def test_queue_with_delay(self):
connection = Connection(host='hbase-docker', port=9090)
queue = HBaseQueue(connection,
1,
b'queue',
use_snappy=False,
drop=True)
r5 = r3.copy()
crawl_at = int(time()) + 1000
r5.meta[b'crawl_at'] = crawl_at
batch = [(r5.meta[b'fingerprint'], 0.5, r5, True)]
queue.schedule(batch)
with mock.patch('frontera.contrib.backends.hbase.time') as mocked_time:
mocked_time.return_value = time()
assert queue.get_next_requests(10,
0,
min_requests=3,
min_hosts=1,
max_requests_per_host=10) == []
mocked_time.return_value = crawl_at + 1
assert set([
r.url
for r in queue.get_next_requests(10,
0,
min_requests=3,
min_hosts=1,
max_requests_per_host=10)
]) == set([r5.url])
def test_queue(self):
connection = Connection(host='hbase-docker', port=9090)
queue = HBaseQueue(connection, 2, b'queue', True)
batch = [('10', 0.5, r1, True), ('11', 0.6, r2, True),
('12', 0.7, r3, True)]
queue.schedule(batch)
assert set([r.url for r in queue.get_next_requests(10, 0, min_requests=3, min_hosts=1,
max_requests_per_host=10)]) == set([r3.url])
assert set([r.url for r in queue.get_next_requests(10, 1, min_requests=3, min_hosts=1,
max_requests_per_host=10)]) == set([r1.url, r2.url])
def test_queue(self):
connection = Connection(host='hbase-docker', port=9090)
queue = HBaseQueue(connection, 2, b'queue', True)
batch = [('10', 0.5, r1, True), ('11', 0.6, r2, True),
('12', 0.7, r3, True)]
queue.schedule(batch)
assert set([r.url for r in queue.get_next_requests(10, 0, min_requests=3, min_hosts=1,
max_requests_per_host=10)]) == set([r3.url])
assert set([r.url for r in queue.get_next_requests(10, 1, min_requests=3, min_hosts=1,
max_requests_per_host=10)]) == set([r1.url, r2.url])
def test_queue_with_delay(self):
connection = Connection(host='hbase-docker', port=9090)
queue = HBaseQueue(connection, 1, b'queue', True)
r5 = r3.copy()
r5.meta[b'crawl_at'] = int(time()) + 1
batch = [(r5.meta[b'fingerprint'], 0.5, r5, True)]
queue.schedule(batch)
assert queue.get_next_requests(10, 0, min_requests=3, min_hosts=1,
max_requests_per_host=10) == []
sleep(1.5)
assert set([r.url for r in queue.get_next_requests(10, 0, min_requests=3, min_hosts=1,
max_requests_per_host=10)]) == set([r5.url])
def test_queue_with_delay(self):
connection = Connection(host='hbase-docker', port=9090)
queue = HBaseQueue(connection, 1, b'queue', use_snappy=False, drop=True)
r5 = r3.copy()
crawl_at = int(time()) + 1000
r5.meta[b'crawl_at'] = crawl_at
batch = [(r5.meta[b'fingerprint'], 0.5, r5, True)]
queue.schedule(batch)
with mock.patch('frontera.contrib.backends.hbase.time') as mocked_time:
mocked_time.return_value = time()
assert queue.get_next_requests(10, 0, min_requests=3, min_hosts=1,
max_requests_per_host=10) == []
mocked_time.return_value = crawl_at + 1
assert set([r.url for r in queue.get_next_requests(10, 0, min_requests=3, min_hosts=1,
max_requests_per_host=10)]) == set([r5.url])
def test_queue_with_delay(self):
connection = Connection(host='hbase-docker', port=9090)
queue = HBaseQueue(connection, 1, b'queue', True)
r5 = r3.copy()
r5.meta[b'crawl_at'] = int(time()) + 1
batch = [(r5.meta[b'fingerprint'], 0.5, r5, True)]
queue.schedule(batch)
assert queue.get_next_requests(10,
0,
min_requests=3,
min_hosts=1,
max_requests_per_host=10) == []
sleep(1.5)
assert set([
r.url for r in queue.get_next_requests(
10, 0, min_requests=3, min_hosts=1, max_requests_per_host=10)
]) == set([r5.url])
def test_queue_with_post_request(self):
connection = Connection(host='hbase-docker', port=9090)
queue = HBaseQueue(connection,
1,
b'queue',
drop=True,
use_snappy=False)
batch = [('10', 0.5, r1, True)]
queue.schedule(batch)
requests = queue.get_next_requests(10,
0,
min_requests=3,
min_hosts=1,
max_requests_per_host=10)
self.assertEqual(b'POST', requests[0].method)
self.assertEqual(data, requests[0].body)