Skip to main content

validate_dlpack_tensor

scirs2_numpy::dlpack

Function validate_dlpack_tensor 

Source 
pub fn validate_dlpack_tensor(
    tensor: &DLTensor,
) -> Result<DLTensorInfo, DlpackError>
Expand description

Validate a DLTensor and extract structured metadata.

This is the entry point for consuming tensors produced by DLPack-aware frameworks (PyTorch, JAX, CuPy, etc.). It checks that:

  • data is non-null,
  • the device type is parseable,
  • the dtype code is recognised.

On success, returns a DLTensorInfo with all metadata decoded.

§Safety

tensor.shape must point to at least tensor.ndim valid i64 values. The caller must ensure the tensor is not concurrently mutated.

§Examples

use scirs2_numpy::dlpack::{dlpack_from_slice, validate_dlpack_tensor, DLDataTypeCode};

let data = vec![1.0_f64, 2.0, 3.0];
let shape = vec![3_i64];
let tensor = dlpack_from_slice(&data, &shape);

let info = validate_dlpack_tensor(&tensor).unwrap();
assert_eq!(info.shape, vec![3_i64]);
assert_eq!(info.dtype_bits, 64);
assert_eq!(info.dtype_code, DLDataTypeCode::Float);