def test_validade_not_cached(self):
os = mock.Mock()
os.exists = mock.Mock(return_value=False)
test_ds = DataSet(os, "/local/path", "test_id",
"http://source/to/file", "test dataset", "zip")
self.assertFalse(test_ds.is_cached())
os.exists.assert_called_with("/local/path")
def test_construct_dataset(self):
test_ds_zip = DataSet(OSFS("."), "/local/path", "test_id",
"http://source/to/file", "test dataset", "zip")
self.assertEquals("zip", test_ds_zip.compression)
test_ds = DataSet(OSFS("."), "/local/path", "test_id",
"http://source/to/file", "test dataset")
self.assertEquals(None, test_ds.compression)
def test_unzip_local_data(self):
os = OSFS(".")
os_remove = os.remove
os.remove = mock.Mock(return_value=None)
os.copy("./tests/resources/local_data/base_train.zip",
"./tests/resources/local_data/train.zip")
test_local = DataSet(os, "/local/path", "train",
"./tests/resources/local_data/train.zip",
"test dataset", "zip")
test_local.unzip_file()
result = os.exists("./tests/resources/local_data/train/train.csv")
os.remove = os_remove
os.remove("./tests/resources/local_data/train/train.csv")
os.remove("./tests/resources/local_data/train.zip")
os.removedir("./tests/resources/local_data/train")
self.assertTrue(result)
def test_path_to_read_in_dir(self):
os = mock.Mock()
os.root_path = "."
os.listdir = mock.Mock(return_value=["something.json"])
os.isdir = mock.Mock(return_value=True)
test_local = DataSet(os, "/local/path/test_id", "test_id",
"./source/to/file", "test dataset")
test_online = DataSet(os, "/local/path/test_id2", "test_id2",
"http://source/to/file", "test dataset")
expected_local = "././source/to/file/something.json"
self.assertEqual(expected_local, test_local.uri)
expected_online = "/local/path/test_id2/something.json"
self.assertEqual(expected_online, test_online.uri)
def test_is_ftp_source(self):
test_online = DataSet(OSFS("."), "/local/path", "test_id",
"ftp://source/to/file", "test dataset")
test_local = DataSet(OSFS("."), "/local/path", "test_id",
"./source/to/file", "test dataset")
self.assertTrue(test_online.is_online_source())
self.assertFalse(test_local.is_online_source())
def test_dont_download_if_cached(self):
test_online_cached = DataSet(OSFS("."), "/local/path", "test_id",
"http://source/to/file", "test dataset",
"zip")
test_online_cached._download = mock.Mock()
test_online_cached.is_cached = mock.Mock(return_value=True)
test_online_cached.download()
test_online_cached._download.assert_not_called()
def test_validate_is_zip(self):
test_ds_zip = DataSet(OSFS("."), "/local/path", "test_id",
"http://source/to/file", "test dataset", "zip")
self.assertEquals(True, test_ds_zip.is_zipped())
test_ds = DataSet(OSFS("."), "/local/path", "test_id",
"http://source/to/file", "test dataset")
self.assertEquals(False, test_ds.is_zipped())
def test_prepare_dataset(self):
os = mock.Mock()
test_ds = DataSet(os, "/local/path/test_id2", "test_id2",
"http://source/to/file", "test dataset")
test_ds.download = mock.Mock()
test_ds.unzip_file = mock.Mock()
test_ds.prepare()
test_ds.download.assert_called_once_with()
test_ds.unzip_file.assert_called_once_with()
def get_datasets(self):
"""Returns a dict with all datasets informations.
Returns:
dict: The key is the identifier and the value is a dict
with the configurations. The identifier is the name of the
configuration file.
"""
datasets = self.__get_datasets()
dataset = {}
for k in datasets:
d = datasets[k]
source = d.pop("source")
description = d.pop("description")
compression = d.pop("compression", None)
dataset[k] = DataSet(self.__fs, os.path.join(self.__local_path, k),
k, source, description, compression, **d)
return dataset
def test_zip_download(self):
test_ds = DataSet(OSFS("."), "/local/path", "test_id",
"http://source/to/file", "test dataset", "zip")
test_ds._download = mock.Mock()
test_ds.download()
test_ds._download.assert_called_with("/local/path.zip")
def test_should_download_zipped_csv(self):
os = OSFS("./tests/test_integration/resources/")
file_name = "test_csv_zipped"
test_zip_file = 'http://localhost:8001/local_data/base_train.zip'
test_ds_zip = DataSet(os, file_name, "test_id", test_zip_file,
"test dataset", "zip")
test_ds_zip.download()
test_ds_zip.unzip_file()
df = pd.read_csv(test_ds_zip.uri)
self.assertEqual((2, 2), df.shape)
os.remove(file_name + "/train.csv")
os.removedir(file_name)
## only download
os = OSFS("./tests/test_integration/resources/")
file_name = "train.csv"
test_file = 'http://localhost:8001/local_data/train.csv'
test_ds = DataSet(os, file_name, "test_id", test_file, "test dataset")
test_ds.download()
test_ds.unzip_file()
df = pd.read_csv(test_ds.uri)
self.assertEqual((2, 2), df.shape)
os.remove(file_name)