def setUp(self):
self._start_container()
self.container.start_rel_from_url(rel_url='res/deploy/r2eoi.yml')
self.dams_cli = DataAcquisitionManagementServiceClient()
self.dpms_cli = DataProductManagementServiceClient()
eda = ExternalDatasetAgent()
self.eda_id = self.dams_cli.create_external_dataset_agent(eda)
eda_inst = ExternalDatasetAgentInstance()
self.eda_inst_id = self.dams_cli.create_external_dataset_agent_instance(
eda_inst, external_dataset_agent_id=self.eda_id)
self._setup_ncom()
proc_name = self.ncom_ds_id + '_worker'
config = {}
config['process'] = {'name': proc_name, 'type': 'agent'}
config['process']['eoa'] = {'dataset_id': self.ncom_ds_id}
pid = self.container.spawn_process(
name=proc_name,
module='eoi.agent.external_observatory_agent',
cls='ExternalObservatoryAgent',
config=config)
queue_id = "%s.%s" % (self.container.id, pid)
log.debug(
"Spawned worker process ==> proc_name: %s\tproc_id: %s\tqueue_id: %s"
% (proc_name, pid, queue_id))
self._agent_cli = ResourceAgentClient(self.ncom_ds_id,
name=pid,
process=FakeProcess())
log.debug("Got a ResourceAgentClient: res_id=%s" %
self._agent_cli.resource_id)
def _setup_resources(self):
# TODO: some or all of this (or some variation) should move to DAMS'
# Build the test resources for the dataset
dms_cli = DatasetManagementServiceClient()
dams_cli = DataAcquisitionManagementServiceClient()
dpms_cli = DataProductManagementServiceClient()
rr_cli = ResourceRegistryServiceClient()
pubsub_cli = PubsubManagementServiceClient()
eda = ExternalDatasetAgent(handler_module=self.DVR_CONFIG['dvr_mod'],
handler_class=self.DVR_CONFIG['dvr_cls'])
eda_id = dams_cli.create_external_dataset_agent(eda)
eda_inst = ExternalDatasetAgentInstance()
eda_inst_id = dams_cli.create_external_dataset_agent_instance(eda_inst, external_dataset_agent_id=eda_id)
# Create and register the necessary resources/objects
# Create DataProvider
dprov = ExternalDataProvider(institution=Institution(), contact=ContactInformation())
dprov.contact.individual_names_given = 'Christopher Mueller'
dprov.contact.email = '*****@*****.**'
# Create DataSource
dsrc = DataSource(protocol_type='FILE', institution=Institution(), contact=ContactInformation())
dsrc.connection_params['base_data_url'] = ''
dsrc.contact.individual_names_given = 'Tim Giguere'
dsrc.contact.email = '*****@*****.**'
# Create ExternalDataset
ds_name = 'slocum_test_dataset'
dset = ExternalDataset(name=ds_name, dataset_description=DatasetDescription(), update_description=UpdateDescription(), contact=ContactInformation())
dset.dataset_description.parameters['base_url'] = 'test_data/slocum/'
dset.dataset_description.parameters['list_pattern'] = 'ru05-2012-021-0-0-sbd.dat'
dset.dataset_description.parameters['date_pattern'] = '%Y %j'
dset.dataset_description.parameters['date_extraction_pattern'] = 'ru05-([\d]{4})-([\d]{3})-\d-\d-sbd.dat'
dset.dataset_description.parameters['temporal_dimension'] = None
dset.dataset_description.parameters['zonal_dimension'] = None
dset.dataset_description.parameters['meridional_dimension'] = None
dset.dataset_description.parameters['vertical_dimension'] = None
dset.dataset_description.parameters['variables'] = [
'c_wpt_y_lmc',
'sci_water_cond',
'm_y_lmc',
'u_hd_fin_ap_inflection_holdoff',
'sci_m_present_time',
'm_leakdetect_voltage_forward',
'sci_bb3slo_b660_scaled',
'c_science_send_all',
'm_gps_status',
'm_water_vx',
'm_water_vy',
'c_heading',
'sci_fl3slo_chlor_units',
'u_hd_fin_ap_gain',
'm_vacuum',
'u_min_water_depth',
'm_gps_lat',
'm_veh_temp',
'f_fin_offset',
'u_hd_fin_ap_hardover_holdoff',
'c_alt_time',
'm_present_time',
'm_heading',
'sci_bb3slo_b532_scaled',
'sci_fl3slo_cdom_units',
'm_fin',
'x_cycle_overrun_in_ms',
'sci_water_pressure',
'u_hd_fin_ap_igain',
'sci_fl3slo_phyco_units',
'm_battpos',
'sci_bb3slo_b470_scaled',
'm_lat',
'm_gps_lon',
'sci_ctd41cp_timestamp',
'm_pressure',
'c_wpt_x_lmc',
'c_ballast_pumped',
'x_lmc_xy_source',
'm_lon',
'm_avg_speed',
'sci_water_temp',
'u_pitch_ap_gain',
'm_roll',
'm_tot_num_inflections',
'm_x_lmc',
'u_pitch_ap_deadband',
'm_final_water_vy',
'm_final_water_vx',
'm_water_depth',
'm_leakdetect_voltage',
'u_pitch_max_delta_battpos',
'm_coulomb_amphr',
'm_pitch', ]
# Create DataSourceModel
dsrc_model = DataSourceModel(name='slocum_model')
# dsrc_model.model = 'SLOCUM'
dsrc_model.data_handler_module = 'N/A'
dsrc_model.data_handler_class = 'N/A'
## Run everything through DAMS
ds_id = dams_cli.create_external_dataset(external_dataset=dset)
ext_dprov_id = dams_cli.create_external_data_provider(external_data_provider=dprov)
ext_dsrc_id = dams_cli.create_data_source(data_source=dsrc)
ext_dsrc_model_id = dams_cli.create_data_source_model(dsrc_model)
# Register the ExternalDataset
dproducer_id = dams_cli.register_external_data_set(external_dataset_id=ds_id)
# Or using each method
dams_cli.assign_data_source_to_external_data_provider(data_source_id=ext_dsrc_id, external_data_provider_id=ext_dprov_id)
dams_cli.assign_data_source_to_data_model(data_source_id=ext_dsrc_id, data_source_model_id=ext_dsrc_model_id)
dams_cli.assign_external_dataset_to_data_source(external_dataset_id=ds_id, data_source_id=ext_dsrc_id)
dams_cli.assign_external_dataset_to_agent_instance(external_dataset_id=ds_id, agent_instance_id=eda_inst_id)
# dams_cli.assign_external_data_agent_to_agent_instance(external_data_agent_id=self.eda_id, agent_instance_id=self.eda_inst_id)
#create temp streamdef so the data product can create the stream
pc_list = []
for pc_k, pc in self._create_parameter_dictionary().iteritems():
pc_list.append(dms_cli.create_parameter_context(pc_k, pc[1].dump()))
pdict_id = dms_cli.create_parameter_dictionary('slocum_param_dict', pc_list)
streamdef_id = pubsub_cli.create_stream_definition(name="slocum_stream_def", description="stream def for slocum testing", parameter_dictionary_id=pdict_id)
# dpms_cli.create_data_product()
# Generate the data product and associate it to the ExternalDataset
tdom, sdom = time_series_domain()
tdom, sdom = tdom.dump(), sdom.dump()
dprod = IonObject(RT.DataProduct,
name='slocum_parsed_product',
description='parsed slocum product',
temporal_domain=tdom,
spatial_domain=sdom)
dproduct_id = dpms_cli.create_data_product(data_product=dprod,
stream_definition_id=streamdef_id)
dams_cli.assign_data_product(input_resource_id=ds_id, data_product_id=dproduct_id)
stream_id, assn = rr_cli.find_objects(subject=dproduct_id, predicate=PRED.hasStream, object_type=RT.Stream, id_only=True)
stream_id = stream_id[0]
log.info('Created resources: {0}'.format({'ExternalDataset': ds_id, 'ExternalDataProvider': ext_dprov_id, 'DataSource': ext_dsrc_id, 'DataSourceModel': ext_dsrc_model_id, 'DataProducer': dproducer_id, 'DataProduct': dproduct_id, 'Stream': stream_id}))
# Create the logger for receiving publications
_, stream_route, _ = self.create_stream_and_logger(name='slocum', stream_id=stream_id)
self.EDA_RESOURCE_ID = ds_id
self.EDA_NAME = ds_name
self.DVR_CONFIG['dh_cfg'] = {
'TESTING': True,
'stream_id': stream_id,
'stream_route': stream_route,
'stream_def': streamdef_id,
'external_dataset_res': dset,
'data_producer_id': dproducer_id, # CBM: Should this be put in the main body of the config - with mod & cls?
'max_records': 20,
}
def _setup_resources(self):
# TODO: some or all of this (or some variation) should move to DAMS'
# Build the test resources for the dataset
dms_cli = DatasetManagementServiceClient()
dams_cli = DataAcquisitionManagementServiceClient()
dpms_cli = DataProductManagementServiceClient()
rr_cli = ResourceRegistryServiceClient()
pubsub_cli = PubsubManagementServiceClient()
eda = ExternalDatasetAgent(name='example eda',handler_module=self.DVR_CONFIG['dvr_mod'],
handler_class=self.DVR_CONFIG['dvr_cls'])
eda_id = dams_cli.create_external_dataset_agent(eda)
eda_inst = ExternalDatasetAgentInstance(name='example eda instance')
eda_inst_id = dams_cli.create_external_dataset_agent_instance(eda_inst,
external_dataset_agent_id=eda_id)
# Create and register the necessary resources/objects
# Create DataProvider
dprov = ExternalDataProvider(name='example data provider', institution=Institution(),
contact=ContactInformation())
dprov.contact.individual_names_given = 'Christopher Mueller'
dprov.contact.email = '*****@*****.**'
# Create DataSource
dsrc = DataSource(name='example datasource', protocol_type='DAP', institution=Institution(),
contact=ContactInformation())
dsrc.connection_params['base_data_url'] = ''
dsrc.contact.individual_names_given = 'Tim Giguere'
dsrc.contact.email = '*****@*****.**'
# Create ExternalDataset
ds_name = 'usgs_test_dataset'
dset = ExternalDataset(name=ds_name,
dataset_description=DatasetDescription(),
update_description=UpdateDescription(),
contact=ContactInformation())
# The usgs.nc test dataset is a download of the R1 dataset found here:
# http://thredds-test.oceanobservatories.org/thredds/dodsC/ooiciData/E66B1A74-A684-454A-9ADE-8388C2C634E5.ncml
dset.dataset_description.parameters['dataset_path'] = 'test_data/usgs.nc'
dset.dataset_description.parameters['temporal_dimension'] = 'time'
dset.dataset_description.parameters['zonal_dimension'] = 'lon'
dset.dataset_description.parameters['meridional_dimension'] = 'lat'
dset.dataset_description.parameters['vertical_dimension'] = 'z'
dset.dataset_description.parameters['variables'] = [
'water_temperature',
'streamflow',
'water_temperature_bottom',
'water_temperature_middle',
'specific_conductance',
'data_qualifier', ]
# Create DataSourceModel
dsrc_model = DataSourceModel(name='dap_model')
#dsrc_model.model = 'DAP'
dsrc_model.data_handler_module = 'N/A'
dsrc_model.data_handler_class = 'N/A'
## Run everything through DAMS
ds_id = dams_cli.create_external_dataset(external_dataset=dset)
ext_dprov_id = dams_cli.create_external_data_provider(external_data_provider=dprov)
ext_dsrc_id = dams_cli.create_data_source(data_source=dsrc)
ext_dsrc_model_id = dams_cli.create_data_source_model(dsrc_model)
# Register the ExternalDataset
dproducer_id = dams_cli.register_external_data_set(external_dataset_id=ds_id)
# Or using each method
dams_cli.assign_data_source_to_external_data_provider(data_source_id=ext_dsrc_id, external_data_provider_id=ext_dprov_id)
dams_cli.assign_data_source_to_data_model(data_source_id=ext_dsrc_id, data_source_model_id=ext_dsrc_model_id)
dams_cli.assign_external_dataset_to_data_source(external_dataset_id=ds_id, data_source_id=ext_dsrc_id)
dams_cli.assign_external_dataset_to_agent_instance(external_dataset_id=ds_id, agent_instance_id=eda_inst_id)
# dams_cli.assign_external_data_agent_to_agent_instance(external_data_agent_id=self.eda_id, agent_instance_id=self.eda_inst_id)
#create temp streamdef so the data product can create the stream
pc_list = []
#Get 'time' parameter context
pc_list.append(dms_cli.read_parameter_context_by_name('time', id_only=True))
for pc_k, pc in self._create_parameter_dictionary().iteritems():
pc_list.append(dms_cli.create_parameter_context(pc_k, pc[1].dump()))
pdict_id = dms_cli.create_parameter_dictionary('netcdf_param_dict', pc_list)
#create temp streamdef so the data product can create the stream
streamdef_id = pubsub_cli.create_stream_definition(name="netcdf", description="netcdf", parameter_dictionary_id=pdict_id)
tdom, sdom = time_series_domain()
tdom, sdom = tdom.dump(), sdom.dump()
dprod = IonObject(RT.DataProduct,
name='usgs_parsed_product',
description='parsed usgs product',
temporal_domain=tdom,
spatial_domain=sdom)
# Generate the data product and associate it to the ExternalDataset
dproduct_id = dpms_cli.create_data_product(data_product=dprod,
stream_definition_id=streamdef_id)
dams_cli.assign_data_product(input_resource_id=ds_id, data_product_id=dproduct_id)
stream_id, assn = rr_cli.find_objects(subject=dproduct_id, predicate=PRED.hasStream, object_type=RT.Stream, id_only=True)
stream_id = stream_id[0]
log.info('Created resources: {0}'.format({'ExternalDataset': ds_id, 'ExternalDataProvider': ext_dprov_id, 'DataSource': ext_dsrc_id, 'DataSourceModel': ext_dsrc_model_id, 'DataProducer': dproducer_id, 'DataProduct': dproduct_id, 'Stream': stream_id}))
# Create the logger for receiving publications
_, stream_route, _ = self.create_stream_and_logger(name='usgs', stream_id=stream_id)
self.EDA_RESOURCE_ID = ds_id
self.EDA_NAME = ds_name
self.DVR_CONFIG['dh_cfg'] = {
'TESTING': True,
'stream_id': stream_id,
'stream_route': stream_route,
'stream_def': streamdef_id,
'data_producer_id': dproducer_id, # CBM: Should this be put in the main body of the config - with mod & cls?
'max_records': 1,
}
def _setup_resources(self):
# TODO: some or all of this (or some variation) should move to DAMS'
# Build the test resources for the dataset
dms_cli = DatasetManagementServiceClient()
dams_cli = DataAcquisitionManagementServiceClient()
dpms_cli = DataProductManagementServiceClient()
rr_cli = ResourceRegistryServiceClient()
pubsub_cli = PubsubManagementServiceClient()
eda = ExternalDatasetAgent(name='example data agent',
handler_module=self.DVR_CONFIG['dvr_mod'],
handler_class=self.DVR_CONFIG['dvr_cls'])
eda_id = dams_cli.create_external_dataset_agent(eda)
eda_inst = ExternalDatasetAgentInstance(
name='example dataset agent instance')
eda_inst_id = dams_cli.create_external_dataset_agent_instance(
eda_inst, external_dataset_agent_id=eda_id)
# Create and register the necessary resources/objects
# Create DataProvider
dprov = ExternalDataProvider(name='example data provider',
institution=Institution(),
contact=ContactInformation())
dprov.contact.individual_names_given = 'Christopher Mueller'
dprov.contact.email = '*****@*****.**'
# Create DataSource
dsrc = DataSource(name='example datasource',
protocol_type='FILE',
institution=Institution(),
contact=ContactInformation())
dsrc.connection_params['base_data_url'] = ''
dsrc.contact.individual_names_given = 'Tim Giguere'
dsrc.contact.email = '*****@*****.**'
# Create ExternalDataset
ds_name = 'ruv_test_dataset'
dset = ExternalDataset(name=ds_name,
dataset_description=DatasetDescription(),
update_description=UpdateDescription(),
contact=ContactInformation())
dset.dataset_description.parameters['base_url'] = 'test_data/ruv/'
dset.dataset_description.parameters[
'list_pattern'] = 'RDLi_SEAB_2011_08_24_1600.ruv'
dset.dataset_description.parameters['date_pattern'] = '%Y %m %d %H %M'
dset.dataset_description.parameters[
'date_extraction_pattern'] = 'RDLi_SEAB_([\d]{4})_([\d]{2})_([\d]{2})_([\d]{2})([\d]{2}).ruv'
dset.dataset_description.parameters['temporal_dimension'] = None
dset.dataset_description.parameters['zonal_dimension'] = None
dset.dataset_description.parameters['meridional_dimension'] = None
dset.dataset_description.parameters['vertical_dimension'] = None
dset.dataset_description.parameters['variables'] = []
# Create DataSourceModel
dsrc_model = DataSourceModel(name='ruv_model')
#dsrc_model.model = 'RUV'
dsrc_model.data_handler_module = 'N/A'
dsrc_model.data_handler_class = 'N/A'
## Run everything through DAMS
ds_id = dams_cli.create_external_dataset(external_dataset=dset)
ext_dprov_id = dams_cli.create_external_data_provider(
external_data_provider=dprov)
ext_dsrc_id = dams_cli.create_data_source(data_source=dsrc)
ext_dsrc_model_id = dams_cli.create_data_source_model(dsrc_model)
# Register the ExternalDataset
dproducer_id = dams_cli.register_external_data_set(
external_dataset_id=ds_id)
# Or using each method
dams_cli.assign_data_source_to_external_data_provider(
data_source_id=ext_dsrc_id, external_data_provider_id=ext_dprov_id)
dams_cli.assign_data_source_to_data_model(
data_source_id=ext_dsrc_id, data_source_model_id=ext_dsrc_model_id)
dams_cli.assign_external_dataset_to_data_source(
external_dataset_id=ds_id, data_source_id=ext_dsrc_id)
dams_cli.assign_external_dataset_to_agent_instance(
external_dataset_id=ds_id, agent_instance_id=eda_inst_id)
pdict = ParameterDictionary()
t_ctxt = ParameterContext(
'data',
param_type=QuantityType(value_encoding=numpy.dtype('int64')))
t_ctxt.axis = AxisTypeEnum.TIME
t_ctxt.uom = 'seconds since 01-01-1970'
pdict.add_context(t_ctxt)
#create temp streamdef so the data product can create the stream
pc_list = []
for pc_k, pc in pdict.iteritems():
pc_list.append(dms_cli.create_parameter_context(
pc_k, pc[1].dump()))
pdict_id = dms_cli.create_parameter_dictionary('ruv_param_dict',
pc_list)
streamdef_id = pubsub_cli.create_stream_definition(
name="ruv",
description="stream def for ruv testing",
parameter_dictionary_id=pdict_id)
dprod = IonObject(RT.DataProduct,
name='ruv_parsed_product',
description='parsed ruv product')
# Generate the data product and associate it to the ExternalDataset
dproduct_id = dpms_cli.create_data_product(
data_product=dprod, stream_definition_id=streamdef_id)
dams_cli.assign_data_product(input_resource_id=ds_id,
data_product_id=dproduct_id)
stream_id, assn = rr_cli.find_objects(subject=dproduct_id,
predicate=PRED.hasStream,
object_type=RT.Stream,
id_only=True)
stream_id = stream_id[0]
log.info('Created resources: {0}'.format({
'ExternalDataset': ds_id,
'ExternalDataProvider': ext_dprov_id,
'DataSource': ext_dsrc_id,
'DataSourceModel': ext_dsrc_model_id,
'DataProducer': dproducer_id,
'DataProduct': dproduct_id,
'Stream': stream_id
}))
#CBM: Eventually, probably want to group this crap somehow - not sure how yet...
# Create the logger for receiving publications
_, stream_route, _ = self.create_stream_and_logger(name='ruv',
stream_id=stream_id)
self.EDA_RESOURCE_ID = ds_id
self.EDA_NAME = ds_name
self.DVR_CONFIG['dh_cfg'] = {
'TESTING': True,
'stream_id': stream_id,
'stream_route': stream_route,
'external_dataset_res': dset,
'param_dictionary': pdict.dump(),
'data_producer_id':
dproducer_id, # CBM: Should this be put in the main body of the config - with mod & cls?
'max_records': 20,
}
def _setup_hfr(self):
# TODO: some or all of this (or some variation) should move to DAMS
# Create and register the necessary resources/objects
eda = ExternalDatasetAgent()
eda_id = self.dams_cli.create_external_dataset_agent(eda)
eda_inst = ExternalDatasetAgentInstance()
eda_inst_id = self.dams_cli.create_external_dataset_agent_instance(
eda_inst, external_dataset_agent_id=eda_id)
# Create DataProvider
dprov = ExternalDataProvider(institution=Institution(),
contact=ContactInformation())
# dprov.institution.name = "HFR UCSD"
# Create DataSource
dsrc = DataSource(protocol_type="DAP",
institution=Institution(),
contact=ContactInformation())
dsrc.connection_params[
"base_data_url"] = "http://hfrnet.ucsd.edu:8080/thredds/dodsC/"
# Create ExternalDataset
dset = ExternalDataset(name="UCSD HFR",
dataset_description=DatasetDescription(),
update_description=UpdateDescription(),
contact=ContactInformation())
dset.dataset_description.parameters[
"dataset_path"] = "HFRNet/USEGC/6km/hourly/RTV"
# dset.dataset_description.parameters["dataset_path"] = "test_data/hfr.nc"
dset.dataset_description.parameters["temporal_dimension"] = "time"
dset.dataset_description.parameters["zonal_dimension"] = "lon"
dset.dataset_description.parameters["meridional_dimension"] = "lat"
# Create DataSourceModel
dsrc_model = DataSourceModel(name="dap_model")
dsrc_model.model = "DAP"
dsrc_model.data_handler_module = "eoi.agent.handler.dap_external_data_handler"
dsrc_model.data_handler_class = "DapExternalDataHandler"
## Run everything through DAMS
ds_id = self.hfr_ds_id = self.dams_cli.create_external_dataset(
external_dataset=dset)
ext_dprov_id = self.dams_cli.create_external_data_provider(
external_data_provider=dprov)
ext_dsrc_id = self.dams_cli.create_data_source(data_source=dsrc)
ext_dsrc_model_id = self.dams_cli.create_data_source_model(dsrc_model)
# Register the ExternalDataset
dproducer_id = self.dams_cli.register_external_data_set(
external_dataset_id=ds_id)
## Associate everything
# Convenience method
# self.dams_cli.assign_eoi_resources(external_data_provider_id=ext_dprov_id, data_source_id=ext_dsrc_id, data_source_model_id=ext_dsrc_model_id, external_dataset_id=ds_id, external_data_agent_id=eda_id, agent_instance_id=eda_inst_id)
# Or using each method
self.dams_cli.assign_data_source_to_external_data_provider(
data_source_id=ext_dsrc_id, external_data_provider_id=ext_dprov_id)
self.dams_cli.assign_data_source_to_data_model(
data_source_id=ext_dsrc_id, data_source_model_id=ext_dsrc_model_id)
self.dams_cli.assign_external_dataset_to_data_source(
external_dataset_id=ds_id, data_source_id=ext_dsrc_id)
self.dams_cli.assign_external_dataset_to_agent_instance(
external_dataset_id=ds_id, agent_instance_id=eda_inst_id)
# self.dams_cli.assign_external_dataset_agent_to_data_model(external_data_agent_id=eda_id, data_source_model_id=ext_dsrc_model_id)
# self.dams_cli.assign_external_data_agent_to_agent_instance(external_data_agent_id=eda_id, agent_instance_id=eda_inst_id)
# Generate the data product and associate it to the ExternalDataset
dprod = DataProduct(name='hfr_product', description='raw hfr product')
dproduct_id = self.dpms_cli.create_data_product(data_product=dprod)
self.dams_cli.assign_data_product(input_resource_id=ds_id,
data_product_id=dproduct_id,
create_stream=True)
def _setup_ncom(self):
# TODO: some or all of this (or some variation) should move to DAMS
# Create and register the necessary resources/objects
eda = ExternalDatasetAgent()
eda_id = self.dams_cli.create_external_dataset_agent(eda)
eda_inst = ExternalDatasetAgentInstance()
eda_inst_id = self.dams_cli.create_external_dataset_agent_instance(
eda_inst, external_dataset_agent_id=eda_id)
# Create DataProvider
dprov = ExternalDataProvider(institution=Institution(),
contact=ContactInformation())
# dprov.institution.name = "OOI CGSN"
dprov.contact.name = "Robert Weller"
dprov.contact.email = "*****@*****.**"
# Create DataSource
dsrc = DataSource(protocol_type="DAP",
institution=Institution(),
contact=ContactInformation())
# dsrc.connection_params["base_data_url"] = "http://ooi.whoi.edu/thredds/dodsC/"
dsrc.connection_params["base_data_url"] = ""
dsrc.contact.name = "Rich Signell"
dsrc.contact.email = "*****@*****.**"
# Create ExternalDataset
dset = ExternalDataset(name="test",
dataset_description=DatasetDescription(),
update_description=UpdateDescription(),
contact=ContactInformation())
# dset.dataset_description.parameters["dataset_path"] = "ooi/AS02CPSM_R_M.nc"
dset.dataset_description.parameters[
"dataset_path"] = "test_data/ncom.nc"
dset.dataset_description.parameters["temporal_dimension"] = "time"
dset.dataset_description.parameters["zonal_dimension"] = "lon"
dset.dataset_description.parameters["meridional_dimension"] = "lat"
# Create DataSourceModel
dsrc_model = DataSourceModel(name="dap_model")
dsrc_model.model = "DAP"
dsrc_model.data_handler_module = "eoi.agent.handler.dap_external_data_handler"
dsrc_model.data_handler_class = "DapExternalDataHandler"
## Run everything through DAMS
ds_id = self.ncom_ds_id = self.dams_cli.create_external_dataset(
external_dataset=dset)
ext_dprov_id = self.dams_cli.create_external_data_provider(
external_data_provider=dprov)
ext_dsrc_id = self.dams_cli.create_data_source(data_source=dsrc)
ext_dsrc_model_id = self.dams_cli.create_data_source_model(dsrc_model)
# Register the ExternalDataset
dproducer_id = self.dams_cli.register_external_data_set(
external_dataset_id=ds_id)
## Associate everything
# Convenience method
# self.dams_cli.assign_eoi_resources(external_data_provider_id=ext_dprov_id, data_source_id=ext_dsrc_id, data_source_model_id=ext_dsrc_model_id, external_dataset_id=ds_id, external_data_agent_id=eda_id, agent_instance_id=eda_inst_id)
# Or using each method
self.dams_cli.assign_data_source_to_external_data_provider(
data_source_id=ext_dsrc_id, external_data_provider_id=ext_dprov_id)
self.dams_cli.assign_data_source_to_data_model(
data_source_id=ext_dsrc_id, data_source_model_id=ext_dsrc_model_id)
self.dams_cli.assign_external_dataset_to_data_source(
external_dataset_id=ds_id, data_source_id=ext_dsrc_id)
self.dams_cli.assign_external_dataset_to_agent_instance(
external_dataset_id=ds_id, agent_instance_id=eda_inst_id)
# self.dams_cli.assign_external_dataset_agent_to_data_model(external_data_agent_id=eda_id, data_source_model_id=ext_dsrc_model_id)
# self.dams_cli.assign_external_data_agent_to_agent_instance(external_data_agent_id=eda_id, agent_instance_id=eda_inst_id)
# Generate the data product and associate it to the ExternalDataset
dprod = DataProduct(name='ncom_product',
description='raw ncom product')
dproduct_id = self.dpms_cli.create_data_product(data_product=dprod)
self.dams_cli.assign_data_product(input_resource_id=ds_id,
data_product_id=dproduct_id,
create_stream=True)
def _setup_resources(self):
# TODO: some or all of this (or some variation) should move to DAMS'
# Build the test resources for the dataset
dams_cli = DataAcquisitionManagementServiceClient()
dpms_cli = DataProductManagementServiceClient()
rr_cli = ResourceRegistryServiceClient()
eda = ExternalDatasetAgent()
eda_id = dams_cli.create_external_dataset_agent(eda)
eda_inst = ExternalDatasetAgentInstance()
eda_inst_id = dams_cli.create_external_dataset_agent_instance(eda_inst, external_dataset_agent_id=eda_id)
# Create and register the necessary resources/objects
# Create DataProvider
dprov = ExternalDataProvider(institution=Institution(), contact=ContactInformation())
dprov.contact.name = 'Christopher Mueller'
dprov.contact.email = '*****@*****.**'
# Create DataSource
dsrc = DataSource(protocol_type='FILE', institution=Institution(), contact=ContactInformation())
dsrc.connection_params['base_data_url'] = ''
dsrc.contact.name='Tim Giguere'
dsrc.contact.email = '*****@*****.**'
# Create ExternalDataset
ds_name = 'slocum_test_dataset'
dset = ExternalDataset(name=ds_name, dataset_description=DatasetDescription(), update_description=UpdateDescription(), contact=ContactInformation())
dset.dataset_description.parameters['dataset_path'] = 'test_data/ru05-2012-021-0-0-sbd.dat'
dset.dataset_description.parameters['temporal_dimension'] = None
dset.dataset_description.parameters['zonal_dimension'] = None
dset.dataset_description.parameters['meridional_dimension'] = None
dset.dataset_description.parameters['vertical_dimension'] = None
dset.dataset_description.parameters['variables'] = [
'c_wpt_y_lmc',
'sci_water_cond',
'm_y_lmc',
'u_hd_fin_ap_inflection_holdoff',
'sci_m_present_time',
'm_leakdetect_voltage_forward',
'sci_bb3slo_b660_scaled',
'c_science_send_all',
'm_gps_status',
'm_water_vx',
'm_water_vy',
'c_heading',
'sci_fl3slo_chlor_units',
'u_hd_fin_ap_gain',
'm_vacuum',
'u_min_water_depth',
'm_gps_lat',
'm_veh_temp',
'f_fin_offset',
'u_hd_fin_ap_hardover_holdoff',
'c_alt_time',
'm_present_time',
'm_heading',
'sci_bb3slo_b532_scaled',
'sci_fl3slo_cdom_units',
'm_fin',
'x_cycle_overrun_in_ms',
'sci_water_pressure',
'u_hd_fin_ap_igain',
'sci_fl3slo_phyco_units',
'm_battpos',
'sci_bb3slo_b470_scaled',
'm_lat',
'm_gps_lon',
'sci_ctd41cp_timestamp',
'm_pressure',
'c_wpt_x_lmc',
'c_ballast_pumped',
'x_lmc_xy_source',
'm_lon',
'm_avg_speed',
'sci_water_temp',
'u_pitch_ap_gain',
'm_roll',
'm_tot_num_inflections',
'm_x_lmc',
'u_pitch_ap_deadband',
'm_final_water_vy',
'm_final_water_vx',
'm_water_depth',
'm_leakdetect_voltage',
'u_pitch_max_delta_battpos',
'm_coulomb_amphr',
'm_pitch',
]
# Create DataSourceModel
dsrc_model = DataSourceModel(name='slocum_model')
dsrc_model.model = 'SLOCUM'
dsrc_model.data_handler_module = 'N/A'
dsrc_model.data_handler_class = 'N/A'
## Run everything through DAMS
ds_id = dams_cli.create_external_dataset(external_dataset=dset)
ext_dprov_id = dams_cli.create_external_data_provider(external_data_provider=dprov)
ext_dsrc_id = dams_cli.create_data_source(data_source=dsrc)
ext_dsrc_model_id = dams_cli.create_data_source_model(dsrc_model)
# Register the ExternalDataset
dproducer_id = dams_cli.register_external_data_set(external_dataset_id=ds_id)
# Or using each method
dams_cli.assign_data_source_to_external_data_provider(data_source_id=ext_dsrc_id, external_data_provider_id=ext_dprov_id)
dams_cli.assign_data_source_to_data_model(data_source_id=ext_dsrc_id, data_source_model_id=ext_dsrc_model_id)
dams_cli.assign_external_dataset_to_data_source(external_dataset_id=ds_id, data_source_id=ext_dsrc_id)
dams_cli.assign_external_dataset_to_agent_instance(external_dataset_id=ds_id, agent_instance_id=eda_inst_id)
# dams_cli.assign_external_data_agent_to_agent_instance(external_data_agent_id=self.eda_id, agent_instance_id=self.eda_inst_id)
# Generate the data product and associate it to the ExternalDataset
dprod = DataProduct(name='slocum_parsed_product', description='parsed slocum product')
dproduct_id = dpms_cli.create_data_product(data_product=dprod)
dams_cli.assign_data_product(input_resource_id=ds_id, data_product_id=dproduct_id, create_stream=True)
stream_id, assn = rr_cli.find_objects(subject=dproduct_id, predicate=PRED.hasStream, object_type=RT.Stream, id_only=True)
stream_id = stream_id[0]
log.info('Created resources: {0}'.format({'ExternalDataset':ds_id, 'ExternalDataProvider':ext_dprov_id, 'DataSource':ext_dsrc_id, 'DataSourceModel':ext_dsrc_model_id, 'DataProducer':dproducer_id, 'DataProduct':dproduct_id, 'Stream':stream_id}))
#CBM: Use CF standard_names
ttool = TaxyTool()
ttool.add_taxonomy_set('c_wpt_y_lmc'),
ttool.add_taxonomy_set('sci_water_cond'),
ttool.add_taxonomy_set('m_y_lmc'),
ttool.add_taxonomy_set('u_hd_fin_ap_inflection_holdoff'),
ttool.add_taxonomy_set('sci_m_present_time'),
ttool.add_taxonomy_set('m_leakdetect_voltage_forward'),
ttool.add_taxonomy_set('sci_bb3slo_b660_scaled'),
ttool.add_taxonomy_set('c_science_send_all'),
ttool.add_taxonomy_set('m_gps_status'),
ttool.add_taxonomy_set('m_water_vx'),
ttool.add_taxonomy_set('m_water_vy'),
ttool.add_taxonomy_set('c_heading'),
ttool.add_taxonomy_set('sci_fl3slo_chlor_units'),
ttool.add_taxonomy_set('u_hd_fin_ap_gain'),
ttool.add_taxonomy_set('m_vacuum'),
ttool.add_taxonomy_set('u_min_water_depth'),
ttool.add_taxonomy_set('m_gps_lat'),
ttool.add_taxonomy_set('m_veh_temp'),
ttool.add_taxonomy_set('f_fin_offset'),
ttool.add_taxonomy_set('u_hd_fin_ap_hardover_holdoff'),
ttool.add_taxonomy_set('c_alt_time'),
ttool.add_taxonomy_set('m_present_time'),
ttool.add_taxonomy_set('m_heading'),
ttool.add_taxonomy_set('sci_bb3slo_b532_scaled'),
ttool.add_taxonomy_set('sci_fl3slo_cdom_units'),
ttool.add_taxonomy_set('m_fin'),
ttool.add_taxonomy_set('x_cycle_overrun_in_ms'),
ttool.add_taxonomy_set('sci_water_pressure'),
ttool.add_taxonomy_set('u_hd_fin_ap_igain'),
ttool.add_taxonomy_set('sci_fl3slo_phyco_units'),
ttool.add_taxonomy_set('m_battpos'),
ttool.add_taxonomy_set('sci_bb3slo_b470_scaled'),
ttool.add_taxonomy_set('m_lat'),
ttool.add_taxonomy_set('m_gps_lon'),
ttool.add_taxonomy_set('sci_ctd41cp_timestamp'),
ttool.add_taxonomy_set('m_pressure'),
ttool.add_taxonomy_set('c_wpt_x_lmc'),
ttool.add_taxonomy_set('c_ballast_pumped'),
ttool.add_taxonomy_set('x_lmc_xy_source'),
ttool.add_taxonomy_set('m_lon'),
ttool.add_taxonomy_set('m_avg_speed'),
ttool.add_taxonomy_set('sci_water_temp'),
ttool.add_taxonomy_set('u_pitch_ap_gain'),
ttool.add_taxonomy_set('m_roll'),
ttool.add_taxonomy_set('m_tot_num_inflections'),
ttool.add_taxonomy_set('m_x_lmc'),
ttool.add_taxonomy_set('u_pitch_ap_deadband'),
ttool.add_taxonomy_set('m_final_water_vy'),
ttool.add_taxonomy_set('m_final_water_vx'),
ttool.add_taxonomy_set('m_water_depth'),
ttool.add_taxonomy_set('m_leakdetect_voltage'),
ttool.add_taxonomy_set('u_pitch_max_delta_battpos'),
ttool.add_taxonomy_set('m_coulomb_amphr'),
ttool.add_taxonomy_set('m_pitch'),
#CBM: Eventually, probably want to group this crap somehow - not sure how yet...
# Create the logger for receiving publications
self.create_stream_and_logger(name='slocum',stream_id=stream_id)
self.EDA_RESOURCE_ID = ds_id
self.EDA_NAME = ds_name
self.DVR_CONFIG['dh_cfg'] = {
'TESTING':True,
'stream_id':stream_id,
'external_dataset_res':dset,
'taxonomy':ttool.dump(),
'data_producer_id':dproducer_id,#CBM: Should this be put in the main body of the config - with mod & cls?
'max_records':20,
}
def _setup_resources(self):
# TODO: some or all of this (or some variation) should move to DAMS'
# Build the test resources for the dataset
dams_cli = DataAcquisitionManagementServiceClient()
dpms_cli = DataProductManagementServiceClient()
rr_cli = ResourceRegistryServiceClient()
eda = ExternalDatasetAgent()
eda_id = dams_cli.create_external_dataset_agent(eda)
eda_inst = ExternalDatasetAgentInstance()
eda_inst_id = dams_cli.create_external_dataset_agent_instance(
eda_inst, external_dataset_agent_id=eda_id)
# Create and register the necessary resources/objects
# Create DataProvider
dprov = ExternalDataProvider(institution=Institution(),
contact=ContactInformation())
dprov.contact.name = 'Christopher Mueller'
dprov.contact.email = '*****@*****.**'
# Create DataSource
dsrc = DataSource(protocol_type='DAP',
institution=Institution(),
contact=ContactInformation())
dsrc.connection_params['base_data_url'] = ''
dsrc.contact.name = 'Tim Giguere'
dsrc.contact.email = '*****@*****.**'
# Create ExternalDataset
ds_name = 'usgs_test_dataset'
dset = ExternalDataset(name=ds_name,
dataset_description=DatasetDescription(),
update_description=UpdateDescription(),
contact=ContactInformation())
# The usgs.nc test dataset is a download of the R1 dataset found here:
# http://thredds-test.oceanobservatories.org/thredds/dodsC/ooiciData/E66B1A74-A684-454A-9ADE-8388C2C634E5.ncml
dset.dataset_description.parameters[
'dataset_path'] = 'test_data/usgs.nc'
dset.dataset_description.parameters['temporal_dimension'] = 'time'
dset.dataset_description.parameters['zonal_dimension'] = 'lon'
dset.dataset_description.parameters['meridional_dimension'] = 'lat'
dset.dataset_description.parameters['vertical_dimension'] = 'z'
dset.dataset_description.parameters['variables'] = [
'water_temperature',
'streamflow',
'water_temperature_bottom',
'water_temperature_middle',
'specific_conductance',
'data_qualifier',
]
# Create DataSourceModel
dsrc_model = DataSourceModel(name='dap_model')
dsrc_model.model = 'DAP'
dsrc_model.data_handler_module = 'N/A'
dsrc_model.data_handler_class = 'N/A'
## Run everything through DAMS
ds_id = dams_cli.create_external_dataset(external_dataset=dset)
ext_dprov_id = dams_cli.create_external_data_provider(
external_data_provider=dprov)
ext_dsrc_id = dams_cli.create_data_source(data_source=dsrc)
ext_dsrc_model_id = dams_cli.create_data_source_model(dsrc_model)
# Register the ExternalDataset
dproducer_id = dams_cli.register_external_data_set(
external_dataset_id=ds_id)
# Or using each method
dams_cli.assign_data_source_to_external_data_provider(
data_source_id=ext_dsrc_id, external_data_provider_id=ext_dprov_id)
dams_cli.assign_data_source_to_data_model(
data_source_id=ext_dsrc_id, data_source_model_id=ext_dsrc_model_id)
dams_cli.assign_external_dataset_to_data_source(
external_dataset_id=ds_id, data_source_id=ext_dsrc_id)
dams_cli.assign_external_dataset_to_agent_instance(
external_dataset_id=ds_id, agent_instance_id=eda_inst_id)
# dams_cli.assign_external_data_agent_to_agent_instance(external_data_agent_id=self.eda_id, agent_instance_id=self.eda_inst_id)
# Generate the data product and associate it to the ExternalDataset
dprod = DataProduct(name='usgs_parsed_product',
description='parsed usgs product')
dproduct_id = dpms_cli.create_data_product(data_product=dprod)
dams_cli.assign_data_product(input_resource_id=ds_id,
data_product_id=dproduct_id,
create_stream=True)
stream_id, assn = rr_cli.find_objects(subject=dproduct_id,
predicate=PRED.hasStream,
object_type=RT.Stream,
id_only=True)
stream_id = stream_id[0]
log.info('Created resources: {0}'.format({
'ExternalDataset': ds_id,
'ExternalDataProvider': ext_dprov_id,
'DataSource': ext_dsrc_id,
'DataSourceModel': ext_dsrc_model_id,
'DataProducer': dproducer_id,
'DataProduct': dproduct_id,
'Stream': stream_id
}))
#CBM: Use CF standard_names
ttool = TaxyTool()
ttool.add_taxonomy_set('time', 'time')
ttool.add_taxonomy_set('lon', 'longitude')
ttool.add_taxonomy_set('lat', 'latitude')
ttool.add_taxonomy_set('z', 'water depth')
ttool.add_taxonomy_set('water_temperature',
'average water temperature')
ttool.add_taxonomy_set('water_temperature_bottom',
'water temperature at bottom of water column')
ttool.add_taxonomy_set('water_temperature_middle',
'water temperature at middle of water column')
ttool.add_taxonomy_set('streamflow', 'flow velocity of stream')
ttool.add_taxonomy_set('specific_conductance',
'specific conductance of water')
ttool.add_taxonomy_set('data_qualifier', 'data qualifier flag')
ttool.add_taxonomy_set('coords',
'This group contains coordinate parameters')
ttool.add_taxonomy_set('data', 'This group contains data parameters')
# Create the logger for receiving publications
self.create_stream_and_logger(name='usgs', stream_id=stream_id)
self.EDA_RESOURCE_ID = ds_id
self.EDA_NAME = ds_name
self.DVR_CONFIG['dh_cfg'] = {
'TESTING': True,
'stream_id': stream_id,
'taxonomy': ttool.dump(),
'data_producer_id':
dproducer_id, #CBM: Should this be put in the main body of the config - with mod & cls?
'max_records': 4,
}