = get_artist_list(lineup)
    get_people_added(artist_list)

requirements.txt

beautifulsoup4==4.3.2
requests==2.6.0