def patient_query_factory(
runner: Runner,
data_source: str,
code_system: tp.Optional[str] = None,
**kwargs: tp.Dict[tp.Any, tp.Any],
) -> base.PatientQuery:
"""Returns the right instance of `PatientQuery` based on `data_source`.
Args:
runner: The runner to use for making data queries
data_source: The definition of the source, e.g., directory containing
Parquet files or a BigQuery dataset.
kwargs: A dictionary of runner specific parameters.
Returns:
The created instance.
Raises:
ValueError: When the input `data_source` is malformed or not implemented.
"""
if runner == Runner.SPARK:
return query_lib_spark.SparkPatientQuery(data_source, code_system)
if runner == Runner.BIG_QUERY:
if "project_name" not in kwargs:
raise ValueError(
"'project_name' should be provided in kwargs while using "
"a BigQuery data source")
return query_lib_big_query.BigQueryPatientQuery(
project_name=kwargs['project_name'],
bq_dataset=data_source,
code_system=code_system,
)
raise ValueError("Query engine {} is not supported yet.".format(runner))
def test_single_code_with_values(self):
patient_query = query_lib_spark.SparkPatientQuery("", "")
patient_query.include_obs_values_in_time_range('TEST_CODE',
['VAL1', 'VAL2'],
'2021-06-01',
'2021-07-10')
sql_constraint = patient_query._all_constraints_sql()
self.assertEqual(
sql_constraint,
('((dateTime >= "2021-06-01" AND dateTime <= "2021-07-10" AND '
'coding.code="TEST_CODE" AND '
'valueCoding.code IN ("VAL1","VAL2") AND '
'valueCoding.system IS NULL))'
' AND TRUE AND TRUE AND TRUE'))
def test_two_codes_with_values_and_range(self):
patient_query = query_lib_spark.SparkPatientQuery("", "")
patient_query.include_obs_values_in_time_range('TEST_CODE1',
['VAL1', 'VAL2'],
'2021-06-01',
'2021-07-10')
patient_query.include_obs_in_value_and_time_range(
'TEST_CODE2', 0.1, None, '2021-07-09', '2021-07-10')
sql_constraint = patient_query._all_constraints_sql()
self.assertEqual(
sql_constraint,
('((dateTime >= "2021-06-01" AND dateTime <= "2021-07-10" AND '
'coding.code="TEST_CODE1" AND '
'valueCoding.code IN ("VAL1","VAL2") AND '
'valueCoding.system IS NULL) '
'OR (dateTime >= "2021-07-09" AND dateTime <= "2021-07-10" AND '
'coding.code="TEST_CODE2" AND value.quantity.value >= 0.1 ))'
' AND TRUE AND TRUE AND TRUE'))
def test_single_code_with_values_and_encounter(self):
patient_query = query_lib_spark.SparkPatientQuery("", "")
patient_query.include_obs_values_in_time_range('TEST_CODE',
['VAL1', 'VAL2'],
'2021-06-01',
'2021-07-10')
patient_query.encounter_constraints(locationId=['L1', 'L2'],
typeCode=['TC1', 'TC2'],
typeSystem='TS')
sql_constraint = patient_query._all_constraints_sql().strip()
self.assertEqual(
sql_constraint,
('((dateTime >= "2021-06-01" AND dateTime <= "2021-07-10" AND '
'coding.code="TEST_CODE" AND '
'valueCoding.code IN ("VAL1","VAL2") AND '
'valueCoding.system IS NULL)) '
'AND locationId IN ("L1","L2") AND '
'arrays_overlap(encTypeCode, array("TC1","TC2")) AND '
'array_contains(encTypeSystem, "TS")'))