2009-09-07 11:14:43 -03:00
|
|
|
from unittest import TestCase
|
|
|
|
|
2023-01-25 14:25:15 -06:00
|
|
|
from scrapy.http import Request, Response
|
2015-04-21 13:07:24 -03:00
|
|
|
from scrapy.spidermiddlewares.depth import DepthMiddleware
|
2015-05-09 04:20:09 -03:00
|
|
|
from scrapy.spiders import Spider
|
2015-04-23 13:07:48 -03:00
|
|
|
from scrapy.statscollectors import StatsCollector
|
2012-03-06 08:59:20 -02:00
|
|
|
from scrapy.utils.test import get_crawler
|
2009-09-07 11:14:43 -03:00
|
|
|
|
|
|
|
|
|
|
|
class TestDepthMiddleware(TestCase):
|
|
|
|
def setUp(self):
|
2014-07-31 04:12:12 -03:00
|
|
|
crawler = get_crawler(Spider)
|
|
|
|
self.spider = crawler._create_spider("scrapytest.org")
|
2009-09-07 11:14:43 -03:00
|
|
|
|
2014-07-31 04:12:12 -03:00
|
|
|
self.stats = StatsCollector(crawler)
|
2010-09-06 11:04:27 -03:00
|
|
|
self.stats.open_spider(self.spider)
|
2009-09-07 11:14:43 -03:00
|
|
|
|
2011-05-18 11:04:48 -03:00
|
|
|
self.mw = DepthMiddleware(1, self.stats, True)
|
2009-09-07 11:14:43 -03:00
|
|
|
|
|
|
|
def test_process_spider_output(self):
|
|
|
|
req = Request("http://scrapytest.org")
|
|
|
|
resp = Response("http://scrapytest.org")
|
|
|
|
resp.request = req
|
|
|
|
result = [Request("http://scrapytest.org")]
|
|
|
|
|
|
|
|
out = list(self.mw.process_spider_output(resp, result, self.spider))
|
2017-08-07 18:29:36 +03:00
|
|
|
self.assertEqual(out, result)
|
2009-09-07 11:14:43 -03:00
|
|
|
|
2010-09-06 11:04:27 -03:00
|
|
|
rdc = self.stats.get_value("request_depth_count/1", spider=self.spider)
|
2017-08-07 18:29:36 +03:00
|
|
|
self.assertEqual(rdc, 1)
|
2009-09-07 11:14:43 -03:00
|
|
|
|
|
|
|
req.meta["depth"] = 1
|
|
|
|
|
|
|
|
out2 = list(self.mw.process_spider_output(resp, result, self.spider))
|
2017-08-07 18:29:36 +03:00
|
|
|
self.assertEqual(out2, [])
|
2009-09-07 11:14:43 -03:00
|
|
|
|
2010-09-06 11:04:27 -03:00
|
|
|
rdm = self.stats.get_value("request_depth_max", spider=self.spider)
|
2017-08-07 18:29:36 +03:00
|
|
|
self.assertEqual(rdm, 1)
|
2015-04-21 13:07:24 -03:00
|
|
|
|
2009-09-07 11:14:43 -03:00
|
|
|
def tearDown(self):
|
2010-09-06 11:04:27 -03:00
|
|
|
self.stats.close_spider(self.spider, "")
|