Пример #1
0
    def test_stop_resume(self):
        """
        Test the ability to stop and restart the process
        """
        # create some data to parse
        self.clear_async_data()
        self.create_sample_data('test_data_1.txt', "DATA001.txt")
        self.create_sample_data('test_data_3.txt', "DATA002.txt")

        state = {
            'DATA001.txt':
            self.get_file_state('/tmp/dsatest/DATA001.txt', True),
            'DATA002.txt':
            self.get_file_state('/tmp/dsatest/DATA002.txt', False, 209),
        }
        state['DATA002.txt']['parser_state']['timestamp'] = 3583861265.0

        self.driver = HypmCTDPFDataSetDriver(
            self._driver_config()['startup_config'], state, self.data_callback,
            self.state_callback, self.event_callback, self.exception_callback)

        self.driver.start_sampling()

        # verify data is produced
        self.assert_data(CtdpfParserDataParticle,
                         'test_data_3.txt.partial_results.yml',
                         count=5,
                         timeout=10)
        # verify we got the rest of the file
        self.assert_file_ingested("DATA002.txt")
Пример #2
0
 def test_harvester_config_exception(self):
     """
     Start the a driver with a bad configuration.  Should raise
     an exception.
     """
     with self.assertRaises(ConfigurationException):
         self.driver = HypmCTDPFDataSetDriver({},
             self.memento,
             self.data_callback,
             self.state_callback,
             self.exception_callback)
Пример #3
0
    def test_stop_resume(self):
        """
        Test the ability to stop and restart the process
        """
        # create some data to parse
        self.clear_async_data()
        self.create_sample_data('test_data_1.txt', "DATA001.txt")
        self.create_sample_data('test_data_3.txt', "DATA002.txt")

        state = {
            'DATA001.txt': self.get_file_state('/tmp/dsatest/DATA001.txt', True),
            'DATA002.txt': self.get_file_state('/tmp/dsatest/DATA002.txt', False, 209),
        }
        state['DATA002.txt']['parser_state']['timestamp'] = 3583861265.0

        self.driver = HypmCTDPFDataSetDriver(
            self._driver_config()['startup_config'],
            state,
            self.data_callback,
            self.state_callback,
            self.event_callback,
            self.exception_callback)

        self.driver.start_sampling()

        # verify data is produced
        self.assert_data(CtdpfParserDataParticle, 'test_data_3.txt.partial_results.yml', count=5, timeout=10)
        # verify we got the rest of the file
        self.assert_file_ingested("DATA002.txt")
Пример #4
0
    def test_stop_resume(self):
        """
        Test the ability to stop and restart the process
        """
        # Create and store the new driver state
        self.memento = {DataSourceConfigKey.HARVESTER: '/tmp/dsatest/DATA001.txt',
                        DataSourceConfigKey.PARSER: {'position': 209, 'timestamp': 3583886465.0}}
        self.driver = HypmCTDPFDataSetDriver(
            self._driver_config()['startup_config'],
            self.memento,
            self.data_callback,
            self.state_callback,
            self.exception_callback)

        # create some data to parse
        self.clear_async_data()
        self.create_sample_data('test_data_1.txt', "DATA001.txt")
        self.create_sample_data('test_data_3.txt', "DATA002.txt")

        self.driver.start_sampling()

        # verify data is produced
        self.assert_data(CtdpfParserDataParticle, 'test_data_3.txt.partial_results.yml', count=5, timeout=10)
Пример #5
0
class IntegrationTest(DataSetIntegrationTestCase):
    def test_get(self):
        """
        Test that we can get data from files.  Verify that the driver sampling
        can be started and stopped.
        """
        self.clear_sample_data()

        # Start sampling and watch for an exception
        self.driver.start_sampling()

        self.clear_async_data()
        self.create_sample_data('test_data_1.txt', "DATA001.txt")
        self.assert_data(CtdpfParserDataParticle, 'test_data_1.txt.result.yml', count=1, timeout=10)
        self.assert_file_ingested("DATA001.txt")

        self.clear_async_data()
        self.create_sample_data('test_data_3.txt', "DATA002.txt")
        self.assert_data(CtdpfParserDataParticle, 'test_data_3.txt.result.yml', count=8, timeout=10)
        self.assert_file_ingested("DATA002.txt")

        self.clear_async_data()
        self.create_sample_data('DATA003.txt')
        self.assert_data(CtdpfParserDataParticle, count=436, timeout=20)
        self.assert_file_ingested("DATA003.txt")

        self.driver.stop_sampling()
        self.driver.start_sampling()

        self.clear_async_data()
        self.create_sample_data('test_data_1.txt', "DATA004.txt")
        self.assert_data(CtdpfParserDataParticle, count=1, timeout=10)
        self.assert_file_ingested("DATA004.txt")

    def test_harvester_config_exception(self):
        """
        Start the a driver with a bad configuration.  Should raise
        an exception.
        """
        with self.assertRaises(ConfigurationException):
            self.driver = HypmCTDPFDataSetDriver({},
                self.memento,
                self.data_callback,
                self.state_callback,
                self.event_callback,
                self.exception_callback)

    def test_stop_resume(self):
        """
        Test the ability to stop and restart the process
        """
        # create some data to parse
        self.clear_async_data()
        self.create_sample_data('test_data_1.txt', "DATA001.txt")
        self.create_sample_data('test_data_3.txt', "DATA002.txt")

        state = {
            'DATA001.txt': self.get_file_state('/tmp/dsatest/DATA001.txt', True),
            'DATA002.txt': self.get_file_state('/tmp/dsatest/DATA002.txt', False, 209),
        }
        state['DATA002.txt']['parser_state']['timestamp'] = 3583861265.0

        self.driver = HypmCTDPFDataSetDriver(
            self._driver_config()['startup_config'],
            state,
            self.data_callback,
            self.state_callback,
            self.event_callback,
            self.exception_callback)

        self.driver.start_sampling()

        # verify data is produced
        self.assert_data(CtdpfParserDataParticle, 'test_data_3.txt.partial_results.yml', count=5, timeout=10)
        # verify we got the rest of the file
        self.assert_file_ingested("DATA002.txt")

    def test_stop_start_ingest(self):
        """
        Test the ability to stop and restart sampling, and ingesting files in the correct order
        """
        # create some data to parse
        self.clear_async_data()
        
        self.driver.start_sampling()
        
        self.create_sample_data('test_data_1.txt', "DATA001.txt")
        self.create_sample_data('test_data_3.txt', "DATA002.txt")
        self.assert_data(CtdpfParserDataParticle, 'test_data_1.txt.result.yml', count=1, timeout=10)
        self.assert_file_ingested("DATA001.txt")
        self.assert_file_not_ingested("DATA002.txt")
        
        self.driver.stop_sampling()
        self.driver.start_sampling()
        
        self.assert_data(CtdpfParserDataParticle, 'test_data_3.txt.result.yml', count=8, timeout=10)
        self.assert_file_ingested("DATA002.txt")

    def test_bad_sample(self):
        """
        Test a bad sample.  To do this we set a state to the middle of a record
        """
        self.clear_sample_data()

        # Start sampling and watch for an exception
        self.driver.start_sampling()

        self.clear_async_data()
        self.create_sample_data('bad_sample_data.txt', "DATA001.txt")
        self.assert_data(CtdpfParserDataParticle, 'bad_sample_data.txt.result.yml', count=2, timeout=10)
        gevent.sleep(5)
        self.assert_file_ingested("DATA001.txt")
Пример #6
0
from mi.dataset.parser.ctdpf import CtdpfParserDataParticle
from pyon.agent.agent import ResourceAgentState

from interface.objects import ResourceAgentErrorEvent

DATADIR='/tmp/dsatest'
STORAGEDIR='/tmp/stored_dsatest'
RESOURCE_ID='ctdpf'

DataSetTestCase.initialize(
    driver_module='mi.dataset.driver.hypm.ctd.driver',
    driver_class="HypmCTDPFDataSetDriver",

    agent_resource_id = '123xyz',
    agent_name = 'Agent007',
    agent_packet_config = HypmCTDPFDataSetDriver.stream_config(),
    startup_config = {
        DataSourceConfigKey.RESOURCE_ID: RESOURCE_ID,
        DataSourceConfigKey.HARVESTER:
        {
            DataSetDriverConfigKeys.DIRECTORY: DATADIR,
            DataSetDriverConfigKeys.STORAGE_DIRECTORY: STORAGEDIR,
            DataSetDriverConfigKeys.PATTERN: '*.txt',
            DataSetDriverConfigKeys.FREQUENCY: 1,
        },
        DataSourceConfigKey.PARSER: {}
    }
)

SAMPLE_STREAM='ctdpf_parsed'
    
Пример #7
0
class IntegrationTest(DataSetIntegrationTestCase):
    def test_get(self):
        """
        Test that we can get data from files.  Verify that the driver sampling
        can be started and stopped.
        """
        self.clear_sample_data()

        # Start sampling and watch for an exception
        self.driver.start_sampling()

        self.clear_async_data()
        self.create_sample_data('test_data_1.txt', "DATA001.txt")
        self.assert_data(CtdpfParserDataParticle, 'test_data_1.txt.result.yml', count=1, timeout=10)

        self.clear_async_data()
        self.create_sample_data('test_data_3.txt', "DATA002.txt")
        self.assert_data(CtdpfParserDataParticle, 'test_data_3.txt.result.yml', count=8, timeout=10)

        self.clear_async_data()
        self.create_sample_data('DATA003.txt')
        self.assert_data(CtdpfParserDataParticle, count=436, timeout=20)

        self.driver.stop_sampling()
        self.driver.start_sampling()

        self.clear_async_data()
        self.create_sample_data('test_data_1.txt', "DATA004.txt")
        self.assert_data(CtdpfParserDataParticle, count=1, timeout=10)

    def test_harvester_config_exception(self):
        """
        Start the a driver with a bad configuration.  Should raise
        an exception.
        """
        with self.assertRaises(ConfigurationException):
            self.driver = HypmCTDPFDataSetDriver({},
                self.memento,
                self.data_callback,
                self.state_callback,
                self.exception_callback)

    def test_harvester_new_file_exception(self):
        """
        Test an exception raised after the driver is started during
        the file read.  Should call the exception callback.
        """
        self.clear_sample_data()

        # create the file so that it is unreadable
        self.create_sample_data('DATA003.txt', mode=000)

        # Start sampling and watch for an exception
        self.driver.start_sampling()

        self.assert_exception(IOError)

        # At this point the harvester thread is dead.  The agent
        # exception handle should handle this case.

    def test_stop_resume(self):
        """
        Test the ability to stop and restart the process
        """
        # Create and store the new driver state
        self.memento = {DataSourceConfigKey.HARVESTER: '/tmp/dsatest/DATA001.txt',
                        DataSourceConfigKey.PARSER: {'position': 209, 'timestamp': 3583886465.0}}
        self.driver = HypmCTDPFDataSetDriver(
            self._driver_config()['startup_config'],
            self.memento,
            self.data_callback,
            self.state_callback,
            self.exception_callback)

        # create some data to parse
        self.clear_async_data()
        self.create_sample_data('test_data_1.txt', "DATA001.txt")
        self.create_sample_data('test_data_3.txt', "DATA002.txt")

        self.driver.start_sampling()

        # verify data is produced
        self.assert_data(CtdpfParserDataParticle, 'test_data_3.txt.partial_results.yml', count=5, timeout=10)

    def test_parameters(self):
        """
        Verify that we can get, set, and report all driver parameters.
        """
        expected_params = [DriverParameter.BATCHED_PARTICLE_COUNT, DriverParameter.PUBLISHER_POLLING_INTERVAL, DriverParameter.RECORDS_PER_SECOND]
        (res_cmds, res_params) = self.driver.get_resource_capabilities()

        # Ensure capabilities are as expected
        self.assertEqual(len(res_cmds), 0)
        self.assertEqual(len(res_params), len(expected_params))
        self.assertEqual(sorted(res_params), sorted(expected_params))

        # Verify default values are as expected.
        params = self.driver.get_resource(DriverParameter.ALL)
        log.debug("Get Resources Result: %s", params)
        self.assertEqual(params[DriverParameter.BATCHED_PARTICLE_COUNT], 1)
        self.assertEqual(params[DriverParameter.PUBLISHER_POLLING_INTERVAL], 1)
        self.assertEqual(params[DriverParameter.RECORDS_PER_SECOND], 60)

        # Try set resource individually
        self.driver.set_resource({DriverParameter.BATCHED_PARTICLE_COUNT: 2})
        self.driver.set_resource({DriverParameter.PUBLISHER_POLLING_INTERVAL: 2})
        self.driver.set_resource({DriverParameter.RECORDS_PER_SECOND: 59})

        params = self.driver.get_resource(DriverParameter.ALL)
        log.debug("Get Resources Result: %s", params)
        self.assertEqual(params[DriverParameter.BATCHED_PARTICLE_COUNT], 2)
        self.assertEqual(params[DriverParameter.PUBLISHER_POLLING_INTERVAL], 2)
        self.assertEqual(params[DriverParameter.RECORDS_PER_SECOND], 59)

        # Try set resource in bulk
        self.driver.set_resource(
            {DriverParameter.BATCHED_PARTICLE_COUNT: 1,
             DriverParameter.PUBLISHER_POLLING_INTERVAL: .1,
             DriverParameter.RECORDS_PER_SECOND: 60})

        params = self.driver.get_resource(DriverParameter.ALL)
        log.debug("Get Resources Result: %s", params)
        self.assertEqual(params[DriverParameter.BATCHED_PARTICLE_COUNT], 1)
        self.assertEqual(params[DriverParameter.PUBLISHER_POLLING_INTERVAL], .1)
        self.assertEqual(params[DriverParameter.RECORDS_PER_SECOND], 60)

        # Set with some bad values
        with self.assertRaises(InstrumentParameterException):
            self.driver.set_resource({DriverParameter.BATCHED_PARTICLE_COUNT: 'a'})
        with self.assertRaises(InstrumentParameterException):
            self.driver.set_resource({DriverParameter.BATCHED_PARTICLE_COUNT: -1})
        with self.assertRaises(InstrumentParameterException):
            self.driver.set_resource({DriverParameter.BATCHED_PARTICLE_COUNT: 0})

        # Try to configure with the driver startup config
        driver_config = self._driver_config()['startup_config']
        cfg = {
            DataSourceConfigKey.HARVESTER: driver_config.get(DataSourceConfigKey.HARVESTER),
            DataSourceConfigKey.PARSER: driver_config.get(DataSourceConfigKey.PARSER),
            DataSourceConfigKey.DRIVER: {
                DriverParameter.PUBLISHER_POLLING_INTERVAL: .2,
                DriverParameter.RECORDS_PER_SECOND: 3,
                DriverParameter.BATCHED_PARTICLE_COUNT: 3,
            }
        }
        self.driver = HypmCTDPFDataSetDriver(
            cfg,
            self.memento,
            self.data_callback,
            self.state_callback,
            self.exception_callback)

        params = self.driver.get_resource(DriverParameter.ALL)
        log.debug("Get Resources Result: %s", params)
        self.assertEqual(params[DriverParameter.BATCHED_PARTICLE_COUNT], 3)
        self.assertEqual(params[DriverParameter.PUBLISHER_POLLING_INTERVAL], .2)
        self.assertEqual(params[DriverParameter.RECORDS_PER_SECOND], 3)

        # Finally verify we get a KeyError when sending in bad config keys
        cfg[DataSourceConfigKey.DRIVER] = {
            DriverParameter.PUBLISHER_POLLING_INTERVAL: .2,
            DriverParameter.RECORDS_PER_SECOND: 3,
            DriverParameter.BATCHED_PARTICLE_COUNT: 3,
            'something_extra': 1
        }

        with self.assertRaises(KeyError):
            self.driver = HypmCTDPFDataSetDriver(
                cfg,
                self.memento,
                self.data_callback,
                self.state_callback,
                self.exception_callback)

    def test_sequences(self):
        """
        Test new sequence flags are set correctly
        """

        ###
        #   One file, no breaks, should only have 1 new sequence flag
        #   New sequence flag when a new file is read
        ###
        self.clear_sample_data()

        self.driver.start_sampling()

        self.clear_async_data()
        self.create_sample_data('test_data_1.txt', "DATA001.txt")
        self.assert_data(CtdpfParserDataParticle, 'test_data_1.txt.result.yml', count=1, timeout=10)

        self.clear_async_data()
        self.create_sample_data('test_data_3.txt', "DATA002.txt")
        self.assert_data(CtdpfParserDataParticle, 'test_data_3.txt.result.yml', count=8, timeout=10)

        ###
        #   New sequence flag when noise if detected between records
        ###
        self.clear_async_data()
        self.create_sample_data('test_data_4.txt', "DATA004.txt")
        self.assert_data(CtdpfParserDataParticle, 'test_data_4.txt.result.yml', count=8, timeout=10)
Пример #8
0
    def test_parameters(self):
        """
        Verify that we can get, set, and report all driver parameters.
        """
        expected_params = [DriverParameter.BATCHED_PARTICLE_COUNT, DriverParameter.PUBLISHER_POLLING_INTERVAL, DriverParameter.RECORDS_PER_SECOND]
        (res_cmds, res_params) = self.driver.get_resource_capabilities()

        # Ensure capabilities are as expected
        self.assertEqual(len(res_cmds), 0)
        self.assertEqual(len(res_params), len(expected_params))
        self.assertEqual(sorted(res_params), sorted(expected_params))

        # Verify default values are as expected.
        params = self.driver.get_resource(DriverParameter.ALL)
        log.debug("Get Resources Result: %s", params)
        self.assertEqual(params[DriverParameter.BATCHED_PARTICLE_COUNT], 1)
        self.assertEqual(params[DriverParameter.PUBLISHER_POLLING_INTERVAL], 1)
        self.assertEqual(params[DriverParameter.RECORDS_PER_SECOND], 60)

        # Try set resource individually
        self.driver.set_resource({DriverParameter.BATCHED_PARTICLE_COUNT: 2})
        self.driver.set_resource({DriverParameter.PUBLISHER_POLLING_INTERVAL: 2})
        self.driver.set_resource({DriverParameter.RECORDS_PER_SECOND: 59})

        params = self.driver.get_resource(DriverParameter.ALL)
        log.debug("Get Resources Result: %s", params)
        self.assertEqual(params[DriverParameter.BATCHED_PARTICLE_COUNT], 2)
        self.assertEqual(params[DriverParameter.PUBLISHER_POLLING_INTERVAL], 2)
        self.assertEqual(params[DriverParameter.RECORDS_PER_SECOND], 59)

        # Try set resource in bulk
        self.driver.set_resource(
            {DriverParameter.BATCHED_PARTICLE_COUNT: 1,
             DriverParameter.PUBLISHER_POLLING_INTERVAL: .1,
             DriverParameter.RECORDS_PER_SECOND: 60})

        params = self.driver.get_resource(DriverParameter.ALL)
        log.debug("Get Resources Result: %s", params)
        self.assertEqual(params[DriverParameter.BATCHED_PARTICLE_COUNT], 1)
        self.assertEqual(params[DriverParameter.PUBLISHER_POLLING_INTERVAL], .1)
        self.assertEqual(params[DriverParameter.RECORDS_PER_SECOND], 60)

        # Set with some bad values
        with self.assertRaises(InstrumentParameterException):
            self.driver.set_resource({DriverParameter.BATCHED_PARTICLE_COUNT: 'a'})
        with self.assertRaises(InstrumentParameterException):
            self.driver.set_resource({DriverParameter.BATCHED_PARTICLE_COUNT: -1})
        with self.assertRaises(InstrumentParameterException):
            self.driver.set_resource({DriverParameter.BATCHED_PARTICLE_COUNT: 0})

        # Try to configure with the driver startup config
        driver_config = self._driver_config()['startup_config']
        cfg = {
            DataSourceConfigKey.HARVESTER: driver_config.get(DataSourceConfigKey.HARVESTER),
            DataSourceConfigKey.PARSER: driver_config.get(DataSourceConfigKey.PARSER),
            DataSourceConfigKey.DRIVER: {
                DriverParameter.PUBLISHER_POLLING_INTERVAL: .2,
                DriverParameter.RECORDS_PER_SECOND: 3,
                DriverParameter.BATCHED_PARTICLE_COUNT: 3,
            }
        }
        self.driver = HypmCTDPFDataSetDriver(
            cfg,
            self.memento,
            self.data_callback,
            self.state_callback,
            self.exception_callback)

        params = self.driver.get_resource(DriverParameter.ALL)
        log.debug("Get Resources Result: %s", params)
        self.assertEqual(params[DriverParameter.BATCHED_PARTICLE_COUNT], 3)
        self.assertEqual(params[DriverParameter.PUBLISHER_POLLING_INTERVAL], .2)
        self.assertEqual(params[DriverParameter.RECORDS_PER_SECOND], 3)

        # Finally verify we get a KeyError when sending in bad config keys
        cfg[DataSourceConfigKey.DRIVER] = {
            DriverParameter.PUBLISHER_POLLING_INTERVAL: .2,
            DriverParameter.RECORDS_PER_SECOND: 3,
            DriverParameter.BATCHED_PARTICLE_COUNT: 3,
            'something_extra': 1
        }

        with self.assertRaises(KeyError):
            self.driver = HypmCTDPFDataSetDriver(
                cfg,
                self.memento,
                self.data_callback,
                self.state_callback,
                self.exception_callback)
Пример #9
0
from mi.dataset.parser.ctdpf import CtdpfParserDataParticle
from pyon.agent.agent import ResourceAgentState

from interface.objects import CapabilityType
from interface.objects import AgentCapability
from interface.objects import ResourceAgentErrorEvent
from interface.objects import ResourceAgentConnectionLostErrorEvent

DataSetTestCase.initialize(
    driver_module='mi.dataset.driver.hypm.ctd.driver',
    driver_class="HypmCTDPFDataSetDriver",

    agent_resource_id = '123xyz',
    agent_name = 'Agent007',
    agent_packet_config = HypmCTDPFDataSetDriver.stream_config(),
    startup_config = {
        'harvester':
        {
            'directory': '/tmp/dsatest',
            'pattern': '*.txt',
            'frequency': 1,
        },
        'parser': {}
    }
)

SAMPLE_STREAM='ctdpf_parsed'
    
###############################################################################
#                                UNIT TESTS                                   #
Пример #10
0
class IntegrationTest(DataSetIntegrationTestCase):
    def test_get(self):
        """
        Test that we can get data from files.  Verify that the driver sampling
        can be started and stopped.
        """
        self.clear_sample_data()

        # Start sampling and watch for an exception
        self.driver.start_sampling()

        self.clear_async_data()
        self.create_sample_data('test_data_1.txt', "DATA001.txt")
        self.assert_data(CtdpfParserDataParticle,
                         'test_data_1.txt.result.yml',
                         count=1,
                         timeout=10)
        self.assert_file_ingested("DATA001.txt")

        self.clear_async_data()
        self.create_sample_data('test_data_3.txt', "DATA002.txt")
        self.assert_data(CtdpfParserDataParticle,
                         'test_data_3.txt.result.yml',
                         count=8,
                         timeout=10)
        self.assert_file_ingested("DATA002.txt")

        self.clear_async_data()
        self.create_sample_data('DATA003.txt')
        self.assert_data(CtdpfParserDataParticle, count=436, timeout=20)
        self.assert_file_ingested("DATA003.txt")

        self.driver.stop_sampling()
        self.driver.start_sampling()

        self.clear_async_data()
        self.create_sample_data('test_data_1.txt', "DATA004.txt")
        self.assert_data(CtdpfParserDataParticle, count=1, timeout=10)
        self.assert_file_ingested("DATA004.txt")

    def test_harvester_config_exception(self):
        """
        Start the a driver with a bad configuration.  Should raise
        an exception.
        """
        with self.assertRaises(ConfigurationException):
            self.driver = HypmCTDPFDataSetDriver({}, self.memento,
                                                 self.data_callback,
                                                 self.state_callback,
                                                 self.event_callback,
                                                 self.exception_callback)

    def test_stop_resume(self):
        """
        Test the ability to stop and restart the process
        """
        # create some data to parse
        self.clear_async_data()
        self.create_sample_data('test_data_1.txt', "DATA001.txt")
        self.create_sample_data('test_data_3.txt', "DATA002.txt")

        state = {
            'DATA001.txt':
            self.get_file_state('/tmp/dsatest/DATA001.txt', True),
            'DATA002.txt':
            self.get_file_state('/tmp/dsatest/DATA002.txt', False, 209),
        }
        state['DATA002.txt']['parser_state']['timestamp'] = 3583861265.0

        self.driver = HypmCTDPFDataSetDriver(
            self._driver_config()['startup_config'], state, self.data_callback,
            self.state_callback, self.event_callback, self.exception_callback)

        self.driver.start_sampling()

        # verify data is produced
        self.assert_data(CtdpfParserDataParticle,
                         'test_data_3.txt.partial_results.yml',
                         count=5,
                         timeout=10)
        # verify we got the rest of the file
        self.assert_file_ingested("DATA002.txt")

    def test_stop_start_ingest(self):
        """
        Test the ability to stop and restart sampling, and ingesting files in the correct order
        """
        # create some data to parse
        self.clear_async_data()

        self.driver.start_sampling()

        self.create_sample_data('test_data_1.txt', "DATA001.txt")
        self.create_sample_data('test_data_3.txt', "DATA002.txt")
        self.assert_data(CtdpfParserDataParticle,
                         'test_data_1.txt.result.yml',
                         count=1,
                         timeout=10)
        self.assert_file_ingested("DATA001.txt")
        self.assert_file_not_ingested("DATA002.txt")

        self.driver.stop_sampling()
        self.driver.start_sampling()

        self.assert_data(CtdpfParserDataParticle,
                         'test_data_3.txt.result.yml',
                         count=8,
                         timeout=10)
        self.assert_file_ingested("DATA002.txt")

    def test_bad_sample(self):
        """
        Test a bad sample.  To do this we set a state to the middle of a record
        """
        self.clear_sample_data()

        # Start sampling and watch for an exception
        self.driver.start_sampling()

        self.clear_async_data()
        self.create_sample_data('bad_sample_data.txt', "DATA001.txt")
        self.assert_data(CtdpfParserDataParticle,
                         'bad_sample_data.txt.result.yml',
                         count=2,
                         timeout=10)
        gevent.sleep(5)
        self.assert_file_ingested("DATA001.txt")
Пример #11
0
class IntegrationTest(DataSetIntegrationTestCase):
    def test_get(self):
        """
        Test that we can get data from files.  Verify that the driver sampling
        can be started and stopped.
        """
        self.clear_sample_data()

        # Start sampling and watch for an exception
        self.driver.start_sampling()

        self.clear_async_data()
        self.create_sample_data('test_data_1.txt', "DATA001.txt")
        self.assert_data(CtdpfParserDataParticle, 'test_data_1.txt.result.yml', count=1, timeout=10)

        self.clear_async_data()
        self.create_sample_data('test_data_3.txt', "DATA002.txt")
        self.assert_data(CtdpfParserDataParticle, 'test_data_3.txt.result.yml', count=8, timeout=10)

        self.clear_async_data()
        self.create_sample_data('DATA003.txt')
        self.assert_data(CtdpfParserDataParticle, count=436, timeout=20)

        self.driver.stop_sampling()
        self.driver.start_sampling()

        self.clear_async_data()
        self.create_sample_data('test_data_1.txt', "DATA004.txt")
        self.assert_data(CtdpfParserDataParticle, count=1, timeout=10)

    def test_harvester_config_exception(self):
        """
        Start the a driver with a bad configuration.  Should raise
        an exception.
        """
        with self.assertRaises(ConfigurationException):
            self.driver = HypmCTDPFDataSetDriver({},
                self.memento,
                self.data_callback,
                self.state_callback,
                self.exception_callback)

    def test_stop_resume(self):
        """
        Test the ability to stop and restart the process
        """
        # Create and store the new driver state
        self.memento = {DataSourceConfigKey.HARVESTER: '/tmp/dsatest/DATA001.txt',
                        DataSourceConfigKey.PARSER: {'position': 209, 'timestamp': 3583861265.0}}
        self.driver = HypmCTDPFDataSetDriver(
            self._driver_config()['startup_config'],
            self.memento,
            self.data_callback,
            self.state_callback,
            self.exception_callback)

        # create some data to parse
        self.clear_async_data()
        self.create_sample_data('test_data_1.txt', "DATA001.txt")
        self.create_sample_data('test_data_3.txt', "DATA002.txt")

        self.driver.start_sampling()

        # verify data is produced
        self.assert_data(CtdpfParserDataParticle, 'test_data_3.txt.partial_results.yml', count=5, timeout=10)