def test_spider_breadth(self):
# Assert BREADTH cross-domain preference.
v = web.Spider(links=["http://www.clips.ua.ac.be/"], delay=10)
while len(v.visited) < 4:
v.crawl(throttle=0.1, cached=False, method=web.BREADTH)
self.assertTrue(v.history.keys()[0] != v.history.keys()[1])
self.assertTrue(v.history.keys()[0] != v.history.keys()[2])
self.assertTrue(v.history.keys()[1] != v.history.keys()[2])
print "pattern.web.Spider.crawl(method=BREADTH)"
def test_spider_crawl(self):
# Assert domain filter.
v = web.Spider(links=["http://www.clips.ua.ac.be/"], domains=["clips.ua.ac.be"], delay=0.5)
while len(v.visited) < 4:
v.crawl(throttle=0.1, cached=False)
for url in v.visited:
self.assertTrue("clips.ua.ac.be" in url)
self.assertTrue(len(v.history) == 1)
print "pattern.web.Spider.crawl()"
def test_spider_delay(self):
# Assert delay for several crawls to a single domain.
v = web.Spider(links=["http://www.clips.ua.ac.be/"], domains=["clips.ua.ac.be"], delay=1.0)
v.crawl()
t = time.time()
while not v.crawl(throttle=0.1, cached=False):
pass
t = time.time() - t
self.assertTrue(t > 1.0)
print "pattern.web.Spider.delay"