Yahoo email validation

validate_yahoo.py

import sys
import time
import urllib
import requests


class Yahoo(object):
    """
    Hokey way of Yahoo validation since their SMTP check returns
    250 for every request.
    """

    def __init__(self):
        """
        Putting together some constants needed or maybe for PEP8 fashion.
        """
        ua = []
        ua.append('Mozilla/5.0 (Macintosh; Intel Mac OS X 10_13_1)')
        ua.append('AppleWebKit/537.36 (KHTML, like Gecko)')
        ua.append('Chrome/63.0.3239.108 Safari/537.36')
        content_type = 'application/x-www-form-urlencoded; charset=UTF-8'
        url = 'https://login.yahoo.com/account/module/create'
        params = 'validateField=yid'
        self.url = '{0}?{1}'.format(url, params)
        self.headers = {'origin': 'https://login.yahoo.com',
                        'accept-encoding': 'gzip, deflate, br',
                        'accept-language': 'en-US,en;q=0.9',
                        'user-agent': ' '.join(ua),
                        'content-type': content_type,
                        'accept': '*/*',
                        'referer': self.url,
                        'authority': 'login.yahoo.com',
                        'x-requested-with': 'XMLHttpRequest'}

    def _valid_response_(self, response):
        """ Internal function for checking the response code.
        Args:
            response (:obj:): Requests response object.
        Returns:
            bool: True or False based on if 2xx status code.
        """
        if response and response.status_code:
            if response.status_code in range(200, 299):
                return True
        return False

    def _extract_crumb(self, cookies):
        """ Crumb needed for the post body element.
        Args:
            cookies (:obj:): Requests.cookies response object.
        Returns:
            acrumb (str): Parsed out string from AS cookie for body.
        """
        acrumb = None
        for cookie in cookies:
            if cookie.name == 'AS':
                dataset = cookie.value.split('&')
                for data in dataset:
                    if data.startswith('s='):
                        acrumb = data[len('s='):]
        return acrumb

    def _browser_data(self, acrumb, mailbox):
        """ Construct of the post return data for second request.
        Args:
            acrumb (str): Hash value returned to post body.
            mailbox (str): {mailbox}@yahoo.com also needed for post body.
        Returns:
            browser, dict_args (tuple): 2 sets of prepared data for response.
        """
        ftime_start = int(time.time())
        ftime_end = ftime_start + 10
        hash1 = '24700f9f1986800ab4fcc880530dd0ed'
        hash2 = 'd52a1516cfb5f1c2d8a427c14bc3645f'
        browser_fp_data = {"language": "en-US",
                           "colorDepth": 24,
                           "deviceMemory": 8,
                           "pixelRatio": 2,
                           "hardwareConcurrency": 8,
                           "resolution": {"w": 1440,
                                          "h": 900},
                           "availableResolution": {"w": 1440,
                                                   "h": 846},
                           "timezoneOffset": 300,
                           "timezone": "America/Chicago",
                           "sessionStorage": 1,
                           "localStorage": 1,
                           "indexedDb": 1,
                           "openDatabase": 1,
                           "cpuClass": "unknown",
                           "platform": "MacIntel",
                           "doNotTrack": 1,
                           "canvas": "canvas winding:yes~canvas",
                           "webgl": 1,
                           "webglVendorAndRenderer": "Google Inc. (ATI Technologies Inc.)~ANGLE (ATI Technologies Inc., AMD Radeon Pro 555 OpenGL Engine, OpenGL 4.1)",
                           "adblock": 0,
                           "hasLiedLanguages": 0,
                           "hasLiedResolution": 0,
                           "hasLiedOs": 0,
                           "hasLiedBrowser": 0,
                           "touch_support": {"points": 0,
                                             "event": 0,
                                             "start": 0},
                           "plugins": {"count": 0,
                                       "hash": hash1},
                           "fonts": {"count": 27,
                                     "hash": hash2},
                           "audio": "124.04347657808103",
                           "ts": {"serve": ftime_start,
                                  "render": ftime_end}}
        dict_args = {'specId': 'yidregsimplified',
                     'cacheStored': 'true',
                     'crumb': 'QFM5cNikkUV',
                     'acrumb': acrumb,
                     'sessionIndex': 'QQ--',
                     'done': 'https%3A%2F%2Fwww.yahoo.com',
                     'attrSetIndex': 0,
                     'googleIdToken': '',
                     'authCode': '',
                     'tos0': 'oath_freereg|us|en-US',
                     'firstName': '',
                     'lastName': '',
                     'attrSetIndex': 0,
                     'userid-domain': 'yahoo',
                     'userId': mailbox,
                     'yidDomainDefault': 'yahoo.com',
                     'yidDomain': 'yahoo.com',
                     'password': '',
                     'shortCountryCode': 'US',
                     'phone': '',
                     'mm': '',
                     'dd': '',
                     'yyyy': '',
                     'multiDomain': '',
                     'signup': '',
                     'freeformGender': ''}

        return browser_fp_data, dict_args

    def _reshape_cookie(self, cookies):
        """ Returns string representation of cookie body
        Args:
            cookies (:obj:`list`): List of cookies
        Returns:
            string: Concatenated string of cookie name+body
        """
        string = []
        for cookie in cookies:
            string.append('{0}={1}'.format(cookie.name, cookie.value))
        return ';'.join(string)

    def _digest_response(self, response):
        """ Tests response output if mailbox exists.
        Args:
            response (:obj:): Requests response object.json()
        Returns:
            string: "exists" or "does not exist"
        """
        output = None
        if 'errors' in response:
            for error in response['errors']:
                if error['name'] == 'userId':
                    output = error['error']
        if output == 'IDENTIFIER_EXISTS':
            return 'exists'
        return 'does not exist'

    def _clean_mailbox(self, mailbox):
        """ Incase someone passes in a full email in instead of a mailbox
        Args:
            mailbox (str): "[email protected]"
        Returns:
            mailbox (str): "test" without the "@yahoo.com"
        """
        if '@' in mailbox:
            return mailbox[:mailbox.find("@") - 1]
        return mailbox

    def validate(self, mailbox):
        """ Validation method that runs the check against the validation.
        Args:
            mailbox (str): mailbox to check validation against.
        Returns:
            string: "{mailbox} exists" or "{mailbox} does not exist"
        """
        mailbox = self._clean_mailbox(mailbox)
        r = requests.get(self.url)
        if not self._valid_response_(r):
            return {'Error': 'Initial request error'}
        acrumb = self._extract_crumb(r.cookies)
        self.headers['cookie'] = self._reshape_cookie(r.cookies)
        browser_fp_data, dict_args = self._browser_data(acrumb, mailbox)
        payload = {'browser-fp-data': browser_fp_data}
        payload.update(dict_args)
        params = {'validateField': 'userId'}
        req = requests.post(self.url, headers=self.headers, cookies=r.cookies,
                            data=payload, params=params)
        if not self._valid_response_(r):
            return {'Error': 'Validation call error'}
        return '{0} {1}'.format(mailbox, self._digest_response(req.json()))


if __name__ == '__main__':
    yahoo = Yahoo()
    mailbox = 'krazybean1'
    if len(sys.argv) == 2:
        mailbox = sys.argv[1]
    print(yahoo.validate(mailbox))