// Copyright (c) 2023 Huawei Technologies Co., Ltd
// All rights reserved.
//
// Licensed under the BSD 3-Clause License  (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// https://opensource.org/licenses/BSD-3-Clause
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

#include "op_plugin/AclOpsInterface.h"
#include "op_plugin/OpApiInterface.h"
#include "op_plugin/utils/op_api_common.h"
#include "op_plugin/utils/KernelNpuOutputSize.h"

namespace op_api {
using small_vector = c10::SmallVector<int64_t, op_infer::SIZE>;

at::Tensor &avg_pool2d_out_npu_nocheck_opapi(at::Tensor &result, const at::Tensor &self, at::IntArrayRef kernel,
                                             at::IntArrayRef stride, at::IntArrayRef padding, bool ceil_mode,
                                             bool count_include_pad, c10::optional<int64_t> divisor_override)
{
    int64_t s_divisor_override = 0;
    if (divisor_override.has_value()) {
        s_divisor_override = divisor_override.value();
        TORCH_CHECK(s_divisor_override != 0, "divisor must be not zero", OPS_ERROR(ErrCode::VALUE));
    }

    const int8_t cube_math_type = at_npu::native::OpPreparation::get_cube_math_type(false);
    EXEC_NPU_CMD(aclnnAvgPool2d, self, kernel, stride, padding, ceil_mode, count_include_pad, s_divisor_override,
                 cube_math_type, result);

    return result;
}

small_vector calc_output_size_with_generalized_attrs(const at::Tensor &self, at::IntArrayRef kernel_size,
                                                     at::IntArrayRef stride, at::IntArrayRef padding, bool ceil_mode,
                                                     bool count_include_pad, c10::optional<int64_t> divisor_override)
{
    // generalize kernels, strides and paddings to 2D-shape
    TORCH_CHECK(!kernel_size.empty(), "kernel_size must either be a single int, or a tuple of two ints",
        OPS_ERROR(ErrCode::PARAM));
    const int64_t k_h = kernel_size[0];
    const int64_t k_w = kernel_size.size() == 1 ? k_h : kernel_size[1];
    c10::SmallVector<int64_t, op_infer::SIZE> kernel_sizes = {k_h, k_w};
    at::IntArrayRef kernels = at::IntArrayRef(kernel_sizes);

    const int64_t s_h = stride.empty() ? k_h : stride[0];
    const int64_t s_w = stride.empty() ? k_w : stride.size() == 1 ? s_h : stride[1];
    c10::SmallVector<int64_t, op_infer::SIZE> stride_sizes = {s_h, s_w};
    TORCH_CHECK(s_h != 0 && s_w != 0, "stride should not be zero", OPS_ERROR(ErrCode::VALUE));
    at::IntArrayRef strides = at::IntArrayRef(stride_sizes);

    const int64_t pad_h = padding[0];
    const int64_t pad_w = padding.size() == 1 ? pad_h : padding[1];
    c10::SmallVector<int64_t, op_infer::SIZE> padding_sizes = {pad_h, pad_w};
    TORCH_CHECK(pad_h >= 0 && pad_w >= 0, "pad should not be less than 0", OPS_ERROR(ErrCode::VALUE));
    TORCH_CHECK(pad_h <= k_h / 2 && pad_w <= k_w / 2, "pad should be smaller than or equal to half of kernel size",
        OPS_ERROR(ErrCode::VALUE));
    at::IntArrayRef paddings = at::IntArrayRef(padding_sizes);

    auto output_size = op_infer::avg_pool2d_npu_output_size(self, kernels, strides, paddings, ceil_mode);

    return output_size;
}

at::Tensor &avg_pool2d_out(const at::Tensor &self, at::IntArrayRef kernel_size, at::IntArrayRef stride,
                           at::IntArrayRef padding, bool ceil_mode, bool count_include_pad,
                           c10::optional<int64_t> divisor_override, at::Tensor &out)
{
    c10::SmallVector<int64_t, op_infer::SIZE> output_size = calc_output_size_with_generalized_attrs(
        self, kernel_size, stride, padding, ceil_mode, count_include_pad, divisor_override);
    at_npu::native::OpPreparation::check_tensor({self}, out, self, output_size);

    DO_COMPATIBILITY(aclnnAvgPool2d, acl_op::avg_pool2d_out(self, kernel_size, stride, padding, ceil_mode,
                                                            count_include_pad, divisor_override, out));

    avg_pool2d_out_npu_nocheck_opapi(out, self, kernel_size, stride, padding, ceil_mode, count_include_pad,
                                     divisor_override);

    return out;
}

at::Tensor avg_pool2d(const at::Tensor &self, at::IntArrayRef kernel_size, at::IntArrayRef stride,
                      at::IntArrayRef padding, bool ceil_mode, bool count_include_pad,
                      c10::optional<int64_t> divisor_override)
{
    c10::SmallVector<int64_t, op_infer::SIZE> output_size = calc_output_size_with_generalized_attrs(
        self, kernel_size, stride, padding, ceil_mode, count_include_pad, divisor_override);

    at::Tensor result = at_npu::native::OpPreparation::apply_tensor_without_format(self, output_size);

    DO_COMPATIBILITY(aclnnAvgPool2d, acl_op::avg_pool2d(self, kernel_size, stride, padding, ceil_mode,
                                                        count_include_pad, divisor_override));

    avg_pool2d_out_npu_nocheck_opapi(result, self, kernel_size, stride, padding, ceil_mode, count_include_pad,
                                     divisor_override);

    return result;
}

}