def test_delay(self):
spider = FollowAllSpider()
yield docrawl(spider, {"DOWNLOAD_DELAY": 1})
t = spider.times[0]
for t2 in spider.times[1:]:
self.assertTrue(t2-t > 0.45, "download delay too small: %s" % (t2-t))
t = t2
def test_closespider_pagecount(self):
spider = FollowAllSpider()
close_on = 5
yield docrawl(spider, {'CLOSESPIDER_PAGECOUNT': close_on})
reason = spider.meta['close_reason']
self.assertEqual(reason, 'closespider_pagecount')
pagecount = spider.crawler.stats.get_value('response_received_count')
self.assertTrue(pagecount >= close_on)
def _test_delay(self, delay, randomize):
settings = {"DOWNLOAD_DELAY": delay, 'RANDOMIZE_DOWNLOAD_DELAY': randomize}
spider = FollowAllSpider(maxlatency=delay * 2)
yield docrawl(spider, settings)
t = spider.times
totaltime = t[-1] - t[0]
avgd = totaltime / (len(t) - 1)
tolerance = 0.6 if randomize else 0.2
self.assertTrue(avgd > delay * (1 - tolerance),
"download delay too small: %s" % avgd)
def test_closespider_timeout(self):
spider = FollowAllSpider(total=1000000)
close_on = 0.1
yield docrawl(spider, {'CLOSESPIDER_TIMEOUT': close_on})
reason = spider.meta['close_reason']
self.assertEqual(reason, 'closespider_timeout')
stats = spider.crawler.stats
start = stats.get_value('start_time')
stop = stats.get_value('finish_time')
diff = stop - start
total_seconds = diff.seconds + diff.microseconds
self.assertTrue(total_seconds >= close_on)
def test_follow_all(self):
spider = FollowAllSpider()
yield docrawl(spider)
self.assertEqual(len(spider.urls_visited), 11) # 10 + start_url
def run(self, args, opts):
with MockServer():
spider = FollowAllSpider(total=100000)
self.crawler.crawl(spider)
self.crawler.start()
def run(self, args, opts):
with MockServer():
spider = FollowAllSpider(total=100000)
crawler = self.crawler_process.create_crawler()
crawler.crawl(spider)
self.crawler_process.start()
