// Copyright (c) 2023 Huawei Technologies Co., Ltd
// Copyright (c) 2019, Facebook CORPORATION.
// 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/utils/OpAdapter.h"

namespace acl_op {
const int DIMENSION_3D = 3;
const int DIMENSION_4D = 4;
const int OUTPUT_SIZE = 2;
using npu_preparation = at_npu::native::OpPreparation;
using npu_utils = at_npu::native::NpuUtils;

namespace {
inline void adaptive_max_pool2d_check(const at::Tensor& self, at::IntArrayRef output_size)
{
    for (int64_t i = 0; i < self.dim(); i++) {
        TORCH_CHECK(
            self.size(i) > 0,
            "adaptive_max_pooling2d(): expected input to have non-empty spatial dimensions, "
            "but input has sizes ",
            self.sizes(),
            " with dimension ",
            i,
            " being "
            "empty" + OPS_ERROR(ErrCode::PARAM));
    }
    TORCH_CHECK(
        (self.dim() == DIMENSION_3D || self.dim() == DIMENSION_4D),
        "non-empty 3D or 4D (batch mode) tensor expected for input" + OPS_ERROR(ErrCode::PARAM));
    TORCH_CHECK(
        (output_size.size() == OUTPUT_SIZE),
        "adaptive_max_pool2d: internal error: output_size.size() must be 2" + OPS_ERROR(ErrCode::PARAM));
}

std::tuple<c10::SmallVector<int64_t, SIZE>, c10::SmallVector<int64_t, SIZE>> adaptive_max_pool2d_infer_size(
    const at::Tensor& self,
    at::IntArrayRef output_size)
{
    int64_t n = self.size(0);
    int64_t c = self.size(1);
    int64_t h = self.size(2);
    int64_t w = self.size(3);
    TORCH_CHECK(output_size[0] != 0 && output_size[1] != 0, "out put size cannot not be Zero"
        + OPS_ERROR(ErrCode::PARAM));
    int64_t stride_h = h / output_size[0];
    int64_t stride_w = w / output_size[1];
    int64_t kernel_size_h = h - (output_size[0] - 1) * stride_h;
    int64_t kernel_size_w = w - (output_size[1] - 1) * stride_w;
    int64_t Ho = output_size[0];
    int64_t Wo = output_size[1];
    c10::SmallVector<int64_t, SIZE> output_sizes = {n, c, Ho, Wo};
    const int64_t BLOCKSIZE = 16;
    int64_t mask_h = kernel_size_h * kernel_size_w;
    int64_t mask_w = (op_infer::CeilDiv(Ho * Wo, BLOCKSIZE) + 1);
    c10::SmallVector<int64_t, SIZE> indices_size = {n, c, mask_h, mask_w};

    return std::tuple<c10::SmallVector<int64_t, SIZE>, c10::SmallVector<int64_t, SIZE>>(output_sizes, indices_size);
}

std::tuple<at::Tensor&, at::Tensor&> adaptive_max_pool2d_out_nocheck(
    at::Tensor& output,
    at::Tensor& indices,
    const at::Tensor& self,
    at::IntArrayRef output_size)
{
    npu_preparation::CheckMemory({self}, {output, indices});
    auto inputsize = self.sizes();
    c10::SmallVector<int64_t, N> input_size;
    if (inputsize.size() == DIMENSION_3D) {
        c10::SmallVector<int64_t, N> size = {inputsize[1], inputsize[2]};
        input_size = at::IntArrayRef(size);
    } else if (inputsize.size() == DIMENSION_4D) {
        c10::SmallVector<int64_t, N> size = {inputsize[2], inputsize[3]};
        input_size = at::IntArrayRef(size);
    }

    // H and W can not be divided, temporarily reported error processing
    TORCH_CHECK(input_size[0] % output_size[0] == 0 && input_size[1] % output_size[1] == 0,
        "H and W must be divisible" + OPS_ERROR(ErrCode::PARAM));

    int64_t kernel_size[2];
    int64_t stride[2];
    int64_t padding[2];
    int64_t stride_h = input_size[0] / output_size[0];
    int64_t stride_w = input_size[1] / output_size[1];
    int64_t kernel_size_h = input_size[0] - (output_size[0] - 1) * stride_h;
    int64_t kernel_size_w = input_size[1] - (output_size[1] - 1) * stride_w;
    stride[0] = stride_h;
    stride[1] = stride_w;
    kernel_size[0] = kernel_size_h;
    kernel_size[1] = kernel_size_w;
    padding[0] = padding[1] = 0;
    c10::SmallVector<int64_t, N> kernel_sizes = {1, kernel_size[0], kernel_size[1], 1};
    c10::SmallVector<int64_t, N> strides_size = {1, stride[0], stride[1], 1};
    c10::SmallVector<int64_t, N> paddings = {1, padding[0], padding[1], 1};
    c10::SmallVector<int64_t, N> dilations = {1, 1, 1, 1};
    bool ceil_mode = false;

    at_npu::native::OpCommand cmd;
    cmd.Name("MaxPoolWithArgmaxV1")
        .Input(self, "x")
        .Output(output, "y")
        .Output(indices, "argmax", c10::nullopt, "uint16")
        .Attr("ksize", kernel_sizes)
        .Attr("strides", strides_size)
        .Attr("pads", paddings)
        .Attr("dilation", dilations)
        .Attr("ceil_mode", ceil_mode)
        .Run();

    return std::tuple<at::Tensor&, at::Tensor&>(output, indices);
}
} // namespace

std::tuple<at::Tensor&, at::Tensor&> adaptive_max_pool2d_out(
    const at::Tensor& self,
    at::IntArrayRef output_size,
    at::Tensor& out,
    at::Tensor& indices)
{
    adaptive_max_pool2d_check(self, output_size);
    c10::SmallVector<int64_t, SIZE> output_sizes = std::get<0>(adaptive_max_pool2d_infer_size(self, output_size));
    c10::SmallVector<int64_t, SIZE> indices_size = std::get<1>(adaptive_max_pool2d_infer_size(self, output_size));

    npu_preparation::CheckOut(
        {self},
        out,
        self,
        output_sizes);
    npu_preparation::CheckOut(
        {self},
        indices,
        ACL_FORMAT_NC1HWC0,
        at::ScalarType::Long,
        indices_size);

    bool out_match = npu_utils::check_match(&out);
    bool indices_match = npu_utils::check_match(&indices);
    if (!(out_match && indices_match)) {
        at::Tensor contiguous_out = out_match ? out : npu_utils::format_contiguous(out);
        at::Tensor contiguous_indices = indices_match ? indices : npu_utils::format_contiguous(indices);
        adaptive_max_pool2d_out_nocheck(contiguous_out, contiguous_indices, self, output_size);
        if (!out_match) {
            npu_utils::format_fresh_view(out, contiguous_out);
        }
        if (!indices_match) {
            npu_utils::format_fresh_view(indices, contiguous_indices);
        }
    } else {
        adaptive_max_pool2d_out_nocheck(out, indices, self, output_size);
    }

    return std::tuple<at::Tensor&, at::Tensor&>(out, indices);
}

std::tuple<at::Tensor, at::Tensor> adaptive_max_pool2d(
    const at::Tensor& self,
    at::IntArrayRef output_size)
{
    adaptive_max_pool2d_check(self, output_size);
    c10::SmallVector<int64_t, SIZE> output_sizes = std::get<0>(adaptive_max_pool2d_infer_size(self, output_size));
    c10::SmallVector<int64_t, SIZE> indices_size = std::get<1>(adaptive_max_pool2d_infer_size(self, output_size));
    at::Tensor output = npu_preparation::apply_tensor(self, output_sizes);
    at::Tensor indices = npu_preparation::apply_tensor_with_format(indices_size,
        self.options().dtype(at::kShort), ACL_FORMAT_NC1HWC0);
    adaptive_max_pool2d_out_nocheck(output, indices, self, output_size);

    return std::tuple<at::Tensor, at::Tensor>(output, indices);
}
} // namespace acl_op