Uncovering a Twitter handle using Twitter API

get_follower_ids.py

import sys
import json
import base64
import tweepy
import time

def is_valid_handle(handle):
    #checks if given handle matches olivia handle specs
    return True if re.compile("^[a-z]{10,}$").match(handle) else False

def is_valid_date(date):
    #might be 2012, she said sophomore year...but said 2011
    #did she skip a grade? smart cookie
    return True if (date.month == 1 and (date.year == 2011)) else False

def is_valid_user(user):
    #olivia probably has default profile and more than 10 favorites
    #does she have profile picture? 
    return is_valid_date(user.created_at) and is_valid_handle(user.screen_name) and not user.following and user.default_profile and user.favourites_count > 10

secrets_file = 'secrets.json'

with open(secrets_file) as secrets_file:
    keys = json.load(secrets_file)

print 'Connecting to Twitter and authenticating'
auth = tweepy.OAuthHandler(keys['consumer_key'], keys['consumer_secret'])
auth.set_access_token(keys['access_token'], keys['access_token_secret'])
api = tweepy.API(auth)

handles = ['Royals', 'FiveThirtyEight', 'Buster_ESPN', 'MLB', 'FoxTV', 'NPR','NYTimes'] #possibly silver's personal handle as well

def wait_a_minute():
    print '\tSleeping for a minute to meet rate limit restrictions'
    sys.stdout.write('\t')
    for i in range(60):
        sys.stdout.write('.')
        sys.stdout.flush()
        time.sleep(1)
    sys.stdout.write('\n')

for handle in handles[1:]:
    print 'Processing', handle
    file_name = handle+'.ids'
    fyle = open(file_name, 'w')
    fyle.write('{')

    cursor = -1
    page_count = 1
    print '\tQuerying for the first page of follower IDs...'
    response = api.followers_ids(id=handle, cursor=cursor)

    while cursor != 0:
        response = api.followers_ids(id=handle, cursor=cursor)
        try:
            ids = response[0]
            fyle.write(str(ids)[1:-1] + ',\n')
            fyle.flush()
            print '\tRecieved page',page_count,'with',len(ids),'IDs!'
            cursor = response[1][1]
            wait_a_minute()
            page_count += 1
        except:
            print 'Excepted!', page_count, cursor, handle
            wait_a_minute()


    fyle.close()

get_follower_ids_old.py

import json
import base64
import requests

def is_valid(handle):
    #checks if given handle matches olivia handle specs
    return True if re.compile("^[a-z]{10,}$").match(handle) else False

get_bearer_token = False
secrets_file = 'secrets.json'
bearer_token_file = 'bearer_token.json'

if get_bearer_token:
    with open(secrets_file) as secrets_file:
        keys = json.load(secrets_file)

    credentials = keys['consumer_key'] + ':' + keys['consumer_secret']
    print credentials
    base64_bearer_token_credentials = base64.b64encode(credentials)
    print base64_bearer_token_credentials

    url = 'https://api.twitter.com/oauth2/token'
    headers = {'authorization':'Basic ' + base64_bearer_token_credentials, 'content-type':'application/x-www-form-urlencoded;charset=UTF-8'}
    payload = 'grant_type=client_credentials'
    response = requests.post(url, headers=headers,data=payload)

    json_bearer_token = json.loads(response.text)
    if 'access_token' in json_bearer_token:
        bearer_token = json_bearer_token['access_token']
        with open(bearer_token_file, 'w') as bearer_token_file_obj:
            bearer_token_file_obj.write("{\"bearer_token\":\"" + bearer_token + "\"}")
        print 'Using bearer token', bearer_token
    else:
        print 'Cannot get bearer token'
        sys.exit(1)
else:
    with open(bearer_token_file) as bearer_token_file:
        bearer_token = json.load(bearer_token_file)['bearer_token']


    print 'Using bearer token:', bearer_token

    handle_1 = '@NYTimes'
    handle_2 = '@FiveThirtyEight' #possibly silver's personal handle as well
    handle_3 = '@MLB'
    handle_4 = '@Buster_ESPN'
    handle_5 = '@FoxTV'
    handle_6 = '@NPR'
    handle_7 = '@Royals'
    handle_8 = '@NPR'

    get_followers(handle_8)

def get_followers(user_id):
    cursor = -1
    authorization_header = 'Bearer ' + bearer_token
    headers = {"authortization":authorization_header}
    url = "https://api.twitter.com/1.1/followers/ids.json?cursor=" + str(cursor) + "&user_id="
    response = requests.get(url, 

I was challenged this summer by a close someone to take a bet. The terms? To find her Twitter handle using less than 10 reasonable questions, or otherwise take her to dinner. I myself stood to gain half a goat, a donation through Oxfam she promised to make should I win the bet.

With my pride and tangible items on the line, I thought I’d use my only trick: Twitter’s API.

How I used my questions

To get a pool of handles to search through, I first asked her if she followed New York Times, FiveThirtyEight, and a few other large names accounts that a college-age student might follow. Then, to narrow down the pool, I used my remaining available questions to ask specifics of the handle: does the handle of more than 10 characters? Does it contain an ‘e’? Does it contain an ‘n’?

The execution

Using Twitter’s API, I first queried and saved the large pools of user IDs that followed the news sources I knew she followed. Twitter, unfortunately, rate limited certain query types, so I had to do a round-about query of first the user ID’s. Then intersect these ID pools, to get a smaller pool of potential ID’s. Then query each user ID to get the corresponding Twitter handle. I filtered these handles (roughly 3,000) based on the answers to the above three questions, leading to roughly 20 handles which I manually examined and filtered. Lucky for me, despite the lack of any personally identifiable information on any of the twenty accounts, there was only one account that was clearly hers.

The gists below were templates I made for querying Twitter API. Most of the set intersection and filtering was done in iPython, output of which was not saved (sorry).

Also relevant:

https://gist.github.com/skoppula/1a16e77a13c1e6fc9fe7
https://gist.github.com/skoppula/cd1ec62aa7248bd2ad97