2009-04-27 06:47:26 +00:00
|
|
|
from unittest import TestCase
|
|
|
|
|
2015-05-09 04:20:09 -03:00
|
|
|
from scrapy.spiders import Spider
|
2009-04-27 06:47:26 +00:00
|
|
|
from scrapy.http import Request
|
2015-04-20 21:23:05 -03:00
|
|
|
from scrapy.downloadermiddlewares.useragent import UserAgentMiddleware
|
2010-09-22 16:09:13 -03:00
|
|
|
from scrapy.utils.test import get_crawler
|
2009-04-27 06:47:26 +00:00
|
|
|
|
|
|
|
|
2009-08-24 07:57:00 -03:00
|
|
|
class UserAgentMiddlewareTest(TestCase):
|
2009-04-27 06:47:26 +00:00
|
|
|
|
2010-09-22 16:09:13 -03:00
|
|
|
def get_spider_and_mw(self, default_useragent):
|
2014-07-31 04:12:12 -03:00
|
|
|
crawler = get_crawler(Spider, {'USER_AGENT': default_useragent})
|
|
|
|
spider = crawler._create_spider('foo')
|
2012-08-21 17:27:45 -03:00
|
|
|
return spider, UserAgentMiddleware.from_crawler(crawler)
|
2009-04-27 06:47:26 +00:00
|
|
|
|
|
|
|
def test_default_agent(self):
|
2010-09-22 16:09:13 -03:00
|
|
|
spider, mw = self.get_spider_and_mw('default_useragent')
|
2009-04-27 06:47:26 +00:00
|
|
|
req = Request('http://scrapytest.org/')
|
2010-09-22 16:09:13 -03:00
|
|
|
assert mw.process_request(req, spider) is None
|
2015-09-01 04:11:33 +05:00
|
|
|
self.assertEquals(req.headers['User-Agent'], b'default_useragent')
|
2009-04-27 06:47:26 +00:00
|
|
|
|
2010-09-22 16:09:13 -03:00
|
|
|
def test_remove_agent(self):
|
|
|
|
# settings UESR_AGENT to None should remove the user agent
|
|
|
|
spider, mw = self.get_spider_and_mw('default_useragent')
|
2012-08-21 17:27:45 -03:00
|
|
|
spider.user_agent = None
|
2013-10-08 18:18:49 -02:00
|
|
|
mw.spider_opened(spider)
|
2009-04-27 16:56:16 +00:00
|
|
|
req = Request('http://scrapytest.org/')
|
2010-09-22 16:09:13 -03:00
|
|
|
assert mw.process_request(req, spider) is None
|
|
|
|
assert req.headers.get('User-Agent') is None
|
2009-04-27 16:56:16 +00:00
|
|
|
|
2009-04-27 06:47:26 +00:00
|
|
|
def test_spider_agent(self):
|
2010-09-22 16:09:13 -03:00
|
|
|
spider, mw = self.get_spider_and_mw('default_useragent')
|
2012-08-21 17:27:45 -03:00
|
|
|
spider.user_agent = 'spider_useragent'
|
2013-10-08 18:18:49 -02:00
|
|
|
mw.spider_opened(spider)
|
2009-04-27 06:47:26 +00:00
|
|
|
req = Request('http://scrapytest.org/')
|
2010-09-22 16:09:13 -03:00
|
|
|
assert mw.process_request(req, spider) is None
|
2015-09-01 04:11:33 +05:00
|
|
|
self.assertEquals(req.headers['User-Agent'], b'spider_useragent')
|
2009-04-27 06:47:26 +00:00
|
|
|
|
|
|
|
def test_header_agent(self):
|
2010-09-22 16:09:13 -03:00
|
|
|
spider, mw = self.get_spider_and_mw('default_useragent')
|
2012-08-21 17:27:45 -03:00
|
|
|
spider.user_agent = 'spider_useragent'
|
2013-10-08 18:18:49 -02:00
|
|
|
mw.spider_opened(spider)
|
2015-09-01 04:11:33 +05:00
|
|
|
req = Request('http://scrapytest.org/',
|
|
|
|
headers={'User-Agent': 'header_useragent'})
|
2010-09-22 16:09:13 -03:00
|
|
|
assert mw.process_request(req, spider) is None
|
2015-09-01 04:11:33 +05:00
|
|
|
self.assertEquals(req.headers['User-Agent'], b'header_useragent')
|
2009-04-27 06:47:26 +00:00
|
|
|
|
|
|
|
def test_no_agent(self):
|
2010-09-22 16:09:13 -03:00
|
|
|
spider, mw = self.get_spider_and_mw(None)
|
2012-08-21 17:27:45 -03:00
|
|
|
spider.user_agent = None
|
2013-10-08 18:18:49 -02:00
|
|
|
mw.spider_opened(spider)
|
2009-04-27 06:47:26 +00:00
|
|
|
req = Request('http://scrapytest.org/')
|
2010-09-22 16:09:13 -03:00
|
|
|
assert mw.process_request(req, spider) is None
|
2009-04-27 06:47:26 +00:00
|
|
|
assert 'User-Agent' not in req.headers
|