def test_base_collection_is_iterable(self):
base = BaseCollection()
base.load_data(self.data)
for i in self.data: # loop over data objects
self.assertIsInstance(i, dict) # returns
def test_None_to_nan_conversion_all_none(self):
BaseCollection.serializer_class = InternalDtypeTestSerializer
b = BaseCollection()
b.load_data(self.dtyp_test_data_all_none)
df = b.to_dataframe() # would raise here
self.assertIsInstance(df, pd.DataFrame)
def test_base_collection_to_dataframe(self):
base = BaseCollection()
base.load_data(self.data)
test = base.to_dataframe()
assert_frame_equal(test, pd.DataFrame().from_dict(self.data))
def test_base_collection_concatenation(self):
base = BaseCollection()
base.load_data(self.data)
base2 = BaseCollection()
base2.load_data(self.data)
new_base = base + base2
def test_load_data_from_dataframe(self):
df = pd.DataFrame(self.data)
base = BaseCollection()
base.load_data(df)
for i in base._data:
self.assertIsInstance(i, InternalObject)
def test_empty_collection_raises_CollectionLoadError_if_passed_empty_record_collection(
self):
BaseCollection.serializer_class = InternalSerializer
records = []
b = BaseCollection()
with self.assertRaises(CollectionLoadError):
b.load_data(records, raise_on_empty=True)
def test_empty_collection_returns_empty_dataframe_in_to_dataframe(self):
BaseCollection.serializer_class = InternalSerializer
records = []
b = BaseCollection()
b.load_data(records)
df = b.to_dataframe()
self.assertIsInstance(df, pd.DataFrame)
self.assertEqual(len(df), 0)
def test_base_collection_dataframe_with_dtypes(self):
BaseCollection.serializer_class = InternalDtypeTestSerializer # NOTE patching a different serializer here
base = BaseCollection()
base.load_data(self.dtype_test_data)
base2 = BaseCollection()
base2.load_data(self.dtype_test_data_none)
df = base2.to_dataframe()
self.assertTrue(df.isnull().values.any())
BaseCollection.serializer_class = InternalSerializer #NOTE must patch this back here
def test_non_required_fields_not_present_do_not_raise_key_error_in_to_dataframe(
self):
BaseCollection.serializer_class = InternalSerializer # these fields are not required
records = [{'bdbid': 1}, {'bdbid': 2}]
b = BaseCollection()
b.load_data(records)
df = b.to_dataframe()
self.assertEqual(records, df.to_dict('records'))
def test_non_required_date_fields_do_not_raiseTypeError_in_to_dataframe(
self):
class TestDateSerializer(BaseSerializer):
test_date = fields.Date('%Y-%m-%d', allow_none=True)
BaseCollection.serializer_class = TestDateSerializer # these fields are not required
records = [{'test_date': None}, {'test_date': '2017-07-01'}]
b = BaseCollection()
b.load_data(records)
df = b.to_dataframe()
check = df.to_dict('records')
self.assertEqual(str(check[0]['test_date']), 'NaT')
self.assertEqual(pd.Timestamp('2017-07-01'), check[1]['test_date'])
def test_pandas_timestamp_correctly_parsed_by_load_data(self):
BaseCollection.serializer_class = DateStringFormatTestSerializer
records = [
{
'a': 1,
'b': pd.Timestamp(2017, 5, 4, 10, 10, 10),
'c': pd.Timestamp(2017, 5, 4)
},
{
'a': 2,
'b': pd.Timestamp(2017, 6, 4, 10, 10, 10),
'c': pd.Timestamp(2018, 5, 4)
},
{
'a': 3,
'b': pd.Timestamp(2017, 7, 4, 10, 10, 10),
'c': pd.Timestamp(2019, 5, 4)
},
]
b = BaseCollection()
b.load_data(records)
test = [{
'a': 1,
'b': '2017-05-04 10:10:10',
'c': '2017-05-04'
}, {
'a': 2,
'b': '2017-06-04 10:10:10',
'c': '2018-05-04'
}, {
'a': 3,
'b': '2017-07-04 10:10:10',
'c': '2019-05-04'
}]
self.assertListEqual(test, b.data)
# testing on a dataframe
df = pd.DataFrame.from_records(records)
b = BaseCollection()
b.load_data(records)
self.assertListEqual(b.data, test)
def test_non_required_int_fields_do_not_raise_TypeError_in_to_dataframe(
self):
class TestIntSerializer(BaseSerializer):
test_id = fields.Integer(allow_none=True)
BaseCollection.serializer_class = TestIntSerializer # these fields are not required
records = [{'test_id': None}, {'test_id': 2}]
#expected_result = [{'test_id': np.nan}, {'test_id': 2.0}]
b = BaseCollection()
b.load_data(records)
df = b.to_dataframe()
check = df.to_dict('records')
self.assertTrue(np.isnan(
check[0]['test_id'])) #NOTE coerced to nan and float
self.assertEqual(check[1]['test_id'], 2.0)
def test_non_required_datetimes_not_present_do_not_raise_utils_key_error(
self):
# if a date field was not required and not provided a KeyError was being raised
# in RecordUtils. We swallow that error and only parse datefields that are in the
# loaded data
BaseCollection.serializer_class = DateStringFormatTestSerializer
records = [
{
'a': 1,
'b': datetime(2017, 5, 4, 10, 10, 10)
},
{
'a': 2,
'c': date(2018, 5, 4)
},
{
'a': 3,
'b': datetime(2017, 7, 4, 10, 10, 10),
'c': date(2019, 5, 4)
},
]
b = BaseCollection()
b.load_data(records)
test = [{
'a': 1,
'b': '2017-05-04 10:10:10'
}, {
'a': 2,
'c': '2018-05-04'
}, {
'a': 3,
'b': '2017-07-04 10:10:10',
'c': '2019-05-04'
}]
self.assertListEqual(test, b.data)
def test_base_collection_raises_ValidationError(self):
base = BaseCollection()
# test 3 cases where data is bad
with self.assertRaises(ValidationError):
base.load_data(self.data_with_none)
with self.assertRaises(ValidationError):
base.load_data(self.data_with_missing_field)
with self.assertRaises(ValidationError):
base.load_data(self.data_bad_input)
def test_base_collection_correctly_loads_good_data(self):
base = BaseCollection()
base.load_data(self.data)
for i in base._data: # creates InternalObject Instances
self.assertIsInstance(i, InternalObject)
def test_base_collection_returns_len(self):
base = BaseCollection()
base.load_data(self.data)
self.assertEqual(len(base), len(self.data))
def test_base_collection_raises_CollectionLoadError(self):
base = BaseCollection()
base._serializer = None # patching to None
with self.assertRaises(CollectionLoadError):
base.load_data(self.data)