/**
 * Copyright (c) 2026 Huawei Technologies Co., Ltd.
 * This file is a part of the CANN Open Software.
 * Licensed under CANN Open Software License Agreement Version 1.0 (the "License").
 * Please refer to the License for details. You may not use this file except in compliance with the License.
 * THIS SOFTWARE IS PROVIDED ON AN "AS IS" BASIS, WITHOUT WARRANTIES OF ANY KIND, EITHER EXPRESS OR IMPLIED,
 * INCLUDING BUT NOT LIMITED TO NON-INFRINGEMENT, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.
 * See LICENSE in the root of the software repository for the full text of the License.
 */

/*!
 * \file test_aclnn_sparse_flash_attention_v2.cpp
 * \brief
 */

#include <iostream>
#include <vector>
#include <cmath>
#include <numeric>
#include "acl/acl.h"
#include "aclnn/opdev/fp16_t.h"
#include "aclnnop/aclnn_sparse_flash_attention_v2.h"

#define CHECK_RET(cond, return_expr) \
    do {                               \
        if (!(cond)) {                   \
            return_expr;                   \
        }                                \
    } while (0)

#define LOG_PRINT(message, ...)     \
    do {                              \
        printf(message, ##__VA_ARGS__); \
    } while (0)

int64_t GetShapeSize(const std::vector<int64_t>& shape) {
    int64_t shapeSize = 1;
    for (auto i : shape) {
        shapeSize *= i;
    }
    return shapeSize;
}

void PrintOutResult(std::vector<int64_t> &shape, void** deviceAddr) {
    auto size = GetShapeSize(shape);
    std::vector<short> resultData(size, 0);
    auto ret = aclrtMemcpy(resultData.data(), resultData.size() * sizeof(resultData[0]),
                           *deviceAddr, size * sizeof(resultData[0]), ACL_MEMCPY_DEVICE_TO_HOST);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("copy result from device to host failed. ERROR: %d\n", ret); return);
    for (int64_t i = 0; i < size; i++) {
        LOG_PRINT("mean result[%ld] is: %e\n", i, resultData[i]);
    }
}

int Init(int32_t deviceId, aclrtContext* context, aclrtStream* stream) {
    // 固定写法，AscendCL初始化
    auto ret = aclInit(nullptr);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclInit failed. ERROR: %d\n", ret); return ret);
    ret = aclrtSetDevice(deviceId);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtSetDevice failed. ERROR: %d\n", ret); return ret);
    ret = aclrtCreateContext(context, deviceId);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtCreateContext failed. ERROR: %d\n", ret); return ret);
    ret = aclrtSetCurrentContext(*context);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtSetCurrentContext failed. ERROR: %d\n", ret); return ret);
    ret = aclrtCreateStream(stream);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtCreateStream failed. ERROR: %d\n", ret); return ret);
    return 0;
}

template <typename T>
int CreateAclTensor(const std::vector<T>& hostData, const std::vector<int64_t>& shape, void** deviceAddr,
                    aclDataType dataType, aclTensor** tensor) {
    auto size = GetShapeSize(shape) * aclDataTypeSize(dataType);
    // 调用aclrtMalloc申请device侧内存
    auto ret = aclrtMalloc(deviceAddr, size, ACL_MEM_MALLOC_HUGE_FIRST);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtMalloc failed. ERROR: %d\n", ret); return ret);
    // 调用aclrtMemcpy将host侧数据拷贝到device侧内存上
    ret = aclrtMemcpy(*deviceAddr, size, hostData.data(), size, ACL_MEMCPY_HOST_TO_DEVICE);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtMemcpy failed. ERROR: %d\n", ret); return ret);

    // 计算连续tensor的strides
    std::vector<int64_t> strides(shape.size(), 1);
    for (int64_t i = shape.size() - 2; i >= 0; i--) {
        strides[i] = shape[i + 1] * strides[i + 1];
    }

    // 调用aclCreateTensor接口创建aclTensor
    *tensor = aclCreateTensor(shape.data(), shape.size(), dataType, strides.data(), 0, aclFormat::ACL_FORMAT_ND,
                              shape.data(), shape.size(), *deviceAddr);
    return 0;
}

int main() {
    // 1. （固定写法）device/context/stream初始化，参考AscendCL对外接口列表
    // 根据自己的实际device填写deviceId
    int32_t deviceId = 0;
    aclrtContext context;
    aclrtStream stream;
    auto ret = Init(deviceId, &context, &stream);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("Init acl failed. ERROR: %d\n", ret); return ret);

    // 2. 构造输入与输出，需要根据API的接口自定义构造
    std::vector<int64_t> qShape = {1, 16, 512};                // T1, N1, D
    std::vector<int64_t> kShape = {2048, 1, 512};              // T2, N2, D
    std::vector<int64_t> vShape = {2048, 1, 512};              // T2, N2, D
    std::vector<int64_t> sparseIndicesShape = {1, 1, 2048};    // T1, N2, K
    std::vector<int64_t> outShape = {1, 16, 512};             // T1, N1, D
    std::vector<int64_t> softmaxMaxShape = {1, 1, 16};        // N2, T1, G
    std::vector<int64_t> softmaxSumShape = {1, 1, 16};        // N2, T1, G
    std::vector<int64_t> actSeqQLenshape = {1};               // B
    std::vector<int64_t> actSeqKvLenshape = {1};           // B
    std::vector<int64_t> qRopeShape = {1, 16, 64};            // T1, N1, Drope
    std::vector<int64_t> kRopeShape = {2048, 1, 64};          // T2, N2, Drope
    std::vector<int64_t> sinksShape = {16};                   // N1

    void* qDeviceAddr = nullptr;
    void* kDeviceAddr = nullptr;
    void* vDeviceAddr = nullptr;
    void* sparseIndicesDeviceAddr = nullptr;
    void* outDeviceAddr = nullptr;
    void* softmaxMaxDeviceAddr = nullptr;
    void* softmaxSumDeviceAddr = nullptr;
    void* actSeqQLenDeviceAddr = nullptr;
    void* actSeqKvLenDeviceAddr = nullptr;
    void* qRopeDeviceAddr = nullptr;
    void* kRopeDeviceAddr = nullptr;
    void* sinksDeviceAddr = nullptr;

    aclTensor* q = nullptr;
    aclTensor* k = nullptr;
    aclTensor* v = nullptr;
    aclTensor* sparseIndices = nullptr;
    aclTensor* out = nullptr;
    aclTensor* softmaxMax = nullptr;
    aclTensor* softmaxSum = nullptr;
    aclTensor* actSeqQLen = nullptr;
    aclTensor* actSeqKvLen = nullptr;
    aclTensor* qRope = nullptr;
    aclTensor* kRope = nullptr;
    aclTensor* sinks = nullptr;

    std::vector<op::fp16_t> qHostData(1 * 16 * 512, 1.0);
    std::vector<op::fp16_t> kHostData(2048 * 1 * 512, 1.0);
    std::vector<op::fp16_t> vHostData(2048 * 1 * 512, 1.0);
    std::vector<int32_t> sparseIndicesHostData(2048);
    std::iota(sparseIndicesHostData.begin(), sparseIndicesHostData.end(), 0);
    std::vector<op::fp16_t> outHostData(1 * 16 * 512, 1.0);
    std::vector<float> softmaxMaxHostData(16, 3.0);
    std::vector<float> softmaxSumHostData(16, 3.0);
    std::vector<int32_t> actSeqQLenHostData(1, 1);
    std::vector<int32_t> actSeqKvLenHostData(1, 2048);
    std::vector<op::fp16_t> qRopeHostData(1 * 16 * 64, 1.0);
    std::vector<op::fp16_t> kRopeHostData(2048 * 1 * 64, 1.0);
    std::vector<float> sinksHostData(16, 0.0);

    ret = CreateAclTensor(qHostData, qShape, &qDeviceAddr, aclDataType::ACL_FLOAT16, &q);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(kHostData, kShape, &kDeviceAddr, aclDataType::ACL_FLOAT16, &k);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(vHostData, vShape, &vDeviceAddr, aclDataType::ACL_FLOAT16, &v);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(sparseIndicesHostData, sparseIndicesShape,
                          &sparseIndicesDeviceAddr, aclDataType::ACL_INT32, &sparseIndices);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(outHostData, outShape, &outDeviceAddr, aclDataType::ACL_FLOAT16, &out);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(softmaxMaxHostData, softmaxMaxShape,
                          &softmaxMaxDeviceAddr, aclDataType::ACL_FLOAT, &softmaxMax);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(softmaxSumHostData, softmaxSumShape,
                          &softmaxSumDeviceAddr, aclDataType::ACL_FLOAT, &softmaxSum);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(actSeqQLenHostData, actSeqQLenshape,
                          &actSeqQLenDeviceAddr, aclDataType::ACL_INT32, &actSeqQLen);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(actSeqKvLenHostData, actSeqKvLenshape,
                          &actSeqKvLenDeviceAddr, aclDataType::ACL_INT32, &actSeqKvLen);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(qRopeHostData, qRopeShape, &qRopeDeviceAddr, aclDataType::ACL_FLOAT16, &qRope);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(kRopeHostData, kRopeShape, &kRopeDeviceAddr, aclDataType::ACL_FLOAT16, &kRope);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(sinksHostData, sinksShape, &sinksDeviceAddr, aclDataType::ACL_FLOAT, &sinks);
    CHECK_RET(ret == ACL_SUCCESS, return ret);

    double scaleValue = 0.0416666666666667;
    int64_t sparseBlockSize = 1;
    int64_t sparseMode = 0;
    int64_t attentionMode = 2;
    int64_t preTokens = 9223372036854775807;
    int64_t nextTokens = 9223372036854775807;
    // bool deterministic = false;
    char layoutQuery[5] = {'T', 'N', 'D', 0};
    char layoutKey[5] = {'T', 'N', 'D', 0};

    // 3. 调用CANN算子库API，需要修改为具体的Api名称
    uint64_t workspaceSize = 0;
    aclOpExecutor* executor = nullptr;
    bool returnSoftmaxLse = false;

    // 调用aclnnSparseFlashAttentionV2第一段接口
    ret = aclnnSparseFlashAttentionV2GetWorkspaceSize(q, k, v, sparseIndices, nullptr, actSeqQLen, actSeqKvLen,
                                                      qRope, kRope, sinks, scaleValue, sparseBlockSize, layoutQuery,
                                                      layoutKey, sparseMode, preTokens, nextTokens, attentionMode,
                                                      returnSoftmaxLse, out, softmaxMax, softmaxSum, &workspaceSize,
                                                      &executor);
    CHECK_RET(ret == ACL_SUCCESS,
        LOG_PRINT("aclnnSparseFlashAttentionV2GetWorkspaceSize failed. ERROR: %d\n", ret); return ret);

    // 根据第一段接口计算出的workspaceSize申请device内存
    void* workspaceAddr = nullptr;
    if (workspaceSize > 0) {
        ret = aclrtMalloc(&workspaceAddr, workspaceSize, ACL_MEM_MALLOC_HUGE_FIRST);
        CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("allocate workspace failed. ERROR: %d\n", ret); return ret);
    }

    // 调用aclnnSparseFlashAttentionV2第二段接口
    ret = aclnnSparseFlashAttentionV2(workspaceAddr, workspaceSize, executor, stream);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnSparseFlashAttentionV2 failed. ERROR: %d\n", ret); return ret);

    // 4. （固定写法）同步等待任务执行结束
    ret = aclrtSynchronizeStream(stream);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtSynchronizeStream failed. ERROR: %d\n", ret); return ret);

    // 5. 释放aclTensor和aclScalar，需要根据具体API的接口定义修改
    aclDestroyTensor(q);
    aclDestroyTensor(k);
    aclDestroyTensor(v);
    aclDestroyTensor(sparseIndices);
    aclDestroyTensor(out);
    aclDestroyTensor(softmaxMax);
    aclDestroyTensor(softmaxSum);
    aclDestroyTensor(actSeqQLen);
    aclDestroyTensor(actSeqKvLen);
    aclDestroyTensor(qRope);
    aclDestroyTensor(kRope);
    aclDestroyTensor(sinks);

    // 6. 释放device资源
    aclrtFree(qDeviceAddr);
    aclrtFree(kDeviceAddr);
    aclrtFree(vDeviceAddr);
    aclrtFree(sparseIndicesDeviceAddr);
    aclrtFree(softmaxMaxDeviceAddr);
    aclrtFree(softmaxSumDeviceAddr);
    aclrtFree(outDeviceAddr);
    aclrtFree(actSeqQLenDeviceAddr);
    aclrtFree(actSeqKvLenDeviceAddr);
    aclrtFree(qRopeDeviceAddr);
    aclrtFree(kRopeDeviceAddr);
    aclrtFree(sinksDeviceAddr);
    if (workspaceSize > 0) {
        aclrtFree(workspaceAddr);
    }
    aclrtDestroyStream(stream);
    aclrtDestroyContext(context);
    aclrtResetDevice(deviceId);
    aclFinalize();

    return 0;
}