tfds-nightly 4.9.9.dev202507290047__py3-none-any.whl → 4.9.9.dev202507310045__py3-none-any.whl

tensorflow_datasets/core/dataset_builders/croissant_builder.py

@@ -58,6 +58,7 @@ from tensorflow_datasets.core.features import image_feature
 from tensorflow_datasets.core.features import sequence_feature
 from tensorflow_datasets.core.features import tensor_feature
 from tensorflow_datasets.core.features import text_feature
+from tensorflow_datasets.core.features import video_feature
 from tensorflow_datasets.core.utils import conversion_utils
 from tensorflow_datasets.core.utils import croissant_utils
 from tensorflow_datasets.core.utils import type_utils
@@ -195,6 +196,8 @@ def datatype_converter(
     feature = audio_feature.Audio(
         doc=field.description, sample_rate=field.source.sampling_rate
     )
+  elif field_data_type == mlc.DataType.VIDEO_OBJECT:
+    feature = video_feature.Video(doc=field.description)
   else:
     raise ValueError(
         f'Unknown data type: {field_data_type} for field {field.id}.'

tensorflow_datasets/core/dataset_builders/croissant_builder_test.py

@@ -28,6 +28,7 @@ from tensorflow_datasets.core.features import image_feature
 from tensorflow_datasets.core.features import sequence_feature
 from tensorflow_datasets.core.features import tensor_feature
 from tensorflow_datasets.core.features import text_feature
+from tensorflow_datasets.core.features import video_feature
 from tensorflow_datasets.core.utils.lazy_imports_utils import mlcroissant as mlc
 
 FileFormat = file_adapters.FileFormat
@@ -225,6 +226,14 @@ def test_datatype_converter_bbox_with_invalid_format():
             bounding_boxes.BBoxFeature,
             None,
         ),
+        (
+            mlc.Field(
+                data_types=mlc.DataType.VIDEO_OBJECT,
+                description="Video feature",
+            ),
+            video_feature.Video,
+            None,
+        ),
         (
             mlc.Field(
                 id="person",

tensorflow_datasets/core/features/video_feature.py

@@ -129,7 +129,7 @@ class Video(sequence_feature.Sequence):
       frame_shape = shape[1:]
     self._encoding_format = encoding_format
     self._extra_ffmpeg_args = list(ffmpeg_extra_args or [])
-    super(Video, self).__init__(
+    super().__init__(
         image_feature.Image(
             shape=frame_shape,
             dtype=dtype,
@@ -137,6 +137,7 @@ class Video(sequence_feature.Sequence):
             use_colormap=use_colormap,
         ),
         length=shape[0] if shape else None,
+        doc=doc,
     )
 
   def _ffmpeg_decode(self, path_or_fobj):

tensorflow_datasets/core/features/video_feature_test.py

@@ -26,6 +26,7 @@ from tensorflow_datasets import testing
 from tensorflow_datasets.core import features
 
 
+
 class VideoFeatureTest(testing.FeatureExpectationsTestCase):
 
   @property
@@ -34,9 +35,10 @@ class VideoFeatureTest(testing.FeatureExpectationsTestCase):
 
   def test_video_numpy(self):
     np_video = np.random.randint(256, size=(128, 64, 64, 3), dtype=np.uint8)
-
+    doc = 'This is a test video.'
+    feature = features.Video(shape=(None, 64, 64, 3), doc=doc)
     self.assertFeature(
-        feature=features.Video(shape=(None, 64, 64, 3)),
+        feature=feature,
         shape=(None, 64, 64, 3),
         dtype=tf.uint8,
         tests=[
@@ -135,5 +137,6 @@ class VideoFeatureTest(testing.FeatureExpectationsTestCase):
           ],
       )
 
+
 if __name__ == '__main__':
   testing.test_main()

tensorflow_datasets/core/utils/conversion_utils.py

@@ -128,6 +128,18 @@ def to_tfds_value(value: Any, feature: feature_lib.FeatureConnector) -> Any:
   match feature:
     case feature_lib.ClassLabel() | feature_lib.Scalar():
       return value
+    case feature_lib.Video():
+      match value:
+        case dict():
+          if 'path' in value and value['path']:
+            return value['path']
+          elif 'bytes' in value and value['bytes']:
+            return value['bytes']
+          else:
+            raise ValueError(
+                'Dictionary-like video features must have either a `path` or'
+                ' `bytes` key.'
+            )
     case feature_lib.FeaturesDict():
       return {
           name: to_tfds_value(value.get(name), inner_feature)

tensorflow_datasets/core/utils/conversion_utils_test.py

@@ -81,7 +81,7 @@ def test_convert_value_raises(value, feature):
 @pytest.mark.parametrize(
     'value,feature,expected_value',
     [
-        # datetime
+        # Datetime.
         (
             datetime.datetime(1970, 1, 1, tzinfo=datetime.timezone.utc),
             feature_lib.Scalar(dtype=np.int64),
@@ -92,14 +92,14 @@ def test_convert_value_raises(value, feature):
             feature_lib.Scalar(dtype=np.int64),
             86400,
         ),
-        # scalar
+        # Scalar.
         (42, feature_lib.Scalar(dtype=np.int64), 42),
         (42, feature_lib.Scalar(dtype=np.int32), 42),
         ('abc', feature_lib.Scalar(dtype=np.object_), 'abc'),
         (True, feature_lib.Scalar(dtype=np.bool_), True),
         (False, feature_lib.Scalar(dtype=np.bool_), False),
         (42.0, feature_lib.Scalar(dtype=np.float32), 42.0),
-        # sequence
+        # Sequence.
         ([42], feature_lib.Sequence(feature=tf.int64), [42]),
         (42, feature_lib.Sequence(feature=tf.int64), [42]),
         (None, feature_lib.Sequence(feature=tf.int64), []),
@@ -111,7 +111,7 @@ def test_convert_value_raises(value, feature):
             ),
             {'someint': [b'', 'string', b'']},
         ),
-        # image
+        # Image.
         (
             lazy_imports_lib.lazy_imports.PIL_Image.new(mode='L', size=(4, 4)),
             feature_lib.Image(),
@@ -119,7 +119,7 @@ def test_convert_value_raises(value, feature):
                 mode='RGB', size=(4, 4)
             ),
         ),
-        # dict
+        # Dict.
         (
             {
                 'de': b'Hallo Welt',
@@ -148,7 +148,18 @@ def test_convert_value_raises(value, feature):
             }),
             {'name': b'Name', 'age': 100},
         ),
-        # nan, but the feature type is not float
+        # Video.
+        (
+            {'path': 'path/to/video.avi', 'bytes': None},
+            feature_lib.Video(),
+            'path/to/video.avi',
+        ),
+        (
+            {'path': None, 'bytes': b'video_bytes'},
+            feature_lib.Video(),
+            b'video_bytes',
+        ),
+        # nan, but the feature type is not float.
         (
             np.nan,
             feature_lib.Text(),

{tfds_nightly-4.9.9.dev202507290047.dist-info → tfds_nightly-4.9.9.dev202507310045.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: tfds-nightly
-Version: 4.9.9.dev202507290047
+Version: 4.9.9.dev202507310045
 Summary: tensorflow/datasets is a library of datasets ready to use with TensorFlow.
 Home-page: https://github.com/tensorflow/datasets
 Download-URL: https://github.com/tensorflow/datasets/tags

{tfds_nightly-4.9.9.dev202507290047.dist-info → tfds_nightly-4.9.9.dev202507310045.dist-info}/RECORD

@@ -141,8 +141,8 @@ tensorflow_datasets/core/data_sources/python_test.py,sha256=O3yqMPx40JlHN0uFfZPN
 tensorflow_datasets/core/dataset_builders/__init__.py,sha256=StTA3euephqDZdpTzJQgfWNqB5inZosrAhaWg2BOeio,1945
 tensorflow_datasets/core/dataset_builders/adhoc_builder.py,sha256=QVE8wWGPOgILPTC27Q28QZ3KIi5N64OGOfKpTq4W4_0,9216
 tensorflow_datasets/core/dataset_builders/adhoc_builder_test.py,sha256=yhRwrznK78MvHeWGRggnMTiyx_SlR1z30iD5VU3Gweo,13096
-tensorflow_datasets/core/dataset_builders/croissant_builder.py,sha256=sWSjouj98I2yCUtr8KbmHtR_dp8tgnbBA7BvBKZeI1Q,16918
-tensorflow_datasets/core/dataset_builders/croissant_builder_test.py,sha256=AgNmnmnHOAhsdIzuwN7EugeoDe83_8G_95WSmDXJIIA,14803
+tensorflow_datasets/core/dataset_builders/croissant_builder.py,sha256=XmnbIKiEN9OnY_RC8P7-83hbUfvtuJhbm24HfNFpiQs,17088
+tensorflow_datasets/core/dataset_builders/croissant_builder_test.py,sha256=42HpBr3pANVKiok4lcx6xqwf0fY7kma6WIGA8WehNSs,15072
 tensorflow_datasets/core/dataset_builders/huggingface_dataset_builder.py,sha256=Loq3qeGk1Ias-d2oT_dK47BRNgTA4LKJchNGh7aA4a0,18313
 tensorflow_datasets/core/dataset_builders/huggingface_dataset_builder_test.py,sha256=6N3DLsry9LhDqhpleaoXrrhaGiLJMBgUlwDnAji-1fI,4389
 tensorflow_datasets/core/dataset_builders/view_builder.py,sha256=eaCtjN5Vg4rK8JD3auA4PhF9mjH5HvQ9dslDX8LbwyM,11907
@@ -213,8 +213,8 @@ tensorflow_datasets/core/features/top_level_feature.py,sha256=JeOnaBUqp-xFLuPxUt
 tensorflow_datasets/core/features/top_level_feature_test.py,sha256=JutGHU-08tg5KWiB3mIB6Q3a80CvS5_F6jG0bfAYXWM,3628
 tensorflow_datasets/core/features/translation_feature.py,sha256=Qmx39XwMJy18u9eoZlT3Spc0VT0qtqsTHahWoETLZZo,8284
 tensorflow_datasets/core/features/translation_feature_test.py,sha256=iK8ckwApuMu13BS1-vkny-m_NV6uNTz6ky5gbZEfxoc,6060
-tensorflow_datasets/core/features/video_feature.py,sha256=o4tuH4HxkG_sYCYUx8R-LVHcuIgFcfCNe8n0cE5MISk,7813
-tensorflow_datasets/core/features/video_feature_test.py,sha256=k0qpYvZIC_-xXTG9EN1Mo2d9RA77CcvJYl9c0U3oXcQ,4255
+tensorflow_datasets/core/features/video_feature.py,sha256=vtKbhT04u-Ne6BImn9VBetYwo9IAmvwKLyGxpBYAhTA,7819
+tensorflow_datasets/core/features/video_feature_test.py,sha256=Zp5wuGvSajMjxiayXoFYJyYHj_M977pHq8smFIdwOqc,4321
 tensorflow_datasets/core/folder_dataset/__init__.py,sha256=Pn2mSU-CPxC89lvywHAD-XrhQj0mvAaqZogpekjr-bs,1515
 tensorflow_datasets/core/folder_dataset/compute_split_utils.py,sha256=Ob_ZaqfS00zViAtRhHK_ff7R8eJAtYDDh6XjQGXdcP4,13515
 tensorflow_datasets/core/folder_dataset/compute_split_utils_test.py,sha256=XBo4UC1IydAPuIP1SY2psrEVeEr3y0KPmCEje6yQWhs,3784
@@ -243,8 +243,8 @@ tensorflow_datasets/core/utils/benchmark_test.py,sha256=WUls_DapQTGs5P3DYJrcFldS
 tensorflow_datasets/core/utils/bool_utils.py,sha256=299rj0bCG4zytHfsTEbptCV1M0RG_xN6d0SvmQZ1Sg4,1066
 tensorflow_datasets/core/utils/bool_utils_test.py,sha256=rwFRcYV0wBknvYODjeTgRDqwUifzwLGCwejGdAjo61Q,1559
 tensorflow_datasets/core/utils/colormap.csv,sha256=DDayUU9R19cxhcG3fj4cFwhI46W20U7ofBG0kToUHOw,2732
-tensorflow_datasets/core/utils/conversion_utils.py,sha256=71nt3pU9SX9WgTCW8QT7FER0gTGFZkPPYJLC4VLuV0I,6340
-tensorflow_datasets/core/utils/conversion_utils_test.py,sha256=6bASnI9oOk-nrf1n0GAc0XBzyPaydtOMDP8O4z5KjfU,5043
+tensorflow_datasets/core/utils/conversion_utils.py,sha256=V8kFmJu38op7-8ufZvEn0fLOH8FMkjQebQ1NstIMRYo,6747
+tensorflow_datasets/core/utils/conversion_utils_test.py,sha256=rP_nbzQWzmZc_GXp3Y6TirwIGJqiQbF-JtY3B1tOuN0,5346
 tensorflow_datasets/core/utils/croissant_utils.py,sha256=9C8sScaEqSRsThqpQQc48GDNR1KFmDkS8hmKIvfZCB0,5181
 tensorflow_datasets/core/utils/croissant_utils_test.py,sha256=UdkAVYDTPm1L0zmMESScurV_IMA5K3qAKmL_umeMJZI,4497
 tensorflow_datasets/core/utils/docs.py,sha256=0AeGyhPXQc_DMsTX7ocbb8IuJv_RTrG7StYdfZiwSzU,1549
@@ -2468,10 +2468,10 @@ tensorflow_datasets/vision_language/wit/wit_test.py,sha256=PXS8DMNW-MDrT2p5oy4Ic
 tensorflow_datasets/vision_language/wit_kaggle/__init__.py,sha256=vGwSGeM8WE4Q-l0-eEE1sBojmk6YT0l1OO60AWa4Q40,719
 tensorflow_datasets/vision_language/wit_kaggle/wit_kaggle.py,sha256=q-vX_FBzIwsFxL4sY9vuyQ3UQD2PLM4yhUR4U6l-qao,16903
 tensorflow_datasets/vision_language/wit_kaggle/wit_kaggle_test.py,sha256=ZymHT1NkmD-pUnh3BmM3_g30c5afsWYnmqDD9dVyDSA,1778
-tfds_nightly-4.9.9.dev202507290047.dist-info/licenses/AUTHORS,sha256=nvBG4WwfgjuOu1oZkuQKw9kg7X6rve679ObS-YDDmXg,309
-tfds_nightly-4.9.9.dev202507290047.dist-info/licenses/LICENSE,sha256=z8d0m5b2O9McPEK1xHG_dWgUBT6EfBDz6wA0F7xSPTA,11358
-tfds_nightly-4.9.9.dev202507290047.dist-info/METADATA,sha256=NLuI8GsdBpZwdvehltKjE-XSxD0UXFgt6YsvMYATTts,11694
-tfds_nightly-4.9.9.dev202507290047.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-tfds_nightly-4.9.9.dev202507290047.dist-info/entry_points.txt,sha256=eHEL7nF5y1uCY2FgkuYIdE062epJXlAQTSdq89px4p4,73
-tfds_nightly-4.9.9.dev202507290047.dist-info/top_level.txt,sha256=bAevmk9209s_oxVZVlN6hSDIVS423qrMQvmcWSvW4do,20
-tfds_nightly-4.9.9.dev202507290047.dist-info/RECORD,,
+tfds_nightly-4.9.9.dev202507310045.dist-info/licenses/AUTHORS,sha256=nvBG4WwfgjuOu1oZkuQKw9kg7X6rve679ObS-YDDmXg,309
+tfds_nightly-4.9.9.dev202507310045.dist-info/licenses/LICENSE,sha256=z8d0m5b2O9McPEK1xHG_dWgUBT6EfBDz6wA0F7xSPTA,11358
+tfds_nightly-4.9.9.dev202507310045.dist-info/METADATA,sha256=Z9I4QyzIXa3XoZSMv2SnWkVzCZQ52gELmAvGkUYWiJ0,11694
+tfds_nightly-4.9.9.dev202507310045.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+tfds_nightly-4.9.9.dev202507310045.dist-info/entry_points.txt,sha256=eHEL7nF5y1uCY2FgkuYIdE062epJXlAQTSdq89px4p4,73
+tfds_nightly-4.9.9.dev202507310045.dist-info/top_level.txt,sha256=bAevmk9209s_oxVZVlN6hSDIVS423qrMQvmcWSvW4do,20
+tfds_nightly-4.9.9.dev202507310045.dist-info/RECORD,,