Enable streaming for some datasets and add video value

Streaming only uses _NO_AUTH urls. Video bytes are added for forward compatibility. m4a is not supported by datasets.Video.

vox_celeb.py

@@ -32,6 +32,8 @@ import pandas as pd
 import requests
 
 import datasets
+import fsspec as fs
+from io import BytesIO 
 
 _CITATION = """\
 @Article{Nagrani19,
@@ -108,43 +110,25 @@ _URLS = {
 
 _NO_AUTH_URLS = {
     "video": {
-        "placeholder": "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_dev_mp4",
-        "dev": (
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_dev_mp4_partaa",
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_dev_mp4_partab",
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_dev_mp4_partac",
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_dev_mp4_partad",
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_dev_mp4_partae",
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_dev_mp4_partaf",
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_dev_mp4_partag",
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_dev_mp4_partah",
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_dev_mp4_partai",
-        ),
-        "test": "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_test_mp4.zip",
+        "dev": {
+            1:"https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_mp4_1.zip",
+            2:"https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_mp4_2.zip",
+            3:"https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_mp4_3.zip",
+            4:"https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_mp4_4.zip",
+            5:"https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_mp4_5.zip",
+            6:"https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_mp4_6.zip",
+
+        },
+        "test": {1:"https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_test_mp4.zip"}
     },
     "audio1": {
-        "placeholder": "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox1/vox1_dev_wav",
-        "dev": (
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox1/vox1_dev_wav_partaa",
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox1/vox1_dev_wav_partab",
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox1/vox1_dev_wav_partac",
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox1/vox1_dev_wav_partad",
-        ),
-        "test": "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox1/vox1_test_wav.zip",
+        "dev": {1:"https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox1/vox1_dev_wav.zip"},
+        "test": {1:"https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox1/vox1_test_wav.zip"},
     },
     "audio2": {
-        "placeholder": "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_dev_aac",
-        "dev": (
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_dev_aac_partaa",
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_dev_aac_partab",
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_dev_aac_partac",
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_dev_aac_partad",
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_dev_aac_partae",
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_dev_aac_partaf",
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_dev_aac_partag",
-            "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_dev_aac_partah",
-        ),
-        "test": "https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_test_aac.zip",
+        "dev": 
+            {1:"https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_aac_1.zip",2:"https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_aac_2.zip"},
+        "test": {1:"https://huggingface.co/datasets/ProgramComputer/voxceleb/resolve/main/vox2/vox2_test_aac.zip"},
     },
 }
 
@@ -152,7 +136,7 @@ _DATASET_IDS = {"video": "vox2", "audio1": "vox1", "audio2": "vox2"}
 
 _PLACEHOLDER_MAPS = dict(
     value
-    for urls in (*_URLS.values(), *_NO_AUTH_URLS.values())
+    for urls in _URLS.values()
     for value in ((urls["placeholder"], urls["dev"]), (urls["test"], (urls["test"],)))
 )
 
@@ -217,11 +201,12 @@ class VoxCeleb(datasets.GeneratorBasedBuilder):
             "video_id": datasets.Value("string"),
             "clip_index": datasets.Value("int32"),
         }
-        if self.config.name == "audio1":
+        if self.config.name.startswith("audio"):
             features["speaker_name"] = datasets.Value("string")
             features["speaker_nationality"] = datasets.Value("string")
-        if self.config.name.startswith("audio"):
-            features["audio"] = datasets.Audio(sampling_rate=16000)
+            features["audio"] = datasets.Audio()
+        if self.config.name.startswith("video"):
+            features["video"] = datasets.Value("large_binary")
 
         return datasets.DatasetInfo(
             description=_DESCRIPTION,
@@ -232,8 +217,8 @@ class VoxCeleb(datasets.GeneratorBasedBuilder):
         )
 
     def _split_generators(self, dl_manager):
-        if dl_manager.is_streaming:
-            raise TypeError("Streaming is not supported for VoxCeleb")
+        if dl_manager.is_streaming and self.config.name != "audio1" and self.config.name != "video":
+            raise TypeError("Streaming is only supported for vox1 wav and vox2 mp4 features")
         targets = (
             ["audio1", "audio2"] if self.config.name == "audio" else [self.config.name]
         )
@@ -248,7 +233,7 @@ class VoxCeleb(datasets.GeneratorBasedBuilder):
         creds_path = Path(hf_dir) / f"voxceleb_{self.VERSION}_credentials"
         all_urls = _URLS
 
-        if cred_key is None:
+        if cred_key is None and not dl_manager.is_streaming:
             if creds_path.exists():
                 with open(creds_path, "r") as creds:
                     cred_key = json.load(creds)
@@ -366,10 +351,7 @@ class VoxCeleb(datasets.GeneratorBasedBuilder):
                 for target in targets
             )
         )
-
-        mapped_paths = dl_manager.extract(
-            dl_manager.download_custom(
-                dict(
+        target_dict =        dict(
                     (
                         placeholder_key,
                         dict(
@@ -377,17 +359,29 @@ class VoxCeleb(datasets.GeneratorBasedBuilder):
                             for target in targets
                         ),
                     )
-                    for placeholder_key in ("placeholder", "test")
-                ),
+                    for placeholder_key in ("dev" if cred_key is None else "placeholder", "test")
+                )
+        download_paths =  dl_manager.download_custom(
+         target_dict,
                 download_custom,
+            ) if cred_key is not None else dl_manager.download(
+         target_dict
             )
+        
+        mapped_paths = dl_manager.extract(
+target_dict
         )
+        
+        
+        apply_function_recursive = lambda d, f: {k: apply_function_recursive(v, f) if isinstance(v, dict) else f(v) for k, v in d.items()}
+
+        mapped_paths = apply_function_recursive(mapped_paths, dl_manager.iter_files) 
 
         return [
             datasets.SplitGenerator(
                 name="train",
                 gen_kwargs={
-                    "paths": mapped_paths["placeholder"],
+                    "paths": mapped_paths["placeholder"] if not dl_manager.is_streaming else mapped_paths["dev"] ,
                     "meta_paths": metadata,
                 },
             ),
@@ -400,8 +394,9 @@ class VoxCeleb(datasets.GeneratorBasedBuilder):
             ),
         ]
 
-    def _generate_examples(self, paths, meta_paths):
+    def _generate_examples(self,paths, meta_paths):
         key = 0
+        cred_key = os.environ.get("HUGGING_FACE_VOX_CELEB_KEY")
         for conf in paths:
             dataset_id = "vox1" if conf == "audio1" else "vox2"
             meta = pd.read_csv(
@@ -410,17 +405,18 @@ class VoxCeleb(datasets.GeneratorBasedBuilder):
                 index_col=0,
                 engine="python",
             )
-            dataset_path = next(Path(paths[conf]).iterdir())
-            dataset_format = dataset_path.name
-            for speaker_path in dataset_path.iterdir():
-                speaker = speaker_path.name
-                speaker_info = meta.loc[speaker]
-                for video in speaker_path.iterdir():
-                    video_id = video.name
-                    for clip in video.iterdir():
-                        clip_index = int(clip.stem)
+            if cred_key is None:
+                for path in paths[conf].values():
+                    for file in path:
+                        try:
+                            t =  tuple(file.split("::")[0].split("/")[2:])
+                            _,dataset_format,speaker,video_id,clip_index=  (None,) * (5 - len(t)) + t
+                        except Exception:
+                            raise Exception(file.split("::")[0].split("/")[2:])
+                        speaker_info = meta.loc[speaker]
+                        clip_index = int(Path(clip_index).stem)
                         info = {
-                            "file": str(clip),
+                            "file": file,
                             "file_format": dataset_format,
                             "dataset_id": dataset_id,
                             "speaker_id": speaker,
@@ -433,5 +429,38 @@ class VoxCeleb(datasets.GeneratorBasedBuilder):
                             info["speaker_nationality"] = speaker_info["Nationality"]
                         if conf.startswith("audio"):
                             info["audio"] = info["file"]
+                        if conf.startswith("video"):
+                            with fs.open(info["file"], 'rb') as f:
+                                info["video"] = BytesIO(f.read()).getvalue()
                         yield key, info
                         key += 1
+            else:
+                dataset_path = next(Path(paths[conf]).iterdir())
+                dataset_format = dataset_path.name
+                for speaker_path in dataset_path.iterdir():
+                    speaker = speaker_path.name
+                    speaker_info = meta.loc[speaker]
+                    for video in speaker_path.iterdir():
+                        video_id = video.name
+                        for clip in video.iterdir():
+                            clip_index = int(clip.stem)
+                            clip = os.path.join(local_extracted_archive, clip.name) if local_extracted_archive else clip
+                            info = {
+                                "file": str(clip),
+                                "file_format": dataset_format,
+                                "dataset_id": dataset_id,
+                                "speaker_id": speaker,
+                                "speaker_gender": speaker_info["Gender"],
+                                "video_id": video_id,
+                                "clip_index": clip_index,
+                            }
+                            if dataset_id == "vox1":
+                                info["speaker_name"] = speaker_info["VGGFace1 ID"]
+                                info["speaker_nationality"] = speaker_info["Nationality"]
+                            if conf.startswith("audio"):
+                                info["audio"] = info["file"]
+                            if conf.startswith("video"):
+                                with fs.open(info["file"], 'rb') as f:
+                                    info["video"] = BytesIO(f.read()).getvalue()
+                            yield key, info
+                            key += 1