mindspore.numpy doesn't support create array from numpy.ndarray

[leonwanghui]

Environment

Hardware Environment(Ascend/GPU/CPU):

/device cpu

Software Environment:

• MindSpore version (source or binary): v1.2.0
• Python version (e.g., Python 3.7.5): Python 3.7.4
• OS platform and distribution (e.g., Linux Ubuntu 16.04): Windows 10
• GCC/Compiler version (if compiled from source):

Describe the current behavior

When I want to convert a numpy ndarray to Tensor using the mindspore.numpy module, some errors occurred:

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _  
D:\Applications\Anaconda3\lib\site-packages\mindspore\numpy\array_creations.py:157: in asarray
    return asarray_const(a, dtype)
D:\Applications\Anaconda3\lib\site-packages\mindspore\ops\primitive.py:188: in __call__
    return _run_op(self, self.name, args)
D:\Applications\Anaconda3\lib\site-packages\mindspore\common\api.py:75: in wrapper
    results = fn(*arg, **kwargs)
D:\Applications\Anaconda3\lib\site-packages\mindspore\ops\primitive.py:556: in _run_op
    output = real_run_op(obj, op_name, args)
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _  

obj = array([[ 10.,   4.,   2.,   5.,  18., -10.,  -4.,   3.,   7.]],
      dtype=float32)

    def get_obj_type(obj):
        """Get the obj type."""
        obj_type = RESOLVE_TYPE_INVALID
        if obj is None:
            obj_type = RESOLVE_TYPE_NONE
        elif isinstance(obj, types.FunctionType):
            obj_type = RESOLVE_TYPE_FUNCTION
        elif isinstance(obj, types.MethodType):
            obj_type = RESOLVE_TYPE_METHOD
        elif isinstance(obj, type):
            obj_type = RESOLVE_TYPE_CLASS_TYPE
        elif _is_class_instance(obj):
            obj_type = RESOLVE_TYPE_CLASS_INSTANCE
        else:
            # here for ndarray, just print its shape (in case of the array to large and print many data in screen)
            is_ndarray = type(obj).__name__ == 'ndarray' and hasattr(obj, 'shape')
>           raise TypeError(f'Invalid object with type `{type(obj)}` and {"shape" if is_ndarray else "value"} '
                            f'`{obj.shape if is_ndarray else obj}`.')
E           TypeError: Invalid object with type `<class 'numpy.ndarray'>` and shape `(1, 9)`.

D:\Applications\Anaconda3\lib\site-packages\mindspore\_extends\parse\parser.py:252: TypeError

Same errors occurred when I use numpy.array(input) method.

Describe the expected behavior

Fix the errors.

Steps to reproduce the issue

1. from mindspore import numpy as np
2. import numpy as onp
3. np.asarray(onp.ones((1, 10)))

Related log / screenshot

Special notes for this issue

[cxl-ll]

Hello, I suggest you upgrade python to version 3.7.5 and try again.

[yanglf1121]

Hi, actually mindspore.numpy.array and mindspore.numpy.asarray cannot accept original numpy array as input, please see the description here: https://www.mindspore.cn/doc/api_python/zh-CN/r1.2/mindspore/numpy/mindspore.numpy.array.html?highlight=array#mindspore.numpy.array

[leonwanghui]

@yanglf1121 Got it, thanks for the elaboration!