Compare commits

...

16 Commits

14 changed files with 187 additions and 104 deletions
+3
View File
@@ -87,6 +87,7 @@ class MongoDbManager:
collection_to_use = self.db[LINKS_TO_VALIDATE]
updated_at = time.strftime("%H:%M:%S", time.localtime())
_ip_country = "FR"
serial = "requests"
# find ip_country info
for _contact in _all_contact_list:
if _contact.mail == mail_address:
@@ -96,6 +97,7 @@ class MongoDbManager:
collection_to_use.replace_one(filter={'_id': mail_address, }, replacement={
u'url': link,
u'email': mail_address,
u'serial': serial,
u'ip_country': _ip_country,
"updated_at": updated_at
},
@@ -103,6 +105,7 @@ class MongoDbManager:
else:
collection_to_use.replace_one(filter={'_id': link, }, replacement={
u'url': link,
u'serial': serial,
u'ip_country': _ip_country,
"updated_at": updated_at
},
+1 -4
View File
@@ -2,9 +2,6 @@ import time
from workers.link_validator_with_provided_list import validate_all_links
# if __name__ == '__main__':
# # link_list = MONGO_STORE_MANAGER.get_links_to_validate()
# validate_with_FR_ip(segment_position=2)
if __name__ == '__main__':
# generate test data
@@ -12,4 +9,4 @@ if __name__ == '__main__':
print("call validate_all_links()")
validate_all_links([])
print("wait for 30 seconds")
time.sleep(30)
time.sleep(10)
+36 -18
View File
@@ -86,47 +86,65 @@ def create_imap(login: str):
if DOMAIN_163 in login:
imap = IMAPClient(IMAP_SERVER_163, use_uid=True)
elif DOMAIN_YAHOO in login:
imap = imaplib.IMAP4_SSL(YAHOO_IMAP_SERVER)
# imap = imaplib.IMAP4_SSL(YAHOO_IMAP_SERVER)
imap = IMAPClient(YAHOO_IMAP_SERVER, use_uid=True)
elif DOMAIN_FIREMAIL_DE in login:
# imap = imaplib.IMAP4_SSL(SERVER_FIREMAIL_DE)
imap = IMAPClient(SERVER_FIREMAIL_DE, use_uid=True)
elif DOMAIN_GMX in login or DOMAIN_GMX_FR in login or DOMAIN_GMX_US in login or DOMAIN_GMX_CH in login or DOMAIN_GMX_PT in login or DOMAIN_GMX_SG in login:
imap = imaplib.IMAP4_SSL(SERVER_GMX)
# imap = imaplib.IMAP4_SSL(SERVER_GMX)
imap = IMAPClient(SERVER_GMX, use_uid=True)
elif DOMAIN_SINA in login:
# imap = imaplib.IMAP4_SSL(IMAP_SERVER_SINA)
imap = IMAPClient(IMAP_SERVER_SINA, use_uid=True)
elif DOMAIN_HOTMAIL in login or DOMAIN_OUTLOOK_COM in login:
imap = imaplib.IMAP4_SSL(HOTMAIL_IMAP_SERVER)
# imap = imaplib.IMAP4_SSL(HOTMAIL_IMAP_SERVER)
imap = IMAPClient(HOTMAIL_IMAP_SERVER, use_uid=True)
elif DOMAIN_RAMBLER_RU in login:
imap = imaplib.IMAP4_SSL(RAMBLER_IMAP_SERVER)
# imap = imaplib.IMAP4_SSL(RAMBLER_IMAP_SERVER)
imap = IMAPClient(RAMBLER_IMAP_SERVER, use_uid=True)
elif DOMAN_BTVM_NE_JP in login:
imap = imaplib.IMAP4_SSL(BTVM_NE_JP)
# imap = imaplib.IMAP4_SSL(BTVM_NE_JP)
imap = IMAPClient(BTVM_NE_JP, use_uid=True)
elif DOMAN_GMAIL in login:
imap = imaplib.IMAP4_SSL(SEREVER_GMAIL, port=993)
# imap = imaplib.IMAP4_SSL(SEREVER_GMAIL, port=993)
imap = IMAPClient(SEREVER_GMAIL, use_uid=True)
elif DOMAIN_ONET in login:
imap = IMAPClient(SERVER_IMAGE_ONET, use_uid=True)
elif DOMAIN_TIM_IT in login:
imap = imaplib.IMAP4(TIME_IT_SERVER)
# imap = imaplib.IMAP4(TIME_IT_SERVER)
imap = IMAPClient(TIME_IT_SERVER, use_uid=True)
elif DOMAIN_ALICE_IT in login:
imap = imaplib.IMAP4(ALICE_IMAP_SERVER, port=143)
# imap = imaplib.IMAP4(ALICE_IMAP_SERVER, port=143)
imap = IMAPClient(ALICE_IMAP_SERVER, use_uid=True)
elif DOMAIN_MARS_DTI_NE_JP in login:
imap = imaplib.IMAP4(MARS_DTI_NE_JP_SERVER, port=143)
# imap = imaplib.IMAP4(MARS_DTI_NE_JP_SERVER, port=143)
imap = IMAPClient(MARS_DTI_NE_JP_SERVER, use_uid=True)
elif DOMAN_AURORA_DTI_NE_JP in login:
imap = imaplib.IMAP4(MARS_DTI_NE_JP_SERVER, port=143)
# imap = imaplib.IMAP4(MARS_DTI_NE_JP_SERVER, port=143)
imap = IMAPClient(MARS_DTI_NE_JP_SERVER, use_uid=True)
elif DOMAIN_NAVER in login:
imap = imaplib.IMAP4_SSL(NAVER_SERVER, port=993)
# imap = imaplib.IMAP4_SSL(NAVER_SERVER, port=993)
imap = IMAPClient(NAVER_SERVER, use_uid=True)
elif DOMAIN_GMX_DE in login or DOMAIN_GMX_NET in login:
imap = imaplib.IMAP4_SSL(SERVER_GMX_NET, port=993)
# imap = imaplib.IMAP4_SSL(SERVER_GMX_NET, port=993)
imap = IMAPClient(SERVER_GMX_NET, use_uid=True)
elif DOMAIN_GMX_AT in login:
imap = imaplib.IMAP4_SSL(SERVER_GMX_AT, port=993)
# imap = imaplib.IMAP4_SSL(SERVER_GMX_AT, port=993)
imap = IMAPClient(SERVER_GMX_AT, use_uid=True)
elif DOMAIN_GAZETA_PL in login:
imap = imaplib.IMAP4_SSL(IMAP_SERVER_DOMAIN_GAZETA_PL, port=993)
# imap = imaplib.IMAP4_SSL(IMAP_SERVER_DOMAIN_GAZETA_PL, port=993)
imap = IMAPClient(IMAP_SERVER_DOMAIN_GAZETA_PL, use_uid=True)
elif DOMAIN_INBOX_LV in login:
imap = imaplib.IMAP4_SSL(INBOX_LV, port=993)
# imap = imaplib.IMAP4_SSL(INBOX_LV, port=993)
imap = IMAPClient(INBOX_LV, use_uid=True)
elif DOMAIN_WEB_DE in login:
imap = imaplib.IMAP4_SSL(SERVER_WEB_DE, port=993)
# imap = imaplib.IMAP4_SSL(SERVER_WEB_DE, port=993)
imap = IMAPClient(SERVER_WEB_DE, use_uid=True)
elif DOMAIN_PISS_MAIL in login or DOMAIN_CHILD_PIZZA in login or DOMAIN_DMC_CHAT in login or DOMAIN_GENOCIDE_FUN in login or DOMAIN_HATESJE_WS in login or DOMAIN_INCEL_EMAIL in login or DOMAIN_SHITPOSTING_EXPERT in login:
imap = imaplib.IMAP4_SSL(SERVER_PISS_MAIL, port=993)
# imap = imaplib.IMAP4_SSL(SERVER_PISS_MAIL, port=993)
imap = IMAPClient(SERVER_PISS_MAIL, use_uid=True)
else:
imap = imaplib.IMAP4_SSL(AOL_IMAP_SERVER)
# imap = imaplib.IMAP4_SSL(AOL_IMAP_SERVER)
imap = IMAPClient(AOL_IMAP_SERVER, use_uid=True)
return imap
+41 -7
View File
@@ -9,6 +9,7 @@ from email.message import Message
from imapclient import IMAPClient
from db.mongo_manager import MONGO_STORE_MANAGER
from excel_reader import read_contacts
from mail.mail_constants import DOMAIN_HOTMAIL, create_imap
from models.mail_pojo import MailPojo, MailAddress
@@ -48,6 +49,11 @@ def find_from_mail(param):
return from_address.strip(" ").strip(">").strip("<")
def extract_email_from_from_address(content: str):
_match = re.search(r'[\w.+-]+@[\w-]+\.[\w.-]+', content)
return _match.group(0)
class MailReader():
def __init__(self, login, password):
self.login = login
@@ -170,23 +176,27 @@ class MailReader():
email_message = email.message_from_bytes(message_data[b'RFC822'])
from_address = email_message.get('FROM')
subject = email_message.get('subject')
# print("{}, {},{}".format(from_address, subject, email_message))
body = ""
if "no-reply@hermes.com" in from_address or "appointment2022@aol.com":
hermes_mail_adress = "no-reply@hermes.com"
if hermes_mail_adress in from_address or "outlook.com" in from_address or "hotmail" in from_address:
for part in email_message.walk():
print(part.get_content_type())
if part.get_content_type() == "text/html":
body = body + part.get_payload(decode=True).decode("utf-8")
elif part.get_content_type() == "text/plain":
body = body + part.get_payload()
if VALIDATION_URL_SUBJECT_fr in subject or VALIDATION_URL_SUBJECT_EN in subject or "Validation=20de=20votre=20demande=20de=20rendez-vous" in subject:
if VALIDATION_URL_SUBJECT_fr in subject or VALIDATION_URL_SUBJECT_EN in subject or "Votre=20demande=20de=20rendez-vous" in subject or "Votre demande de rendez-vous" in body:
mail = MailPojo(subject=subject, body=body, from_address=from_address)
mail.isImapClient = True
print("email is {}".format(self.login))
print("body is {}".format(body))
print("subject is {}".format(subject))
if len(mail.to_address) == 0:
mail.to_address = self.login
if "outlook.com" in from_address or "hotmail.com" in from_address:
# it is a transferred email
mail.to_address = extract_email_from_from_address(from_address)
else:
mail.to_address = self.login
mail_messages.append(mail)
except Exception as error:
print(error)
@@ -262,7 +272,31 @@ def find_links_to_validate_from_mail_list(mail_list: list, logger):
if __name__ == '__main__':
mail_address1 = MailAddress(mail="tzfhygwwjiwnf@hotmail.com", password="rqO4Pjm7NQ")
mail_list = [mail_address1]
# mail_address1 = MailAddress(mail="tinagonzales685585@aol.com", password="yhihvdkrbxnksema")
# mail_list = [mail_address1]
contact_to_book_list = read_contacts(
# file_name="/Users/rdv/Desktop/contact_list_not_used_contacts.xlsx")
# file_name="/Users/lpan/Desktop/contact_list_not_used_contacts.xlsx")
# file_name="/Users/rdv/Desktop/real_name_contacts_100_hotmail.xlsx")
file_name="/Users/rdv/Desktop/contact_list_2025-06-23.xlsx")
# file_name="/Users/rdv/Desktop/contact_list_all_studio.xlsx")
# file_name="/Users/rdv/Desktop/contact_list_all_studo_gmx_us.xlsx")
# file_name="/Users/rdv/Desktop/contact_list_2025-05-24.xlsx")
all_mail_list = MONGO_STORE_MANAGER.get_destination_emails()
mail_list_to_check = []
for contact in contact_to_book_list:
for mail in all_mail_list:
if contact.mail == mail.mail:
mail_list_to_check.append(mail)
logger = logging.getLogger()
find_links_to_validate_from_mail_list(mail_list, logger)
_all_links = MONGO_STORE_MANAGER.get_links_to_validate()
filter_mail = []
for mail_pojo in mail_list_to_check:
_to_add = True
for _link in _all_links:
if _link.email == mail_pojo.mail:
_to_add = False
if _to_add:
filter_mail.append(mail_pojo)
filter_mail.append(MailAddress("saigecong1990@pissmail.com", "cvExXKOP8oY1D@"))
find_links_to_validate_from_mail_list(filter_mail, logger)
+1 -1
View File
@@ -3,6 +3,6 @@ from queue_message.CookiesPublisher import MORNING_DATA_CACHE_2, MORNING_DATA_CA
from workers.MessagerTransporter import migrate_message_to_queue, empty_message_just_to
if __name__ == '__main__':
migrate_message_to_queue(from_queue=MORNING_DATA_CACHE_BAK, to_queue=MORNING_DATA_CACHE)
migrate_message_to_queue(from_queue=MORNING_DATA_CACHE, to_queue=MORNING_DATA_CACHE_BAK)
# empty_message_just_to(2000, queue_name=MORNING_DATA_CACHE_BAK)
# migrate_message_to_queue(from_queue=MORNING_DATA_CACHE_BAK,to_queue=REQUEST_DATA_QUEUE_DE)
+3 -1
View File
@@ -11,7 +11,9 @@ class LinkPojo():
def from_firestore_dict(source):
updated_at = source['updated_at']
email = source['email']
model = source['model']
model = ""
if 'model' in source:
model = source['model']
serial = source['serial']
url = source['url']
ip_country = "FR"
+2 -2
View File
@@ -29,6 +29,7 @@ class ReserveResultPojo:
slot_position = None
sim_position = None
ccid: str = ""
model: str = ""
source_from: str = socket.gethostname()
store_type = 0
url_validated = None
@@ -122,8 +123,7 @@ class ReserveResultPojo:
u'email': self.email,
u'passport': self.passport,
u'url': self.url,
# u'sim_position': self.sim_position,
# u'slot_position': self.slot_position,
u'model': self.model,
u'source_from': self.source_from,
u'hostName': self.source_from,
u'created_at': self.created_at,
+29 -15
View File
@@ -1,5 +1,8 @@
import logging
import random
import requests
FR_ASOCKS_MOBILE_PROXY = {
'http': 'http://11797317-mobile-country-FR:nv958134x@190.2.151.110:14046',
'https': 'http://11797317-mobile-country-FR:nv958134x@190.2.151.110:14046',
@@ -16,8 +19,8 @@ FR_PROXY_RES_PARIS_OXY = {
}
FR_PROXY_MOB_OXY = {
'http': 'http://customer-rendezvousmob-cc-FR:Rdv202220212023@pr.oxylabs.io:7777',
'https': 'http://customer-rendezvousmob-cc-FR:Rdv202220212023@pr.oxylabs.io:7777'
'http': 'http://customer-rendezvousmob-cc-FR:Rdv+20222021@pr.oxylabs.io:7777',
'https': 'http://customer-rendezvousmob-cc-FR:Rdv+20222021@pr.oxylabs.io:7777'
}
FR_PROXY_ASOCK_RES_2 = {
@@ -25,8 +28,8 @@ FR_PROXY_ASOCK_RES_2 = {
'https': 'http://zd6fbrujot-res-country-FR-hold-query:8k8avNlnLHQaMsWg@217.23.6.161:9999'
}
FR_PROXY_MOB_OXY_STICKY = {
'http': 'http://customer-rendezvousmob-cc-FR:Rdv202220212023@fr-pr.oxylabs.io:{}',
'https': 'http://customer-rendezvousmob-cc-FR:Rdv202220212023@fr-pr.oxylabs.io:{}'
'http': 'http://customer-rendezvousmob-cc-fr-sessid-0{}-sesstime-2:Rdv+202220212023@pr.oxylabs.io:7777',
'https': 'http://customer-rendezvousmob-cc-fr-sessid-0{}-sesstime-2:Rdv+202220212023@pr.oxylabs.io:7777'
}
FR_PROXY_DATA_IMPULSE_STICKY = {
@@ -52,17 +55,16 @@ FR_MOBILE_ANY_IP_ROTATING = {
'https': 'http://user_6a7f21,type_residential,country_FR:d5c051@portal.anyip.io:1080',
}
# 八分之一用data_impulse
MOBILE_PROXY_LIST = [FR_PROXY_MOB_OXY_STICKY, FR_PROXY_MOB_OXY_STICKY, FR_PROXY_MOB_OXY_STICKY, FR_PROXY_MOB_OXY_STICKY,
FR_PROXY_MOB_OXY_STICKY, FR_MOBILE_ANY_IP_STICKY, FR_MOBILE_ANY_IP_STICKY, FR_MOBILE_ANY_IP_STICKY,
FR_MOBILE_ANY_IP_STICKY, FR_MOBILE_ANY_IP_STICKY,
FR_PROXY_RES_OXY_STICKY,
FR_PROXY_DATA_IMPULSE_STICKY]
# MOBILE_PROXY_LIST = [FR_PROXY_MOB_OXY_STICKY, FR_PROXY_MOB_OXY_STICKY, FR_PROXY_MOB_OXY_STICKY, FR_PROXY_MOB_OXY_STICKY,
# FR_PROXY_MOB_OXY_STICKY, FR_MOBILE_ANY_IP_STICKY, FR_MOBILE_ANY_IP_STICKY, FR_MOBILE_ANY_IP_STICKY,
# FR_MOBILE_ANY_IP_STICKY, FR_MOBILE_ANY_IP_STICKY,
# FR_PROXY_RES_OXY_STICKY,
# FR_PROXY_DATA_IMPULSE_STICKY]
# MOBILE_PROXY_LIST = [FR_MOBILE_ANY_IP_STICKY]
# MOBILE_PROXY_LIST = [FR_MOBILE_ANY_IP_STICKY]
# MOBILE_PROXY_LIST = [FR_MOBILE_ANY_IP_STICKY, FR_PROXY_MOB_OXY_STICKY]
# MOBILE_PROXY_LIST = [FR_PROXY_MOB_OXY_STICKY]
MOBILE_PROXY_LIST = [FR_MOBILE_ANY_IP_STICKY]
class ProxyManager:
@@ -72,7 +74,8 @@ class ProxyManager:
def get_link_validate_proxy(self, links_to_validate: list) -> list:
# return [FR_PROXY_RES_PARIS_OXY]
return [FR_MOBILE_ANY_IP_ROTATING]
# return [FR_MOBILE_ANY_IP_ROTATING]
return [FR_PROXY_RES_OXY]
# if len(links_to_validate) > 15:
# return [FR_PROXY_RES_OXY, FR_PROXY_MOB_OXY, FR_PROXY_ASOCK_RES_2, FR_DATA_IMPULSE_RES]
# # return [FR_PROXY_RES_OXY, FR_PROXY_ASOCK_RES_2, FR_DATA_IMPULSE_RES, FR_ASOCKS_MOBILE_PROXY]
@@ -88,7 +91,7 @@ class ProxyManager:
_chosen_proxy = random.choice(MOBILE_PROXY_LIST)
if "oxylabs" in _chosen_proxy["http"]:
self.logger.info("use oxylabs proxy")
_port = random.randint(40001, 49999)
_port = random.randint(900000000, 995869818)
elif "anyip" in _chosen_proxy["http"]:
self.logger.info("use anyip proxy")
_port = random.randint(40001, 49999)
@@ -100,3 +103,14 @@ class ProxyManager:
_proxy_to_use["http"] = _chosen_proxy["http"].format(_port)
_proxy_to_use["https"] = _chosen_proxy["https"].format(_port)
return _proxy_to_use
if __name__ == '__main__':
_logger = logging.getLogger()
_proxy = ProxyManager(logger=_logger)
_proxise = _proxy.get_proxy_for_appointment_request()
print(_proxise)
response = requests.get(
"https://ip.oxylabs.io/location",
proxies=_proxise)
print(response.content)
@@ -239,6 +239,7 @@ class AppointmentRequestSender(threading.Thread):
self.channel.start_consuming()
def retrieve_invalidate_urls(self):
self.list_to_retrieve_mails = self.initial_contact_list
# 如果没有已读邮件,而且需要读邮件的联系人表不为空,就读取未读邮件
if not self.already_read_emails and len(self.list_to_retrieve_mails) > 0:
self.logger.info("will retrieve validate urls")
+3 -3
View File
@@ -71,10 +71,10 @@ def send_appointment_request(message_queue_name, _contact_list):
def start_send_requests():
print("start send requests")
contacts_file_path = '~/Desktop/11_05_to_test.xlsx'
_contact_list = read_contacts(contacts_file_path)[:-1]
contacts_file_path = '~/Desktop/06_05_to_test.xlsx'
_contact_list = read_contacts(contacts_file_path)[:1]
_contact_list_to_book = filter_contacts(_contact_list)
_segment_number = 5
_segment_number = 1
logger.info("{} contacts to book".format(len(_contact_list_to_book)))
last_thread = None
for i in range(0, _segment_number):
+2 -2
View File
@@ -103,6 +103,6 @@ if __name__ == '__main__':
# '~/Desktop/contact_list_2024-05-21.xlsx',
# '~/Desktop/15_05_to_test.xlsx']
# file_list = ['~/Desktop/15_05_to_test.xlsx', '~/Desktop/16_05_to_test.xlsx']
file_list = ['~/Desktop/contact_list_2024-10-05_gmx_net_100.xlsx']
send_request_for_file_list(file_list=file_list, thread_number=1,
file_list = ['~/Desktop/contact_list_2025-06-09_2.xlsx']
send_request_for_file_list(file_list=file_list, thread_number=10,
data_queue_name=MORNING_DATA_CACHE, stop_at_hour=19, stop_at_mins=50)
+2 -4
View File
@@ -6,10 +6,8 @@ from request_sender_test import send_request_for_file_list
def start_book_appointment():
file_list = ['~/Desktop/contact_list_2024-09-06.xlsx',
'~/Desktop/contact_list_all.xlsx',
'~/Desktop/contact_list_2024-05-27.xlsx']
send_request_for_file_list(file_list=file_list, thread_number=80,
file_list = ['~/Desktop/contact_list_2025-06-09_2.xlsx']
send_request_for_file_list(file_list=file_list, thread_number=30,
data_queue_name=MORNING_DATA_CACHE, stop_at_hour=11, stop_at_mins=20)
+56 -40
View File
@@ -19,7 +19,7 @@ from utils.AppLogging import init_logger
from utils.user_agent_helper import generate_headers_from_request_message
DOUBLE_MESSAGE = "Une demande de rendez-vous a déjà été enregistrée avec ces coordonnées"
INVALID = "Depuis plus de 130 ans,"
INVALID = "nos conseillers vous accueillent sur rendez-vous dans nos magasins"
init_logger()
logger = logging.getLogger()
@@ -28,7 +28,7 @@ logger = logging.getLogger()
def filter_link_pojo_list_with_serial(_received_dict, link_to_validate_list):
_serial = _received_dict["serial"]
_model = _received_dict["model"]
_bit_browser_list = list(filter(lambda link_pojo: link_pojo.serial == "bitbrowser", link_to_validate_list))
_bit_browser_list = list(filter(lambda link_pojo: link_pojo.serial == "requests", link_to_validate_list))
if len(_bit_browser_list) > 0:
return _bit_browser_list
else:
@@ -149,44 +149,51 @@ class LinkValidatorWithProvidedList(threading.Thread):
_message_in_queue_count = self.cookiesPublisher.message_count()
print("message count in queue is {}".format(_message_in_queue_count))
_received_object = body.decode("UTF-8")
_received_dict = json.loads(_received_object)
_received_cookies = _received_dict["cookiesStr"]
self.cookie_str = _received_cookies
_links_to_validate = filter_link_pojo_list_with_serial(_received_dict, self.link_to_validate_list)
random.shuffle(_links_to_validate)
if len(_links_to_validate) > 0 and _message_in_queue_count >= self.limit:
print("{}:links number is {}".format(threading.currentThread().name, len(_links_to_validate)))
can_continue = None
for link_to_validate in _links_to_validate:
print(link_to_validate)
self.proxy_to_use = random.choice(
self.proxy_manager.get_link_validate_proxy(_links_to_validate))
print("proxy to use is {}".format(self.proxy_to_use))
can_continue = self.send_request(link_to_validate, _received_dict)
# remove the tested link from link list
self.link_to_validate_list.remove(link_to_validate)
if can_continue == RequestResult.BLOCKED:
print("cannot continue, blocked, then skip")
break
else:
if can_continue == RequestResult.BAD_GATEWAY:
time.sleep(30)
try:
_received_dict = json.loads(_received_object)
_received_cookies = _received_dict["cookiesStr"]
self.cookie_str = _received_cookies
_links_to_validate = filter_link_pojo_list_with_serial(_received_dict, self.link_to_validate_list)
random.shuffle(_links_to_validate)
if len(_links_to_validate) > 0 and _message_in_queue_count >= self.limit:
print("{}:links number is {}".format(threading.currentThread().name, len(_links_to_validate)))
can_continue = None
for link_to_validate in _links_to_validate:
print(link_to_validate)
self.proxy_to_use = random.choice(
self.proxy_manager.get_link_validate_proxy(_links_to_validate))
print("proxy to use is {}".format(self.proxy_to_use))
can_continue = self.send_request(link_to_validate, _received_dict)
# remove the tested link from link list
self.link_to_validate_list.remove(link_to_validate)
if can_continue == RequestResult.BLOCKED:
print("cannot continue, blocked, then skip")
break
time.sleep(random.randint(2, 5))
print("can continue, continue")
if can_continue == RequestResult.BAD_GATEWAY or can_continue == RequestResult.PROXY_ERROR or can_continue == RequestResult.SUCCESS:
print("will requeue the message")
ch.basic_reject(delivery_tag=method.delivery_tag, requeue=True)
else:
if can_continue == RequestResult.BAD_GATEWAY:
time.sleep(30)
break
time.sleep(random.randint(2, 5))
print("can continue, continue")
if can_continue == RequestResult.BAD_GATEWAY or can_continue == RequestResult.PROXY_ERROR or can_continue == RequestResult.SUCCESS:
print("will requeue the message")
ch.basic_reject(delivery_tag=method.delivery_tag, requeue=True)
else:
print("will ack")
print("will wait for 40s")
time.sleep(20)
ch.basic_ack(delivery_tag=method.delivery_tag)
else:
print("will ack")
print("empty list, will republish message")
time.sleep(0)
print("body in json:{}".format(json.dumps(_received_dict)))
self.cookiesPublisher.publish_body(json.dumps(_received_dict))
ch.basic_ack(delivery_tag=method.delivery_tag)
else:
print("empty list, will republish message")
time.sleep(0)
print("body in json:{}".format(json.dumps(_received_dict)))
self.cookiesPublisher.publish_body(json.dumps(_received_dict))
self.channel.stop_consuming()
except Exception as error:
print(error)
print("not json format will ack")
ch.basic_ack(delivery_tag=method.delivery_tag)
self.channel.stop_consuming()
def filter_with_ip_country(self):
_link_list_to_click = []
@@ -210,12 +217,18 @@ def validate_links(cookiesPublisher, queue_name: str, link_list: list, _contact_
def validate_all_links(_contact_serial_list):
print("will get all links")
all_link_list = MONGO_STORE_MANAGER.get_links_to_validate()
link_to_validated = []
for _link in all_link_list:
# print("serial is "+_link.serial)
print("email is " + _link.email)
if _link.serial == "requests":
link_to_validated.append(_link)
# get the first 50 links
if len(all_link_list) == 0:
if len(link_to_validated) == 0:
return
# divided = 1
divided = 1
# default_segment_number = 20
# _first_25_percent_links = all_link_list[0:(int(len(all_link_list) / divided))]
_first_25_percent_links = link_to_validated[0:(int(len(all_link_list) / divided))]
_first_25_percent_links = all_link_list
_queue_name = MORNING_DATA_CACHE
# _queue_name = MORNING_DATA_CACHE_BAK
@@ -228,7 +241,10 @@ def validate_all_links(_contact_serial_list):
# else:
# _segment_number = 1
_thread_list = []
_segment_number = 1
if len(_first_25_percent_links) >= 10:
_segment_number = 10
else:
_segment_number = len(_first_25_percent_links)
for i in range(0, _segment_number):
logger.info("{}:{} links to validate".format(threading.currentThread().name, len(_first_25_percent_links)))
logger.info("segment is {}".format(i))
+7 -7
View File
@@ -1,7 +1,5 @@
import datetime
import json
import logging
import random
import re
import time
from http.cookies import SimpleCookie
@@ -40,7 +38,8 @@ class Sender:
self.proxy_to_use = proxy_to_use
self.cookie.load(self.cookie_str)
def publish_message_to_queue(self, contact: ContactPojo, status: PublishType, url: str, store_type: str):
def publish_message_to_queue(self, contact: ContactPojo, status: PublishType, url: str, store_type: str,
model: str = ""):
# create the message
if url == "https://rendezvousparis.hermes.com/client/welcome":
return
@@ -61,11 +60,12 @@ class Sender:
result.proxy = "data_impulse"
result.id = id
result.store_type = store_type
result.model = model
result.created_at = time.strftime("%H:%M:%S", time.localtime())
collection_name = str(datetime.date.today())
MONGO_STORE_MANAGER.insert_reserve_result(collection_name=collection_name, reserve=result)
def apply_redirect(self, response, old_headers, contact, js_data, selected_store):
def apply_redirect(self, response, old_headers, contact, js_data, selected_store, model=""):
# /client/register/5XD2E2
_res_headers = response.headers
_location = _res_headers['location']
@@ -89,7 +89,7 @@ class Sender:
contact.current_ip = get_address_ip(proxy_to_use=self.proxy_to_use)
_appointment_url = _redirect_url
self.publish_message_to_queue(contact, status=PublishType.SUCCESS, url=_appointment_url,
store_type=selected_store)
store_type=selected_store, model=model)
self.cookie.load(_cookies_to_set)
new_cookies = {k: v.value for k, v in self.cookie.items()}
new_coolies_str = ""
@@ -162,7 +162,7 @@ class Sender:
headers['sec-ch-ua-full-version-list'] = _chrome_ua_version
data = {'check': '', '_csrf': _csrf, 'prefer': _selected_store,
'surname': contact.last_name.capitalize(), 'name': contact.first_name.capitalize(),
'phone_country': "FR", 'phone_number': "+33 " + contact.phone, 'email': contact.mail,
'phone_country': "FR", 'phone_number': "0" + contact.phone, 'email': contact.mail,
'passport_id': contact.passport, 'processing': 'on', 'cgu': 'on'}
self.logger.info(data)
self.logger.info("try to request for mail:{}".format(contact.mail))
@@ -177,7 +177,7 @@ class Sender:
# add to mongodb
self.logger.info(response.text)
self.apply_redirect(response=response, old_headers=headers, contact=contact, js_data=js_data,
selected_store=_selected_store)
selected_store=_selected_store, model=model)
# self.logger.info("{}:{}".format(contact.mail, response.url))
# contact.ua = js_data.ua
# contact.current_ip = get_address_ip(proxy_to_use=proxy_to_use)