5270840792

pending completion

Build # 5270840792

Build Type

Pull #1436

github

Committed by

web-flow

Commit Message

Merge 3ef181e6c into c51b3ce2b

Pull Request Pull Request #1436: New implementation of dpnp.outer

Run Details

5807 of 24608 branches covered (23.6%)

Branch coverage included in aggregate %.

13 of 14 new or added lines in 1 file covered. (92.86%)

1 existing line in 1 file now uncovered.

11408 of 17583 relevant lines covered (64.88%)

39494.46 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

74.6

/dpnp/dpnp_utils/dpnp_algo_utils.pyx

# cython: language_level=3
# cython: linetrace=True
# -*- coding: utf-8 -*-
# *****************************************************************************
# Copyright (c) 2016-2023, Intel Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
# - Redistributions of source code must retain the above copyright notice,
#   this list of conditions and the following disclaimer.
# - Redistributions in binary form must reproduce the above copyright notice,
#   this list of conditions and the following disclaimer in the documentation
#   and/or other materials provided with the distribution.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
# ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
# LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
# CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
# SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
# INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
# CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
# ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
# THE POSSIBILITY OF SUCH DAMAGE.
# *****************************************************************************

"""Module Utilities

This module contains differnt helpers and utilities

"""

import numpy

import dpctl
import dpctl.utils as dpu
import dpctl.tensor._copy_utils as dpt_cu
import dpctl.tensor._tensor_impl as dpt_ti

import dpnp.config as config
import dpnp.dpnp_container as dpnp_container
import dpnp

from dpnp.dpnp_array import dpnp_array
from dpnp.dpnp_algo.dpnp_algo cimport (
    dpnp_DPNPFuncType_to_dtype,
    dpnp_dtype_to_DPNPFuncType,
    get_dpnp_function_ptr
)

from libcpp cimport bool as cpp_bool
from libcpp.complex cimport complex as cpp_complex

cimport cpython
cimport cython
cimport numpy


"""
Python import functions
"""
__all__ = [
    "call_origin",
    "checker_throw_axis_error",
    "checker_throw_index_error",
    "checker_throw_runtime_error",
    "checker_throw_type_error",
    "checker_throw_value_error",
    "create_output_descriptor_py",
    "convert_item",
    "dpnp_descriptor",
    "get_axis_indeces",
    "get_axis_offsets",
    "get_usm_allocations",
    "_get_linear_index",
    "map_dtype_to_device",
    "normalize_axis",
    "_object_to_tuple",
    "unwrap_array",
    "use_origin_backend"
]

cdef ERROR_PREFIX = "DPNP error:"


def convert_item(item):
    if getattr(item, "__sycl_usm_array_interface__", False):
        item_converted = dpnp.asnumpy(item)
    elif getattr(item, "__array_interface__", False):  # detect if it is a container (TODO any better way?)
        mod_name = getattr(item, "__module__", 'none')
        if (mod_name != 'numpy'):
            item_converted = dpnp.asnumpy(item)
        else:
            item_converted = item
    elif isinstance(item, list):
        item_converted = convert_list_args(item)
    elif isinstance(item, tuple):
        item_converted = tuple(convert_list_args(item))
    else:
        item_converted = item

    return item_converted


def convert_list_args(input_list):
    result_list = []
    for item in input_list:
        item_converted = convert_item(item)
        result_list.append(item_converted)

    return result_list


def copy_from_origin(dst, src):
    """Copy origin result to output result."""
    if hasattr(dst, "__sycl_usm_array_interface__"):
        if src.size:
            dst_dpt = unwrap_array(dst)
            dst_dpt[...] = src
    else:
        for i in range(dst.size):
            dst.flat[i] = src.item(i)


def call_origin(function, *args, **kwargs):
    """
    Call fallback function for unsupported cases
    """

    allow_fallback = kwargs.pop("allow_fallback", False)

    if not allow_fallback and config.__DPNP_RAISE_EXCEPION_ON_NUMPY_FALLBACK__ == 1:
        raise NotImplementedError(f"Requested funtion={function.__name__} with args={args} and kwargs={kwargs} "
                                   "isn't currently supported and would fall back on NumPy implementation. "
                                   "Define enviroment variable `DPNP_RAISE_EXCEPION_ON_NUMPY_FALLBACK` to `0` "
                                   "if the fall back is required to be supported without rasing an exception.")

    dpnp_inplace = kwargs.pop("dpnp_inplace", False)
    sycl_queue = kwargs.pop("sycl_queue", None)
    # print(f"DPNP call_origin(): Fallback called. \n\t function={function}, \n\t args={args}, \n\t kwargs={kwargs}, \n\t dpnp_inplace={dpnp_inplace}")

    kwargs_out = kwargs.get("out", None)
    alloc_queues = [sycl_queue] if sycl_queue else []
    if (kwargs_out is not None):
        if isinstance(kwargs_out, numpy.ndarray):
            kwargs["out"] = kwargs_out
        else:
            if hasattr(kwargs_out, "sycl_queue"):
                alloc_queues.append(kwargs_out.sycl_queue)
            kwargs["out"] = dpnp.asnumpy(kwargs_out)

    args_new_list = []
    for arg in args:
        if hasattr(arg, "sycl_queue"):
            alloc_queues.append(arg.sycl_queue)
        argx = convert_item(arg)
        args_new_list.append(argx)
    args_new = tuple(args_new_list)

    kwargs_new = {}
    for key, kwarg in kwargs.items():
        if hasattr(kwarg, "sycl_queue"):
            alloc_queues.append(kwarg.sycl_queue)
        kwargx = convert_item(kwarg)
        kwargs_new[key] = kwargx

    exec_q = dpu.get_execution_queue(alloc_queues)
    if exec_q is None:
        exec_q = dpnp.get_normalized_queue_device(sycl_queue=sycl_queue)
    # print(f"DPNP call_origin(): bakend called. \n\t function={function}, \n\t args_new={args_new}, \n\t kwargs_new={kwargs_new}, \n\t dpnp_inplace={dpnp_inplace}")
    # TODO need to put array memory into NumPy call
    result_origin = function(*args_new, **kwargs_new)
    # print(f"DPNP call_origin(): result from backend. \n\t result_origin={result_origin}, \n\t args_new={args_new}, \n\t kwargs_new={kwargs_new}, \n\t dpnp_inplace={dpnp_inplace}")
    result = result_origin
    if dpnp_inplace:
        # enough to modify only first argument in place
        if args and args_new:
            arg, arg_new = args[0], args_new[0]
            if isinstance(arg_new, numpy.ndarray):
                copy_from_origin(arg, arg_new)
            elif isinstance(arg_new, list):
                for i, val in enumerate(arg_new):
                    arg[i] = val

    elif isinstance(result, numpy.ndarray):
        if kwargs_out is None:
            # use dtype from input arguments if present or from the result otherwise
            result_dtype = kwargs.get("dtype", None) or result_origin.dtype

            if exec_q is not None:
                result_dtype = map_dtype_to_device(result_origin.dtype, exec_q.sycl_device)

            result = dpnp_container.empty(result_origin.shape, dtype=result_dtype, sycl_queue=exec_q)
        else:
            result = kwargs_out

        copy_from_origin(result, result_origin)

    elif isinstance(result, tuple):
        # convert tuple(fallback_array) to tuple(result_array)
        result_list = []
        for res_origin in result:
            res = res_origin
            if isinstance(res_origin, numpy.ndarray):
                res = dpnp_container.empty(res_origin.shape, dtype=res_origin.dtype, sycl_queue=exec_q)
                copy_from_origin(res, res_origin)
            result_list.append(res)

        result = tuple(result_list)

    return result


def unwrap_array(x1):
    """
    Get array from input object.
    """
    if isinstance(x1, dpnp_array):
        return x1.get_array()

    return x1


def _get_coerced_usm_type(objects):
    types_in_use = [obj.usm_type for obj in objects if hasattr(obj, "usm_type")]
    if len(types_in_use) == 0:
        return None
    elif len(types_in_use) == 1:
        return types_in_use[0]

    common_usm_type = dpu.get_coerced_usm_type(types_in_use)
    if common_usm_type is None:
        raise ValueError("Input arrays must have coerced USM types")
    return common_usm_type


def _get_common_allocation_queue(objects):
    queues_in_use = [obj.sycl_queue for obj in objects if hasattr(obj, "sycl_queue")]
    if len(queues_in_use) == 0:
        return None
    elif len(queues_in_use) == 1:
        return queues_in_use[0]

    common_queue = dpu.get_execution_queue(queues_in_use)
    if common_queue is None:
        raise ValueError("Input arrays must be allocated on the same SYCL queue")
    return common_queue


def get_usm_allocations(objects):
    """
    Given a list of objects returns a tuple of USM type and SYCL queue
    which can be used for a memory allocation and to follow compute follows data paradigm,
    or returns `(None, None)` if the default USM type and SYCL queue can be used.
    An exception will be raised, if the paradigm is broked for the given list of objects.

    """

    if not isinstance(objects, (list, tuple)):
        raise TypeError("Expected a list or a tuple, got {}".format(type(objects)))

    if len(objects) == 0:
        return (None, None)
    return (_get_coerced_usm_type(objects), _get_common_allocation_queue(objects))


def map_dtype_to_device(dtype, device):
    """
    Map an input ``dtype`` with type ``device`` may use
    """

    dtype = dpnp.dtype(dtype)
    if not hasattr(dtype, 'char'):
        raise TypeError(f"Invalid type of input dtype={dtype}")
    elif not isinstance(device, dpctl.SyclDevice):
        raise TypeError(f"Invalid type of input device={device}")

    dtc = dtype.char
    if dtc == "?" or dpnp.issubdtype(dtype, dpnp.integer):
        # bool or integer type
        return dtype

    if dpnp.issubdtype(dtype, dpnp.floating):
        if dtc == "f":
            # float32 type
            return dtype
        elif dtc == "d":
            # float64 type
            if device.has_aspect_fp64:
                return dtype
        elif dtc == "e":
            # float16 type
            if device.has_aspect_fp16:
                return dtype
        # float32 is default floating type
        return dpnp.dtype("f4")

    if dpnp.issubdtype(dtype, dpnp.complexfloating):
        if dtc == "F":
            # complex64 type
            return dtype
        elif dtc == "D":
            # complex128 type
            if device.has_aspect_fp64:
                return dtype
        # complex64 is default complex type
        return dpnp.dtype("c8")
    raise RuntimeError(f"Unrecognized type of input dtype={dtype}")


cpdef checker_throw_axis_error(function_name, param_name, param, expected):
    err_msg = f"{ERROR_PREFIX} in function {function_name}()"
    err_msg += f" axes '{param_name}' expected `{expected}`, but '{param}' provided"
    raise numpy.AxisError(err_msg)


cpdef checker_throw_index_error(function_name, index, size):
    raise IndexError(
        f"{ERROR_PREFIX} in function {function_name}() index {index} is out of bounds. dimension size `{size}`")


cpdef checker_throw_runtime_error(function_name, message):
    raise RuntimeError(f"{ERROR_PREFIX} in function {function_name}(): '{message}'")


cpdef checker_throw_type_error(function_name, given_type):
    raise TypeError(f"{ERROR_PREFIX} in function {function_name}() type '{given_type}' is not supported")


cpdef checker_throw_value_error(function_name, param_name, param, expected):
    # import sys
    # sys.tracebacklimit = 0
    err_msg = f"{ERROR_PREFIX} in function {function_name}() paramenter '{param_name}'"
    err_msg += f" expected `{expected}`, but '{param}' provided"
    raise ValueError(err_msg)


cpdef dpnp_descriptor create_output_descriptor_py(shape_type_c output_shape,
                                                  d_type,
                                                  requested_out,
                                                  device=None,
                                                  usm_type="device",
                                                  sycl_queue=None):
    py_type = dpnp.default_float_type() if d_type is None else d_type

    cdef DPNPFuncType c_type = dpnp_dtype_to_DPNPFuncType(py_type)

    return create_output_descriptor(output_shape,
                                    c_type,
                                    requested_out,
                                    device=device,
                                    usm_type=usm_type,
                                    sycl_queue=sycl_queue)


cpdef tuple get_axis_indeces(idx, shape):
    """
    Compute axis indices of an element in array from array linear index
    """

    ids = []
    remainder = idx
    offsets = get_axis_offsets(shape)
    for i in offsets:
        quotient, remainder = divmod(remainder, i)
        ids.append(quotient)

    return _object_to_tuple(ids)


cpdef tuple get_axis_offsets(shape):
    """
    Compute axis offsets in the linear array memory
    """

    res_size = len(shape)
    result = [0] * res_size
    acc = 1
    for i in range(res_size - 1, -1, -1):
        result[i] = acc
        acc *= shape[i]

    return _object_to_tuple(result)


cpdef long _get_linear_index(key, tuple shape, int ndim):
    """
    Compute linear index of an element in memory from array indices
    """

    if isinstance(key, tuple):
        li = 0
        m = 1
        for i in range(ndim - 1, -1, -1):
            li += key[i] * m
            m *= shape[i]
    else:
        li = key
    return li


cdef tuple get_shape_dtype(object input_obj):
    cdef shape_type_c return_shape  # empty shape means scalar
    return_dtype = None

    # TODO replace with checking "shape" and "dtype" attributes
    if hasattr(input_obj, "shape") and hasattr(input_obj, "dtype"):
        return (input_obj.shape, input_obj.dtype)

    cdef shape_type_c elem_shape
    cdef shape_type_c list_shape
    if isinstance(input_obj, (list, tuple)):
        for elem in input_obj:
            elem_shape, elem_dtype = get_shape_dtype(elem)

            if return_shape.empty():
                return_shape = elem_shape
                return_dtype = elem_dtype

            # shape and dtype does not match with siblings.
            if ((return_shape != elem_shape) or (return_dtype != elem_dtype)):
                return (elem_shape, dpnp.dtype(numpy.object_))

        list_shape.push_back(len(input_obj))
        list_shape.insert(list_shape.end(), return_shape.begin(), return_shape.end())
        return (list_shape, return_dtype)

    # assume scalar or object
    return (return_shape, dpnp.dtype(type(input_obj)))


cpdef find_common_type(object x1_obj, object x2_obj):
    _, x1_dtype = get_shape_dtype(x1_obj)
    _, x2_dtype = get_shape_dtype(x2_obj)

    cdef list array_types = []
    cdef list scalar_types = []

    if dpnp.isscalar(x1_obj):
        scalar_types.append(x1_dtype)
    else:
        array_types.append(x1_dtype)

    if dpnp.isscalar(x2_obj):
        scalar_types.append(x2_dtype)
    else:
        array_types.append(x2_dtype)

    return numpy.find_common_type(array_types, scalar_types)


cdef shape_type_c get_common_shape(shape_type_c input1_shape, shape_type_c input2_shape) except *:
    cdef shape_type_c input1_shape_orig = input1_shape
    cdef shape_type_c input2_shape_orig = input2_shape
    cdef shape_type_c result_shape

    # ex (8, 1, 6, 1) and (7, 1, 5) -> (8, 1, 6, 1) and (1, 7, 1, 5)
    cdef size_t max_shape_size = max(input1_shape.size(), input2_shape.size())
    input1_shape.insert(input1_shape.begin(), max_shape_size - input1_shape.size(), 1)
    input2_shape.insert(input2_shape.begin(), max_shape_size - input2_shape.size(), 1)

    # ex result (8, 7, 6, 5)
    for it in range(max_shape_size):
        if input1_shape[it] == input2_shape[it]:
            result_shape.push_back(input1_shape[it])
        elif input1_shape[it] == 1:
            result_shape.push_back(input2_shape[it])
        elif input2_shape[it] == 1:
            result_shape.push_back(input1_shape[it])
        else:
            err_msg = f"{ERROR_PREFIX} in function get_common_shape(): "
            err_msg += f"operands could not be broadcast together with shapes {input1_shape_orig} {input2_shape_orig}"
            raise ValueError(err_msg)

    return result_shape


cdef shape_type_c get_reduction_output_shape(shape_type_c input_shape, object axis, cpp_bool keepdims):
    cdef shape_type_c result_shape
    cdef tuple axis_tuple = _object_to_tuple(axis)

    if axis is not None:
        for it in range(input_shape.size()):
            if it not in axis_tuple:
                result_shape.push_back(input_shape[it])
            elif keepdims is True:
                result_shape.push_back(1)
    elif keepdims is True:
        for it in range(input_shape.size()):
            result_shape.push_back(1)

    return result_shape


cdef DPNPFuncType get_output_c_type(DPNPFuncName funcID,
                                    DPNPFuncType input_array_c_type,
                                    object requested_out,
                                    object requested_dtype):

    if requested_out is None:
        if requested_dtype is None:
            """ get recommended result type by function ID """
            kernel_data = get_dpnp_function_ptr(funcID, input_array_c_type, input_array_c_type)
            return kernel_data.return_type
        else:
            """ return type by request """
            return dpnp_dtype_to_DPNPFuncType(requested_dtype)
    else:
        if requested_dtype is None:
            """ determined by 'out' parameter """
            return dpnp_dtype_to_DPNPFuncType(requested_out.dtype)

    checker_throw_value_error("get_output_c_type", "dtype and out", requested_dtype, requested_out)


cdef dpnp_descriptor create_output_descriptor(shape_type_c output_shape,
                                              DPNPFuncType c_type,
                                              dpnp_descriptor requested_out,
                                              device=None,
                                              usm_type="device",
                                              sycl_queue=None):
    cdef dpnp_descriptor result_desc

    if requested_out is None:
        result = None
        if sycl_queue is not None:
            device = None
        result_dtype = dpnp_DPNPFuncType_to_dtype(< size_t > c_type)
        result_obj = dpnp_container.empty(output_shape,
                                          dtype=result_dtype,
                                          device=device,
                                          usm_type=usm_type,
                                          sycl_queue=sycl_queue)
        result_desc = dpnp_descriptor(result_obj)
    else:
        """ Based on 'out' parameter """
        if (output_shape != requested_out.shape):
            checker_throw_value_error("create_output_descriptor", "out.shape", requested_out.shape, output_shape)

        if isinstance(requested_out, dpnp_descriptor):
            result_desc = requested_out
        else:
            result_desc = dpnp_descriptor(requested_out)

    return result_desc


cpdef shape_type_c normalize_axis(object axis_obj, size_t shape_size_inp):
    """
    Conversion of the transformation shape axis [-1, 0, 1] into [2, 0, 1] where numbers are `id`s of array shape axis
    """

    cdef shape_type_c axis = _object_to_tuple(axis_obj)  # axis_obj might be a scalar
    cdef ssize_t shape_size = shape_size_inp  # convert type for comparison with axis id

    cdef size_t axis_size = axis.size()
    cdef shape_type_c result = shape_type_c(axis_size, 0)
    for i in range(axis_size):
        if (axis[i] >= shape_size) or (axis[i] < -shape_size):
            checker_throw_axis_error("normalize_axis", "axis", axis[i], shape_size - 1)

        if (axis[i] < 0):
            result[i] = shape_size + axis[i]
        else:
            result[i] = axis[i]

    return result


@cython.profile(False)
cdef inline int _normalize_order(order, cpp_bool allow_k=True) except? 0:
    """ Converts memory order letters to some common view

    """

    cdef int order_type
    order_type = b'C' if len(order) == 0 else ord(order[0])

    if order_type == b'K' or order_type == b'k':
        if not allow_k:
            raise ValueError("DPNP _normalize_order(): order \'K\' is not permitted")
        order_type = b'K'
    elif order_type == b'A' or order_type == b'a':
        order_type = b'A'
    elif order_type == b'C' or order_type == b'c':
        order_type = b'C'
    elif order_type == b'F' or order_type == b'f':
        order_type = b'F'
    else:
        raise TypeError("DPNP _normalize_order(): order is not understood")

    return order_type


@cython.profile(False)
cpdef inline tuple _object_to_tuple(object obj):
    """ Converts Python object into tuple

    """

    if obj is None:
        return ()

    if cpython.PySequence_Check(obj):
        return tuple(obj)

    if dpnp.isscalar(obj):
        return (obj, )

    raise ValueError("DPNP object_to_tuple(): 'obj' should be 'None', collections.abc.Sequence, or 'int'")


cpdef cpp_bool use_origin_backend(input1=None, size_t compute_size=0):
    """
    This function needs to redirect particular computation cases to original backend
    Parameters:
        input1: One of the input parameter of the API function
        compute_size: Some amount of total compute size of the task
    Return:
        True - computations are better to be executed on original backend
        False - it is better to use this SW to compute
    """

    if (config.__DPNP_ORIGIN__):
        return True

    return False


cdef shape_type_c strides_to_vector(object strides, object shape) except *:
    """
    Get or calculate srtides based on shape.
    """
    cdef shape_type_c res
    if strides is None:
        res = get_axis_offsets(shape)
    else:
        res = strides

    return res


cdef tuple get_common_usm_allocation(dpnp_descriptor x1, dpnp_descriptor x2):
    """Get common USM allocation in the form of (sycl_device, usm_type, sycl_queue)."""
    array1_obj = x1.get_array()
    array2_obj = x2.get_array()

    common_usm_type = dpctl.utils.get_coerced_usm_type((array1_obj.usm_type, array2_obj.usm_type))
    if common_usm_type is None:
        raise ValueError(
            "could not recognize common USM type for inputs of USM types {} and {}"
            "".format(array1_obj.usm_type, array2_obj.usm_type))

    common_sycl_queue = dpu.get_execution_queue((array1_obj.sycl_queue, array2_obj.sycl_queue))
    if common_sycl_queue is None:
        raise ValueError(
            "could not recognize common SYCL queue for inputs in SYCL queues {} and {}"
            "".format(array1_obj.sycl_queue, array2_obj.sycl_queue))

    return (common_sycl_queue.sycl_device, common_usm_type, common_sycl_queue)


cdef class dpnp_descriptor:
    def __init__(self, obj, dpnp_descriptor orig_desc=None):
        """ Initialze variables """
        self.origin_pyobj = None
        self.origin_desc = None
        self.descriptor = None
        self.dpnp_descriptor_data_size = 0
        self.dpnp_descriptor_is_scalar = True

        """ Accure DPCTL data container storage """
        self.descriptor = getattr(obj, "__sycl_usm_array_interface__", None)
        if self.descriptor is None:

            """ Accure main data storage """
            self.descriptor = getattr(obj, "__array_interface__", None)
            if self.descriptor is None:
                return

            if self.descriptor["version"] != 3:
                return

        self.origin_pyobj = obj

        """ Keep track of a descriptor with original data """
        if orig_desc is not None and orig_desc.is_valid:
            self.origin_desc = orig_desc

        """ array size calculation """
        cdef Py_ssize_t shape_it = 0
        self.dpnp_descriptor_data_size = 1
        for shape_it in self.shape:
            # TODO need to use common procedure from utils to calculate array size by shape
            if shape_it < 0:
                raise ValueError(f"{ERROR_PREFIX} dpnp_descriptor::__init__() invalid value {shape_it} in 'shape'")
            self.dpnp_descriptor_data_size *= shape_it

        """ set scalar property """
        self.dpnp_descriptor_is_scalar = False

    @property
    def is_valid(self):
        if self.descriptor is None:
            return False
        return True

    @property
    def shape(self):
        if self.is_valid:
            return self.descriptor["shape"]
        return None

    @property
    def strides(self):
        if self.is_valid:
            return self.descriptor["strides"]
        return None

    @property
    def ndim(self):
        if self.is_valid:
            return len(self.shape)
        return 0

    @property
    def size(self):
        if self.is_valid:
            return self.dpnp_descriptor_data_size
        return 0

    @property
    def dtype(self):
        if self.is_valid:
            type_str = self.descriptor["typestr"]
            return dpnp.dtype(type_str)
        return None

    @property
    def offset(self):
        if self.is_valid:
            return self.descriptor.get('offset', 0)
        return 0

    @property
    def is_scalar(self):
        return self.dpnp_descriptor_is_scalar

    @property
    def is_temporary(self):
        """
        Non-none descriptor of original data means the current descriptor
        holds a temporary allocated data.
        """
        return self.origin_desc is not None

    @property
    def data(self):
        if self.is_valid:
            data_tuple = self.descriptor["data"]
            return data_tuple[0]
        return None

    @property
    def descr(self):
        if self.is_valid:
            return self.descriptor["descr"]
        return None

    @property
    def __array_interface__(self):
        # print(f"====dpnp_descriptor::__array_interface__====self.descriptor={ < size_t > self.descriptor}")
        if self.descriptor is None:
            return None

        # TODO need to think about feature compatibility
        interface_dict = {
            "data": self.descriptor["data"],
            "strides": self.descriptor["strides"],
            "descr": self.descriptor["descr"],
            "typestr": self.descriptor["typestr"],
            "shape": self.descriptor["shape"],
            "version": self.descriptor["version"]
        }

        return interface_dict

    def _copy_array_from(self, other_desc):
        """
        Fill array data with usm_ndarray of the same shape from other DPNP descriptor
        """
        if not isinstance(other_desc, dpnp_descriptor):
            raise TypeError("expected dpnp_descriptor, got {}".format(type(other_desc)))

        dpt_cu._copy_same_shape(self.get_array(), other_desc.get_array())

    def get_pyobj(self):
        return self.origin_pyobj

    def get_array(self):
        if isinstance(self.origin_pyobj, dpctl.tensor.usm_ndarray):
            return self.origin_pyobj
        if isinstance(self.origin_pyobj, dpnp_array):
            return self.origin_pyobj.get_array()

        raise TypeError(
            "expected either dpctl.tensor.usm_ndarray or dpnp.dpnp_array.dpnp_array, got {}"
            "".format(type(self.origin_pyobj)))

    def get_result_desc(self, result_desc=None):
        """
        Copy the result data into an original array
        """
        if self.is_temporary:
            # Original descriptor is not None, so copy the array data into it and return
            from_desc = self if result_desc is None else result_desc
            self.origin_desc._copy_array_from(from_desc)
            return self.origin_desc
        elif result_desc is not None:
            # A temporary result descriptor was allocated, needs to copy data back into 'out' descriptor
            self._copy_array_from(result_desc)
        return self

    def is_array_overlapped(self, other_desc):
        """
        Check if usm_ndarray overlaps an array from other DPNP descriptor
        """
        if not isinstance(other_desc, dpnp_descriptor):
            raise TypeError("expected dpnp_descriptor, got {}".format(type(other_desc)))

        return dpt_ti._array_overlap(self.get_array(), other_desc.get_array())

    cdef void * get_data(self):
        cdef Py_ssize_t item_size = 0
        cdef Py_ssize_t elem_offset = 0
        cdef char *data_ptr = NULL
        cdef size_t val = self.data

        if self.offset > 0:
            item_size = self.origin_pyobj.itemsize
            elem_offset = self.offset
            data_ptr = <char *>(val) + item_size * elem_offset
            return < void * > data_ptr

        return < void * > val

    cdef cpp_bool match_ctype(self, DPNPFuncType ctype):
        return self.dtype == dpnp_DPNPFuncType_to_dtype(< size_t > ctype)

    def __bool__(self):
        return self.is_valid

    def __str__(self):
        return str(self.descriptor)

IntelPython / dpnp / 5270840792

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous