def test_givenAFasttextEmbeddingsNotLocal_whenDownloadFasttextEmbeddings_thenDownloadIt(
self, isfile_mock):
# since we create a local fake file, the file exist, so we mock that the file doest not exist.
isfile_mock.return_value = False
create_file(self.a_fasttext_file_name_path,
content="Fake fasttext embedding content")
# we create a fake fasttext archive
with gzip.open(self.a_fasttext_gz_file_name_path, "wb") as f:
f.write(self.a_fasttext_file_name_path.encode("utf-8"))
with patch("deepparse.fasttext_tools.download_gz_model") as _:
actual = download_fasttext_embeddings(self.a_directory_path)
expected = self.a_fasttext_file_name_path
self.assertEqual(actual, expected)
def test_givenAFasttextLightEmbeddingsNotLocal_whenDownloadFasttextEmbeddingsNoVerbose_thenNoVerbose(
self, isfile_mock):
self._capture_output()
# since we create a local fake file, the file exist, so we mock that the file doest not exist.
isfile_mock.return_value = False
create_file(self.a_fasttext_light_name_path,
content="Fake fasttext embedding content")
# we create a fake fasttext archive
with gzip.open(self.a_fasttext_light_gz_file_name_path, "wb") as f:
f.write(self.a_fasttext_light_name_path.encode("utf-8"))
with patch("deepparse.fasttext_tools.download_from_url"):
download_fasttext_magnitude_embeddings(self.a_directory_path,
verbose=False)
expected = ""
actual = self.test_out.getvalue().strip()
self.assertEqual(expected, actual)
def populate_directory(self, with_retrain_parser: bool = False):
os.makedirs(os.path.join(self.fake_directory, "a_directory"),
exist_ok=True)
create_file(os.path.join(self.fake_directory, "afile.txt"),
"a content")
create_file(os.path.join(self.fake_directory, "another_file.txt"),
"a content")
create_file(os.path.join(self.fake_directory, "random_file.txt"),
"a content")
checkpoints_dir_path = os.path.join(self.fake_directory,
"checkpoints_dir")
os.makedirs(checkpoints_dir_path, exist_ok=True)
create_file(os.path.join(checkpoints_dir_path, "random_file.txt"),
"a content")
if with_retrain_parser:
create_file(
os.path.join(checkpoints_dir_path,
"retrained_fasttext_address_parser.ckpt"),
"a content",
)
def create_cache_version(self, model_name, content):
version_file_path = os.path.join(self.fake_cache_path,
model_name + ".version")
create_file(version_file_path, content)
def populate_directory(self):
create_file(
os.path.join(self.a_logging_path,
"retrained_fasttext_address_parser.ckpt"),
"a content",
)
