Delete Tweets

delete_tweets.py

#!/usr/bin/env python3
# -*-*- encoding: utf-8 -*-*-
# Created: Tue, 26 Sep 2017 20:05:39 -0300

"""
Delete old tweets
split -l $((1 + $(wc -l < tweets.csv) / 3)) tweets.csv
"""


import argparse
import sys
import csv
import time
import requests
from twython import Twython, TwythonRateLimitError
from multiprocessing.dummy import Pool


__author__ = "Julio Batista Silva"
__copyright__ = "Copyright (c) 2017, Julio Batista Silva"
__license__ = "GPL v3"
__version__ = "1.0"
__email__ = "julio@juliobs.com"


# Source:
# https://unsupervisedlearning.wordpress.com/2014/07/06/scraping-your-twitter-homepage-with-python-and-mongodb/
def handle_rate_limiting(twitter):
    app_status = {'remaining': 1}
    while True:
        if app_status['remaining'] > 0:
            status = twitter.get_application_rate_limit_status(
                resources=['statuses', 'application'])
            app_status = status['resources']['application']['/application/rate_limit_status']
            home_status = status['resources']['statuses']['/statuses/home_timeline']
            if home_status['remaining'] == 0:
                wait = max(
                    home_status['reset'] - time.time(),
                    0) + 1  # addding 1 second pad
                print("Sleeping {} seconds.".format(wait))
                time.sleep(wait)
            else:
                return
        else:
            wait = max(app_status['reset'] - time.time(), 0) + 10
            print("Sleeping for {} seconds.".format(wait))
            time.sleep(wait)


# Delete old tweets
def delete_tweets(twitter, archive_csv):
    with open(archive_csv) as csv_file:
        read_csv = csv.reader(csv_file, delimiter=",")
        for row in read_csv:
            id = row[0]

            # Check header.
            # 404: tweet already deleted
            # 302: RT
            url = "https://twitter.com/USUARIO/status/" + str(id)

            try:
                r = requests.head(url)
                if r.status_code == 200 or r.status_code == 302:

                    try:
                        handle_rate_limiting(twitter)

                        tweet = twitter.show_status(id=id)
                        favs = tweet["favorite_count"]
                        rts = tweet["retweet_count"]

                        if (rts < 2 and favs < 2) or r.status_code == 302:
                            print("Deleting id: {}".format(id))
                            twitter.destroy_status(id=id)
                        else:
                            print("Keeping id: {}".format(id))

                    except TwythonRateLimitError as e:
                        print("[Exception Raised] Rate limit exceeded")
                        reset = int(
                            twitter.get_lastfunction_header('x-rate-limit-reset'))
                        wait = max(reset - time.time(), 0) + \
                            10  # adding 10 second pad
                        print("Waiting {} seconds.".format(wait))
                        time.sleep(wait)
                    except Exception as e:
                        print("Error: {}".format(e))
                        continue

                else:
                    print(r.status_code)

            except requests.ConnectionError:
                print("failed to connect")


def main():
    # Auth and API
    consumer_key = ''
    consumer_secret = ''
    access_token = ''
    access_token_secret = ''
    twitter1 = Twython(
        consumer_key,
        consumer_secret,
        access_token,
        access_token_secret)

    # Auth and API 2
    consumer_key = ''
    consumer_secret = ''
    access_token = ''
    access_token_secret = ''
    twitter2 = Twython(
        consumer_key,
        consumer_secret,
        access_token,
        access_token_secret)

    # Auth and API 3
    consumer_key = ''
    consumer_secret = ''
    access_token = ''
    access_token_secret = ''
    twitter3 = Twython(
        consumer_key,
        consumer_secret,
        access_token,
        access_token_secret)

    # Auth and API 4
    consumer_key = ''
    consumer_secret = ''
    access_token = ''
    access_token_secret = ''
    twitter4 = Twython(
        consumer_key,
        consumer_secret,
        access_token,
        access_token_secret)

    # Auth and API 5
    consumer_key = ''
    consumer_secret = ''
    access_token = ''
    access_token_secret = ''
    twitter5 = Twython(
        consumer_key,
        consumer_secret,
        access_token,
        access_token_secret)

    # Threads
    with Pool() as pool:
        pool.starmap(delete_tweets, [(twitter1, "xaa"), (twitter2, "xab"),
                                     (twitter3, "xac"), (twitter4, "xad"), (twitter5, "xae")])

    return 0


if __name__ == "__main__":
    parser = argparse.ArgumentParser(description='Delete old tweets')
    parser.add_argument('--version', action='version',
                        version='%(prog)s v' + __version__)
    # parser.add_argument("archive_csv", type=str, help="twitter archive csv file")
    args = parser.parse_args()

    sys.exit(main())