def test_simple(self):
"""
This test reads in a small number of particles and verifies the result of one of the particles.
"""
file_path = os.path.join(RESOURCE_PATH, 'flcdr_1_20131124T005004_458.mpk')
stream_handle = open(file_path, 'rb')
parser = FlcdrXMmpCdsParser(self.config, None, stream_handle,
self.state_callback, self.pub_callback)
particles = parser.get_records(6)
for particle in particles:
print particle.generate_dict()
test_data = self.get_dict_from_yml('good_cdr.yml')
self.assert_result(test_data['data'][0], particles[5])
test_data = self.get_dict_from_yml('second_cdr.yml')
self.assert_result(test_data['data'][0], particles[1])
test_data = self.get_dict_from_yml('first_cdr.yml')
self.assert_result(test_data['data'][0], particles[0])
stream_handle.close()
def test_get_many(self):
"""
This test exercises retrieving 20 particles, verifying the 20th particle, then retrieves 30 particles
and verifies the 30th particle.
"""
file_path = os.path.join(RESOURCE_PATH, 'flcdr_1_20131124T005004_458.mpk')
stream_handle = open(file_path, 'rb')
parser = FlcdrXMmpCdsParser(self.config, None, stream_handle,
self.state_callback, self.pub_callback)
particles = parser.get_records(20)
# Should end up with 20 particles
self.assertTrue(len(particles) == 20)
test_data = self.get_dict_from_yml('first_cdr.yml')
self.assert_result(test_data['data'][0], particles[0])
test_data = self.get_dict_from_yml('get_many_one_cdr.yml')
self.assert_result(test_data['data'][0], particles[19])
particles = parser.get_records(30)
# Should end up with 30 particles
self.assertTrue(len(particles) == 30)
test_data = self.get_dict_from_yml('get_many_two_cdr.yml')
self.assert_result(test_data['data'][0], particles[29])
stream_handle.close()
def test_mid_state_start(self):
"""
This test exercises setting the state past one chunk, retrieving particles and verify the result of one
of the particles.
"""
# Using two concatenated msgpack files to simulate two chunks.
file_path = os.path.join(RESOURCE_PATH, 'set_state_cdr.mpk')
stream_handle = open(file_path, 'rb')
stat_info = os.stat(file_path)
# Moving the file position to the end of the first chunk
state = {StateKey.PARTICLES_RETURNED: 20}
parser = FlcdrXMmpCdsParser(self.config, state, stream_handle,
self.state_callback, self.pub_callback)
particles = parser.get_records(4)
log.info(len(particles))
# Should end up with 4 particles
self.assertTrue(len(particles) == 4)
test_data = self.get_dict_from_yml('set_state_cdr.yml')
self.assert_result(test_data['data'][20], particles[0])
test_data = self.get_dict_from_yml('set_state_cdr.yml')
self.assert_result(test_data['data'][23], particles[3])
stream_handle.close()
def _build_parser(self, parser_state, infile, data_key=None):
"""
Build and return the parser
"""
parser = None
if data_key == DataParticleType.FLNTU_X_MMP_CDS_INSTRUMENT:
config = self._parser_config.get(
DataParticleType.FLNTU_X_MMP_CDS_INSTRUMENT)
config.update({
DataSetDriverConfigKeys.PARTICLE_MODULE:
'mi.dataset.parser.flntu_x_mmp_cds',
DataSetDriverConfigKeys.PARTICLE_CLASS:
'FlntuXMmpCdsParserDataParticle'
})
log.debug("My Config: %s", config)
parser = FlntuXMmpCdsParser(
config, parser_state, infile,
lambda state, ingested: self._save_parser_state(
state, data_key, ingested), self._data_callback,
self._sample_exception_callback)
elif data_key == DataParticleType.FLCDR_X_MMP_CDS_INSTRUMENT:
config = self._parser_config.get(
DataParticleType.FLCDR_X_MMP_CDS_INSTRUMENT)
config.update({
DataSetDriverConfigKeys.PARTICLE_MODULE:
'mi.dataset.parser.flcdr_x_mmp_cds',
DataSetDriverConfigKeys.PARTICLE_CLASS:
'FlcdrXMmpCdsParserDataParticle'
})
log.debug("My Config: %s", config)
parser = FlcdrXMmpCdsParser(
config, parser_state, infile,
lambda state, ingested: self._save_parser_state(
state, data_key, ingested), self._data_callback,
self._sample_exception_callback)
else:
raise ConfigurationException(
'flntu/flcdr parser not built due to missing key')
if parser is None:
raise ConfigurationException(
'flntu/flcdr parser not built due to failed instantiation')
return parser
def _build_parser(self, stream_handle):
parser_config = {
DataSetDriverConfigKeys.PARTICLE_MODULE:
'mi.dataset.parser.flcdr_x_mmp_cds',
DataSetDriverConfigKeys.PARTICLE_CLASS:
'FlcdrXMmpCdsParserDataParticle'
}
parser = FlcdrXMmpCdsParser(parser_config, None, stream_handle,
lambda state, ingested: None,
lambda data: None,
self._exception_callback)
return parser
def test_bad_data_two(self):
"""
This test verifies that a SampleException is raised when an entire msgpack buffer is not msgpack.
"""
file_path = os.path.join(RESOURCE_PATH, 'not-msg-pack.mpk')
stream_handle = open(file_path, 'rb')
parser = FlcdrXMmpCdsParser(self.config, None, stream_handle,
self.state_callback, self.pub_callback)
with self.assertRaises(SampleException):
parser.get_records(1)
stream_handle.close()
def test_bad_data_one(self):
"""
This test verifies that a SampleException is raised when msgpack data is malformed.
"""
file_path = os.path.join(RESOURCE_PATH, 'flcdr_1_20131124T005004_458-BAD.mpk')
stream_handle = open(file_path, 'rb')
parser = FlcdrXMmpCdsParser(self.config, None, stream_handle,
self.state_callback, self.pub_callback)
with self.assertRaises(SampleException):
parser.get_records(1)
stream_handle.close()
def test_long_stream(self):
"""
This test exercises retrieve approximately 200 particles.
"""
# Using two concatenated msgpack files to simulate two chunks to get more particles.
file_path = os.path.join(RESOURCE_PATH, 'flcdr_concat.mpk')
stream_handle = open(file_path, 'rb')
parser = FlcdrXMmpCdsParser(self.config, None, stream_handle,
self.state_callback, self.pub_callback)
# Attempt to retrieve 200 particles, but we will retrieve less
particles = parser.get_records(400)
# Should end up with 172 particles
self.assertTrue(len(particles) == 394)
stream_handle.close()
def test_verify_with_yml(self):
"""
This test exercises retrieving 4 particles and verifies them using a yml file
"""
file_path = os.path.join(RESOURCE_PATH, 'flcdr_1_20131124T005004_459.mpk')
stream_handle = open(file_path, 'rb')
parser = FlcdrXMmpCdsParser(self.config, None, stream_handle,
self.state_callback, self.pub_callback)
particles = parser.get_records(4)
# Should end up with 4 particles
self.assertTrue(len(particles) == 4)
self.assert_particles(particles, 'first_four_cdr.yml', RESOURCE_PATH)
stream_handle.close()
def test_set_state(self):
"""
This test exercises setting the state past one chunk, retrieving particles, verifying one
of the particles, and then setting the state back to the beginning, retrieving a few particles, and
verifying one of the particles.
"""
# Using the default mspack test file.
file_path = os.path.join(RESOURCE_PATH, 'set_state_cdr.mpk')
stream_handle = open(file_path, 'rb')
parser = FlcdrXMmpCdsParser(self.config, None, stream_handle,
self.state_callback, self.pub_callback)
particles = parser.get_records(4)
# Should end up with 4 particles
self.assertTrue(len(particles) == 4)
log.info(parser._state)
test_data = self.get_dict_from_yml('set_state_cdr.yml')
self.assert_result(test_data['data'][0], particles[0])
self.assert_result(test_data['data'][2], particles[2])
self.assert_result(test_data['data'][3], particles[3])
state = copy.copy(parser._state)
log.info(state)
parser = FlcdrXMmpCdsParser(self.config, state, stream_handle,
self.state_callback, self.pub_callback)
particles = parser.get_records(4)
# Should end up with 4 particles
self.assertTrue(len(particles) == 4)
self.assert_result(test_data['data'][6], particles[2])
self.assert_result(test_data['data'][7], particles[3])
# Give a bad position which will be ignored
state = {StateKey.PARTICLES_RETURNED: 0}
parser = FlcdrXMmpCdsParser(self.config, state, stream_handle,
self.state_callback, self.pub_callback)
particles = parser.get_records(1)
self.assertTrue(len(particles) == 1)
# Give a bad position which will be ignored
state = {StateKey.PARTICLES_RETURNED: 0}
parser = FlcdrXMmpCdsParser(self.config, state, stream_handle,
self.state_callback, self.pub_callback)
particles = parser.get_records(1000)
self.assertTrue(len(particles) == 30)
self.assert_result(test_data['data'][29], particles[29])
self.assert_result(test_data['data'][20], particles[20])
self.assert_result(test_data['data'][0], particles[0])
# Provide a bad particles returned
state = {StateKey.PARTICLES_RETURNED: 80}
parser = FlcdrXMmpCdsParser(self.config, state, stream_handle,
self.state_callback, self.pub_callback)
particles = parser.get_records(1)
self.assertTrue(len(particles) == 0)
stream_handle.close()