2011-09-11 08:27:05 -03:00
|
|
|
import os
|
|
|
|
from datetime import datetime
|
|
|
|
from twisted.trial import unittest
|
|
|
|
|
2015-04-21 13:48:28 -03:00
|
|
|
from scrapy.extensions.spiderstate import SpiderState
|
2015-05-09 04:20:09 -03:00
|
|
|
from scrapy.spiders import Spider
|
2016-01-26 16:33:24 +01:00
|
|
|
from scrapy.exceptions import NotConfigured
|
|
|
|
from scrapy.utils.test import get_crawler
|
2011-09-11 08:27:05 -03:00
|
|
|
|
|
|
|
|
|
|
|
class SpiderStateTest(unittest.TestCase):
|
|
|
|
|
|
|
|
def test_store_load(self):
|
|
|
|
jobdir = self.mktemp()
|
|
|
|
os.mkdir(jobdir)
|
2013-12-28 00:47:32 +06:00
|
|
|
spider = Spider(name='default')
|
2011-09-11 08:27:05 -03:00
|
|
|
dt = datetime.now()
|
|
|
|
|
|
|
|
ss = SpiderState(jobdir)
|
|
|
|
ss.spider_opened(spider)
|
|
|
|
spider.state['one'] = 1
|
|
|
|
spider.state['dt'] = dt
|
|
|
|
ss.spider_closed(spider)
|
|
|
|
|
2013-12-28 00:47:32 +06:00
|
|
|
spider2 = Spider(name='default')
|
2011-09-11 08:27:05 -03:00
|
|
|
ss2 = SpiderState(jobdir)
|
|
|
|
ss2.spider_opened(spider2)
|
|
|
|
self.assertEqual(spider.state, {'one': 1, 'dt': dt})
|
|
|
|
ss2.spider_closed(spider2)
|
2011-09-23 12:56:44 -03:00
|
|
|
|
|
|
|
def test_state_attribute(self):
|
|
|
|
# state attribute must be present if jobdir is not set, to provide a
|
2013-12-28 00:47:32 +06:00
|
|
|
# consistent interface
|
|
|
|
spider = Spider(name='default')
|
2011-09-23 12:56:44 -03:00
|
|
|
ss = SpiderState()
|
|
|
|
ss.spider_opened(spider)
|
|
|
|
self.assertEqual(spider.state, {})
|
|
|
|
ss.spider_closed(spider)
|
2016-01-26 16:33:24 +01:00
|
|
|
|
|
|
|
def test_not_configured(self):
|
|
|
|
crawler = get_crawler(Spider)
|
|
|
|
self.assertRaises(NotConfigured, SpiderState.from_crawler, crawler)
|