def CheckSlice(warnings, slice, dsid):
_CheckIdPresent(warnings, 'Slice', slice)
slice_id = GetSchemaId(slice)
_CheckType(warnings, 'Slice', slice, ['DataSlice'])
_CheckUrlPresent(warnings, 'Slice', slice, 'dataset',
'required for id ' + slice_id, dsid)
_CheckPropertyPresent(warnings, 'Slice', slice, 'dimension', 'required')
dims = AsList(GetSchemaProp(slice, 'dimension'))
for dim in dims:
url = GetUrl(dim)
if url is None:
warnings.append(
f'Slice property "dimension" values must have URLs for {slice_id}'
)
_CheckPropertyPresent(warnings, 'Slice', slice, 'measure', 'required')
measures = AsList(GetSchemaProp(slice, 'measure'))
for measure in measures:
url = GetUrl(measure)
if url is None:
warnings.append(
f'Slice property "measure" values must have URLs for {slice_id}'
)
_CheckPropertyPresent(warnings, 'Slice', slice, 'data', 'required')
data = GetSchemaProp(slice, 'data')
if not isinstance(data, str):
if isinstance(data, dict):
CheckSliceData(warnings, data, slice_id)
elif isinstance(data, list):
for datum in data:
CheckSliceData(warnings, datum, slice_id)
def CheckStatisticalDataset(warnings, dataset):
_CheckType(warnings, 'StatisticalDataset', dataset, ['StatisticalDataset'])
_CheckIdPresent(warnings, 'StatisticalDataset', dataset)
_CheckPropertyPresent(warnings, 'StatisticalDataset', dataset, 'dimension',
'required')
for dim in AsList(GetSchemaProp(dataset, 'dimension')):
CheckDimension(warnings, dim, GetSchemaId(dataset))
_CheckPropertyPresent(warnings, 'StatisticalDataset', dataset, 'measure',
'required')
for measure in AsList(GetSchemaProp(dataset, 'measure')):
CheckMeasure(warnings, measure, GetSchemaId(dataset))
_CheckPropertyPresent(warnings, 'StatisticalDataset', dataset, 'slice',
'required')
for slice in AsList(GetSchemaProp(dataset, 'slice')):
CheckSlice(warnings, slice, GetSchemaId(dataset))
def _ExpandCodeList(self, dim):
"""Load a code list from CSV and return a list of JSON-LD objects."""
codeList = []
with self.getter.Fetch(GetSchemaProp(dim, 'codeList')) as f:
reader = DictReader(f)
for row in reader:
if GetSchemaProp(dim, 'equivalentType'):
row['@type'] = ['DimensionValue', GetSchemaProp(dim, 'equivalentType')]
else:
row['@type'] = 'DimensionValue'
row['@id'] = GetSchemaId(dim) + '='
row['@id'] += row['codeValue']
row['dimension'] = GetSchemaId(dim)
codeList.append(row)
return codeList
def _CheckUrlPresent(warnings, name, obj, prop, category, expected=None):
val = GetUrl(GetSchemaProp(obj, prop))
if val is None:
warnings.append(f'{name} property "{prop}" is {category}')
elif expected and val != expected:
warnings.append(
f'{name} property "{prop}" has value "{val}" but expected "{expected}"'
)
def _ExpandCodeList(self, dim):
"""Load a code list from CSV and return a list of JSON-LD objects."""
codeList = []
dimProps = []
tableMappings = {}
for dimProp in AsList(GetSchemaProp(dim, 'dimensionProperty')):
dimProps.append(dimProp)
for tableMapping in AsList(GetSchemaProp(dim, 'tableMapping')):
tableMappings[GetUrl(tableMapping['sourceEntity'])] = tableMapping
with self.getter.Fetch(GetSchemaProp(dim, 'codeList')) as f:
reader = DictReader(f)
for row in reader:
entry = {k: v for k, v in row.items()}
if GetSchemaProp(dim, 'equivalentType'):
entry['@type'] = ['DimensionValue']
entry['@type'] += AsList(
GetSchemaProp(dim, 'equivalentType'))
else:
entry['@type'] = 'DimensionValue'
entry['@id'] = GetSchemaId(dim) + '='
entry['@id'] += row['codeValue']
entry['dimension'] = GetSchemaId(dim)
for dimProp in dimProps:
propId = GetSchemaProp(dimProp, 'propertyID')
value = dimProp.get('value')
if propId:
if value:
entry[dimProp['propertyID']] = value
continue
columnId = propId
dimPropId = GetSchemaId(dimProp)
if dimPropId:
tableMapping = tableMappings.get(dimPropId)
if tableMapping and 'columnIdentifier' in tableMapping:
columnId = tableMapping.get('columnIdentifier')
else:
columnId = propId
for field in row:
if field == columnId:
if columnId != propId:
entry[propId] = entry[columnId]
del entry[columnId]
elif field.startswith(columnId + '.'):
entry[columnId] = entry.get(
columnId,
{'@type': dimProp['propertyType']})
if isinstance(entry[columnId], str):
entry[columnId] = {
'@type': dimProp['propertyType'],
'name': row['columnId']
}
entry[columnId][field[len(columnId) +
1:]] = entry[field]
del entry[field]
codeList.append(entry)
return codeList
def _ExpandSliceData(self, slice, dim_defs_by_id):
data = []
with self.getter.Fetch(GetSchemaProp(slice, 'data')) as f:
reader = DictReader(f)
for row in reader:
val = {}
val['@type'] = 'Observation'
val['slice'] = GetSchemaId(slice)
val['dimensionValues'] = []
val['measureValues'] = []
for dim in AsList(GetSchemaProp(slice, 'dimension')):
dim = GetUrl(dim)
fragment = urlparse(dim).fragment
dim_val = {
'@type': 'DimensionValue',
'dimension': dim,
}
dim_def = dim_defs_by_id.get(dim)
if dim_def:
if GetSchemaProp(dim_def, '@type') == 'CategoricalDimension':
dim_val['codeValue'] = row[fragment]
elif GetSchemaProp(dim_def, '@type') == 'TimeDimension':
if GetSchemaProp(dim_def, 'equivalentType'):
dim_val['value'] = {
'@type': GetSchemaProp(dim_def, 'equivalentType'),
'@value': row[fragment]
}
else:
val['dimensionValues'][-1]['value'] = row[fragment]
val['dimensionValues'].append(dim_val)
for measure in AsList(GetSchemaProp(slice, 'measure')):
measure = GetUrl(measure)
fragment = urlparse(measure).fragment
val['measureValues'].append({
'@type': 'MeasureValue',
'measure': measure,
'value': row[fragment]
})
if row.get(fragment + '*'):
val['measureValues'][-1]['footnote'] = [
{
'@type': 'StatisticalAnnotation',
'codeValue': footnote
}
for footnote in row[fragment + '*'].split(';')
]
data.append(val)
return data
def Expand(self):
json_val = FrameGraph(self.getter.graph, frame=_DataFileFrame)
for dim in AsList(GetSchemaProp(json_val, 'dimension')):
if isinstance(dim.get('codeList'), str):
dim['codeList'] = self._ExpandCodeList(dim)
if isinstance(GetSchemaProp(json_val, 'footnote'), str):
json_val['footnote'] = self._ExpandFootnotes(
GetSchemaProp(json_val, 'footnote'), json_val)
for slice in AsList(GetSchemaProp(json_val, 'slice')):
dim_defs_by_id = MakeIdKeyedDict(
AsList(GetSchemaProp(json_val, 'dimension')))
if isinstance(GetSchemaProp(slice, 'data'), str):
slice['data'] = self._ExpandSliceData(slice, dim_defs_by_id)
return json_val
def _CheckAnyPropertyPresent(warnings, name, obj, props, category):
if not any(GetSchemaProp(obj, prop) for prop in props):
warnings.append(f'{name}: One property of {props} is {category}')
def test_GetSchemaProp(self):
self.assertEqual(GetSchemaProp({'id': 'val'}, 'id'), 'val')
self.assertEqual(GetSchemaProp({'schema:id': 'val'}, 'id'), 'val')
def _ExpandSliceData(self, slice, dim_defs_by_id, meas_defs_by_id):
data = []
tableMappings = {}
for tableMapping in AsList(GetSchemaProp(slice, 'tableMapping')):
tableMappings[GetUrl(tableMapping['sourceEntity'])] = tableMapping
with self.getter.Fetch(GetSchemaProp(slice, 'data')) as f:
reader = DictReader(f)
for row in reader:
val = {}
val['@type'] = 'Observation'
val['slice'] = GetSchemaId(slice)
val['dimensionValue'] = []
val['measureValue'] = []
for dim in AsList(GetSchemaProp(slice, 'dimension')):
dim = GetUrl(dim)
dim_def = dim_defs_by_id.get(dim)
if dim_def is None:
raise RuntimeError(
"Unable to find definition for dimension " + dim)
tableMapping = tableMappings.get(dim)
if tableMapping:
col_id = tableMapping['columnIdentifier']
else:
col_id = urlparse(dim).fragment
dim_val = {
'@type': 'DimensionValue',
'dimension': dim,
}
if dim_def:
if GetSchemaProp(dim_def,
'@type') == 'CategoricalDimension':
dim_val['codeValue'] = row[col_id]
elif GetSchemaProp(dim_def,
'@type') == 'TimeDimension':
if GetSchemaProp(dim_def, 'equivalentType'):
dim_val['value'] = {
'@type':
GetSchemaProp(dim_def, 'equivalentType'),
'@value':
row[col_id]
}
else:
dim_val['value'] = row[col_id]
val['dimensionValue'].append(dim_val)
for measure in AsList(GetSchemaProp(slice, 'measure')):
measure = GetUrl(measure)
meas_def = meas_defs_by_id.get(measure)
tableMapping = tableMappings.get(measure)
if tableMapping:
col_id = tableMapping['columnIdentifier']
else:
col_id = urlparse(measure).fragment
val['measureValue'].append({
'@type': 'MeasureValue',
'measure': measure,
'value': row[col_id]
})
if row.get(col_id + '*'):
val['measureValue'][-1]['footnote'] = [{
'@type':
'StatisticalAnnotation',
'codeValue':
footnote
} for footnote in row[col_id + '*'].split(';')]
data.append(val)
return data