Automated Code Change

tensorflow_datasets/graphs/cardiotox/cardiotox.py

@@ -142,13 +142,13 @@ def _split_generators(self, dl_manager: tfds.download.DownloadManager):
     """Returns SplitGenerators."""
 
     return {
-        tfds.Split.TRAIN: self._generate_examples(
+        tfds.Split.TRAIN: self._generate_examples(  # pyrefly: ignore[missing-attribute]
             os.path.join(_DATA_URL, _FILENAME_TRAIN), is_training=True
         ),
-        tfds.Split.VALIDATION: self._generate_examples(
+        tfds.Split.VALIDATION: self._generate_examples(  # pyrefly: ignore[missing-attribute]
             os.path.join(_DATA_URL, _FILENAME_VAL), is_training=False
         ),
-        tfds.Split.TEST: self._generate_examples(
+        tfds.Split.TEST: self._generate_examples(  # pyrefly: ignore[missing-attribute]
             os.path.join(_DATA_URL, _FILENAME_TEST), is_training=False
         ),
         tfds.Split('test2'): self._generate_examples(

tensorflow_datasets/image/lsun.py

@@ -119,14 +119,14 @@ def _info(self):
     )
 
   def _split_generators(self, dl_manager):
-    if self.builder_config.name in _SCENES_CATEGORIES:
+    if self.builder_config.name in _SCENES_CATEGORIES:  # pyrefly: ignore[missing-attribute]
       extracted_dirs = dl_manager.download_and_extract({
           "train": LSUN_SCENE_URL % (self.builder_config.name, "train"),
           "val": LSUN_SCENE_URL % (self.builder_config.name, "val"),
       })
       return [
           tfds.core.SplitGenerator(
-              name=tfds.Split.TRAIN,
+              name=tfds.Split.TRAIN,  # pyrefly: ignore[missing-attribute]
               gen_kwargs={
                   "extracted_dir": extracted_dirs["train"],
                   "file_path": "%s_%s_lmdb" % (
@@ -136,7 +136,7 @@ def _split_generators(self, dl_manager):
               },
           ),
           tfds.core.SplitGenerator(
-              name=tfds.Split.VALIDATION,
+              name=tfds.Split.VALIDATION,  # pyrefly: ignore[missing-attribute]
               gen_kwargs={
                   "extracted_dir": extracted_dirs["val"],
                   "file_path": "%s_%s_lmdb" % (self.builder_config.name, "val"),
@@ -151,7 +151,7 @@ def _split_generators(self, dl_manager):
       )
       return [
           tfds.core.SplitGenerator(
-              name=tfds.Split.TRAIN,
+              name=tfds.Split.TRAIN,  # pyrefly: ignore[missing-attribute]
               gen_kwargs={
                   "extracted_dir": extracted_dirs["train"],
                   "file_path": self.builder_config.name,

tensorflow_datasets/image_classification/cifar100_n/cifar100_n.py

@@ -193,12 +193,12 @@ def _generate_examples(self, split_prefix, filepaths, dl_paths):
           # Note: The human labels are provided according to default shuffling
           # of 'cifar100'. We need to invert the shuffling to assign the right
           # human label to each example.
-          annotations_index = np.where(annotations_order == index)[0][0]
+          annotations_index = np.where(annotations_order == index)[0][0]  # pyrefly: ignore[unbound-name]
 
-          record['noise_label'] = noise_label[annotations_index]
+          record['noise_label'] = noise_label[annotations_index]  # pyrefly: ignore[unbound-name]
           # Worker metadata is shared every 5 samples
-          record['worker_id'] = worker_id[index // 5]
-          record['worker_time'] = worker_time[index // 5]
+          record['worker_id'] = worker_id[index // 5]  # pyrefly: ignore[unbound-name]
+          record['worker_time'] = worker_time[index // 5]  # pyrefly: ignore[unbound-name]
         else:
           # There is no annotator metadata for test split
           record['noise_label'] = -1

tensorflow_datasets/image_classification/cifar10_n/cifar10_n.py

@@ -217,21 +217,21 @@ def _generate_examples(self, split_prefix, filepaths, dl_paths):
           # Note: The human labels are provided according to default shuffling
           # of 'cifar10'. We need to invert the shuffling to assign the right
           # human label to each example.
-          annotations_index = np.where(annotations_order == index)[0][0]
+          annotations_index = np.where(annotations_order == index)[0][0]  # pyrefly: ignore[unbound-name]
 
-          record['worse_label'] = worse_label[annotations_index]
-          record['aggre_label'] = aggre_label[annotations_index]
-          record['random_label1'] = random_label1[annotations_index]
-          record['random_label2'] = random_label2[annotations_index]
-          record['random_label3'] = random_label3[annotations_index]
+          record['worse_label'] = worse_label[annotations_index]  # pyrefly: ignore[unbound-name]
+          record['aggre_label'] = aggre_label[annotations_index]  # pyrefly: ignore[unbound-name]
+          record['random_label1'] = random_label1[annotations_index]  # pyrefly: ignore[unbound-name]
+          record['random_label2'] = random_label2[annotations_index]  # pyrefly: ignore[unbound-name]
+          record['random_label3'] = random_label3[annotations_index]  # pyrefly: ignore[unbound-name]
 
           # Worker metadata is shared every 10 samples
-          record['worker1_id'] = worker1_id[index // 10]
-          record['worker1_time'] = worker1_time[index // 10]
-          record['worker2_id'] = worker2_id[index // 10]
-          record['worker2_time'] = worker2_time[index // 10]
-          record['worker3_id'] = worker3_id[index // 10]
-          record['worker3_time'] = worker3_time[index // 10]
+          record['worker1_id'] = worker1_id[index // 10]  # pyrefly: ignore[unbound-name]
+          record['worker1_time'] = worker1_time[index // 10]  # pyrefly: ignore[unbound-name]
+          record['worker2_id'] = worker2_id[index // 10]  # pyrefly: ignore[unbound-name]
+          record['worker2_time'] = worker2_time[index // 10]  # pyrefly: ignore[unbound-name]
+          record['worker3_id'] = worker3_id[index // 10]  # pyrefly: ignore[unbound-name]
+          record['worker3_time'] = worker3_time[index // 10]  # pyrefly: ignore[unbound-name]
         else:
           # There is no annotator metadata for test split
           record['worse_label'] = -1

tensorflow_datasets/image_classification/controlled_noisy_web_labels/controlled_noisy_web_labels.py

@@ -361,7 +361,7 @@ def _info(self) -> tfds.core.DatasetInfo:
     features = {}
     features['image'] = tfds.features.Image()
     features['label'] = tfds.features.ClassLabel(
-        num_classes=self.builder_config.num_classes
+        num_classes=self.builder_config.num_classes  # pyrefly: ignore[missing-attribute]
     )
     features['is_clean'] = tfds.features.Tensor(shape=(), dtype=np.bool_)
     features['id'] = tfds.features.Text()
@@ -421,16 +421,16 @@ def _split_generators(self, dl_manager: tfds.download.DownloadManager):
 
     split_to_generator = {}
 
-    split_to_generator[tfds.Split.VALIDATION] = self._generate_val_examples(
+    split_to_generator[tfds.Split.VALIDATION] = self._generate_val_examples(  # pyrefly: ignore[missing-attribute]
         val_split_file, dl_manager.iter_archive(val_path)
     )
 
     for percent in _PERCENTS:
-      split_name = tfds.Split.TRAIN + '_' + '{:02d}'.format(percent)
+      split_name = tfds.Split.TRAIN + '_' + '{:02d}'.format(percent)  # pyrefly: ignore[missing-attribute]
       split_file = os.path.join(
           noisy_split_path,
           '{}_noise_nl_{}'.format(
-              self.builder_config.color, str(percent / 100)
+              self.builder_config.color, str(percent / 100)  # pyrefly: ignore[missing-attribute]
           ),
       )
       split_to_generator[split_name] = self._generate_examples(

tensorflow_datasets/image_classification/domainnet/domainnet.py

@@ -432,10 +432,10 @@ def _info(self) -> tfds.core.DatasetInfo:
 
   def _split_generators(self, dl_manager: tfds.download.DownloadManager):
     """Returns SplitGenerators."""
-    img_download_url = f'{self._BASE_URL}/{self.builder_config.img_url}'
+    img_download_url = f'{self._BASE_URL}/{self.builder_config.img_url}'  # pyrefly: ignore[missing-attribute]
     self.img_path = dl_manager.download_and_extract(img_download_url)
 
-    domain = self.builder_config.name
+    domain = self.builder_config.name  # pyrefly: ignore[missing-attribute]
     train_split_url = f'{self._BASE_URL}/domainnet/txt/{domain}_train.txt'
     test_split_url = f'{self._BASE_URL}/domainnet/txt/{domain}_test.txt'
 
@@ -456,7 +456,7 @@ def _generate_examples(
       for i, img_class_line in enumerate(split_file.read().split('\n')):
         if not img_class_line:
           continue
-        key = f'{self.builder_config.name}_{split}_{i:08d}'
+        key = f'{self.builder_config.name}_{split}_{i:08d}'  # pyrefly: ignore[missing-attribute]
 
         example_path, example_class = img_class_line.split(' ')
         example_fullpath = os.path.join(self.img_path, example_path)  # pytype: disable=attribute-error  # gen-stub-imports

tensorflow_datasets/image_classification/i_naturalist2021/i_naturalist2021.py

@@ -109,9 +109,9 @@ def _split_generators(self, dl_manager: tfds.download.DownloadManager):
     generate_dict = {}
     for split, split_file in _SPLIT_FILENAMES.items():
       generate_dict[split] = self._generate_examples(
-          images_archive=dl_manager.iter_archive(output_paths[f'{split}_img']),
+          images_archive=dl_manager.iter_archive(output_paths[f'{split}_img']),  # pyrefly: ignore[bad-index]
           json_file=os.path.join(
-              output_paths[f'{split}_json'], f'{split_file}.json'
+              output_paths[f'{split}_json'], f'{split_file}.json'  # pyrefly: ignore[bad-index]
           ),
       )
 

tensorflow_datasets/question_answering/coqa/coqa.py

@@ -74,13 +74,13 @@ def _split_generators(self, dl_manager):
 
     return [
         tfds.core.SplitGenerator(
-            name=tfds.Split.TRAIN,
+            name=tfds.Split.TRAIN,  # pyrefly: ignore[missing-attribute]
             gen_kwargs={
                 "filepath": downloaded_files["train"],
             },
         ),
         tfds.core.SplitGenerator(
-            name=tfds.Split.TEST,
+            name=tfds.Split.TEST,  # pyrefly: ignore[missing-attribute]
             gen_kwargs={
                 "filepath": downloaded_files["test"],
             },

tensorflow_datasets/ranking/istella/istella.py

@@ -136,19 +136,19 @@ def _info(self) -> tfds.core.DatasetInfo:
 
   def _split_generators(self, dl_manager: tfds.download.DownloadManager):
     """Returns SplitGenerators."""
-    path = dl_manager.download_and_extract(_URLS[self.builder_config.name])
+    path = dl_manager.download_and_extract(_URLS[self.builder_config.name])  # pyrefly: ignore[missing-attribute]
 
     # For some dataset configs, the data is in a subdirectory.
-    if self.builder_config.subdirectory is not None:
-      path = path / self.builder_config.subdirectory
+    if self.builder_config.subdirectory is not None:  # pyrefly: ignore[missing-attribute]
+      path = path / self.builder_config.subdirectory  # pyrefly: ignore[unsupported-operation]
 
     splits = {
         "train": self._generate_examples(path / "train.txt"),
         "test": self._generate_examples(path / "test.txt"),
     }
 
     # For some dataset configs, there is an additional validation split.
-    if self.builder_config.has_vali:
+    if self.builder_config.has_vali:  # pyrefly: ignore[missing-attribute]
       splits["vali"] = self._generate_examples(path / "vali.txt")
 
     return splits

tensorflow_datasets/ranking/mslr_web/mslr_web.py

@@ -267,8 +267,8 @@ def _info(self) -> tfds.core.DatasetInfo:
 
   def _split_generators(self, dl_manager: tfds.download.DownloadManager):
     """Returns SplitGenerators."""
-    size = self.builder_config.size
-    fold = self.builder_config.fold
+    size = self.builder_config.size  # pyrefly: ignore[missing-attribute]
+    fold = self.builder_config.fold  # pyrefly: ignore[missing-attribute]
     path = dl_manager.download_and_extract(_URLS[size])
 
     return {

tensorflow_datasets/ranking/yahoo_ltrc/yahoo_ltrc.py

@@ -99,7 +99,7 @@ def _info(self) -> tfds.core.DatasetInfo:
     encoding = tfds.features.Encoding.ZLIB
     features = {
         "float_features": tfds.features.Tensor(
-            shape=(None, self.builder_config.num_features),
+            shape=(None, self.builder_config.num_features),  # pyrefly: ignore[missing-attribute]
             dtype=np.float64,
             encoding=encoding,
         ),
@@ -126,10 +126,10 @@ def _split_generators(self, dl_manager: tfds.download.DownloadManager):
     archive_path = dl_manager.manual_dir / "ltrc_yahoo.tar.bz2"
     path = dl_manager.extract(archive_path)
 
-    prefix = self.builder_config.prefix
+    prefix = self.builder_config.prefix  # pyrefly: ignore[missing-attribute]
     feature_names = {
         n: f"feature_{n}"
-        for n in range(1, self.builder_config.num_features + 1)
+        for n in range(1, self.builder_config.num_features + 1)  # pyrefly: ignore[missing-attribute]
     }
 
     splits = {

tensorflow_datasets/rl_unplugged/dmlab_dataset.py

@@ -118,8 +118,8 @@ def get_citation(self):
     return _CITATION
 
   def get_file_prefix(self):
-    run = self.builder_config.name
-    task = self.builder_config.task
+    run = self.builder_config.name  # pyrefly: ignore[missing-attribute]
+    task = self.builder_config.task  # pyrefly: ignore[missing-attribute]
     return f'{self._INPUT_FILE_PREFIX}/{task}/{run}/tfrecord'
 
   def num_shards(self):
@@ -129,7 +129,7 @@ def tf_example_to_step_ds(
       self, tf_example: tf.train.Example
   ) -> Dict[str, Any]:
     """Create an episode from a TF example."""
-    episode_length = self.builder_config.episode_length
+    episode_length = self.builder_config.episode_length  # pyrefly: ignore[missing-attribute]
 
     # Parse tf.Example.
     def sequence_feature(shape, dtype=tf.float32):

tensorflow_datasets/rl_unplugged/rlu_atari/rlu_atari.py

@@ -51,12 +51,12 @@ def get_citation(self):
     return atari_utils.citation()
 
   def get_file_prefix(self):
-    run = self.builder_config.run
-    game = self.builder_config.game
+    run = self.builder_config.run  # pyrefly: ignore[missing-attribute]
+    game = self.builder_config.game  # pyrefly: ignore[missing-attribute]
     return atari_utils.file_prefix(self._INPUT_FILE_PREFIX, run, game)
 
   def num_shards(self):
-    return atari_utils.num_shards(self.builder_config.game, self._SHARDS)
+    return atari_utils.num_shards(self.builder_config.game, self._SHARDS)  # pyrefly: ignore[missing-attribute]
 
   def get_episode_id(self, episode):
     return atari_utils.episode_id(episode)

tensorflow_datasets/rl_unplugged/rlu_atari_checkpoints/rlu_atari_checkpoints.py

@@ -56,12 +56,12 @@ def get_citation(self):
     return atari_utils.citation()
 
   def get_file_prefix(self):
-    run = self.builder_config.run
-    game = self.builder_config.game
+    run = self.builder_config.run  # pyrefly: ignore[missing-attribute]
+    game = self.builder_config.game  # pyrefly: ignore[missing-attribute]
     return atari_utils.file_prefix(self._INPUT_FILE_PREFIX, run, game)
 
   def num_shards(self):
-    return atari_utils.num_shards(self.builder_config.game, self._SHARDS)
+    return atari_utils.num_shards(self.builder_config.game, self._SHARDS)  # pyrefly: ignore[missing-attribute]
 
   def get_episode_id(self, episode):
     return atari_utils.episode_id(episode)

tensorflow_datasets/rl_unplugged/rlu_atari_checkpoints_ordered/rlu_atari_checkpoints_ordered.py

@@ -77,12 +77,12 @@ def get_citation(self):
     return atari_utils.citation()
 
   def get_file_prefix(self):
-    run = self.builder_config.run
-    game = self.builder_config.game
+    run = self.builder_config.run  # pyrefly: ignore[missing-attribute]
+    game = self.builder_config.game  # pyrefly: ignore[missing-attribute]
     return atari_utils.file_prefix(self._INPUT_FILE_PREFIX, run, game)
 
   def num_shards(self):
-    return atari_utils.num_shards(self.builder_config.game, self._SHARDS)
+    return atari_utils.num_shards(self.builder_config.game, self._SHARDS)  # pyrefly: ignore[missing-attribute]
 
   def get_episode_id(self, episode):
     # TODO(b/209933106): Ordered datasets need an int id so we cannot use the