def test_check_for_delimiter(self):
comparable = Compare()
comparable.file_name = os.path.join(self.MOCK_DATA_DIR,
"simple_csv_file.csv")
comparable.encoding = "utf-8"
comparable.delimiter = ","
assert input_validator.check_for_delimiter(comparable) is None
def test_check_for_non_comma_delimiter(self):
comparable = Compare()
comparable.file_name = os.path.join(self.MOCK_DATA_DIR,
"none_csv_file.txt")
comparable.encoding = "utf-8"
comparable.delimiter = ","
with pytest.raises(AppErrorHandler):
input_validator.check_for_delimiter(comparable)
def test_verify_file_encoding_when_can_detect_encoding(self):
not_able_to_detect_encoding_names = self.create_mock_data_file_for_encoding_types(
"failing_encoding_names.csv")
comparator_obj = Compare()
for encoding_val in not_able_to_detect_encoding_names:
mock_file_path = os.path.join(self.MOCK_DATA_DIR,
f'{encoding_val}.csv')
comparator_obj.file_name = mock_file_path
comparator_obj.encoding = encoding_val
assert input_validator.check_for_file_encoding(
comparator_obj) is None
def test_verify_file_encoding_when_can_not_detect_encoding(self):
not_able_to_detect_encoding_names = self.create_mock_data_file_for_encoding_types(
"failing_encoding_names.csv")
comparator_obj = Compare()
for encoding_val in not_able_to_detect_encoding_names:
with pytest.raises(AppErrorHandler):
mock_file_path = os.path.join(self.MOCK_DATA_DIR,
f'{encoding_val}.csv')
comparator_obj.file_name = mock_file_path
comparator_obj.encoding = 'ascii'
input_validator.check_for_file_encoding(comparator_obj)
def set_file_encoding(comparable: Compare):
with open(comparable.file_name, 'rb') as file:
detector = UniversalDetector()
for line in file.readlines():
detector.feed(line)
if detector.done:
break
detector.close()
if detector.result["confidence"] < 0.5 or \
detector.result["encoding"] is None:
raise AppErrorHandler(AppErrorHandler.unknown_encoding)
else:
comparable.encoding = detector.result['encoding'].lower()
