torchoutil.extras.hdf.dataset module¶

class HDFDataset( hdf_fpath: str | Path, transform: Callable[[T], U] | None = None, keep_padding: Iterable[str] = (), return_added_columns: bool = False, open_hdf: bool = True, cast: Literal['to_torch_or_builtin', 'to_torch_or_numpy', 'to_builtin', 'to_numpy_src', 'to_torch_src', 'none'] = 'none', file_kwds: Dict[str, Any] | None = None, )[source]¶

Bases: Generic[T, U], DatasetSlicer[U]

property added_columns: List[str]¶: Return the list of columns added by pack_to_hdf function.

property all_columns: List[str]¶: The name of all columns of the dataset.

at(

*args,

**kwargs,

) → Any[source]¶: Deprecated: Use get_item method instead.

property attrs: HDFDatasetAttributes¶

close( ignore_if_closed: bool = False, remove_file: bool = False, ) → None[source]¶

property column_names: List[str]¶: The name of each column of the dataset.

get_attrs() → HDFDatasetAttributes[source]¶

get_column_dtype( column_name: str, ) → dtype[source]¶

get_column_shape( column_name: str, ) → Tuple[int, ...][source]¶

get_columns_shapes() → Dict[str, Tuple[int, ...]][source]¶

get_hdf_fpath() → Path[source]¶

get_hdf_keys() → Tuple[str, ...][source]¶

get_item( index: int, column: None = None, ) → T[source]¶
get_item( index: Iterable[int] | slice | None, column: str, ) → List
get_item( index: Iterable[int] | slice | None, column: List[str] | None = None, ) → Dict[str, List]
get_item( index: Any, column: Any, raw: bool = False, ) → Any

property info: Dict[str, Any]¶: Return the global dataset info.

is_closed() → bool[source]¶

is_open() → bool[source]¶

property item_type: Literal['dict', 'tuple']¶: Return the global dataset info.

property keep_padding: List[str]¶

property num_columns: int¶

property num_rows: int¶

open( ignore_if_opened: bool = False, ) → None[source]¶

property shape: Tuple[int, ...]¶: The shape of the Clotho dataset.

to_dict( raw: bool = False, ) → Dict[str, ndarray][source]¶

property transform: Callable[[T], U] | None¶

property user_attrs: Any¶