goo-gl-inspector.py

#!/usr/bin/python2 -u
#
# REQUIREMENTS:
# - sre_yield
#
# AUTHOR:
# - KhasMek
#
import datetime
import json
import re
import requests
import sre_yield
import time

api_key = ''
characters = '[A-Za-z0-9]'
min_length = 1
max_length = 3
regular_expression = r'{char}{{{min},{max}}}'.format(char=characters, min=min_length, max=max_length)
base_url = 'https://www.googleapis.com/urlshortener/v1/url?shortUrl=http://goo.gl/'
fields = 'id,longUrl,status,analytics(allTime)'
already_checked_file = 'already-checked.txt'
results_file = 'results.json'
delay = 5


def get_response(short_url_tag):
    if fields:
        full_url = str('{base_url}{short_url_tag}&projection=FULL&key={key}&fields={fields}'.format(base_url=base_url, short_url_tag=short_url_tag, key=api_key, fields=fields))
    else:
        full_url = str('{base_url}{short_url_tag}&projection=FULL&key={key}'.format(base_url=base_url, short_url_tag=short_url_tag, key=api_key))
    r = requests.get(full_url)
    return json.loads(r.text)


def main():
    now = {"start": str(datetime.datetime.now())}
    with open(results_file, 'a+') as results:
        json.dump(now, results)
        results.write('\n')
    for short_url_tag in sre_yield.AllMatches(regular_expression):
        checked = False
        with open(already_checked_file, 'a+') as file:
            tag = re.compile("^{}$".format(short_url_tag), re.MULTILINE)
            # I should break this into its own function and just do a check?
            if tag.search(file.read()):
                print(" [!] ALREADY CHECKED:\t{}".format(short_url_tag))
                checked = True

        if not checked:
            print(" [*] NOW CHECKING:\t{}".format(short_url_tag))
            response = get_response(short_url_tag)
            if 'error' in response and response['error']['code'] == 403:
                print(' [!] RATE LIMIT:\tSLEEPING FOR 300 SECONDS')
                time.sleep(300)
                response = get_response(short_url_tag)

            if 'error' in response:  # and response['error']['code'] != 403:
                print(' [!] ERROR:\t\t{}\t{}\t{}'.format(short_url_tag, response['error']['code'], response['error']['message']))
            else:
                print(' [+] SHORT URL FOUND:\t{}\t{}'.format(short_url_tag, response['longUrl']))
                with open(results_file, 'a+') as results:
                    item = {short_url_tag: response}
                    json.dump(item, results)
                    results.write('\n')

            with open(already_checked_file, 'a+') as ac:
                ac.write(short_url_tag + "\n")
            time.sleep(delay)

    with open(results_file, 'a+') as results:
        now = {"end": str(datetime.datetime.now())}
        json.dump(now, results)


if __name__ == "__main__":
    main()