def test_abstract(self):
name = "unregistered_builder"
self.assertEqual(name, UnregisteredBuilder.name)
self.assertNotIn(name, registered.list_builders())
with self.assertRaisesWithPredicateMatch(ValueError, "an abstract class"):
registered.builder(name)
def test_in_development(self):
name = "in_development_dataset_builder"
self.assertEqual(name, InDevelopmentDatasetBuilder.name)
self.assertNotIn(name, registered.list_builders())
with self.assertRaisesWithPredicateMatch(
ValueError,
("Dataset %s is under active development and is not available yet."
) % name):
registered.builder(name)
def test_registered(self):
name = "empty_dataset_builder"
self.assertEqual(name, EmptyDatasetBuilder.name)
self.assertIsInstance(registered.builder(name), EmptyDatasetBuilder)
self.assertIn(name, registered.list_builders())
nonexistent = "nonexistent_foobar_dataset"
with self.assertRaisesWithPredicateMatch(ValueError, "not found"):
registered.builder(nonexistent)
# Prints registered datasets
with self.assertRaisesWithPredicateMatch(ValueError, name):
registered.builder(nonexistent)
def main(_):
builder = registered.builder(FLAGS.dataset_name, data_dir=FLAGS.data_dir)
builder.download_and_prepare(cache_dir=FLAGS.cache_dir)
# TODO(rsepassi): Get splits from info
for split in [splits.Split.TRAIN, splits.Split.TEST]:
compute_stats(builder, split)
def test_builder_with_kwargs(self):
name = "empty_dataset_builder"
name_with_kwargs = name + "/k1=1,k2=1.,k3=foo,k4=True,k5=False"
builder = registered.builder(name_with_kwargs, data_dir="bar")
expectations = [("k1", 1), ("k2", 1.), ("k3", u"foo"), ("k4", True),
("k5", False)]
for k, v in expectations:
self.assertEqual(type(builder.kwargs[k]), type(v))
self.assertEqual(builder.kwargs[k], v)
def main(_):
dm = download_manager.DownloadManager(FLAGS.download_dir)
builder = registered.builder(FLAGS.builder)(data_dir=FLAGS.data_dir,
download_manager=dm)
builder.download_and_prepare()
# TODO(rsepassi): Get splits from info
splits = [dataset_builder.Split.TRAIN, dataset_builder.Split.TEST]
for split in splits:
compute_stats(builder, split)
def test_builder_fullname(self):
fullname = "empty_dataset_builder/conf1-attr:1.0.1/k1=1,k2=2"
builder = registered.builder(fullname, data_dir="bar")
expected = {
"k1": 1,
"k2": 2,
"version": "1.0.1",
"config": "conf1-attr",
"data_dir": "bar"
}
self.assertEqual(expected, builder.kwargs)
def test_notebook_overwrite_dataset(self):
"""Redefining the same builder twice is possible on colab."""
with mock.patch.object(py_utils, "is_notebook", lambda: True):
name = "colab_builder"
self.assertNotIn(name, registered.list_builders())
@six.add_metaclass(registered.RegisteredDataset)
class ColabBuilder(object):
pass
self.assertIn(name, registered.list_builders())
self.assertIsInstance(registered.builder(name), ColabBuilder)
old_colab_class = ColabBuilder
@six.add_metaclass(registered.RegisteredDataset) # pylint: disable=function-redefined
class ColabBuilder(object):
pass
self.assertIsInstance(registered.builder(name), ColabBuilder)
self.assertNotIsInstance(registered.builder(name), old_colab_class)
def test_show_examples(self):
with testing.mock_data():
builder = registered.builder('imagenet2012')
visualization.show_statistics(builder.info)
def test_builder_camel_case(self):
fullname = "EmptyDatasetBuilder"
builder = registered.builder(fullname)
self.assertIsInstance(builder, EmptyDatasetBuilder)