class MiddlewareManagerTest(unittest.TestCase):
def setUp(self):
self.lw = LogWrapper()
self.lw.setUp(level=log.DEBUG)
def tearDown(self):
self.lw.tearDown()
def test_init(self):
mw = TestMiddlewareManager(get_engine())
active = [x.__class__ for x in mw.middlewares]
self.assertListEqual(active, [M1, M2])
logged = self.lw.get_first_line()
self.assertEqual(logged, "[crawlmi] DEBUG: Disabled <class 'crawlmi.tests.test_middleware_manager.MOff'>:")
def test_init2(self):
mw = TestMiddlewareManager(get_engine(), mw_classes=[M1, M2, MOff])
active = [x.__class__ for x in mw.middlewares]
self.assertListEqual(active, [M1, M2])
def test_enabled_setting(self):
mw = TestMiddlewareManager(get_engine(), mw_classes=[M1, M2, MOff])
active = [x.__class__ for x in mw.middlewares]
self.assertListEqual(active, [M1, M2])
self.assertEqual(mw.middlewares[0].enabled_setting, 'M1_ENABLED')
self.assertEqual(mw.middlewares[1].enabled_setting, 'M2_OFF')
mw = TestMiddlewareManager(get_engine(M1_ENABLED=False, M2_OFF=False),
mw_classes=[M1, M2, MOff])
active = [x.__class__ for x in mw.middlewares]
self.assertListEqual(active, [])
class MiddlewareManagerTest(unittest.TestCase):
def setUp(self):
self.lw = LogWrapper()
self.lw.setUp(level=log.DEBUG)
def tearDown(self):
self.lw.tearDown()
def test_init(self):
mw = TestMiddlewareManager(get_engine())
active = [x.__class__ for x in mw.middlewares]
self.assertListEqual(active, [M1, M2])
logged = self.lw.get_first_line()
self.assertEqual(
logged,
"[crawlmi] DEBUG: Disabled <class 'crawlmi.tests.test_middleware_manager.MOff'>:"
)
def test_init2(self):
mw = TestMiddlewareManager(get_engine(), mw_classes=[M1, M2, MOff])
active = [x.__class__ for x in mw.middlewares]
self.assertListEqual(active, [M1, M2])
def test_enabled_setting(self):
mw = TestMiddlewareManager(get_engine(), mw_classes=[M1, M2, MOff])
active = [x.__class__ for x in mw.middlewares]
self.assertListEqual(active, [M1, M2])
self.assertEqual(mw.middlewares[0].enabled_setting, 'M1_ENABLED')
self.assertEqual(mw.middlewares[1].enabled_setting, 'M2_OFF')
mw = TestMiddlewareManager(get_engine(M1_ENABLED=False, M2_OFF=False),
mw_classes=[M1, M2, MOff])
active = [x.__class__ for x in mw.middlewares]
self.assertListEqual(active, [])
def setUp(self):
self.clock = Clock()
self.engine = get_engine(LOG_STATS_INTERVAL=30)
self.engine.signals = SignalManager(self.engine)
self.ls = LogStats(self.engine, clock=self.clock)
self.lw = LogWrapper()
self.lw.setUp()
class LogStatsTest(unittest.TestCase):
def setUp(self):
self.clock = Clock()
self.engine = get_engine(LOG_STATS_INTERVAL=30)
self.engine.signals = SignalManager(self.engine)
self.ls = LogStats(self.engine, clock=self.clock)
self.lw = LogWrapper()
self.lw.setUp()
def tearDown(self):
self.lw.tearDown()
def test_config(self):
self.assertRaises(NotConfigured, LogStats, get_engine(LOG_STATS_INTERVAL=0))
def test_basic(self):
# engine is stopped
self.clock.advance(60)
self.assertEqual(self.lw.get_first_line(), '')
# start the engine
self.engine.signals.send(signals.engine_started)
self.clock.advance(29)
self.assertEqual(self.lw.get_first_line(), '')
self.clock.advance(1)
self.assertEqual(self.lw.get_first_line(), '[crawlmi] INFO: Crawled 0 pages (at 0 pages/min).')
# download some responses
self.engine.signals.send(signals.response_downloaded, response=Response(url=''))
self.engine.signals.send(signals.response_downloaded, response=Response(url=''))
self.engine.signals.send(signals.response_received, response=Response(url=''))
self.clock.advance(30)
self.assertEqual(self.lw.get_first_line(), '[crawlmi] INFO: Crawled 2 pages (at 4 pages/min).')
# stop the engine
self.engine.signals.send(signals.engine_stopped)
self.clock.advance(60)
self.assertEqual(self.lw.get_first_line(), '')
def setUp(self):
orig_spiders_dir = os.path.join(module_dir, 'test_spiders')
self.tmpdir = self.mktemp()
os.mkdir(self.tmpdir)
self.spiders_dir = os.path.join(self.tmpdir, 'test_spiders_xxx')
shutil.copytree(orig_spiders_dir, self.spiders_dir)
sys.path.append(self.tmpdir)
settings = {'SPIDER_MODULES': ['test_spiders_xxx']}
self.manager = SpiderManager(Settings(settings))
self.lw = LogWrapper()
self.lw.setUp()
def setUp(self):
self.clock = Clock()
self.engine = get_engine(LOG_STATS_INTERVAL=30)
self.engine.signals = SignalManager(self.engine)
self.ls = LogStats(self.engine, clock=self.clock)
self.lw = LogWrapper()
self.lw.setUp()
class SpiderManagerTest(unittest.TestCase):
def setUp(self):
orig_spiders_dir = os.path.join(module_dir, 'test_spiders')
self.tmpdir = self.mktemp()
os.mkdir(self.tmpdir)
self.spiders_dir = os.path.join(self.tmpdir, 'test_spiders_xxx')
shutil.copytree(orig_spiders_dir, self.spiders_dir)
sys.path.append(self.tmpdir)
settings = {'SPIDER_MODULES': ['test_spiders_xxx']}
self.manager = SpiderManager(Settings(settings))
self.lw = LogWrapper()
self.lw.setUp()
def tearDown(self):
sys.path.remove(self.tmpdir)
self.lw.tearDown()
def test_get_spiders(self):
self.assertSetEqual(
set(self.manager.get_spiders()),
set(['spider1', 'spider2', 'spider3']))
def test_create_by_name(self):
spider1 = self.manager.create_spider_by_name('spider1')
self.assertEqual(spider1.__class__.__name__, 'Spider1')
spider2 = self.manager.create_spider_by_name('spider2', {'p1': 1, 'p2': 2})
self.assertEqual(spider2.__class__.__name__, 'Spider2')
self.assertEqual(spider2.p1, 1)
self.assertEqual(spider2.p2, 2)
def test_get_spiders_by_url(self):
self.assertEqual(self.manager.get_spiders_by_url('http://crawlmi1.org/test'), ['spider1'])
self.assertEqual(self.manager.get_spiders_by_url('http://crawlmi2.org/test'), ['spider2'])
self.assertEqual(set(self.manager.get_spiders_by_url('http://crawlmi3.org/test')), set(['spider1', 'spider2']))
self.assertEqual(self.manager.get_spiders_by_url('http://crawlmi999.org/test'), [])
self.assertEqual(self.manager.get_spiders_by_url('http://spider3.com'), ['spider3'])
def test_create_spiders_by_url(self):
spider = self.manager.create_spider_by_url('http://crawlmi1.org/test')
self.assertEqual(spider.__class__.__name__, 'Spider1')
spider = self.manager.create_spider_by_url('http://crawlmi3.org/test')
self.assertIsNone(spider)
self.assertTrue(self.lw.get_first_line().startswith('[crawlmi] ERROR: More than one spider can handle:'))
spider = self.manager.create_spider_by_url('http://crawlmi999.org/test')
self.assertIsNone(spider)
self.assertTrue(self.lw.get_first_line().startswith('[crawlmi] ERROR: Unable to find spider that handles:'))
def test_load_spider_module(self):
settings = {'SPIDER_MODULES': ['crawlmi.tests.test_spider_manager.test_spiders.spider1']}
self.manager = SpiderManager(Settings(settings))
self.assertEqual(len(self.manager._spiders), 1)
def test_load_base_spider(self):
settings = {'SPIDER_MODULES': ['crawlmi.tests.test_spider_manager.test_spiders.spider0']}
self.manager = SpiderManager(Settings(settings))
self.assertEqual(len(self.manager._spiders), 0)
class LogStatsTest(unittest.TestCase):
def setUp(self):
self.clock = Clock()
self.engine = get_engine(LOG_STATS_INTERVAL=30)
self.engine.signals = SignalManager(self.engine)
self.ls = LogStats(self.engine, clock=self.clock)
self.lw = LogWrapper()
self.lw.setUp()
def tearDown(self):
self.lw.tearDown()
def test_config(self):
self.assertRaises(NotConfigured, LogStats,
get_engine(LOG_STATS_INTERVAL=0))
def test_basic(self):
# engine is stopped
self.clock.advance(60)
self.assertEqual(self.lw.get_first_line(), '')
# start the engine
self.engine.signals.send(signals.engine_started)
self.clock.advance(29)
self.assertEqual(self.lw.get_first_line(), '')
self.clock.advance(1)
self.assertEqual(self.lw.get_first_line(),
'[crawlmi] INFO: Crawled 0 pages (at 0 pages/min).')
# download some responses
self.engine.signals.send(signals.response_downloaded,
response=Response(url=''))
self.engine.signals.send(signals.response_downloaded,
response=Response(url=''))
self.engine.signals.send(signals.response_received,
response=Response(url=''))
self.clock.advance(30)
self.assertEqual(self.lw.get_first_line(),
'[crawlmi] INFO: Crawled 2 pages (at 4 pages/min).')
# stop the engine
self.engine.signals.send(signals.engine_stopped)
self.clock.advance(60)
self.assertEqual(self.lw.get_first_line(), '')
def setUp(self):
self.lw = LogWrapper()
self.lw.setUp(level=log.DEBUG)
def setUp(self):
self.lw = LogWrapper()
self.lw.setUp(log.INFO, 'utf-8')
class CrawlmiFileLogObserverTest(unittest.TestCase):
def setUp(self):
self.lw = LogWrapper()
self.lw.setUp(log.INFO, 'utf-8')
def tearDown(self):
self.flushLoggedErrors()
self.lw.tearDown()
def test_msg_basic(self):
log.msg('Hello')
self.assertEqual(self.lw.get_first_line(), '[crawlmi] INFO: Hello')
def test_format(self):
log.msg(format='%(hi)s', hi='Hello')
self.assertEqual(self.lw.get_first_line(), '[crawlmi] INFO: Hello')
def test_msg_level1(self):
log.msg('Hello', level=log.WARNING)
self.assertEqual(self.lw.get_first_line(), '[crawlmi] WARNING: Hello')
def test_msg_level2(self):
log.msg('Hello', log.WARNING)
self.assertEqual(self.lw.get_first_line(), '[crawlmi] WARNING: Hello')
def test_msg_wrong_level(self):
log.msg('Hello', level=9999)
self.assertEqual(self.lw.get_first_line(), '[crawlmi] NOLEVEL: Hello')
def test_msg_encoding(self):
log.msg(u'Price: \xa3100')
self.assertEqual(self.lw.get_first_line(), '[crawlmi] INFO: Price: \xc2\xa3100')
def test_msg_ignore_level(self):
log.msg('Hello', level=log.DEBUG)
log.msg('World', level=log.INFO)
self.assertEqual(self.lw.get_first_line(), '[crawlmi] INFO: World')
def test_msg_ignore_system(self):
txlog.msg('Hello')
self.assertEqual(self.lw.get_first_line(), '')
def test_msg_ignore_system_err(self):
txlog.err('Hello')
self.assertEqual(self.lw.get_first_line(), '[-] ERROR: \'Hello\'')
def test_err_noargs(self):
try:
a = 1 / 0
except:
log.err()
logged = self.lw.get_logged()
self.assertIn('Traceback', logged)
self.assertIn('ZeroDivisionError', logged)
def test_err_why(self):
log.err(TypeError('bad type'), 'Wrong type')
logged = self.lw.get_logged(clear=False)
self.assertEqual(self.lw.get_first_line(), '[crawlmi] ERROR: Wrong type')
self.assertIn('TypeError', logged)
self.assertIn('bad type', logged)
def test_error_outside_crawlmi(self):
'''Crawlmi logger should still print outside errors'''
txlog.err(TypeError('bad type'), 'Wrong type')
logged = self.lw.get_logged(clear=False)
self.assertEqual(self.lw.get_first_line(), '[-] ERROR: Wrong type')
self.assertIn('TypeError', logged)
self.assertIn('bad type', logged)
# this test fails in twisted trial observer, not in crawlmi observer
# def test_err_why_encoding(self):
# log.err(TypeError('bad type'), u'\xa3')
# self.assertEqual(self.lw.get_first_line(), '[crawlmi] ERROR: \xc2\xa3')
def test_err_exc(self):
log.err(TypeError('bad type'))
logged = self.lw.get_logged()
self.assertIn('Unhandled Error', logged)
self.assertIn('TypeError', logged)
self.assertIn('bad type', logged)
def test_err_failure(self):
log.err(failure.Failure(TypeError('bad type')))
logged = self.lw.get_logged()
self.assertIn('Unhandled Error', logged)
self.assertIn('TypeError', logged)
self.assertIn('bad type', logged)
def setUp(self):
self.lw = LogWrapper()
self.lw.setUp(level=log.DEBUG)
def setUp(self):
self.lw = LogWrapper()
self.lw.setUp()
def setUp(self):
self.lw = LogWrapper()
self.lw.setUp()
class StatsTest(unittest.TestCase):
def setUp(self):
self.lw = LogWrapper()
self.lw.setUp()
def tearDown(self):
self.lw.tearDown()
def test_memory_stats(self):
stats = MemoryStats(get_engine(STATS_DUMP=True))
self.assertEqual(stats.get_stats(), {})
self.assertEqual(stats.get_value('anything'), None)
self.assertEqual(stats.get_value('anything', 'default'), 'default')
stats.set_value('test', 'value')
self.assertEqual(stats.get_stats(), {'test': 'value'})
stats.set_value('test2', 23)
self.assertEqual(stats.get_stats(), {'test': 'value', 'test2': 23})
self.assertEqual(stats.get_value('test2'), 23)
stats.inc_value('test2')
self.assertEqual(stats.get_value('test2'), 24)
stats.inc_value('test2', 6)
self.assertEqual(stats.get_value('test2'), 30)
stats.max_value('test2', 6)
self.assertEqual(stats.get_value('test2'), 30)
stats.max_value('test2', 40)
self.assertEqual(stats.get_value('test2'), 40)
stats.max_value('test3', 1)
self.assertEqual(stats.get_value('test3'), 1)
stats.min_value('test2', 60)
self.assertEqual(stats.get_value('test2'), 40)
stats.min_value('test2', 35)
self.assertEqual(stats.get_value('test2'), 35)
stats.min_value('test4', 7)
self.assertEqual(stats.get_value('test4'), 7)
stats.add_value('stats', 3)
stats.add_value('stats', 2, 2.0)
statistics = stats.get_value('stats')
self.assertTrue(eq(statistics.average, 7.0/3.0), statistics.average)
self.assertRaises(RuntimeError, stats.add_value, 'test4', 1)
stats.add_sample('samples', 3, 'hello')
stats.add_sample('samples', 2, 'world')
samples = stats.get_value('samples')
self.assertEqual(len(samples), 2)
self.assertListEqual(samples.samples, [(3, 'hello'), (2, 'world')])
self.assertRaises(RuntimeError, stats.add_value, 'test4', 5, '!')
stats.dump_stats()
logged = self.lw.get_first_line(clear=False)
self.assertTrue(logged.startswith('[crawlmi] INFO: Dumping crawlmi stats:'))
logged = self.lw.get_logged()
self.assertIn('test', logged)
self.assertIn('test2', logged)
self.assertIn('test3', logged)
self.assertIn('test3', logged)
self.assertIn('stats', logged)
self.assertIn('samples', logged)
def test_dummy_stats(self):
stats = DummyStats(get_engine())
self.assertEqual(stats.get_stats(), {})
self.assertEqual(stats.get_value('anything'), None)
self.assertEqual(stats.get_value('anything', 'default'), 'default')
stats.set_value('test', 'value')
stats.inc_value('v1')
stats.max_value('v2', 100)
stats.min_value('v3', 100)
stats.set_value('test', 'value')
stats.add_value('stats', 100)
stats.add_value('stats', 100, 12)
stats.add_sample('samples', 3, 'hello')
self.assertEqual(stats.get_stats(), {})
stats.dump_stats()