def test_RequestDatasetCollection(self):
env.DIR_DATA = ocgis.env.DIR_TEST_DATA
daymet = self.test_data.get_rd('daymet_tmax')
tas = self.test_data.get_rd('cancm4_tas')
uris = [daymet.uri,
tas.uri]
variables = ['foo1','foo2']
rdc = RequestDatasetCollection()
for uri,variable in zip(uris,variables):
rd = RequestDataset(uri,variable)
rdc.update(rd)
self.assertEqual([1,2],[rd.did for rd in rdc])
variables = ['foo1','foo1']
rdc = RequestDatasetCollection()
for ii,(uri,variable) in enumerate(zip(uris,variables)):
rd = RequestDataset(uri,variable)
if ii == 1:
with self.assertRaises(KeyError):
rdc.update(rd)
else:
rdc.update(rd)
aliases = ['a1','a2']
for uri,variable,alias in zip(uris,variables,aliases):
rd = RequestDataset(uri,variable,alias=alias)
rdc.update(rd)
for row in rdc:
self.assertIsInstance(row,RequestDataset)
self.assertIsInstance(rdc[0],RequestDataset)
self.assertIsInstance(rdc['a2'],RequestDataset)
def test_set_unique_id(self):
rd = self.test_data.get_rd('cancm4_tas')
field = rd.get()
for element in [rd, field]:
RequestDatasetCollection._set_unique_id_(element, 5)
uid = RequestDatasetCollection._get_unique_id_(element)
self.assertEqual(uid, 5)
def test_iter_request_datasets(self):
rd = self.test_data.get_rd('cancm4_tas')
field = rd.get()
field.name = 'foo'
rdc = RequestDatasetCollection(target=[rd, field])
tt = list(rdc.iter_request_datasets())
self.assertEqual(len(tt), 1)
self.assertEqual(len(rdc), 2)
self.assertIsInstance(tt[0], RequestDataset)
def __init__(self, init_value):
if isinstance(init_value, self.__class__):
# Allow the dataset object to be initialized by an instance of itself.
self.__dict__ = init_value.__dict__
else:
if init_value is not None:
if isinstance(init_value, RequestDatasetCollection):
init_value = deepcopy(init_value)
else:
if isinstance(init_value, (RequestDataset, dict, Field)):
itr = [init_value]
elif type(init_value) in [list, tuple]:
itr = init_value
else:
should_raise = True
try:
import ESMF
except ImportError:
# ESMF is not a required library
ocgis_lh('Could not import ESMF library.', level=logging.WARN)
else:
if isinstance(init_value, ESMF.Field):
from ocgis.regrid.base import get_ocgis_field_from_esmf_field
field = get_ocgis_field_from_esmf_field(init_value)
itr = [field]
should_raise = False
if should_raise:
raise DefinitionValidationError(self, 'Type not accepted: {0}'.format(type(init_value)))
rdc = RequestDatasetCollection()
for rd in itr:
if not isinstance(rd, Field):
rd = deepcopy(rd)
try:
rdc.update(rd)
except NoDimensionedVariablesFound:
if rd._name is None:
msg = messages.M2.format(rd.uri)
raise DefinitionValidationError(self, msg)
else:
raise
init_value = rdc
else:
init_value = init_value
super(Dataset, self).__init__(init_value)
def test_with_overloads(self):
rd = self.test_data.get_rd('cancm4_tas')
field = rd.get()
# loaded calendar should match file metadata
self.assertEqual(field.temporal.calendar, '365_day')
# the overloaded calendar in the request dataset should still be None
self.assertEqual(rd.t_calendar, None)
dataset = [{'time_region': None,
'uri': [rd.uri],
'alias': u'tas',
't_units': u'days since 1940-01-01 00:00:00',
'variable': u'tas',
't_calendar': u'will_not_work'}]
rdc = RequestDatasetCollection(dataset)
rd2 = RequestDataset(**dataset[0])
# the overloaded calendar should be passed to the request dataset
self.assertEqual(rd2.t_calendar, 'will_not_work')
self.assertEqual(rdc.first().t_calendar, 'will_not_work')
# when this bad calendar value is used it should raise an exception
with self.assertRaises(ValueError):
rdc.first().get().temporal.value_datetime
dataset = [{'time_region': None,
'uri': [rd.uri],
'alias': u'tas',
't_units': u'days since 1940-01-01 00:00:00',
'variable': u'tas'}]
rdc = RequestDatasetCollection(dataset)
# ensure the overloaded units are properly passed
self.assertEqual(rdc.first().get().temporal.units, 'days since 1940-01-01 00:00:00')
# the calendar was not overloaded and the value should be read from the metadata
self.assertEqual(rdc.first().get().temporal.calendar, '365_day')
def __init__(self,arg):
if arg is not None:
if isinstance(arg,RequestDatasetCollection):
init_value = arg
else:
if isinstance(arg,RequestDataset):
itr = [arg]
elif isinstance(arg,dict):
itr = [arg]
else:
itr = arg
rdc = RequestDatasetCollection()
for rd in itr:
rdc.update(rd)
init_value = rdc
else:
init_value = arg
super(Dataset,self).__init__(init_value)
def __init__(self,arg):
if arg is not None:
if isinstance(arg,RequestDatasetCollection):
init_value = arg
else:
if isinstance(arg,RequestDataset):
itr = [arg]
elif isinstance(arg,dict):
itr = [arg]
else:
itr = arg
rdc = RequestDatasetCollection()
for rd in itr:
rdc.update(rd)
init_value = rdc
## dereference any prior dataset connections
for rd in init_value:
rd._ds = None
else:
init_value = arg
super(Dataset,self).__init__(init_value)
def test_update(self):
rd = self.test_data.get_rd('cancm4_tas')
rd.did = 10
field = rd.get()
self.assertEqual(field.uid, 10)
field.uid = 20
rdc = RequestDatasetCollection()
rdc.update(rd)
# name is already in collection and should yield a key error
with self.assertRaises(KeyError):
rdc.update(field)
field.name = 'tas2'
rdc.update(field)
# add another object and check the increment
field2 = deepcopy(field)
field2.name = 'hanzel'
field2.uid = None
rdc.update(field2)
self.assertEqual(field2.uid, 21)
def test_get_unique_id(self):
rd = self.test_data.get_rd('cancm4_tas')
field = rd.get()
rd_did = deepcopy(rd)
rd_did.did = 1
field_uid = deepcopy(field)
field_uid.uid = 1
for element in [rd, field, rd_did, field_uid]:
uid = RequestDatasetCollection._get_unique_id_(element)
try:
self.assertEqual(uid, 1)
except AssertionError:
try:
self.assertIsNone(element.did)
except AttributeError:
self.assertIsNone(element.uid)
def test_name_attribute_used_for_keys(self):
rd = self.test_data.get_rd('cancm4_tas')
rd.name = 'hi_there'
rdc = RequestDatasetCollection(target=[rd])
self.assertEqual(rdc.keys(), ['hi_there'])
def test_get_meta_rows(self):
for k in self.iter_keywords():
rdc = RequestDatasetCollection(target=k.target)
rows = rdc._get_meta_rows_()
self.assertTrue(len(rows) >= 1)
