breaking down download of files

P3.py

@@ -41,11 +41,50 @@ _HOMEPAGE = "https://github.com/bigscience-workshop/promptsource"
 _DATA_PATH = "data"
 
 
-def load_cached_task(cache_dir, split):
-    # TODO(Victor): this info.*.json is actually done twice... -> factorize
-    with tf.io.gfile.GFile(os.path.join(cache_dir, f"info.{split}.json")) as f:
-        split_info = json.load(f)
-        features = split_info["features"]
+# def load_cached_task(cache_dir, split):
+#     # TODO(Victor): this info.*.json is actually done twice... -> factorize
+#     with tf.io.gfile.GFile(os.path.join(cache_dir, f"info.{split}.json")) as f:
+#         split_info = json.load(f)
+#         features = split_info["features"]
+
+#     # Use `FixedLenSequenceFeature` for sequences with variable length.
+#     def _feature_config(shape, dtype):
+#         if dtype in ("int32", "bool"):
+#             # int32 and bool are stored as int64 in the tf.train.Example protobuf.
+#             dtype = "int64"
+#         if shape and shape[0] is None:
+#             return tf.io.FixedLenSequenceFeature(
+#                 shape[1:], dtype, allow_missing=True
+#             )
+#         return tf.io.FixedLenFeature(shape, dtype)
+
+#     feature_description = {
+#         feat: _feature_config(**desc) for feat, desc in features.items()
+#     }
+
+#     tfrecords = os.path.join(
+#         cache_dir, f"{split}.tfrecord-*-of-*{split_info['num_shards']}"
+#     )
+#     ds = tf.data.TFRecordDataset(tf.io.gfile.glob(tfrecords))
+#     ds = ds.map(
+#         lambda pb: tf.io.parse_single_example(pb, feature_description),
+#         num_parallel_calls=tf.data.experimental.AUTOTUNE
+#     )
+#     # Cast features back to the types from the info JSON since some features
+#     # must be cast for storage (e.g., in32 is stored as int64).
+#     ds = ds.map(
+#         lambda x: {k: tf.cast(v, features[k]["dtype"]) for k, v in x.items()},
+#         num_parallel_calls=tf.data.experimental.AUTOTUNE
+#     )
+#     return ds
+
+def load_cached_task(features_file, tfrecord, split):
+    # # TODO(Victor): this info.*.json is actually done twice... -> factorize
+    # with tf.io.gfile.GFile(os.path.join(cache_dir, f"info.{split}.json")) as f:
+    #     split_info = json.load(f)
+    #     features = split_info["features"]
+    with tf.io.gfile.GFile(features_file) as f:
+        features = json.load(f)
 
     # Use `FixedLenSequenceFeature` for sequences with variable length.
     def _feature_config(shape, dtype):
@@ -62,10 +101,10 @@ def load_cached_task(cache_dir, split):
         feat: _feature_config(**desc) for feat, desc in features.items()
     }
 
-    tfrecords = os.path.join(
-        cache_dir, f"{split}.tfrecord-*-of-*{split_info['num_shards']}"
-    )
-    ds = tf.data.TFRecordDataset(tf.io.gfile.glob(tfrecords))
+    # tfrecords = os.path.join(
+    #     cache_dir, f"{split}.tfrecord-*-of-*{split_info['num_shards']}"
+    # )
+    ds = tf.data.TFRecordDataset(tf.io.gfile.glob([tfrecord]))
     ds = ds.map(
         lambda pb: tf.io.parse_single_example(pb, feature_description),
         num_parallel_calls=tf.data.experimental.AUTOTUNE
@@ -78,7 +117,6 @@ def load_cached_task(cache_dir, split):
     )
     return ds
 
-
 def find_task_splits_and_features():
     """Find the available tasks under ./data and their available splits and features."""
     task_and_their_splits = defaultdict(dict)
@@ -100,6 +138,7 @@ def find_task_splits_and_features():
             with open(os.path.join(folder_path, f"info.{split_name}.json")) as f:
                 split_info = json.load(f)
                 features = split_info["features"]
+                assert split_info["num_shards"] == 1
 
             # All splits under the same task have the same features dictionary (and thus the same features list)
             if task_and_their_splits[task_name] == {}:
@@ -118,7 +157,16 @@ def find_task_splits_and_features():
 
 
 _TASK_SPLITS_AND_FEATURES = find_task_splits_and_features()
-_URLs = {task_name: f"{_DATA_PATH}/{task_name}" for task_name in _TASK_SPLITS_AND_FEATURES.keys()}
+_URLs = {
+    task_name: {
+        split_name: {
+            "tfrecord": f"{_DATA_PATH}/{task_name}/{split_name}.tfrecord-00000-of-00001",
+            "features_file": f"{_DATA_PATH}/{task_name}/info.{split_name}.json",
+        }
+        for split_name in splits_and_features["splits"]
+    }
+    for task_name, splits_and_features in _TASK_SPLITS_AND_FEATURES.items()
+}
 
 
 class P3Config(datasets.BuilderConfig):
@@ -184,13 +232,13 @@ class P3(datasets.GeneratorBasedBuilder):
     def _split_generators(self, dl_manager):
         split_generators = []
         data_dir = dl_manager.download_and_extract(_URLs)
-        import pdb; pdb.set_trace()
         if "train" in self.config.splits:
             split_generators.append(
                 datasets.SplitGenerator(
                     name=datasets.Split.TRAIN,
                     gen_kwargs={
-                        "data_folder": data_dir,
+                        "features_file": data_dir["features_file"],
+                        "tfrecord": data_dir["tfrecord"],
                         "split": "train",
                     }
                 )
@@ -200,7 +248,8 @@ class P3(datasets.GeneratorBasedBuilder):
                 datasets.SplitGenerator(
                     name=datasets.Split.VALIDATION,
                     gen_kwargs={
-                        "data_folder": data_dir,
+                        "features_file": data_dir["features_file"],
+                        "tfrecord": data_dir["tfrecord"],
                         "split": "validation",
                     }
                 )
@@ -210,7 +259,8 @@ class P3(datasets.GeneratorBasedBuilder):
                 datasets.SplitGenerator(
                     name=datasets.Split.TEST,
                     gen_kwargs={
-                        "data_folder": data_dir,
+                        "features_file": data_dir["features_file"],
+                        "tfrecord": data_dir["tfrecord"],
                         "split": "test",
                     }
                 )
@@ -222,7 +272,8 @@ class P3(datasets.GeneratorBasedBuilder):
                 datasets.SplitGenerator(
                     name=datasets.Split(special_split_name),
                     gen_kwargs={
-                        "data_folder": data_dir,
+                        "features_file": data_dir["features_file"],
+                        "tfrecord": data_dir["tfrecord"],
                         "split": special_split_name,
                     }
                 )
@@ -230,7 +281,7 @@ class P3(datasets.GeneratorBasedBuilder):
         return split_generators
 
 
-    def _generate_examples(self, data_folder, split):
+    def _generate_examples(self, features_file, tfrecord, split):
         """This function returns the examples in the raw (text) form."""
         _FEAT_MAPPING_FUNCTIONS = {
             "answer_choices": lambda x: [choice.decode("utf-8") for choice in x],
@@ -244,7 +295,7 @@ class P3(datasets.GeneratorBasedBuilder):
         }
 
         key = 0
-        ds = load_cached_task(data_folder, split)
+        ds = load_cached_task(features_file, tfrecord, split)
         for ex in ds.as_numpy_iterator():
             ex_dict = {}
             for feat_name, feat_value in ex.items():