def test_deprecation_warning(self):
"""
Make sure deprecation warnings are logged whenever BaseItem is used,
either instantiated or in an isinstance check
"""
with catch_warnings(record=True) as warnings:
BaseItem()
self.assertEqual(len(warnings), 1)
self.assertEqual(warnings[0].category, ScrapyDeprecationWarning)
with catch_warnings(record=True) as warnings:
class SubclassedBaseItem(BaseItem):
pass
SubclassedBaseItem()
self.assertEqual(len(warnings), 1)
self.assertEqual(warnings[0].category, ScrapyDeprecationWarning)
with catch_warnings(record=True) as warnings:
self.assertFalse(isinstance("foo", BaseItem))
self.assertEqual(len(warnings), 1)
self.assertEqual(warnings[0].category, ScrapyDeprecationWarning)
with catch_warnings(record=True) as warnings:
self.assertTrue(isinstance(BaseItem(), BaseItem))
self.assertEqual(len(warnings), 1)
self.assertEqual(warnings[0].category, ScrapyDeprecationWarning)
def test_process_spider_output_stats_legacy(self):
# testing the subclass not handling stats works at runtime
# (i.e. that trying to update stats does not trigger exception)
class LegacyDeltaFetchSubClass(self.mwcls):
def __init__(self, dir, reset=False, *args, **kwargs):
super(LegacyDeltaFetchSubClass, self).__init__(dir=dir,
reset=reset)
self.something = True
self._create_test_db()
mw = LegacyDeltaFetchSubClass(self.temp_dir, reset=False)
mw.spider_opened(self.spider)
response = mock.Mock()
response.request = Request('http://url',
meta={'deltafetch_key': 'key'})
result = []
self.assertEqual(
list(mw.process_spider_output(response, result, self.spider)), [])
self.assertEqual(self.stats.get_stats(), {})
result = [
Request('http://url', meta={'deltafetch_key': 'key'}),
Request('http://url1', meta={'deltafetch_key': 'test_key_1'})
]
# stats should not be updated
self.assertEqual(
list(mw.process_spider_output(response, result, self.spider)),
[result[0]])
self.assertEqual(self.stats.get_value('deltafetch/skipped'), None)
result = [BaseItem(), "not a base item"]
self.assertEqual(
list(mw.process_spider_output(response, result, self.spider)),
result)
self.assertEqual(self.stats.get_value('deltafetch/stored'), None)
def test_process_spider_output(self):
self._create_test_db()
mw = self.mwcls(self.temp_dir, reset=False)
mw.spider_opened(self.spider)
response = mock.Mock()
response.request = Request('http://url',
meta={'deltafetch_key': 'key'})
result = []
self.assertEqual(
list(mw.process_spider_output(response, result, self.spider)), [])
result = [
Request('http://url', meta={'deltafetch_key': 'key1'}),
Request('http://url1', meta={'deltafetch_key': 'test_key_1'})
]
self.assertEqual(
list(mw.process_spider_output(response, result, self.spider)),
[result[0]])
result = [BaseItem(), "not a base item"]
self.assertEqual(
list(mw.process_spider_output(response, result, self.spider)),
result)
self.assertEqual(mw.db.keys(), ['test_key_1', 'key', 'test_key_2'])
assert mw.db['key']
def test_isinstance_check(self):
class SubclassedBaseItem(BaseItem):
pass
class SubclassedItem(Item):
pass
self.assertTrue(isinstance(BaseItem(), BaseItem))
self.assertTrue(isinstance(SubclassedBaseItem(), BaseItem))
self.assertTrue(isinstance(Item(), BaseItem))
self.assertTrue(isinstance(SubclassedItem(), BaseItem))
# make sure internal checks using private _BaseItem class succeed
self.assertTrue(isinstance(BaseItem(), _BaseItem))
self.assertTrue(isinstance(SubclassedBaseItem(), _BaseItem))
self.assertTrue(isinstance(Item(), _BaseItem))
self.assertTrue(isinstance(SubclassedItem(), _BaseItem))
def test_iterate_spider_output(self):
i = BaseItem()
r = Request('http://scrapytest.org')
o = object()
self.assertEqual(list(iterate_spider_output(i)), [i])
self.assertEqual(list(iterate_spider_output(r)), [r])
self.assertEqual(list(iterate_spider_output(o)), [o])
self.assertEqual(list(iterate_spider_output([r, i, o])), [r, i, o])
def test_isinstance_check(self):
class SubclassedBaseItem(BaseItem):
pass
class SubclassedItem(Item):
pass
with catch_warnings():
filterwarnings("ignore", category=ScrapyDeprecationWarning)
self.assertTrue(isinstance(BaseItem(), BaseItem))
self.assertTrue(isinstance(SubclassedBaseItem(), BaseItem))
self.assertTrue(isinstance(Item(), BaseItem))
self.assertTrue(isinstance(SubclassedItem(), BaseItem))
# make sure internal checks using private _BaseItem class succeed
self.assertTrue(isinstance(BaseItem(), _BaseItem))
self.assertTrue(isinstance(SubclassedBaseItem(), _BaseItem))
self.assertTrue(isinstance(Item(), _BaseItem))
self.assertTrue(isinstance(SubclassedItem(), _BaseItem))
def test_process_spider_output_stats(self):
self._create_test_db()
mw = self.mwcls(self.temp_dir, reset=False, stats=self.stats)
mw.spider_opened(self.spider)
response = mock.Mock()
response.request = Request('http://url',
meta={'deltafetch_key': 'key'})
result = []
self.assertEqual(list(mw.process_spider_output(
response, result, self.spider)), [])
self.assertEqual(self.stats.get_stats(), {})
result = [
Request('http://url', meta={'deltafetch_key': 'key'}),
Request('http://url1', meta={'deltafetch_key': 'test_key_1'})
]
self.assertEqual(list(mw.process_spider_output(
response, result, self.spider)), [result[0]])
self.assertEqual(self.stats.get_value('deltafetch/skipped'), 1)
result = [BaseItem(), "not a base item"]
self.assertEqual(list(mw.process_spider_output(
response, result, self.spider)), result)
self.assertEqual(self.stats.get_value('deltafetch/stored'), 1)
def test_process_spider_output(self):
self._create_test_db()
mw = self.mwcls(self.temp_dir, reset=False, stats=self.stats)
mw.spider_opened(self.spider)
response = mock.Mock()
response.request = Request('http://url',
meta={'deltafetch_key': 'key'})
result = []
self.assertEqual(list(mw.process_spider_output(
response, result, self.spider)), [])
result = [
# same URL but with new key --> it should be processed
Request('http://url', meta={'deltafetch_key': 'key1'}),
# 'test_key_1' is already in the test db --> it should be skipped
Request('http://url1', meta={'deltafetch_key': 'test_key_1'})
]
# so only the 1 request should go through
self.assertEqual(list(mw.process_spider_output(
response, result, self.spider)), [result[0]])
# the skipped "http://url1" should be counted in stats
self.assertEqual(self.stats.get_stats(), {'deltafetch/skipped': 1})
# b'key' should not be in the db yet as no item was collected yet
self.assertEqual(set(mw.db.keys()),
set([b'test_key_1',
b'test_key_2']))
# if the spider returns items, the request's key is added in db
result = [BaseItem(), "not a base item"]
self.assertEqual(list(mw.process_spider_output(
response, result, self.spider)), result)
self.assertEqual(set(mw.db.keys()),
set([b'key',
b'test_key_1',
b'test_key_2']))
assert mw.db[b'key']
