from twisted.internet import defer from twisted.trial.unittest import TestCase from scrapy.utils.test import get_crawler from tests.spiders import FollowAllSpider, ItemSpider, ErrorSpider from tests.mockserver import MockServer class TestCloseSpider(TestCase): def setUp(self): self.mockserver = MockServer() self.mockserver.__enter__() def tearDown(self): self.mockserver.__exit__(None, None, None) @defer.inlineCallbacks def test_closespider_itemcount(self): close_on = 5 crawler = get_crawler(ItemSpider, {'CLOSESPIDER_ITEMCOUNT': close_on}) yield crawler.crawl() reason = crawler.spider.meta['close_reason'] self.assertEqual(reason, 'closespider_itemcount') itemcount = crawler.stats.get_value('item_scraped_count') self.assertTrue(itemcount >= close_on) @defer.inlineCallbacks def test_closespider_pagecount(self): close_on = 5 crawler = get_crawler(FollowAllSpider, {'CLOSESPIDER_PAGECOUNT': close_on}) yield crawler.crawl() reason = crawler.spider.meta['close_reason'] self.assertEqual(reason, 'closespider_pagecount') pagecount = crawler.stats.get_value('response_received_count') self.assertTrue(pagecount >= close_on) @defer.inlineCallbacks def test_closespider_errorcount(self): close_on = 5 crawler = get_crawler(ErrorSpider, {'CLOSESPIDER_ERRORCOUNT': close_on}) yield crawler.crawl(total=1000000) reason = crawler.spider.meta['close_reason'] self.assertEqual(reason, 'closespider_errorcount') key = 'spider_exceptions/{name}'\ .format(name=crawler.spider.exception_cls.__name__) errorcount = crawler.stats.get_value(key) self.assertTrue(errorcount >= close_on) @defer.inlineCallbacks def test_closespider_timeout(self): close_on = 0.1 crawler = get_crawler(FollowAllSpider, {'CLOSESPIDER_TIMEOUT': close_on}) yield crawler.crawl(total=1000000) reason = crawler.spider.meta['close_reason'] self.assertEqual(reason, 'closespider_timeout') stats = 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)