gistfile1.py

#!/usr/bin/env python3
# A simple script to download stuff from Game One, a show on MTV.
# This code is public domain (if not applicable, then MIT licensed)
import re, urllib.request, urllib.parse, subprocess, sys, json, posixpath

# regex to get the variables from the page that the flash player uses
variable_re = re.compile(r'var variables = (.*);')
# crude regex to convert JS-dict syntax into JSON syntax
jsonize_re = re.compile(r' (\w+):')
# regex to get the path to the SWF file
swf_re = re.compile(r'swfobject.embedSWF\("(.*?)"')

def get_variables(source):
    """Rips out the JS code with the variables, converts it to JSON
    and parses that"""
    result = variable_re.search(source)
    js = result.group(1)
    jsonized = jsonize_re.sub(r' "\1":', js)
    return json.loads(jsonized)

def get_swf_path(source):
    """Returns the URL to the SWF file that is going to play the file"""
    return swf_re.search(source).group(1)

def main():
    # grab URL to download from
    url = sys.argv[1]
    # construct base url from what we got
    scheme, netloc = urllib.parse.urlparse(url)[:2]
    base_url = "{0}://{1}".format(scheme, netloc)

    # read the page and extract useful information
    with urllib.request.urlopen(url) as page:
        content = page.read().decode('utf-8')
        variables = get_variables(content)
        swf = get_swf_path(content)

    # from this, get the path of the JSON config file
    json_path = base_url + variables['playlistfile']

    # grab that and parse
    with urllib.request.urlopen(json_path) as page:
        content = page.read().decode('utf-8')
        config = json.loads(content)

    streamer = variables['streamer']
    path = config['playlist'][0]['hqv']
    basename = posixpath.basename(path)
    # call rtmpdump with the information that we gathered
    rtmpdump = subprocess.Popen(('rtmpdump',
        '-r', streamer,
        '--swfVfy', swf,
        '-y', 'mp4:' + path,
        '-o', basename))
    rtmpdump.communicate()

    print("Downloaded file to {0}".format(basename))

if __name__ == '__main__':
    main()