Exemple #1
0
 def testInitWithSeconds(self):
     """Tests that sepcifying seconds sets the correct fields."""
     self.requester = requester.Requester(None, None, None, 0.1, seconds=2)
     self.assertEqual(self.requester._time_between_requests, 0.1)
     self.assertEqual(self.requester._last_request_start_time, 0.0)
     self.assertEqual(self.requester._timedelta, 2)
     self.assertFalse(self.requester._use_requests_as_stop_signal)
Exemple #2
0
    def testShouldSendMoreRequestsStopsAfterTimeout(self):
        """Tests that ShouldSendMoreRequests stops after the given timeout value."""
        self.requester = requester.Requester(MockGenerator(),
                                             None,
                                             None,
                                             0.1,
                                             seconds=10)
        # Mock out _ShouldSendMoreRequests so that we can invoke Start to set the
        # start and stop time, while otherwise being a no-op.
        should_send_more_requests = self.requester._ShouldSendMoreRequests
        self.requester._ShouldSendMoreRequests = MockMethod(
            '_ShouldSendMoreRequests', [()], [False])
        current_time = time.time()
        # _GetCurrentTime will get invoked once in Start to set the start time, then
        # on each invocation of _ShouldSendMoreRequests.
        self.requester._GetCurrentTime = MockMethod('_GetCurrentTime', [
            (), (), (), ()
        ], [
            current_time, current_time + 1, current_time + 9, current_time + 10
        ])
        # Sets _start_time and _stop_time.
        self.requester.Start()
        # Check that start and stop times are as expected.
        self.assertEqual(self.requester._start_time, current_time)
        self.assertEqual(self.requester._stop_time, current_time + 10)

        # Restore _ShouldSendMoreRequests so we can test it.
        self.requester._ShouldSendMoreRequests = should_send_more_requests
        # 1 second has passed.
        self.assertTrue(self.requester._ShouldSendMoreRequests())
        # 9 seconds has passed.
        self.assertTrue(self.requester._ShouldSendMoreRequests())
        # 10 seconds has passed.
        self.assertFalse(self.requester._ShouldSendMoreRequests())
Exemple #3
0
 def testInitWithRequests(self):
     """Tests that specifying requests sets the correct fields."""
     self.requester = requester.Requester(None, None, None, 0.1, requests=2)
     self.assertEqual(self.requester._time_between_requests, 0.1)
     self.assertEqual(self.requester._last_request_start_time, 0.0)
     self.assertEqual(self.requester._max_requests, 2)
     self.assertTrue(self.requester._use_requests_as_stop_signal)
Exemple #4
0
 def testGenerateRequest(self):
   generator = MockGenerator()
   self.requester = requester.Requester(generator, None, None, 0.1,
                                        requests=2)
   self.assertEqual(0, self.requester._generated_requests)
   request = self.requester._GenerateRequest()
   self.assertEqual(1, self.requester._generated_requests)
   self.assertEqual(generator.request, request)
Exemple #5
0
 def build_domain(self, url):
     """
     :param domaininfo:
     :return:
     """
     d = parse(url, rule='IRI')
     li = d['authority'].split('.')
     domain = li[len(li) - 2]
     self.is_valid(domain)
     module = self.my_import('domains.' + domain)
     self.fetcher = module.Fetcher(requester.Requester())
Exemple #6
0
 def testShouldSendMoreRequestsStopsOnMaxRequests(self):
   """Tests that ShouldSendMoreRequests stops at the maximum request count."""
   generator = MockGenerator()
   self.requester = requester.Requester(generator, None, None, 0.1,
                                        requests=2)
   # 0 requests sent.
   self.assertTrue(self.requester._ShouldSendMoreRequests())
   self.requester._GenerateRequest()
   # 1 requests sent.
   self.assertTrue(self.requester._ShouldSendMoreRequests())
   self.requester._GenerateRequest()
   # 2 requests sent, we ignore the lock since no multi-threading is going on.
   self.assertFalse(self.requester._ShouldSendMoreRequests())
Exemple #7
0
    def testWait(self):
        """Tests that _Wait sleeps for the correct amount of time."""
        time_to_wait = 0.1
        self.requester = requester.Requester(None,
                                             None,
                                             None,
                                             time_to_wait,
                                             seconds=10)

        # current time = 1.0, last_request_start_time = 0.0 (i.e. first request),
        # time_to_wait = 0.1.
        # Wait shouldn't call _GetCurrentTime because this is the first request.
        # Wait should call sleep with time 0.1.
        time.sleep = MockMethod('sleep', [(0.1, )], [None])
        self.requester._GetCurrentTime = MockMethod('_GetCurrentTime', [], [])
        self.requester._Wait()
        self.assertEqual(1, time.sleep._call_count)

        # current time = 1.04, last_request_start_time = 1.01, time_to_wait = 0.1.
        # Wait should call _GetCurrentTime once to calculate time to wait.
        # Wait should call sleep with time of 0.1 - (1.04 - 1.01) = 0.07.
        last_request_time = 1.01
        self.requester._last_request_start_time = last_request_time
        current_time = 1.04
        self.requester._GetCurrentTime = MockMethod('_GetCurrentTime', [()],
                                                    [current_time])
        time.sleep = MockMethod(
            'sleep', [(time_to_wait - (current_time - last_request_time), )],
            [None])
        self.requester._Wait()
        self.assertEqual(1, time.sleep._call_count)
        self.assertEqual(1, self.requester._GetCurrentTime._call_count)

        # Current time = 1.12, last_request_time = 1.01, time_to_wait = 0.1.
        # Wait should call _GetCurrentTime once to calculate time to wait.
        # Wait should not call sleep since too much time has elapsed.
        current_time = 1.12
        self.requester._GetCurrentTime = MockMethod('_GetCurrentTime', [()],
                                                    [current_time])
        time.sleep = MockMethod('sleep', [], [])
        self.requester._Wait()
        self.assertEqual(0, time.sleep._call_count)
        self.assertEqual(1, self.requester._GetCurrentTime._call_count)
Exemple #8
0
  def testWaitsBetweenRequestInvocations(self):
    """Tests that Requester waits between requests."""
    generator = MockGenerator()
    logger = log.Logger()
    self.requester = requester.Requester(generator, logger, None, 0.1,
                                         seconds=10)
    # Send two requests.
    self.requester._ShouldSendMoreRequests = MockMethod(
        '_ShouldSendMoreRequests', [(), (), ()], [True, True, False])
    # Wait should get called twice.
    self.requester._Wait = MockMethod('_Wait', [(), ()], [None, None])
    request_payload = generator.request.SerializeToString()
    self.requester._sender = MockMethod(
        '_sender',
        [(request_payload,), (request_payload,)],
        [(1, 'return value'), (1, 'return value')])

    self.requester.Start()
    self.assertEqual(2, self.requester._Wait._call_count)
Exemple #9
0
    print(name)
    time.sleep(randint(3, 10))

    #    print(url)
    resp = req(url)
    if resp:
        with open('images/' + name, 'wb') as outfile:
            outfile.write(resp.content)
    else:
        print('BADBADBADBAD')


data = pandas.read_excel('Pots.xlsx')
data.reset_index(inplace=True, drop=True)

req = requester.Requester()
bad = []
allList = []

for i, row in data.iterrows():
    pics = row['Pics'].replace('[', '').replace(']',
                                                '').replace("'",
                                                            '').split(', ')
    pot, *_ = row['Text'].split(',')
    for pic in pics:
        *_, picName, _ = pic.split('/')
        picName = '%s__%s.jpg' % (pot, picName)

        allList.append([picName, pic])

dSet = {x.replace('.jpg', '') for x in os.listdir('images')}
Exemple #10
0
		"headers" : [],
		"cache_dir" : "/var/mod_dsrc_cache",
		"azure_containers" : [
							{
								"account_name" : "",
								"account_key" : "",
								"container" : "",
								"compression" : True,
								"decompression" : False ,
							}
						 ],
	
}
##################################################


queue = Queue.Queue()
_obj_watcher = watcher.Watcher(_config_watcher, True)

_obj_request = requester.Requester(_config_request, True)
while True:
	_obj_watcher.set_test_loop(1)
	_obj_watcher.worker_init(queue)
	_obj_request.set_test_loop(7)
	_obj_request.worker_init(queue)
	queue.join()