def test_dataset_download(self):
with testing.mock_kaggle_api():
with testing.tmp_dir() as tmp_dir:
out_path = kaggle.download_kaggle_data('user/dataset', tmp_dir)
self.assertEqual(out_path, os.path.join(tmp_dir, 'user_dataset'))
with tf.io.gfile.GFile(os.path.join(out_path, 'output.txt')) as f:
self.assertEqual('user/dataset', f.read())
def test_competition_download(self):
with testing.mock_kaggle_api():
with testing.tmp_dir() as tmp_dir:
out_path = kaggle.download_kaggle_data('digit-recognizer', tmp_dir)
self.assertEqual(out_path, os.path.join(tmp_dir, 'digit-recognizer'))
with tf.io.gfile.GFile(os.path.join(out_path, 'output.txt')) as f:
self.assertEqual('digit-recognizer', f.read())
def test_competition_download_error(self):
with testing.mock_kaggle_api(err_msg="Some error"):
with self.assertLogs("install the kaggle API", level="error"):
downloader = kaggle.KaggleCompetitionDownloader(
"digit-recognizer")
with self.assertRaises(subprocess.CalledProcessError):
_ = downloader.competition_files
def test_competition_download_404(self):
with testing.mock_kaggle_api(err_msg="404 - Not found"):
with self.assertLogs("spelled the competition name correctly",
level="error"):
downloader = kaggle.KaggleCompetitionDownloader(
"digit-recognizer")
with self.assertRaises(subprocess.CalledProcessError):
_ = downloader.competition_files
def test_kaggle_api(self):
fname = 'a.csv'
with testing.mock_kaggle_api(filenames=[fname, 'b.txt']):
promise = self.downloader.download(
'kaggle://some-competition/a.csv', self.tmp_dir)
_, dl_size = promise.get()
self.assertEqual(dl_size, len(fname))
with tf.io.gfile.GFile(os.path.join(self.tmp_dir, fname)) as f:
self.assertEqual(fname, f.read())
def test_competition_download(self):
filenames = ["a", "b"]
with testing.mock_kaggle_api(filenames):
downloader = kaggle.KaggleCompetitionDownloader("digit-recognizer")
self.assertEqual(downloader.competition_files, ["a", "b"])
with testing.tmp_dir() as tmp_dir:
for fname in downloader.competition_files:
out_path = downloader.download_file(fname, tmp_dir)
self.assertEqual(out_path, os.path.join(tmp_dir, fname))
with tf.io.gfile.GFile(out_path) as f:
self.assertEqual(fname, f.read())
def test_kaggle_api(self):
fname = 'a.csv'
with testing.mock_kaggle_api(filenames=[fname, 'b.txt']):
# Testing Competition Downloader
promise = self.downloader.download(
'kaggle://competition/some-competition/a.csv', self.tmp_dir)
url_info = promise.get()
self.assertEqual(url_info.size, len(fname))
with tf.io.gfile.GFile(os.path.join(self.tmp_dir, fname)) as f:
self.assertEqual(fname, f.read())
# Testing Dataset Downloader
promise = self.downloader.download(
'kaggle://dataset/some-author/some-dataset/a.csv',
self.tmp_dir)
url_info = promise.get()
self.assertEqual(url_info.size, len(fname))
with tf.io.gfile.GFile(os.path.join(self.tmp_dir, fname)) as f:
self.assertEqual(fname, f.read())
def test_competition_download_404(self):
with testing.mock_kaggle_api(err_msg='404 - Not found'):
with testing.tmp_dir() as tmp_dir:
with self.assertRaisesRegex(
ValueError, 'Please ensure you have spelled the name'):
kaggle.download_kaggle_data('digit-recognize', tmp_dir)
