class DownloaderTest(unittest.TestCase):
default_settings = {
'CONCURRENT_REQUESTS': 2,
'CONCURRENT_REQUESTS_PER_DOMAIN': 1,
'DOWNLOAD_DELAY': 0,
'RANDOMIZE_DOWNLOAD_DELAY': False}
def setUp(self):
self.clock = Clock()
self.request_queue = MemoryQueue()
self.response_queue = ResponseQueue()
self.dwn = Downloader(Settings(self.default_settings), self.request_queue,
self.response_queue,
download_handler=MockDownloaderHandler(Settings()),
clock=self.clock)
self.handler = self.dwn.download_handler
def _update_dwn(self, **kwargs):
'''Update downloader with the new settings.
'''
new_settings = self.default_settings.copy()
new_settings.update(**kwargs)
self.dwn.processing.cancel()
self.dwn = Downloader(Settings(new_settings), self.request_queue, self.response_queue,
download_handler=MockDownloaderHandler(Settings()),
clock=self.clock)
self.handler = self.dwn.download_handler
def test_concurrency(self):
# standard situation
self._update_dwn()
self.assertEqual(self.dwn.total_concurrency, 2)
self.assertEqual(self.dwn.domain_concurrency, 1)
self.assertTrue(self.dwn.use_domain_specific)
# delay set
self._update_dwn(CONCURRENT_REQUESTS=10, CONCURRENT_REQUESTS_PER_DOMAIN=5,
DOWNLOAD_DELAY=5)
self.assertEqual(self.dwn.total_concurrency, 1)
self.assertEqual(self.dwn.domain_concurrency, 1)
self.assertFalse(self.dwn.use_domain_specific)
# domain concurrency is 0
self._update_dwn(CONCURRENT_REQUESTS=10, CONCURRENT_REQUESTS_PER_DOMAIN=0)
self.assertEqual(self.dwn.total_concurrency, 10)
self.assertEqual(self.dwn.domain_concurrency, 10)
self.assertFalse(self.dwn.use_domain_specific)
# domain concurrency is too big
self._update_dwn(CONCURRENT_REQUESTS=5, CONCURRENT_REQUESTS_PER_DOMAIN=10)
self.assertEqual(self.dwn.total_concurrency, 5)
self.assertEqual(self.dwn.domain_concurrency, 5)
self.assertFalse(self.dwn.use_domain_specific)
self._update_dwn(CONCURRENT_REQUESTS=5, CONCURRENT_REQUESTS_PER_DOMAIN=5)
self.assertFalse(self.dwn.use_domain_specific)
def test_get_slot(self):
key, slot = self.dwn._get_slot(Request('http://www.github.com/'))
self.assertEqual(key, 'www.github.com')
key2, slot2 = self.dwn._get_slot(Request('http://www.github.com/hello/world#bla'))
self.assertEqual(key2, 'www.github.com')
self.assertIs(slot2, slot)
key3, slot3 = self.dwn._get_slot(Request('http://sites.github.com/'))
self.assertEqual(key3, 'sites.github.com')
self.assertIsNot(slot3, slot)
self.assertEqual(len(self.dwn.slots), 2)
# don't use domain specific slots
self.dwn.use_domain_specific = False
key, slot = self.dwn._get_slot(Request('http://www.github.com/'))
self.assertEqual(key, '')
key2, slot2 = self.dwn._get_slot(Request('http://sites.github.com/'))
self.assertIs(slot2, slot)
def test_basic(self):
# create 5 requests with slot ids: a, b, a, a, c
requests = [get_request(id)[0] for id in 'abaac']
map(lambda r: self.request_queue.push(r), requests)
self.assertEqual(self.dwn.free_slots, 2)
self.assertTrue(self.dwn.is_idle())
# start downloading first two requests
self.clock.advance(0)
self.assertEqual(self.dwn.free_slots, 0)
self.assertFalse(self.dwn.is_idle())
# no more requests are scheduled, until download is finished
self.clock.advance(20)
self.assertEqual(len(self.request_queue), 3)
# download the first request
self.handler.call(requests[0], Response('hello'))
self.assertEqual(self.dwn.free_slots, 1) # slot is immediately available
# result is also available
result = self.response_queue.peek()
self.assertIs(result.request, requests[0])
self.assertEqual(result.url, 'hello')
# enqueue third request
self.clock.advance(Downloader.QUEUE_CHECK_FREQUENCY)
self.assertEqual(self.dwn.free_slots, 0)
# download second request
self.handler.call(requests[1], Response(''))
# enqueue fourth request
self.clock.advance(Downloader.QUEUE_CHECK_FREQUENCY)
self.assertEqual(self.dwn.free_slots, 0)
# fourth request should not begin download, until 3rd request is done
self.assertRaises(KeyError, self.handler.call, requests[3], Response(''))
# finish
self.handler.call(requests[2], Response(''))
self.handler.call(requests[3], Response(''))
self.clock.advance(Downloader.QUEUE_CHECK_FREQUENCY)
self.handler.call(requests[4], Response(''))
# final checks
self.clock.pump([1] * 10)
self.assertEqual(len(self.response_queue), 5)
self.assertTrue(self.dwn.is_idle())
def test_close(self):
req1 = get_request('a')[0]
req2 = get_request('b')[0]
self.request_queue.push(req1)
self.clock.advance(20)
self.request_queue.push(req2)
# test basic attributes, before and after closing
self.assertTrue(self.dwn.running)
self.assertTrue(self.dwn.processing.is_scheduled())
self.dwn.close()
self.assertFalse(self.dwn.running)
self.assertFalse(self.dwn.processing.is_scheduled())
self.clock.advance(20)
self.assertEqual(len(self.request_queue), 1) # request 2 remains unqueued
# downloader behavior after closing
self.assertEqual(len(self.response_queue), 0)
self.handler.call(req1, Response(''))
self.assertEqual(len(self.response_queue), 0)
def test_fail(self):
self._update_dwn(CONCURRENT_REQUESTS=3, CONCURRENT_REQUESTS_PER_DOMAIN=2)
requests = [get_request(id)[0] for id in 'aab']
map(lambda r: self.request_queue.push(r), requests)
# enqueue requests
self.clock.advance(0)
# fail 1st request
err = ValueError('my bad')
self.handler.fail(requests[0], err)
self.assertEqual(self.dwn.free_slots, 1)
fail = self.response_queue.pop()
self.assertIs(fail.request, requests[0])
self.assertIs(fail.value, err)
# fail 3rd request
self.handler.fail(requests[2], err)
fail = self.response_queue.pop()
self.assertIs(fail.request, requests[2])
self.assertIs(fail.value, err)
# succeed 2nd request
self.handler.call(requests[1], Response('nice!', request=requests[1]))
resp = self.response_queue.pop()
self.assertIs(resp.request, requests[1])
self.assertEqual(resp.url, 'nice!')
def test_clear_slots(self):
requests = [get_request(id)[0] for id in xrange(30)]
for r in requests:
self.request_queue.push(r)
self.clock.advance(Downloader.QUEUE_CHECK_FREQUENCY)
self.handler.call(r, Response(''))
self.assertLessEqual(len(self.dwn.slots), 2 * self.dwn.total_concurrency)
class DownloaderTest(unittest.TestCase):
default_settings = {
'CONCURRENT_REQUESTS': 2,
'CONCURRENT_REQUESTS_PER_DOMAIN': 1,
'DOWNLOAD_DELAY': 0,
'RANDOMIZE_DOWNLOAD_DELAY': False
}
def setUp(self):
self.clock = Clock()
self.request_queue = MemoryQueue()
self.response_queue = ResponseQueue()
self.dwn = Downloader(Settings(self.default_settings),
self.request_queue,
self.response_queue,
download_handler=MockDownloaderHandler(
Settings()),
clock=self.clock)
self.handler = self.dwn.download_handler
def _update_dwn(self, **kwargs):
'''Update downloader with the new settings.
'''
new_settings = self.default_settings.copy()
new_settings.update(**kwargs)
self.dwn.processing.cancel()
self.dwn = Downloader(Settings(new_settings),
self.request_queue,
self.response_queue,
download_handler=MockDownloaderHandler(
Settings()),
clock=self.clock)
self.handler = self.dwn.download_handler
def test_concurrency(self):
# standard situation
self._update_dwn()
self.assertEqual(self.dwn.total_concurrency, 2)
self.assertEqual(self.dwn.domain_concurrency, 1)
self.assertTrue(self.dwn.use_domain_specific)
# delay set
self._update_dwn(CONCURRENT_REQUESTS=10,
CONCURRENT_REQUESTS_PER_DOMAIN=5,
DOWNLOAD_DELAY=3.14)
self.assertEqual(self.dwn.download_delay, 3.14)
self.assertEqual(self.dwn.total_concurrency, 1)
self.assertEqual(self.dwn.domain_concurrency, 1)
self.assertFalse(self.dwn.use_domain_specific)
# domain concurrency is 0
self._update_dwn(CONCURRENT_REQUESTS=10,
CONCURRENT_REQUESTS_PER_DOMAIN=0)
self.assertEqual(self.dwn.total_concurrency, 10)
self.assertEqual(self.dwn.domain_concurrency, 10)
self.assertFalse(self.dwn.use_domain_specific)
# domain concurrency is too big
self._update_dwn(CONCURRENT_REQUESTS=5,
CONCURRENT_REQUESTS_PER_DOMAIN=10)
self.assertEqual(self.dwn.total_concurrency, 5)
self.assertEqual(self.dwn.domain_concurrency, 5)
self.assertFalse(self.dwn.use_domain_specific)
self._update_dwn(CONCURRENT_REQUESTS=5,
CONCURRENT_REQUESTS_PER_DOMAIN=5)
self.assertFalse(self.dwn.use_domain_specific)
def test_get_slot(self):
key, slot = self.dwn._get_slot(Request('http://www.github.com/'))
self.assertEqual(key, 'www.github.com')
key2, slot2 = self.dwn._get_slot(
Request('http://www.github.com/hello/world#bla'))
self.assertEqual(key2, 'www.github.com')
self.assertIs(slot2, slot)
key3, slot3 = self.dwn._get_slot(Request('http://sites.github.com/'))
self.assertEqual(key3, 'sites.github.com')
self.assertIsNot(slot3, slot)
self.assertEqual(len(self.dwn.slots), 2)
# don't use domain specific slots
self.dwn.use_domain_specific = False
key, slot = self.dwn._get_slot(Request('http://www.github.com/'))
self.assertEqual(key, '')
key2, slot2 = self.dwn._get_slot(Request('http://sites.github.com/'))
self.assertIs(slot2, slot)
def test_basic(self):
# create 5 requests with slot ids: a, b, a, a, c
requests = [get_request(id)[0] for id in 'abaac']
map(lambda r: self.request_queue.push(r), requests)
self.assertEqual(self.dwn.free_slots, 2)
self.assertTrue(self.dwn.is_idle())
# start downloading first two requests
self.clock.advance(0)
self.assertEqual(self.dwn.free_slots, 0)
self.assertFalse(self.dwn.is_idle())
# no more requests are scheduled, until download is finished
self.clock.advance(20)
self.assertEqual(len(self.request_queue), 3)
# download the first request
self.handler.call(requests[0], Response('hello'))
self.assertEqual(self.dwn.free_slots,
1) # slot is immediately available
# result is also available
result = self.response_queue.peek()
self.assertIs(result.request, requests[0])
self.assertEqual(result.url, 'hello')
# enqueue third request
self.clock.advance(Downloader.QUEUE_CHECK_FREQUENCY)
self.assertEqual(self.dwn.free_slots, 0)
# download second request
self.handler.call(requests[1], Response(''))
# enqueue fourth request
self.clock.advance(Downloader.QUEUE_CHECK_FREQUENCY)
self.assertEqual(self.dwn.free_slots, 0)
# fourth request should not begin download, until 3rd request is done
self.assertRaises(KeyError, self.handler.call, requests[3],
Response(''))
# finish
self.handler.call(requests[2], Response(''))
self.handler.call(requests[3], Response(''))
self.clock.advance(Downloader.QUEUE_CHECK_FREQUENCY)
self.handler.call(requests[4], Response(''))
# final checks
self.clock.pump([1] * 10)
self.assertEqual(len(self.response_queue), 5)
self.assertTrue(self.dwn.is_idle())
def test_close(self):
req1 = get_request('a')[0]
req2 = get_request('b')[0]
self.request_queue.push(req1)
self.clock.advance(20)
self.request_queue.push(req2)
# test basic attributes, before and after closing
self.assertTrue(self.dwn.running)
self.assertTrue(self.dwn.processing.is_scheduled())
self.dwn.close()
self.assertFalse(self.dwn.running)
self.assertFalse(self.dwn.processing.is_scheduled())
self.clock.advance(20)
self.assertEqual(len(self.request_queue),
1) # request 2 remains unqueued
# downloader behavior after closing
self.assertEqual(len(self.response_queue), 0)
self.handler.call(req1, Response(''))
self.assertEqual(len(self.response_queue), 0)
def test_fail(self):
self._update_dwn(CONCURRENT_REQUESTS=3,
CONCURRENT_REQUESTS_PER_DOMAIN=2)
requests = [get_request(id)[0] for id in 'aab']
map(lambda r: self.request_queue.push(r), requests)
# enqueue requests
self.clock.advance(0)
# fail 1st request
err = ValueError('my bad')
self.handler.fail(requests[0], err)
self.assertEqual(self.dwn.free_slots, 1)
fail = self.response_queue.pop()
self.assertIs(fail.request, requests[0])
self.assertIs(fail.value, err)
# fail 3rd request
self.handler.fail(requests[2], err)
fail = self.response_queue.pop()
self.assertIs(fail.request, requests[2])
self.assertIs(fail.value, err)
# succeed 2nd request
self.handler.call(requests[1], Response('nice!', request=requests[1]))
resp = self.response_queue.pop()
self.assertIs(resp.request, requests[1])
self.assertEqual(resp.url, 'nice!')
def test_clear_slots(self):
requests = [get_request(id)[0] for id in xrange(30)]
for r in requests:
self.request_queue.push(r)
self.clock.advance(Downloader.QUEUE_CHECK_FREQUENCY)
self.handler.call(r, Response(''))
self.assertLessEqual(len(self.dwn.slots),
2 * self.dwn.total_concurrency)
