Mixpanel Delete People

mixpanel-delete-people.py

"""
This is from mixpanel customer service, I just PEP8ified it. Update api key, secret and token.
You can define which users you want to delete on line 125. Right now it will delete users who haven't
been seen in over 7 weeks. You will recieve a confirmation prompt before the people are deleted.
"""

import hashlib
import time
import urllib
import base64
import sys
try:
    import json
except ImportError:
    import simplejson as json
try:
    import eventlet
    from eventlet.green import urllib2
except ImportError:
    print "You need to pip install eventlet. Quitting now..."
    sys.exit()


class Mixpanel(object):

    def __init__(self, api_key, api_secret, token):
        self.api_key = api_key
        self.api_secret = api_secret
        self.token = token

    def request(self, params, format='json'):
        '''let's craft the http request'''
        params['api_key'] = self.api_key
        params['expire'] = int(time.time())+600  # 600 is ten minutes from now
        if 'sig' in params:
            del params['sig']
        params['sig'] = self.hash_args(params)

        request_url = 'http://mixpanel.com/api/2.0/engage/?' + self.unicode_urlencode(params)

        request = urllib.urlopen(request_url)
        data = request.read()
        return data

    def hash_args(self, args, secret=None):
        '''Hash dem arguments in the proper way
        join keys - values and append a secret -> md5 it'''

        for a in args:
            if isinstance(args[a], list):
                args[a] = json.dumps(args[a])

        args_joined = ''
        for a in sorted(args.keys()):
            if isinstance(a, unicode):
                args_joined += a.encode('utf-8')
            else:
                args_joined += str(a)

            args_joined += "="

            if isinstance(args[a], unicode):
                args_joined += args[a].encode('utf-8')
            else:
                args_joined += str(args[a])

        hash = hashlib.md5(args_joined)

        if secret:
            hash.update(secret)
        elif self.api_secret:
            hash.update(self.api_secret)
        return hash.hexdigest()

    def unicode_urlencode(self, params):
        ''' Convert stuff to json format and correctly handle unicode url parameters'''

        if isinstance(params, dict):
            params = params.items()
        for i, param in enumerate(params):
            if isinstance(param[1], list):
                params[i] = (param[0], json.dumps(param[1]),)

        result = urllib.urlencode([(k, isinstance(v, unicode) and v.encode('utf-8') or v) for k, v in params])
        return result

    def update(self, userlist, uparams):
        url = "http://api.mixpanel.com/engage/"
        batch = []
        for user in userlist:
            distinctid = user['$distinct_id']
            tempparams = {
                'token': self.token,
                '$distinct_id': distinctid,
                '$ignore_alias': True
            }
            tempparams.update(uparams)
            batch.append(tempparams)

        payload = {"data": base64.b64encode(json.dumps(batch)), "verbose": 1, "api_key": self.api_key}

        response = urllib2.urlopen(url, urllib.urlencode(payload))
        message = response.read()

        '''if something goes wrong, this will say what'''
        if json.loads(message)['status'] != 1:
            print message

    def batch_update(self, users, params):
        pool = eventlet.GreenPool(size=200)
        while len(users):
            batch = users[:50]
            pool.spawn(self.update, batch, params)
            users = users[50:]
        pool.waitall()
        print "Done!"

if __name__ == '__main__':
    api = Mixpanel(
        api_key='wackawackawackawacka',
        api_secret='wackawackawackawacka',
        token='wackawackawackawacka'
    )
    parameters = {}
    '''Here is the place to define your selector to target only the users that you're after'''
    parameters.update({
        'selector': '(datetime(%s - %s) > properties["$last_seen"])' % (int(round(time.time() * 1000)), (1000 * 60 * 60 * 24 * 7 * 7))
    })
    response = api.request(parameters)

    parameters.update({
        'session_id': json.loads(response)['session_id'],
        'page': 0
    })
    global_total = json.loads(response)['total']

    print "Here are the # of people %d" % global_total
    fname = "Mixpanel_backup-" + str(int(time.time())) + ".txt"
    has_results = True
    total = 0
    print fname
    f = open(fname, 'w')
    x = raw_input("You are about to delete %s users.  Enter YES to continue: " % (str(global_total)))
    while has_results and x == "YES":
        responser = json.loads(response)['results']
        total += len(responser)
        has_results = len(responser) == 1000
        for data in responser:
                f.write(json.dumps(data)+'\n')
        print "%d / %d" % (total, global_total)
        api.batch_update(responser, {'$delete': ''})
        parameters['page'] += 1
        if has_results:
            response = api.request(parameters)

awesome! thanks for sharing!