Upload load.py with huggingface_hub
Browse files
load.py
ADDED
|
@@ -0,0 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from .operator import StreamSource
|
| 2 |
+
from datasets import DatasetDict
|
| 3 |
+
from typing import Union
|
| 4 |
+
|
| 5 |
+
from .catalog import LocalCatalog
|
| 6 |
+
from .artifact import Artifact
|
| 7 |
+
|
| 8 |
+
|
| 9 |
+
def load_stream(source_name_or_path: str) -> StreamSource:
|
| 10 |
+
if Artifact.is_artifact_file(source_name_or_path):
|
| 11 |
+
return Artifact.load(source_name_or_path)
|
| 12 |
+
else:
|
| 13 |
+
return LocalCatalog().load(source_name_or_path)
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
def load_dataset(source: Union[StreamSource, str]) -> DatasetDict:
|
| 17 |
+
assert isinstance(source, (StreamSource, str)), "source must be a StreamSource or a string"
|
| 18 |
+
if isinstance(source, str):
|
| 19 |
+
source = load_stream(source)
|
| 20 |
+
return source().to_dataset()
|