def test_request_counter(self):
import threading
reset_request_counter()
g = build_grab()
g.go(self.server.get_url())
self.assertEqual(g.request_counter, 1)
g.go(self.server.get_url())
self.assertEqual(g.request_counter, 2)
def func():
g = build_grab()
g.go(self.server.get_url())
# Make 10 requests in concurrent threads
threads = []
for x in six.moves.range(10):
th = threading.Thread(target=func)
threads.append(th)
th.start()
for th in threads:
th.join()
g.go(self.server.get_url())
self.assertEqual(g.request_counter, 13)
def test_request_counter(self):
import threading
reset_request_counter()
g = build_grab()
g.go(self.server.get_url())
self.assertEqual(g.request_counter, 1)
g.go(self.server.get_url())
self.assertEqual(g.request_counter, 2)
def func():
g = build_grab()
g.go(self.server.get_url())
# Make 10 requests in concurrent threads
threads = []
for x in six.moves.range(10):
th = threading.Thread(target=func)
threads.append(th)
th.start()
for th in threads:
th.join()
g.go(self.server.get_url())
self.assertEqual(g.request_counter, 13)
def test_log_option(self):
clear_directory(TMP_DIR)
reset_request_counter()
log_file_path = os.path.join(TMP_DIR, 'log.html')
g = build_grab()
g.setup(log_file=log_file_path)
self.server.response['get.data'] = 'omsk'
self.assertEqual(os.listdir(TMP_DIR), [])
g.go(self.server.get_url())
self.assertEqual(os.listdir(TMP_DIR), ['log.html'])
self.assertEqual(open(log_file_path).read(), 'omsk')
def test_log_option(self):
clear_directory(TMP_DIR)
reset_request_counter()
log_file_path = os.path.join(TMP_DIR, 'log.html')
g = build_grab()
g.setup(log_file=log_file_path)
self.server.response['get.data'] = 'omsk'
self.assertEqual(os.listdir(TMP_DIR), [])
g.go(self.server.get_url())
self.assertEqual(os.listdir(TMP_DIR), ['log.html'])
self.assertEqual(open(log_file_path).read(), 'omsk')
def test_log_option(self):
clear_directory(TMP_DIR)
reset_request_counter()
log_file_path = os.path.join(TMP_DIR, 'log.html')
g = build_grab()
g.setup(log_file=log_file_path)
SERVER.RESPONSE['get'] = 'omsk'
self.assertEqual(os.listdir(TMP_DIR), [])
g.go(SERVER.BASE_URL)
self.assertEqual(os.listdir(TMP_DIR), ['log.html'])
self.assertEqual(open(log_file_path).read(), 'omsk')
def test_log_option(self):
clear_directory(TMP_DIR)
reset_request_counter()
log_file_path = os.path.join(TMP_DIR, 'log.html')
g = build_grab()
g.setup(log_file=log_file_path)
SERVER.RESPONSE['get'] = 'omsk'
self.assertEqual(os.listdir(TMP_DIR), [])
g.go(SERVER.BASE_URL)
self.assertEqual(os.listdir(TMP_DIR), ['log.html'])
self.assertEqual(open(log_file_path).read(), 'omsk')
def test_log_option(self):
with temp_dir() as tmp_dir:
reset_request_counter()
log_file_path = os.path.join(tmp_dir, 'log.html')
g = build_grab()
g.setup(log_file=log_file_path)
self.server.response['get.data'] = 'omsk'
self.assertEqual(os.listdir(tmp_dir), [])
g.go(self.server.get_url())
self.assertEqual(os.listdir(tmp_dir), ['log.html'])
self.assertEqual(open(log_file_path).read(), 'omsk')
def test_log_option(self):
with temp_dir() as tmp_dir:
reset_request_counter()
log_file_path = os.path.join(tmp_dir, 'log.html')
g = build_grab()
g.setup(log_file=log_file_path)
self.server.response['get.data'] = 'omsk'
self.assertEqual(os.listdir(tmp_dir), [])
g.go(self.server.get_url())
self.assertEqual(os.listdir(tmp_dir), ['log.html'])
self.assertEqual(open(log_file_path).read(), 'omsk')
def test_log_dir_request_content_is_empty(self):
clear_directory(TMP_DIR)
reset_request_counter()
g = build_grab()
g.setup(log_dir=TMP_DIR)
g.setup(headers={'X-Name': 'spider'}, post='xxxPost')
self.assertEqual(os.listdir(TMP_DIR), [])
g.go(self.server.get_url())
self.assertEqual(sorted(os.listdir(TMP_DIR)), ['01.html', '01.log'])
log_file_content = open(os.path.join(TMP_DIR, '01.log')).read()
self.assertFalse('X-Name' in log_file_content)
self.assertFalse('xxxPost' in log_file_content)
def test_log_dir_request_content_is_empty(self):
clear_directory(TMP_DIR)
reset_request_counter()
g = build_grab()
g.setup(log_dir=TMP_DIR)
g.setup(headers={'X-Name': 'spider'}, post='xxxPost')
self.assertEqual(os.listdir(TMP_DIR), [])
g.go(self.server.get_url())
self.assertEqual(sorted(os.listdir(TMP_DIR)), ['01.html', '01.log'])
log_file_content = open(os.path.join(TMP_DIR, '01.log')).read()
self.assertFalse('X-Name' in log_file_content)
self.assertFalse('xxxPost' in log_file_content)
def test_log_dir_response_content(self):
clear_directory(TMP_DIR)
reset_request_counter()
g = build_grab()
g.setup(log_dir=TMP_DIR)
SERVER.RESPONSE['get'] = 'omsk'
SERVER.RESPONSE['headers'] = [('X-Engine', 'PHP')]
self.assertEqual(os.listdir(TMP_DIR), [])
g.go(SERVER.BASE_URL)
self.assertEqual(sorted(os.listdir(TMP_DIR)), ['01.html', '01.log'])
log_file_content = open(os.path.join(TMP_DIR, '01.log')).read()
self.assertTrue('X-Engine' in log_file_content)
def test_log_dir_response_content(self):
clear_directory(TMP_DIR)
reset_request_counter()
g = build_grab()
g.setup(log_dir=TMP_DIR)
SERVER.RESPONSE['get'] = 'omsk'
SERVER.RESPONSE['headers'] = [('X-Engine', 'PHP')]
self.assertEqual(os.listdir(TMP_DIR), [])
g.go(SERVER.BASE_URL)
self.assertEqual(sorted(os.listdir(TMP_DIR)), ['01.html', '01.log'])
log_file_content = open(os.path.join(TMP_DIR, '01.log')).read()
self.assertTrue('X-Engine' in log_file_content)
def test_log_dir_request_content_is_empty(self):
with temp_dir() as tmp_dir:
reset_request_counter()
g = build_grab()
g.setup(log_dir=tmp_dir)
g.setup(headers={'X-Name': 'spider'}, post='xxxPost')
self.assertEqual(os.listdir(tmp_dir), [])
g.go(self.server.get_url())
self.assertEqual(sorted(os.listdir(tmp_dir)), ['01.html', '01.log'])
log_file_content = open(os.path.join(tmp_dir, '01.log')).read()
self.assertFalse('X-Name' in log_file_content)
self.assertFalse('xxxPost' in log_file_content)
def test_log_dir_response_content(self):
with temp_dir() as tmp_dir:
reset_request_counter()
g = build_grab()
g.setup(log_dir=tmp_dir)
self.server.response['get.data'] = 'omsk'
self.server.response['headers'] = [('X-Engine', 'PHP')]
self.assertEqual(os.listdir(tmp_dir), [])
g.go(self.server.get_url())
self.assertEqual(sorted(os.listdir(tmp_dir)), ['01.html', '01.log'])
log_file_content = open(os.path.join(tmp_dir, '01.log')).read()
self.assertTrue('x-engine' in log_file_content.lower())
def test_log_dir_request_content_headers_and_post(self):
clear_directory(TMP_DIR)
reset_request_counter()
g = build_grab()
g.setup(log_dir=TMP_DIR, debug=True)
g.setup(headers={'X-Name': 'spider'}, post={'xxx': 'Post'})
self.assertEqual(os.listdir(TMP_DIR), [])
g.go(SERVER.BASE_URL)
self.assertEqual(sorted(os.listdir(TMP_DIR)), ['01.html', '01.log'])
log_file_content = open(os.path.join(TMP_DIR, '01.log')).read()
self.assertTrue('X-Name' in log_file_content)
self.assertTrue('xxx=Post' in log_file_content)
def test_log_dir_request_content_headers_and_post(self):
clear_directory(TMP_DIR)
reset_request_counter()
g = build_grab()
g.setup(log_dir=TMP_DIR, debug=True)
g.setup(headers={'X-Name': 'spider'}, post={'xxx': 'Post'})
self.assertEqual(os.listdir(TMP_DIR), [])
g.go(SERVER.BASE_URL)
self.assertEqual(sorted(os.listdir(TMP_DIR)), ['01.html', '01.log'])
log_file_content = open(os.path.join(TMP_DIR, '01.log')).read()
self.assertTrue('X-Name' in log_file_content)
self.assertTrue('xxx=Post' in log_file_content)
def test_log_dir_option(self):
clear_directory(TMP_DIR)
reset_request_counter()
g = build_grab()
g.setup(log_dir=TMP_DIR)
SERVER.RESPONSE_ONCE['get'] = 'omsk1'
SERVER.RESPONSE['get'] = 'omsk2'
self.assertEqual(os.listdir(TMP_DIR), [])
g.go(SERVER.BASE_URL)
g.go(SERVER.BASE_URL)
self.assertEqual(sorted(os.listdir(TMP_DIR)), ['01.html', '01.log', '02.html', '02.log'])
self.assertEqual(open(os.path.join(TMP_DIR, '01.html')).read(), 'omsk1')
self.assertEqual(open(os.path.join(TMP_DIR, '02.html')).read(), 'omsk2')
def test_log_dir_option(self):
clear_directory(TMP_DIR)
reset_request_counter()
g = build_grab()
g.setup(log_dir=TMP_DIR)
SERVER.RESPONSE_ONCE['get'] = 'omsk1'
SERVER.RESPONSE['get'] = 'omsk2'
self.assertEqual(os.listdir(TMP_DIR), [])
g.go(SERVER.BASE_URL)
g.go(SERVER.BASE_URL)
self.assertEqual(sorted(os.listdir(TMP_DIR)), ['01.html', '01.log', '02.html', '02.log'])
self.assertEqual(open(os.path.join(TMP_DIR, '01.html')).read(), 'omsk1')
self.assertEqual(open(os.path.join(TMP_DIR, '02.html')).read(), 'omsk2')
def test_log_dir_request_content_headers_and_post(self):
with temp_dir() as tmp_dir:
reset_request_counter()
g = build_grab()
g.setup(log_dir=tmp_dir, debug=True)
g.setup(headers={'X-Name': 'spider'}, post={'xxx': 'Post'})
self.assertEqual(os.listdir(tmp_dir), [])
g.go(self.server.get_url())
self.assertEqual(sorted(os.listdir(tmp_dir)), ['01.html', '01.log'])
log_file_content = open(os.path.join(tmp_dir, '01.log')).read()
#if not 'x-name' in log_file_content.lower():
# print('CONTENT OF 01.log:')
# print(log_file_content)
self.assertTrue('x-name' in log_file_content.lower())
self.assertTrue('xxx=post' in log_file_content.lower())
def test_log_dir_option(self):
clear_directory(TMP_DIR)
reset_request_counter()
g = build_grab()
g.setup(log_dir=TMP_DIR)
self.server.response_once['get.data'] = 'omsk1'
self.server.response['get.data'] = 'omsk2'
self.assertEqual(os.listdir(TMP_DIR), [])
g.go(self.server.get_url())
g.go(self.server.get_url())
self.assertEqual(sorted(os.listdir(TMP_DIR)),
['01.html', '01.log', '02.html', '02.log'])
self.assertEqual(
open(os.path.join(TMP_DIR, '01.html')).read(), 'omsk1')
self.assertEqual(
open(os.path.join(TMP_DIR, '02.html')).read(), 'omsk2')
def test_log_dir_option(self):
with temp_dir() as tmp_dir:
reset_request_counter()
g = build_grab()
g.setup(log_dir=tmp_dir)
self.server.response_once['get.data'] = 'omsk1'
self.server.response['get.data'] = 'omsk2'
self.assertEqual(os.listdir(tmp_dir), [])
g.go(self.server.get_url())
g.go(self.server.get_url())
self.assertEqual(sorted(os.listdir(tmp_dir)),
['01.html', '01.log', '02.html', '02.log'])
self.assertEqual(open(os.path.join(tmp_dir, '01.html')).read(),
'omsk1')
self.assertEqual(open(os.path.join(tmp_dir, '02.html')).read(),
'omsk2')
def test_log_dir_option(self):
clear_directory(TMP_DIR)
reset_request_counter()
g = build_grab()
g.setup(log_dir=TMP_DIR)
self.server.response_once['get.data'] = 'omsk1'
self.server.response['get.data'] = 'omsk2'
self.assertEqual(os.listdir(TMP_DIR), [])
g.go(self.server.get_url())
g.go(self.server.get_url())
self.assertEqual(sorted(os.listdir(TMP_DIR)),
['01.html', '01.log', '02.html', '02.log'])
self.assertEqual(open(os.path.join(TMP_DIR, '01.html')).read(),
'omsk1')
self.assertEqual(open(os.path.join(TMP_DIR, '02.html')).read(),
'omsk2')
def test_log_dir_option(self):
with temp_dir() as tmp_dir:
reset_request_counter()
g = build_grab()
g.setup(log_dir=tmp_dir)
self.server.response_once['get.data'] = 'omsk1'
self.server.response['get.data'] = 'omsk2'
self.assertEqual(os.listdir(tmp_dir), [])
g.go(self.server.get_url())
g.go(self.server.get_url())
self.assertEqual(sorted(os.listdir(tmp_dir)),
['01.html', '01.log', '02.html', '02.log'])
self.assertEqual(
open(os.path.join(tmp_dir, '01.html')).read(), 'omsk1')
self.assertEqual(
open(os.path.join(tmp_dir, '02.html')).read(), 'omsk2')
def test_log_dir_response_network_error(self):
with temp_dir() as tmp_dir:
reset_request_counter()
g = build_grab()
g.setup(log_dir=tmp_dir, timeout=1, user_agent='Perl', debug=True)
self.server.response['get.data'] = 'omsk'
self.server.response['headers'] = [('X-Engine', 'PHP')]
self.server.response['sleep'] = 2
self.assertEqual(os.listdir(tmp_dir), [])
try:
g.go(self.server.get_url())
except GrabTimeoutError:
pass
self.assertEqual(sorted(os.listdir(tmp_dir)),
['01.html', '01.log'])
log_file_content = open(os.path.join(tmp_dir, '01.log')).read()
self.assertTrue('user-agent: perl' in log_file_content.lower())
def test_log_dir_response_content_thread(self):
with temp_dir() as tmp_dir:
reset_request_counter()
g = build_grab()
g.setup(log_dir=tmp_dir)
self.server.response['get.data'] = 'omsk'
self.server.response['headers'] = [('X-Engine', 'PHP')]
self.assertEqual(os.listdir(tmp_dir), [])
def func():
g.go(self.server.get_url())
th = threading.Thread(target=func)
th.start()
th.join()
files = os.listdir(tmp_dir)
self.assertEqual(2, len([x for x in files if '01-thread' in x]))
fname = [x for x in files if x.endswith('.log')][0]
log_file_content = open(os.path.join(tmp_dir, fname)).read()
self.assertTrue('x-engine' in log_file_content.lower())