Count how many times an archive.org item has been derived.

count_derives.py

#!/usr/bin/env python
import sys
import os
import json

import requests


def get_tasks(identifier, params={}):

    log_in_cookies = {'logged-in-sig': os.environ['LOGGED_IN_SIG'],
                      'logged-in-user': os.environ['LOGGED_IN_USER'],
    }

    params['json'] = 2
    params['output'] = 'json'
    params['callback'] = 'foo'
    params['history'] = '1'
    params['identifier'] = identifier
    url = 'http://www-tracey.us.archive.org/catalog.php'
    r = requests.get(url, params=params, cookies=log_in_cookies)

    tasks_json = json.loads(r.text.strip('foo').strip().strip('()'))
    tasks = []
    for t in tasks_json:
        td = {}
        td['identifier'] = t[0]
        td['server'] = t[1]
        td['command'] = t[2]
        td['time'] = t[3]
        td['submitter'] = t[4]
        td['args'] = t[5]
        td['task_id'] = t[6]
        td['type'] = t[7]
        tasks.append(td)

    return r, tasks

if __name__ == '__main__':
    identifier = sys.argv[1]
    r, tasks = get_tasks(identifier)

    task_commmands = [x['command'] for x in tasks]
    derive_tasks = [x for x in task_commmands if x == 'derive']

    print "%s\t%s" % (identifier, len(derive_tasks))