def _get_date_time_code_list_for_dimension(
self, column_title: str,
new_dimension: NewQbDimension) -> CompositeQbCodeList:
csvw_safe_column_title = csvw_column_name_safe(column_title)
assert isinstance(new_dimension.code_list, NewQbCodeList)
return CompositeQbCodeList(
CatalogMetadata(new_dimension.label),
[
DuplicatedQbConcept(
existing_concept_uri=uritemplate.expand(
self.value,
{csvw_safe_column_title: c.label},
),
label=c.label,
code=c.code,
) for c in new_dimension.code_list.concepts
],
)
def _get_code_list(
column_label: str,
maybe_code_list: Optional[Union[bool, str]],
info_json_parent_dir: Path,
maybe_parent_uri: Optional[str],
column_data: PandasDataTypes,
maybe_property_value_url: Optional[str],
) -> Optional[QbCodeList]:
is_date_time_code_list = (
(maybe_code_list is None or
(isinstance(maybe_code_list, bool) and maybe_code_list))
and maybe_parent_uri
== "http://purl.org/linked-data/sdmx/2009/dimension#refPeriod"
and maybe_property_value_url is not None and
maybe_property_value_url.startswith("http://reference.data.gov.uk/id/")
)
if is_date_time_code_list:
column_name_csvw_safe = csvw_column_name_safe(column_label)
columnar_data = pandas_input_to_columnar_str(column_data)
concepts = [
DuplicatedQbConcept(
existing_concept_uri=uritemplate.expand(
maybe_property_value_url, {column_name_csvw_safe: c}),
label=c,
) for c in sorted(set(columnar_data))
]
return CompositeQbCodeList(
CatalogMetadata(column_label),
concepts,
)
elif maybe_code_list is not None:
if isinstance(maybe_code_list, str):
return ExistingQbCodeList(maybe_code_list)
elif isinstance(maybe_code_list, bool) and not maybe_code_list:
return None
else:
raise Exception(f"Unexpected codelist value '{maybe_code_list}'")
return NewQbCodeListInCsvW(info_json_parent_dir / "codelists" /
f"{uri_safe(column_label)}.csv-metadata.json")
def map_to_multi_measure_dimension(
self, column_title: str,
data: PandasDataTypes) -> QbMultiMeasureDimension:
csvw_column_name = csvw_column_name_safe(column_title)
return QbMultiMeasureDimension.existing_measures_from_data(
data, csvw_column_name, self.value)
def map_to_qb_multi_units(self, data: PandasDataTypes,
column_title: str) -> QbMultiUnits:
return QbMultiUnits.existing_units_from_data(
data, csvw_column_name_safe(column_title), self.value)
def _get_column_for_metadata_config(
column_name: str,
col_config: Optional[Union[dict, bool]],
column_data: PandasDataTypes,
info_json_parent_dir: Path,
) -> CsvColumn:
if isinstance(col_config, dict):
if col_config.get("type") is not None:
return v1point1.map_column_to_qb_component(column_name, col_config,
column_data,
info_json_parent_dir)
csv_safe_column_name = csvw_column_name_safe(column_name)
maybe_dimension_uri = col_config.get("dimension")
maybe_property_value_url = col_config.get("value")
maybe_parent_uri = col_config.get("parent")
maybe_description = col_config.get("description")
maybe_label = col_config.get("label")
maybe_attribute_uri = col_config.get("attribute")
maybe_unit_uri = col_config.get("unit")
maybe_measure_uri = col_config.get("measure")
maybe_data_type = col_config.get("datatype")
if maybe_dimension_uri is not None and maybe_property_value_url is not None:
if maybe_dimension_uri == "http://purl.org/linked-data/cube#measureType":
# multi-measure cube
defined_measure_types: List[str] = col_config.get("types", [])
if maybe_property_value_url is not None:
defined_measure_types = [
uritemplate.expand(maybe_property_value_url,
{csv_safe_column_name: d})
for d in defined_measure_types
]
if len(defined_measure_types) == 0:
raise Exception(
f"Property 'types' was not defined in measure types column '{column_name}'."
)
measures = QbMultiMeasureDimension(
[ExistingQbMeasure(t) for t in defined_measure_types])
return QbColumn(column_name, measures,
maybe_property_value_url)
else:
return QbColumn(
column_name,
ExistingQbDimension(maybe_dimension_uri),
maybe_property_value_url,
)
elif (maybe_parent_uri is not None or maybe_description is not None
or maybe_label is not None):
label: str = column_name if maybe_label is None else maybe_label
code_list = _get_code_list(
label,
col_config.get("codelist"),
info_json_parent_dir,
maybe_parent_uri,
column_data,
maybe_property_value_url,
)
new_dimension = NewQbDimension(
label,
description=maybe_description,
parent_dimension_uri=maybe_parent_uri,
source_uri=col_config.get("source"),
code_list=code_list,
)
csv_column_value_url_template = (None if isinstance(
code_list, CompositeQbCodeList) else maybe_property_value_url)
return QbColumn(
column_name,
new_dimension,
csv_column_value_url_template,
)
elif maybe_attribute_uri is not None and maybe_property_value_url is not None:
if (maybe_attribute_uri ==
"http://purl.org/linked-data/sdmx/2009/attribute#unitMeasure"
):
distinct_unit_uris = [
uritemplate.expand(maybe_property_value_url,
{csv_safe_column_name: u})
for u in set(pandas_input_to_columnar_str(column_data))
]
dsd_component = QbMultiUnits(
[ExistingQbUnit(u) for u in distinct_unit_uris])
else:
dsd_component = ExistingQbAttribute(maybe_attribute_uri)
return QbColumn(column_name, dsd_component,
maybe_property_value_url)
elif maybe_unit_uri is not None and maybe_measure_uri is not None:
measure_component = ExistingQbMeasure(maybe_measure_uri)
unit_component = ExistingQbUnit(maybe_unit_uri)
observation_value = QbSingleMeasureObservationValue(
measure=measure_component,
unit=unit_component,
data_type=maybe_data_type or "decimal",
)
return QbColumn(column_name, observation_value)
elif maybe_data_type is not None:
return QbColumn(column_name,
QbMultiMeasureObservationValue(maybe_data_type))
else:
raise Exception(f"Unmatched column definition: {col_config}")
elif isinstance(col_config, bool) and col_config:
return SuppressedCsvColumn(column_name)
else:
# If not a known/expected type/value (or is a string), treat it as a dimension.
maybe_description: Optional[str] = None
if isinstance(col_config, str):
maybe_description = col_config
new_dimension = NewQbDimension.from_data(column_name,
column_data,
description=maybe_description)
return QbColumn(column_name, new_dimension)