def test_blocking_middleware(self):
settings = Settings()
settings.BACKEND = 'tests.mocks.components.FakeBackend'
settings.MIDDLEWARES = [
'tests.mocks.components.FakeMiddleware',
'tests.mocks.components.FakeMiddlewareModifySeeds',
'tests.mocks.components.FakeMiddlewareBlocking',
'tests.mocks.components.FakeMiddlewareModifyResponse',
'tests.mocks.components.FakeMiddlewareModifyLinks'
]
settings.CANONICAL_SOLVER = 'tests.mocks.components.FakeCanonicalSolver'
fm = FrontierManager.from_settings(settings)
fm.add_seeds([r1, r2, r3])
response = Response(r1.url, request=r1)
fm.page_crawled(response)
fm.links_extracted(r1, links=[r2])
fm.request_error(r3, 'error')
#the seeds, responses, links and errors have not reached the backend.
assert [len(list) for list in fm.backend.lists] == [0] * 4
#the 3 seeds reach the first three middlewares.
assert [len(fm.middlewares[i].seeds) for i in range(3)] == [3] * 3
#the error, response and link reached the first three middlewares.
assert [[len(list) for list in fm.middlewares[i].lists[1:]]
for i in range(3)] == [[1] * 3] * 3
#the values do not reach the bottom 2 middlewares and the canonical solver.
assert [[len(list) for list in fm.middlewares[i].lists]
for i in range(3, 5)] == [[0] * 4] * 2
assert [len(list) for list in fm.canonicalsolver.lists] == [0] * 4
def test_blocking_middleware(self):
settings = Settings()
settings.BACKEND = 'tests.mocks.components.FakeBackend'
settings.MIDDLEWARES = [
'frontera.contrib.middlewares.domain.DomainMiddleware',
'frontera.contrib.middlewares.fingerprint.UrlFingerprintMiddleware',
'tests.mocks.components.FakeMiddleware',
'tests.mocks.components.FakeMiddlewareModifySeeds',
'tests.mocks.components.FakeMiddlewareBlocking',
'tests.mocks.components.FakeMiddlewareModifyResponse',
'tests.mocks.components.FakeMiddlewareModifyLinks'
]
settings.CANONICAL_SOLVER = 'tests.mocks.components.FakeCanonicalSolver'
settings.STRATEGY = 'tests.mocks.components.CrawlingStrategy'
fm = LocalFrontierManager.from_settings(settings)
SEEDS_FILE.seek(0)
fm.add_seeds(SEEDS_FILE)
response = Response(r1.url, request=r1)
fm.page_crawled(response)
fm.links_extracted(r1, links=[r2])
fm.request_error(r3, 'error')
#the seeds, responses, links and errors have not reached the backend.
assert [len(list) for list in fm.backend.lists] == [0] * 4
#the 3 seeds reach the first three middlewares.
assert [len(fm.middlewares[i].requests)
for i in range(2, 5)] == [3] * 3
#the error, response and link reached the first three middlewares.
assert [[len(list) for list in fm.middlewares[i].lists[1:]]
for i in range(2, 5)] == [[1] * 3] * 3
#the values do not reach the bottom 2 middlewares and the canonical solver.
assert [[len(list) for list in fm.middlewares[i].lists]
for i in range(5, 7)] == [[0] * 4] * 2
assert [len(list) for list in fm.canonicalsolver.lists] == [0] * 4
def test_blocking_middleware(self):
settings = Settings()
settings.BACKEND = 'tests.mocks.components.FakeBackend'
settings.MIDDLEWARES = ['tests.mocks.components.FakeMiddleware',
'tests.mocks.components.FakeMiddlewareModifySeeds',
'tests.mocks.components.FakeMiddlewareBlocking',
'tests.mocks.components.FakeMiddlewareModifyResponse',
'tests.mocks.components.FakeMiddlewareModifyLinks']
settings.CANONICAL_SOLVER = 'tests.mocks.components.FakeCanonicalSolver'
fm = FrontierManager.from_settings(settings)
fm.add_seeds([r1, r2, r3])
response = Response(r1.url, request=r1)
fm.page_crawled(response)
fm.links_extracted(r1, links=[r2])
fm.request_error(r3, 'error')
#the seeds, responses, links and errors have not reached the backend.
assert [len(list) for list in fm.backend.lists] == [0]*4
#the 3 seeds reach the first three middlewares.
assert [len(fm.middlewares[i].seeds) for i in range(3)] == [3]*3
#the error, response and link reached the first three middlewares.
assert [[len(list) for list in fm.middlewares[i].lists[1:]] for i in range(3)] == [[1]*3]*3
#the values do not reach the bottom 2 middlewares and the canonical solver.
assert [[len(list) for list in fm.middlewares[i].lists] for i in range(3, 5)] == [[0]*4]*2
assert [len(list) for list in fm.canonicalsolver.lists] == [0]*4
def test_blocking_middleware(self):
settings = Settings()
settings.BACKEND = 'tests.mocks.components.FakeBackend'
settings.MIDDLEWARES = ['frontera.contrib.middlewares.domain.DomainMiddleware',
'frontera.contrib.middlewares.fingerprint.UrlFingerprintMiddleware',
'tests.mocks.components.FakeMiddleware',
'tests.mocks.components.FakeMiddlewareModifySeeds',
'tests.mocks.components.FakeMiddlewareBlocking',
'tests.mocks.components.FakeMiddlewareModifyResponse',
'tests.mocks.components.FakeMiddlewareModifyLinks']
settings.CANONICAL_SOLVER = 'tests.mocks.components.FakeCanonicalSolver'
settings.STRATEGY = 'tests.mocks.components.CrawlingStrategy'
fm = LocalFrontierManager.from_settings(settings)
SEEDS_FILE.seek(0)
fm.add_seeds(SEEDS_FILE)
response = Response(r1.url, request=r1)
fm.page_crawled(response)
fm.links_extracted(r1, links=[r2])
fm.request_error(r3, 'error')
#the seeds, responses, links and errors have not reached the backend.
assert [len(list) for list in fm.backend.lists] == [0]*4
#the 3 seeds reach the first three middlewares.
assert [len(fm.middlewares[i].requests) for i in range(2, 5)] == [3]*3
#the error, response and link reached the first three middlewares.
assert [[len(list) for list in fm.middlewares[i].lists[1:]] for i in range(2, 5)] == [[1]*3]*3
#the values do not reach the bottom 2 middlewares and the canonical solver.
assert [[len(list) for list in fm.middlewares[i].lists] for i in range(5, 7)] == [[0]*4]*2
assert [len(list) for list in fm.canonicalsolver.lists] == [0]*4