webscraping.py

#! /usr/bin/python
# coding=utf-8

import requests

# Track ids to download
tracks = [
	'9319946',
	'9319955',
	...
]
# Different users to login
users = [
    {"email":"[email protected]", "password":"password"},
    {"email":"[email protected]", "password":"password"},
    ...
]
# Different proxys to download tracks
proxies = [
    {"http": "http://XX.XX.XX.XX:PP/"},
    {"http": "http://YY.YY.YY.YY:PP/"},
    ...
]

def login(credentials):
    login_url = 'http://es.wikiloc.com/wikiloc/login.do'
    headers = {
    	'Cache-Control': 'no-cache',
    	'Content-Type': 'application/x-www-form-urlencoded',
    	'Host': 'es.wikiloc.com'
    }
    session = requests.session()
    session.post(login_url)
    session.post(login_url, data=credentials, headers=headers)
    return session

def downloadGpx(session, track_id, proxy):
    download_url = 'http://es.wikiloc.com/wikiloc/downloadToFile.do?event=download&id={}&format=gpx'.format(track_id)
    response = session.get(download_url, stream=True, proxies=proxy)
    status_code = response.status_code
    if status_code == 200:
        f = open("gpx/" + track_id + ".gpx", 'wb')
        inside = False
        for chunk in response.iter_content(chunk_size=1024):
        	inside = True
            if chunk:
                f.write(chunk)
        f.close()
        if inside == False: status_code = -1
    return status_code

def downloadTracks():
    user = 0
    proxy = 0
    session = login(users[user])
    for track_id in tracks:
        try:
            status_code = downloadGpx(session, track_id, proxies[proxy])
        except:
            status_code = 0
            pass
        while status_code != 200:
            if status_code == -1:
                user = 0 if user == len(users)-1 else user = user + 1
                session = login(users[user])
            else:
                proxy = 0 if proxy == len(proxies)-1 else proxy = proxy + 1
            try:
                status_code = downloadGpx(session, track_id, proxies[proxy])
            except:
                status_code = 0
                pass
    print "CHIM-PUM"

def main():
	downloadTracks()

if __name__ == "__main__": main()