def profiling_column(self, variable_metadata: VariableMetadata,
column: pd.Series) -> VariableMetadata:
"""Profiling single column for necessary fields of metadata, if data is present .
Args:
variable_metadata: the original VariableMetadata instance.
column: the column to profile.
Returns:
profiled VariableMetadata instance
"""
if not variable_metadata.name:
variable_metadata.name = column.name
if not variable_metadata.description:
variable_metadata.description = self.profiler.construct_variable_description(
column)
if variable_metadata.named_entity is None:
variable_metadata.named_entity = self.profiler.profile_named_entity(
column)
if variable_metadata.temporal_coverage:
if variable_metadata.temporal_coverage[
'start'] or not variable_metadata.temporal_coverage['end']:
variable_metadata.temporal_coverage = self.profiler.profile_temporal_coverage(
variable_metadata.temporal_coverage, column)
return variable_metadata
def setUp(self):
self.variable_1 = copy.deepcopy(sample_variable_1)
self.variable_2 = copy.deepcopy(sample_variable_2)
self.metadata_1 = VariableMetadata(description=self.variable_1,
datamart_id=0)
self.metadata_2 = VariableMetadata(description=self.variable_2,
datamart_id=10)
def construct_variable_metadata(
self,
description: dict,
global_datamart_id: int,
col_offset: int,
data: pd.DataFrame = None) -> VariableMetadata:
"""Construct variable metadata.
Args:
description: description dict.
global_datamart_id: integer of datamart id.
col_offset: integer, the column index.
data: dataframe of data.
Returns:
VariableMetadata instance
"""
variable_metadata = VariableMetadata.construct_variable(
description, datamart_id=col_offset + global_datamart_id + 1)
if data is not None:
variable_metadata = self._profiling_column(
description, variable_metadata, data.iloc[:, col_offset])
return variable_metadata
def test_add_variable(self):
self.assertEqual(len(self.metadata.variables), 0)
for col_offset, variable_description in enumerate(
self.global_metadata_description["variables"]):
variable_metadata = VariableMetadata(variable_description,
datamart_id=col_offset + 1)
self.metadata.add_variable_metadata(variable_metadata)
self.assertEqual(len(self.metadata.variables),
len(sample_global_metadata_description["variables"]))
self.assertEqual(self.metadata.value, gt["metadata"])
def test_add_variable(self):
print("[Test]{}/test_add_variable".format(self.__class__.__name__))
self.assertEqual(len(self.metadata.variables), 0)
for col_offset, variable_description in enumerate(
self.global_metadata_description["variables"]):
variable_metadata = VariableMetadata(variable_description,
datamart_id=col_offset + 1)
self.metadata.add_variable_metadata(variable_metadata)
self.assertEqual(len(self.metadata.variables),
len(sample_global_metadata_description["variables"]))
self.assertEqual(self.metadata.value, gt["metadata"])
print(colored('.Done', 'red'))
def basic_profiling_column(cls, description: dict,
variable_metadata: VariableMetadata,
column: pd.Series) -> VariableMetadata:
"""Profiling single column for necessary fields of metadata, if data is present .
Args:
description: description dict about the column.
variable_metadata: the original VariableMetadata instance.
column: the column to profile.
Returns:
profiled VariableMetadata instance
"""
if not variable_metadata.name:
variable_metadata.name = str(column.name)
if not variable_metadata.description:
variable_metadata.description = cls.construct_variable_description(
column)
if variable_metadata.named_entity is None:
variable_metadata.named_entity = cls.profile_named_entity(column)
elif variable_metadata.named_entity is False and not description:
if cls.named_entity_column_recognize(column):
variable_metadata.named_entity = cls.profile_named_entity(
column)
if variable_metadata.temporal_coverage is not False:
if not variable_metadata.temporal_coverage[
'start'] or not variable_metadata.temporal_coverage['end']:
variable_metadata.temporal_coverage = cls.profile_temporal_coverage(
column=column,
coverage=variable_metadata.temporal_coverage)
elif not description:
temporal_coverage = cls.profile_temporal_coverage(column=column)
if temporal_coverage:
variable_metadata.temporal_coverage = temporal_coverage
if not variable_metadata.semantic_type:
variable_metadata.semantic_type = cls.profile_semantic_type(column)
return variable_metadata
