def test_ifIsNotSupportedExportFile_raiseValueError(self):
create_csv_file(self.fake_data_path_csv, predict_container=True)
with self.assertRaises(ValueError):
parse.main([
self.a_fasttext_model_type,
self.fake_data_path_csv,
self.a_unsupported_data_path,
"--device",
self.cpu_device,
])
def test_ifIsCSVFile_noColumnName_raiseValueError(self):
create_csv_file(self.fake_data_path_csv, predict_container=True)
with self.assertRaises(ValueError):
parse.main([
self.a_fasttext_model_type,
self.fake_data_path_csv,
self.csv_export_filename,
"--device",
self.cpu_device,
])
def test_integration_cpu(self):
create_pickle_file(self.fake_data_path_pickle, predict_container=True)
parse.main([
self.a_fasttext_model_type,
self.fake_data_path_pickle,
self.pickle_p_export_filename,
"--device",
self.cpu_device,
])
export_path = generate_export_path(self.fake_data_path_pickle,
self.pickle_p_export_filename)
self.assertTrue(os.path.isfile(export_path))
def test_integration_no_logging(self):
with self._caplog.at_level(logging.INFO):
create_pickle_file(self.fake_data_path_pickle,
predict_container=True)
parse.main([
self.a_fasttext_model_type,
self.fake_data_path_pickle,
self.pickle_p_export_filename,
"--device",
self.cpu_device,
"--log",
"False",
])
self.assertEqual(0, len(self._caplog.records))
def test_integration_csv(self):
create_csv_file(self.fake_data_path_csv, predict_container=True)
parse.main([
self.a_fasttext_att_model_type,
self.fake_data_path_csv,
self.csv_export_filename,
"--device",
self.cpu_device,
"--csv_column_name",
"Address",
])
export_path = generate_export_path(self.fake_data_path_csv,
self.csv_export_filename)
self.assertTrue(os.path.isfile(export_path))
def test_ifPathToFastTextRetrainModel_thenUseFastTextRetrainModel(self):
with self._caplog.at_level(logging.INFO):
path_to_retrained_model = self.path_to_retrain_fasttext
create_pickle_file(self.fake_data_path_pickle,
predict_container=True)
parse.main([
self.a_fasttext_model_type,
self.fake_data_path_pickle,
self.pickle_p_export_filename,
"--device",
self.cpu_device,
"--path_to_retrained_model",
path_to_retrained_model,
])
expected_first_message = (
f"Parsing dataset file {self.fake_data_path_pickle} using the parser "
f"FastTextAddressParser")
actual_first_message = self._caplog.records[0].message
self.assertEqual(expected_first_message, actual_first_message)
def test_ifPathToFakeRetrainModel_thenUseFakeRetrainModel(self):
with self._caplog.at_level(logging.INFO):
# We use the default path to fasttext model as a "retrain model path"
path_to_retrained_model = os.path.join(os.path.expanduser("~"),
".cache", "deepparse",
"fasttext.ckpt")
create_pickle_file(self.fake_data_path_pickle,
predict_container=True)
parse.main([
self.a_fasttext_model_type,
self.fake_data_path_pickle,
self.pickle_p_export_filename,
"--device",
self.cpu_device,
"--path_to_retrained_model",
path_to_retrained_model,
])
expected_first_message = (
f"Parsing dataset file {self.fake_data_path_pickle} using the parser "
f"FastTextAddressParser")
actual_first_message = self._caplog.records[0].message
self.assertEqual(expected_first_message, actual_first_message)
def test_integration_logging(self):
with self._caplog.at_level(logging.INFO):
create_pickle_file(self.fake_data_path_pickle,
predict_container=True)
parse.main([
self.a_fasttext_model_type,
self.fake_data_path_pickle,
self.pickle_p_export_filename,
"--device",
self.cpu_device,
])
expected_first_message = (
f"Parsing dataset file {self.fake_data_path_pickle} using the parser "
f"FastTextAddressParser")
actual_first_message = self._caplog.records[0].message
self.assertEqual(expected_first_message, actual_first_message)
export_path = generate_export_path(self.fake_data_path_pickle,
"a_file.p")
expected_second_message = (
f"4 addresses have been parsed.\n"
f"The parsed addresses are outputted here: {export_path}")
actual_second_message = self._caplog.records[1].message
self.assertEqual(expected_second_message, actual_second_message)