Read parquet files - runs remotely on KubeRay cluster workers.
(self, path: Union[str, List[str]], **kwargs)
| 402 | |
| 403 | # Ray Data API methods - wrapped in @ray.remote to execute on cluster workers |
| 404 | def read_parquet(self, path: Union[str, List[str]], **kwargs) -> Any: |
| 405 | """Read parquet files - runs remotely on KubeRay cluster workers.""" |
| 406 | from feast.infra.ray_shared_utils import RemoteDatasetProxy |
| 407 | |
| 408 | @ray.remote |
| 409 | def _remote(file_path, read_kwargs): |
| 410 | import ray |
| 411 | |
| 412 | return ray.data.read_parquet(file_path, **read_kwargs) |
| 413 | |
| 414 | opts = self._get_task_options() |
| 415 | remote_fn = _remote.options(**opts) if opts else _remote |
| 416 | return RemoteDatasetProxy(remote_fn.remote(path, kwargs)) |
| 417 | |
| 418 | def read_csv(self, path: Union[str, List[str]], **kwargs) -> Any: |
| 419 | """Read CSV files - dispatched via @ray.remote to cluster workers.""" |
nothing calls this directly
no test coverage detected