read_array
read_array(
self ,
resource: H5Reader ,
windows: dict[str, tuple[int, int]] ,
snap: bool | None[bool, None] = None,
snap_dims: bool | None[bool, None] = None,
**kwargs ,
)-> ‘np.ndarray’
Slice the data node directly.
Besides the requested block, only the time node’s ends and the header are read (the whole time node for an unfinished file, to count the rows it actually wrote). snap selects the grid, and unlike other formats it decides how many rows there are: snapped, they stop at the last written sample; raw, every stored row counts, as the raw time coordinate does. read calls the same option snap_dims, so both spellings are taken, and snap wins when both are given, as it does in read.