Downloads all participant information from http://www.bmw-berlin-marathon.com/en/results-and-list-of-participants/participant-list.html and stores it in a csv file

berlin_marathon_participants_2014.py

#!/usr/bin/env python
# encoding: utf-8
"""
berlin_marathon_participants_2014.py

Created by Christian Stade-Schuldt on 2013-11-12.
"""

import urllib
import json
import time
import csv

def main():
    """Downloads all participant information from 
    http://www.bmw-berlin-marathon.com/en/results-and-list-of-participants/participant-list.html 
    and stores it in a csv file"""
    base_url = "http://www.bmw-berlin-marathon.com/files/addons/scc_events_data/ajax.teilnehmer.php?ident=MAH,MAL,MAR,MAW,MAI&_search=false&nd=1384301124189&rows=100&page="
    all_runner = []
    for i in range(1, 169):
        url = base_url + "%d&sidx=teilnehmer_name&sord=asc" % i
        print url
        response = urllib.urlopen(url)
        data = json.loads(response.read())
        for row in data["rows"]:
            runner_id = row["cell"][0]
            last_name = row["cell"][2]
            first_name = row["cell"][3]
            country = row["cell"][4]
            birth_date = row["cell"][5]
            team = row["cell"][6]
            city = row["cell"][7]
            runner = (runner_id, last_name, first_name, country, birth_date, team, city)
            all_runner.append(runner)
        time.sleep(1)
    print len(all_runner)

    with open("output.csv", "wb") as f:
        writer = csv.writer(f)
        writer.writerow(["id", "last_name", "first_name", "country", "birth_date", "team", "city"])
        for runner in all_runner:
            writer.writerow([k.encode('utf-8') for k in runner])
         
if __name__ == '__main__':
    main()