def test_create_connection_only_once(self, mocker):
"""Test that two datasets that need to connect to the same db
(but different tables, for example) only create a connection once.
"""
mock_engine = mocker.patch(
"kedro.extras.datasets.pandas.sql_dataset.create_engine")
first = SQLQueryDataSet(sql=SQL_QUERY,
credentials=dict(con=CONNECTION))
assert len(first.engines) == 1
second = SQLQueryDataSet(sql=SQL_QUERY,
credentials=dict(con=CONNECTION))
assert len(second.engines) == 1
assert len(first.engines) == 1
mock_engine.assert_called_once_with(CONNECTION)
def test_empty_con_error(self):
"""Check the error when instantiating with empty connection string"""
pattern = (
r"`con` argument cannot be empty\. Please provide "
r"a SQLAlchemy connection string"
)
with pytest.raises(DataSetError, match=pattern):
SQLQueryDataSet(sql=SQL_QUERY, credentials=dict(con=""))
def test_load_driver_missing(self, mocker):
"""Test that if an unknown module/driver is encountered by SQLAlchemy
then the error should contain the original error message"""
_err = ImportError("No module named 'mysqldb'")
mocker.patch("kedro.extras.datasets.pandas.sql_dataset.create_engine",
side_effect=_err)
with pytest.raises(DataSetError, match=ERROR_PREFIX + "mysqlclient"):
SQLQueryDataSet(sql=SQL_QUERY, credentials=dict(con=CONNECTION))
def test_empty_query_error(self):
"""Check the error when instantiating with empty query or file"""
pattern = (r"`sql` and `filepath` arguments cannot both be empty\."
r"Please provide a sql query or path to a sql query file\.")
with pytest.raises(DataSetError, match=pattern):
SQLQueryDataSet(sql="",
filepath="",
credentials=dict(con=CONNECTION))
def test_invalid_module(self, mocker):
"""Test that if an unknown module/driver is encountered by SQLAlchemy
then the error should contain the original error message"""
_err = ImportError("Invalid module some_module")
mocker.patch("kedro.extras.datasets.pandas.sql_dataset.create_engine",
side_effect=_err)
pattern = ERROR_PREFIX + r"Invalid module some\_module"
with pytest.raises(DataSetError, match=pattern):
SQLQueryDataSet(sql=SQL_QUERY, credentials=dict(con=CONNECTION))
def query_data_set(request):
kwargs = dict(sql=SQL_QUERY, credentials=dict(con=CONNECTION))
kwargs.update(request.param)
return SQLQueryDataSet(**kwargs)
def test_empty_query_error(self):
"""Check the error when instantiating with empty query"""
pattern = r"`sql` argument cannot be empty\. " r"Please provide a sql query"
with pytest.raises(DataSetError, match=pattern):
SQLQueryDataSet(sql="", credentials=dict(con=CONNECTION))
read_diagnoses= read_diagnoses_query(admissions_case,adm_where)
#UNION VIEWS FOR OLD SMCH AND NEW SMCH DATA
read_new_smch_admissions = read_new_smch_admissions_query()
read_new_smch_discharges = read_new_smch_discharges_query()
read_old_smch_admissions = read_old_smch_admissions_query()
read_old_smch_discharges = read_old_smch_discharges_query()
read_old_smch_matched_data = read_old_smch_matched_view_query()
read_new_smch_matched = read_new_smch_matched_query()
#Create A Kedro Data Catalog from which we can easily get a Pandas DataFrame using catalog.load('name_of_dataframe')
catalog = DataCatalog(
{
#Read Admissions
"read_admissions": SQLQueryDataSet(
sql= read_admissions,
credentials=dict(con=con)
),
#Read Raw Discharges
"read_discharges": SQLQueryDataSet(
sql= read_discharges,
credentials=dict(con=con)
),
#Read Derived Admissions
"read_derived_admissions": SQLQueryDataSet(
sql= derived_admissions,
#load_args= dict(index_col="uid"),
credentials=dict(con=con)
),
#Read Derived Discharges
"read_derived_discharges": SQLQueryDataSet(
sql= derived_discharges,
def query_file_data_set(request, sql_file):
kwargs = dict(filepath=sql_file, credentials=dict(con=CONNECTION))
kwargs.update(request.param)
return SQLQueryDataSet(**kwargs)
def test_sql_and_filepath_args(self, sql_file):
"""Test that an error is raised when both `sql` and `filepath` args are given."""
pattern = (r"`sql` and `filepath` arguments cannot both be provided."
r"Please only provide one.")
with pytest.raises(DataSetError, match=pattern):
SQLQueryDataSet(sql=SQL_QUERY, filepath=sql_file)
def test_load_unknown_sql(self):
"""Check the error when unknown SQL dialect is provided
in the connection string"""
pattern = r"The SQL dialect in your connection is not supported by SQLAlchemy"
with pytest.raises(DataSetError, match=pattern):
SQLQueryDataSet(sql=SQL_QUERY, credentials=dict(con=FAKE_CONN_STR))
