Ejemplo n.º 1
0
 def profile_worker(self):
     """
     Given a prefilled profile queue this worker will pop an id
     and fetch the associated profile
     :return:
     """
     while not self._profile_queue.empty():
         profile_id = self._profile_queue.get()
         try:
             # Fetch the profile
             profile = self._crawler.get_profile_by_id(profile_id)
             self._profiles.append(profile)
             log.debug(
                 "The profile for profile_id {profile_id} has been fetched".
                 format(profile_id=profile_id))
             # Mark task as done
             self._profile_queue.task_done()
         except Exception as e:
             log.warn(
                 "Failed to fetch the profile for profile_id {profile_id}".
                 format(profile_id=profile_id))
             self._profile_queue.task_done()
Ejemplo n.º 2
0
 def profile_worker(self):
     """
     Given a prefilled profile queue this worker will pop an id
     and fetch the associated profile
     :return:
     """
     while not self._profile_queue.empty():
         profile_id = self._profile_queue.get()
         try:
             # Fetch the profile
             profile = self._crawler.get_profile_by_id(profile_id)
             self._profiles.append(profile)
             log.debug("The profile for profile_id {profile_id} has been fetched".format(
                 profile_id=profile_id
             ))
             # Mark task as done
             self._profile_queue.task_done()
         except Exception as e:
             log.warn("Failed to fetch the profile for profile_id {profile_id}".format(
                 profile_id=profile_id
             ))
             self._profile_queue.task_done()
Ejemplo n.º 3
0
 def document_worker(self):
     """
     Given a prefilled profile_documents queue this worker will pop an id
     and fetch the associated documents
     :return:
     """
     while not self._profile_documents_queue.empty():
         profile_id = self._profile_documents_queue.get()
         try:
             # Fetch the document
             documents = self._crawler.get_documents_by_profile_id(
                 profile_id)
             self._profile_documents[profile_id] = documents
             log.debug(
                 "{num} documents have been fetched for profile_id {profile_id}"
                 .format(num=len(documents), profile_id=profile_id))
             # Mark task as done
             self._profile_documents_queue.task_done()
         except Exception as e:
             log.warn(
                 "Failed to fetch documents for profile_id {profile_id}".
                 format(profile_id=profile_id))
             self._profile_documents_queue.task_done()
Ejemplo n.º 4
0
 def document_worker(self):
     """
     Given a prefilled profile_documents queue this worker will pop an id
     and fetch the associated documents
     :return:
     """
     while not self._profile_documents_queue.empty():
         profile_id = self._profile_documents_queue.get()
         try:
             # Fetch the document
             documents = self._crawler.get_documents_by_profile_id(profile_id)
             self._profile_documents[profile_id] = documents
             log.debug("{num} documents have been fetched for profile_id {profile_id}".format(
                 num=len(documents),
                 profile_id=profile_id
             ))
             # Mark task as done
             self._profile_documents_queue.task_done()
         except Exception as e:
             log.warn("Failed to fetch documents for profile_id {profile_id}".format(
                 profile_id=profile_id
             ))
             self._profile_documents_queue.task_done()