/*
 * Copyright (c) 2019 ARM Limited.
 *
 * SPDX-License-Identifier: MIT
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to
 * deal in the Software without restriction, including without limitation the
 * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
 * sell copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in all
 * copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 */
#include "arm_compute/core/NEON/kernels/NEConvertQuantizedSignednessKernel.h"

#include "arm_compute/core/Error.h"
#include "arm_compute/core/Helpers.h"
#include "arm_compute/core/ITensor.h"
#include "arm_compute/core/NEON/wrapper/wrapper.h"
#include "arm_compute/core/TensorInfo.h"
#include "arm_compute/core/Validate.h"
#include "arm_compute/core/Window.h"

namespace arm_compute
{
namespace
{
Status validate_arguments(const ITensorInfo *input, const ITensorInfo *output)
{
    ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input, output);
    ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::QASYMM8, DataType::QASYMM8_SIGNED);

    // Validate output if initialized
    if(output->total_size() != 0)
    {
        ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(output, 1, DataType::QASYMM8, DataType::QASYMM8_SIGNED);
        ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DIMENSIONS(input->tensor_shape(), output->tensor_shape());
    }

    return Status{};
}

std::pair<Status, Window> validate_and_configure_window(ITensorInfo *input, ITensorInfo *output)
{
    // Output auto inizialitation if not yet initialized
    {
        const bool                    is_input_signed   = input->data_type() == DataType::QASYMM8_SIGNED;
        const DataType                dt                = is_input_signed ? DataType::QASYMM8 : DataType::QASYMM8_SIGNED;
        const UniformQuantizationInfo qinfo             = input->quantization_info().uniform();
        const int                     offset_correction = is_input_signed ? -128 : 128;
        const QuantizationInfo        corrected_qinfo   = QuantizationInfo(qinfo.scale, qinfo.offset + offset_correction);

        auto_init_if_empty(*output, input->clone()->set_data_type(dt).set_quantization_info(corrected_qinfo));
    }

    return std::make_pair(Status{}, calculate_max_window(*output));
}
} // namespace

NEConvertQuantizedSignednessKernel::NEConvertQuantizedSignednessKernel()
    : _input(nullptr), _output(nullptr)
{
}

void NEConvertQuantizedSignednessKernel::configure(const ITensor *input, ITensor *output)
{
    ARM_COMPUTE_ERROR_ON_NULLPTR(input, output);
    ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(input->info(), output->info()));

    _input  = input;
    _output = output;

    std::pair<Status, Window> win_config = validate_and_configure_window(input->info(), output->info());
    ARM_COMPUTE_ERROR_THROW_ON(win_config.first);
    INEKernel::configure(win_config.second);
}

Status NEConvertQuantizedSignednessKernel::validate(const arm_compute::ITensorInfo *input, const arm_compute::ITensorInfo *output)
{
    ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(input, output));
    ARM_COMPUTE_RETURN_ON_ERROR(validate_and_configure_window(input->clone().get(), output->clone().get()).first);
    return Status{};
}

void NEConvertQuantizedSignednessKernel::run(const Window &window, const ThreadInfo &info)
{
    ARM_COMPUTE_UNUSED(info);
    ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
    ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(INEKernel::window(), window);

    Window win_collapsed = window.collapse_if_possible(window, Window::DimZ);
    win_collapsed.set(Window::DimX, Window::Dimension(0, 1, 1));

    Iterator input(_input, win_collapsed);
    Iterator output(_output, win_collapsed);

    const int  window_step_x  = 16;
    const auto window_start_x = static_cast<int>(window.x().start());
    const auto window_end_x   = static_cast<int>(window.x().end());

    const uint8_t mask  = 128;
    const auto    vmask = wrapper::vdup_n(mask, wrapper::traits::vector_128_tag{});

    execute_window_loop(win_collapsed, [&](const Coordinates &)
    {
        const auto input_ptr  = reinterpret_cast<const uint8_t *>(input.ptr());
        const auto output_ptr = reinterpret_cast<uint8_t *>(output.ptr());

        // Compute S elements per iteration
        int x = window_start_x;
        for(; x <= (window_end_x - window_step_x); x += window_step_x)
        {
            const auto vin = wrapper::vloadq(input_ptr + x);
            wrapper::vstore(output_ptr + x, wrapper::veor(vin, vmask));
        }

        // Compute left-over elements
        for(; x < window_end_x; ++x)
        {
            const uint8_t in  = *(reinterpret_cast<const uint8_t *>(input_ptr + x));
            *(output_ptr + x) = in ^ mask;
        }
    },
    input, output);
}
} // namespace arm_compute