def process(self):
'''Process.
Coroutine.
'''
verdict = self._fetch_rule.check_ftp_request(
self._url_item.url_info, self._url_item.url_record)[0]
if not verdict:
self._url_item.skip()
return
request = Request(self._url_item.url_info.url) # TODO: dependency inject
if self._fetch_rule.ftp_login:
request.username, request.password = self._fetch_rule.ftp_login
dir_name, filename = self._url_item.url_info.split_path()
if self._processor.fetch_params.glob and frozenset(filename) & GLOB_CHARS:
directory_url = to_dir_path_url(request.url_info)
directory_request = copy.deepcopy(request)
directory_request.url = directory_url
request = directory_request
is_file = False
self._glob_pattern = urllib.parse.unquote(filename)
else:
is_file = yield From(self._prepare_request_file_vs_dir(request))
self._file_writer_session.process_request(request)
wait_time = yield From(self._fetch(request, is_file))
if wait_time:
_logger.debug('Sleeping {0}.'.format(wait_time))
yield From(trollius.sleep(wait_time))
def test_login_no_password_required(self):
client = Client()
file = io.BytesIO()
with client.session() as session:
request = Request(self.get_url('/example (copy).txt'))
request.username = '******'
yield From(session.fetch(request))
yield From(session.read_content(file))
def test_login_no_password_required(self):
client = Client()
file = io.BytesIO()
with client.session() as session:
request = Request(self.get_url('/example (copy).txt'))
request.username = '******'
yield From(session.fetch(request))
yield From(session.read_content(file))
def test_fetch_file_restart_not_supported(self):
client = Client()
file = io.BytesIO()
with client.session() as session:
request = Request(self.get_url('/example (copy).txt'))
request.set_continue(99999) # Magic value in the test server
response = yield From(session.fetch(request))
self.assertFalse(response.restart_value)
yield From(session.read_content(file))
self.assertEqual(
'The real treasure is in Smaug’s heart 💗.\n'.encode('utf-8'),
response.body.content())
def test_fetch_file_restart(self):
client = Client()
file = io.BytesIO()
with client.session() as session:
request = Request(self.get_url('/example (copy).txt'))
request.set_continue(10)
response = yield From(session.fetch(request))
self.assertEqual(10, response.restart_value)
yield From(session.read_content(file))
self.assertEqual(
'reasure is in Smaug’s heart 💗.\n'.encode('utf-8'),
response.body.content())
def test_fetch_file_restart_not_supported(self):
client = Client()
file = io.BytesIO()
with client.session() as session:
request = Request(self.get_url('/example (copy).txt'))
request.set_continue(99999) # Magic value in the test server
response = yield From(session.fetch(request))
self.assertFalse(response.restart_value)
yield From(session.read_content(file))
self.assertEqual(
'The real treasure is in Smaug’s heart 💗.\n'.encode('utf-8'),
response.body.content()
)
def test_fetch_file_restart(self):
client = Client()
file = io.BytesIO()
with client.session() as session:
request = Request(self.get_url('/example (copy).txt'))
request.set_continue(10)
response = yield From(session.fetch(request))
self.assertEqual(10, response.restart_value)
yield From(session.read_content(file))
self.assertEqual(
'reasure is in Smaug’s heart 💗.\n'.encode('utf-8'),
response.body.content()
)
def test_duration_timeout(self):
client = Client()
file = io.BytesIO()
with self.assertRaises(DurationTimeout), client.session() as session:
yield From(
session.fetch(Request(self.get_url('/hidden/sleep.txt'))))
yield From(session.read_content(file, duration_timeout=0.1))
def test_to_dict(self):
request = Request('ftp://foofle.com')
request_dict = request.to_dict()
self.assertEqual('ftp://foofle.com', request_dict['url'])
self.assertEqual('ftp', request_dict['protocol'])
response = Response()
response.request = request
response.reply = Reply(code=200, text='Success')
response_dict = response.to_dict()
self.assertEqual('ftp://foofle.com', response_dict['request']['url'])
self.assertEqual('ftp', response_dict['protocol'])
self.assertEqual(200, response_dict['reply']['code'])
self.assertEqual(200, response_dict['response_code'])
self.assertEqual('Success', response_dict['reply']['text'])
self.assertEqual('Success', response_dict['response_message'])
def test_to_dict(self):
request = Request('ftp://foofle.com')
request_dict = request.to_dict()
self.assertEqual('ftp://foofle.com', request_dict['url'])
self.assertEqual('ftp', request_dict['protocol'])
response = Response()
response.request = request
response.reply = Reply(code=200, text='Success')
response_dict = response.to_dict()
self.assertEqual('ftp://foofle.com', response_dict['request']['url'])
self.assertEqual('ftp', response_dict['protocol'])
self.assertEqual(200, response_dict['reply']['code'])
self.assertEqual(200, response_dict['response_code'])
self.assertEqual('Success', response_dict['reply']['text'])
self.assertEqual('Success', response_dict['response_message'])
def test_fetch_no_file(self):
client = Client()
file = io.BytesIO()
with client.session() as session:
try:
yield From(session.fetch(Request(self.get_url('/asdf.txt'))))
yield From(session.read_content(file))
except FTPServerError as error:
self.assertEqual(550, error.reply_code)
else:
self.fail() # pragma: no cover
def test_fetch_file(self):
client = Client()
file = io.BytesIO()
with client.session() as session:
response = yield From(
session.fetch(Request(self.get_url('/example (copy).txt'))))
yield From(session.read_content(file))
self.assertEqual(
'The real treasure is in Smaug’s heart 💗.\n'.encode('utf-8'),
response.body.content())
def test_fetch_listing(self):
client = Client()
file = io.BytesIO()
with client.session() as session:
response = yield From(
session.fetch_file_listing(Request(self.get_url('/'))))
yield From(session.read_listing_content(file))
print(response.body.content())
self.assertEqual(5, len(response.files))
self.assertEqual('junk', response.files[0].name)
self.assertEqual('example1', response.files[1].name)
self.assertEqual('example2💎', response.files[2].name)
self.assertEqual('example (copy).txt', response.files[3].name)
self.assertEqual('readme.txt', response.files[4].name)
def test_fetch_bad_pasv_addr(self):
client = Client()
file = io.BytesIO()
with client.session() as session:
original_func = session._log_in
@trollius.coroutine
def override_func():
yield From(original_func())
yield From(
session._control_stream.write_command(
Command('EVIL_BAD_PASV_ADDR')))
print('Evil awaits')
session._log_in = override_func
with self.assertRaises(ProtocolError):
yield From(
session.fetch(Request(
self.get_url('/example (copy).txt'))))
def test_warc_recorder_ftp(self):
file_prefix = 'asdf'
warc_filename = 'asdf.warc'
warc_recorder = WARCRecorder(file_prefix,
params=WARCRecorderParams(compress=False))
request = FTPRequest('ftp://example.com/example.txt')
request.address = ('0.0.0.0', 80)
response = FTPResponse()
response.reply = FTPReply(200, 'OK')
response.body = Body()
response.data_address = ('0.0.0.0', 12345)
with wpull.util.reset_file_offset(response.body):
response.body.write(b'KITTEH DOGE')
with warc_recorder.session() as session:
session.begin_control(request)
session.request_control_data(b'GIMMEH example.txt')
session.response_control_data(b'200 OK, no need to yell.')
session.pre_response(response)
session.response_data(b'KITTEH DOGE')
session.response(response)
session.end_control(response)
warc_recorder.close()
with open(warc_filename, 'rb') as in_file:
warc_file_content = in_file.read()
self.assertTrue(warc_file_content.startswith(b'WARC/1.0'))
self.assertIn(b'WARC-Type: warcinfo\r\n', warc_file_content)
self.assertIn(b'Content-Type: application/warc-fields',
warc_file_content)
self.assertIn(b'WARC-Date: ', warc_file_content)
self.assertIn(b'WARC-Record-ID: <urn:uuid:', warc_file_content)
self.assertIn(b'WARC-Block-Digest: sha1:', warc_file_content)
self.assertNotIn(b'WARC-Payload-Digest: sha1:', warc_file_content)
self.assertIn(b'WARC-Type: resource\r\n', warc_file_content)
self.assertIn(b'WARC-Target-URI: ftp://', warc_file_content)
self.assertIn(b'Content-Type: application/octet-stream',
warc_file_content)
self.assertIn(b'WARC-Type: metadata', warc_file_content)
self.assertIn(b'WARC-Concurrent-To: <urn:uuid:', warc_file_content)
self.assertIn(b'Content-Type: text/x-ftp-control-conversation',
warc_file_content)
self.assertIn(
'Wpull/{0}'.format(wpull.version.__version__).encode('utf-8'),
warc_file_content)
self.assertIn(
'Python/{0}'.format(wpull.util.python_version()).encode('utf-8'),
warc_file_content)
self.assertIn(b'KITTEH DOGE', warc_file_content)
self.assertIn(b'* Opening control connection to', warc_file_content)
self.assertIn(b'* Kept control connection to', warc_file_content)
self.assertIn(b'* Opened data connection to ', warc_file_content)
self.assertIn(b'* Closed data connection to ', warc_file_content)
self.assertIn(b'> GIMMEH example.txt', warc_file_content)
self.assertIn(b'< 200 OK, no need to yell.', warc_file_content)
# Ignore Concurrent Record ID not seen yet
self.validate_warc(warc_filename, ignore_minor_error=True)
with open(warc_filename, 'r+b') as in_file:
# Intentionally modify the contents
in_file.seek(355)
in_file.write(b'f')
with self.assertRaises(Exception):
# Sanity check that it actually raises error on bad digest
self.validate_warc(warc_filename, ignore_minor_error=True)
def test_warc_recorder_ftp(self):
file_prefix = 'asdf'
warc_filename = 'asdf.warc'
warc_recorder = WARCRecorder(
file_prefix,
params=WARCRecorderParams(compress=False)
)
request = FTPRequest('ftp://example.com/example.txt')
request.address = ('0.0.0.0', 80)
response = FTPResponse()
response.reply = FTPReply(200, 'OK')
response.body = Body()
response.data_address = ('0.0.0.0', 12345)
with wpull.util.reset_file_offset(response.body):
response.body.write(b'KITTEH DOGE')
with warc_recorder.session() as session:
session.begin_control(request)
session.request_control_data(b'GIMMEH example.txt')
session.response_control_data(b'200 OK, no need to yell.')
session.pre_response(response)
session.response_data(b'KITTEH DOGE')
session.response(response)
session.end_control(response)
warc_recorder.close()
with open(warc_filename, 'rb') as in_file:
warc_file_content = in_file.read()
self.assertTrue(warc_file_content.startswith(b'WARC/1.0'))
self.assertIn(b'WARC-Type: warcinfo\r\n', warc_file_content)
self.assertIn(b'Content-Type: application/warc-fields',
warc_file_content)
self.assertIn(b'WARC-Date: ', warc_file_content)
self.assertIn(b'WARC-Record-ID: <urn:uuid:', warc_file_content)
self.assertIn(b'WARC-Block-Digest: sha1:', warc_file_content)
self.assertNotIn(b'WARC-Payload-Digest: sha1:', warc_file_content)
self.assertIn(b'WARC-Type: resource\r\n', warc_file_content)
self.assertIn(b'WARC-Target-URI: ftp://', warc_file_content)
self.assertIn(b'Content-Type: application/octet-stream',
warc_file_content)
self.assertIn(b'WARC-Type: metadata', warc_file_content)
self.assertIn(b'WARC-Concurrent-To: <urn:uuid:', warc_file_content)
self.assertIn(b'Content-Type: text/x-ftp-control-conversation',
warc_file_content)
self.assertIn(
'Wpull/{0}'.format(wpull.version.__version__).encode('utf-8'),
warc_file_content
)
self.assertIn(
'Python/{0}'.format(
wpull.util.python_version()).encode('utf-8'),
warc_file_content
)
self.assertIn(b'KITTEH DOGE', warc_file_content)
self.assertIn(b'* Opening control connection to', warc_file_content)
self.assertIn(b'* Kept control connection to', warc_file_content)
self.assertIn(b'* Opened data connection to ', warc_file_content)
self.assertIn(b'* Closed data connection to ', warc_file_content)
self.assertIn(b'> GIMMEH example.txt', warc_file_content)
self.assertIn(b'< 200 OK, no need to yell.', warc_file_content)
# Ignore Concurrent Record ID not seen yet
self.validate_warc(warc_filename, ignore_minor_error=True)
with open(warc_filename, 'r+b') as in_file:
# Intentionally modify the contents
in_file.seek(355)
in_file.write(b'f')
with self.assertRaises(Exception):
# Sanity check that it actually raises error on bad digest
self.validate_warc(warc_filename, ignore_minor_error=True)