support invalid url
This commit is contained in:
@@ -11,10 +11,11 @@ from models.LinkPojo import LinkPojo
|
||||
from models.result_pojo import RequestResult
|
||||
from proxy_manager.proxy_manager import ProxyManager
|
||||
from queue_message.CookiesPublisher import CookiesPublisher, REQUEST_DATA_QUEUE_TEST
|
||||
from queue_message.appointmentrequestsender import QUEUE_HOST, REQUEST_DATA_QUEUE, credentials
|
||||
from queue_message.appointmentrequestsender import QUEUE_HOST, REQUEST_DATA_QUEUE, credentials, REQUEST_DATA_DE
|
||||
from workers.proxies_constants import PROXY_LIST_FR
|
||||
|
||||
DOUBLE_MESSAGE = "Une demande de rendez-vous a déjà été enregistrée avec ces coordonnées"
|
||||
INVALID = "Depuis plus de 130 ans,"
|
||||
|
||||
|
||||
class LinkValidator(threading.Thread):
|
||||
@@ -63,6 +64,7 @@ class LinkValidator(threading.Thread):
|
||||
_proxy_to_use = random.choice(self.proxy_manager.get_link_validate_proxy(self.link_to_validate_list))
|
||||
print(_proxy_to_use)
|
||||
print("received cookie is " + str(self.cookie_str))
|
||||
print("send request for link: " + linkPojo.url)
|
||||
try:
|
||||
response = requests.get(url=linkPojo.url, headers=headers, verify=False, proxies=_proxy_to_use,
|
||||
timeout=30)
|
||||
@@ -85,6 +87,22 @@ class LinkValidator(threading.Thread):
|
||||
self.cookiesPublisher.publish_body(new_coolies_str)
|
||||
self.cookie_str = new_coolies_str
|
||||
return RequestResult.SUCCESS
|
||||
elif INVALID in _content:
|
||||
|
||||
MONGO_STORE_MANAGER.link_validated_for_result(response.url, linkPojo, is_invalid=True)
|
||||
# set new cookies
|
||||
_cookies_to_set = response.headers['set-cookie']
|
||||
self.cookie.load(_cookies_to_set)
|
||||
new_cookies = {k: v.value for k, v in self.cookie.items()}
|
||||
new_coolies_str = ""
|
||||
for key in new_cookies:
|
||||
new_coolies_str = new_coolies_str + key + "=" + new_cookies[key] + ";"
|
||||
print("will publish to queue {}".format(new_coolies_str))
|
||||
# upload the cookie to queue
|
||||
self.cookiesPublisher.publish_body(new_coolies_str)
|
||||
self.cookie_str = new_coolies_str
|
||||
return RequestResult.SUCCESS
|
||||
|
||||
elif DOUBLE_MESSAGE in _content:
|
||||
print(response.url)
|
||||
MONGO_STORE_MANAGER.link_validated_for_result(response.url, linkPojo, is_duplicated=True)
|
||||
@@ -171,14 +189,15 @@ class LinkValidator(threading.Thread):
|
||||
|
||||
|
||||
def validate_with_FR_ip(segment_position=1):
|
||||
_queue_name = REQUEST_DATA_QUEUE
|
||||
# _queue_name = REQUEST_DATA_QUEUE
|
||||
_queue_name = REQUEST_DATA_DE
|
||||
cookiesPublisher = CookiesPublisher(queue_name=_queue_name)
|
||||
cookiesPublisher.set_up_connection()
|
||||
print("filter links with ip_country")
|
||||
_proxy_manager = ProxyManager()
|
||||
receiver = LinkValidator(cookiesPublisher=cookiesPublisher,
|
||||
proxy_manager=_proxy_manager,
|
||||
queue_to_listen=_queue_name, ip_country="FR", segment_position=segment_position, limit=50)
|
||||
queue_to_listen=_queue_name, ip_country="FR", segment_position=segment_position, limit=0)
|
||||
print("will connect to queue")
|
||||
receiver.set_up_connection()
|
||||
receiver.listen_to_queue(receiver.on_message)
|
||||
|
||||
Reference in New Issue
Block a user