/**
 * Copyright (c) 2025 Huawei Technologies Co., Ltd.
 * This program is free software, you can redistribute it and/or modify it under the terms and conditions of
 * CANN Open Software License Agreement Version 2.0 (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.
 */
#include "aclnn_pow.h"
#include "pow.h"
#include "aclnn_kernels/contiguous.h"
#include "conversion/fill/op_api/fill.h"
#include "math/pows/op_api/pows.h"
#include "aclnn_kernels/cast.h"
#include "math/square/op_api/square.h"
#include "aclnn_kernels/common/op_error_check.h"
#include "op_api/op_api_def_math.h"
#include "op_api/aclnn_check.h"
#include "aclnn/aclnn_base.h"
#include "opdev/common_types.h"
#include "opdev/data_type_utils.h"
#include "opdev/format_utils.h"
#include "opdev/op_dfx.h"
#include "opdev/op_executor.h"
#include "opdev/shape_utils.h"
#include "opdev/tensor_view_utils.h"
#include "opdev/platform.h"
#include "opdev/op_log.h"
#include <cmath>

using namespace op;
#ifdef __cplusplus
extern "C" {
#endif

const float SQRT_EXP = 0.5;
const float NOOP_EXP = 1.0;
const float SQUARE_EXP = 2.0;
const float CUBE_EXP = 3.0;
const float NEGTIVE_SQRT_EXP = -0.5;
const float NEGTIVE_ONE_EXP = -1.0;
const float NEGTIVE_SQUARE_EXP = -2.0;

constexpr float EXP_EPSILON = 1e-6f;

static const std::initializer_list<op::DataType> DTYPE_SUPPORT_LIST = {
    op::DataType::DT_FLOAT, op::DataType::DT_INT32,     op::DataType::DT_INT64,      op::DataType::DT_FLOAT16,
    op::DataType::DT_INT8,  op::DataType::DT_UINT8,     op::DataType::DT_DOUBLE,     op::DataType::DT_BOOL,
    op::DataType::DT_INT16, op::DataType::DT_COMPLEX64, op::DataType::DT_COMPLEX128, op::DataType::DT_BF16};

// out的数据类型支持列表(比输入列表多UINT16/UINT32/UINT64,与API资料一致)
static const std::initializer_list<op::DataType> OUT_DTYPE_SUPPORT_LIST = {
    op::DataType::DT_FLOAT,     op::DataType::DT_INT32,      op::DataType::DT_INT64,  op::DataType::DT_FLOAT16,
    op::DataType::DT_INT8,      op::DataType::DT_UINT8,      op::DataType::DT_DOUBLE, op::DataType::DT_INT16,
    op::DataType::DT_COMPLEX64, op::DataType::DT_COMPLEX128, op::DataType::DT_BF16,   op::DataType::DT_UINT16,
    op::DataType::DT_UINT32,    op::DataType::DT_UINT64};

static const std::initializer_list<op::DataType> SQUARE_NEED_CAST_DTYPE_LIST = {
    op::DataType::DT_INT8, op::DataType::DT_UINT8, op::DataType::DT_BOOL, op::DataType::DT_INT16};

static const std::initializer_list<op::DataType> POWS_DTYPE_SUPPORT_LIST = {
    op::DataType::DT_FLOAT16, op::DataType::DT_FLOAT, op::DataType::DT_BF16};

static op::DataType GetScalarDefaultDtype(const op::DataType input)
{
    if (IsComplexType(input)) {
        return op::DataType::DT_COMPLEX64;
    } else if (IsFloatingType(input)) {
        return op::DataType::DT_FLOAT;
    }
    return input;
}

static op::DataType InnerTypeToComplexType(const op::DataType input)
{
    switch (input) {
        case op::DataType::DT_BF16:
            // BFloat16 has range equivalent to Float,
            // so we map it to ComplexFloat.
            return op::DataType::DT_COMPLEX64;
        case op::DataType::DT_FLOAT16:
            return op::DataType::DT_COMPLEX32;
        case op::DataType::DT_FLOAT:
            return op::DataType::DT_COMPLEX64;
        case op::DataType::DT_DOUBLE:
            return op::DataType::DT_COMPLEX128;
        case op::DataType::DT_COMPLEX32:
            return op::DataType::DT_COMPLEX32;
        case op::DataType::DT_COMPLEX64:
            return op::DataType::DT_COMPLEX64;
        case op::DataType::DT_COMPLEX128:
            return op::DataType::DT_COMPLEX128;
        default:
            OP_LOGE(ACLNN_ERR_PARAM_INVALID, "Unknown Complex ScalarType for [%s]", ToString(input).GetString());
            return op::DataType::DT_UNDEFINED;
    }
}

static op::DataType CombineCategoriesWithComplex(const op::DataType higher, const op::DataType lower)
{
    if (IsComplexType(higher)) {
        return higher;
    } else if (IsComplexType(lower)) {
        // preserve value type of higher if it is floating type.
        if (IsFloatingType(higher)) {
            return InnerTypeToComplexType(higher);
        }
        // in case of integral input
        // lower complex takes precedence.
        return lower;
    } else if (IsFloatingType(higher)) {
        return higher;
    }
    if (higher == op::DataType::DT_BOOL || IsFloatingType(lower)) {
        return op::PromoteType(higher, lower);
    }
    if (higher != op::DataType::DT_UNDEFINED) {
        return higher;
    }
    return lower;
}

static bool CheckPowTensorScalarNotNull(const aclTensor* self, const aclScalar* exponent, const aclTensor* out)
{
    OP_CHECK_NULL(self, return false);
    OP_CHECK_NULL(exponent, return false);
    OP_CHECK_NULL(out, return false);
    return true;
}

static bool CheckPowScalarTensorNotNull(const aclScalar* self, const aclTensor* exponent, const aclTensor* out)
{
    OP_CHECK_NULL(self, return false);
    OP_CHECK_NULL(exponent, return false);
    OP_CHECK_NULL(out, return false);
    return true;
}

static inline bool CheckSocVersionIsSupportBf16(void)
{
    return (GetCurrentPlatformInfo().GetSocVersion() >= SocVersion::ASCEND910B &&
            GetCurrentPlatformInfo().GetSocVersion() <= SocVersion::ASCEND910E) ||
           IsRegBase();
}

// 判断910B芯片上,pow是否走AICPU路径
static inline bool IsPowAiCpuOn910B(const op::DataType dtype, const aclScalar* self, const aclTensor* exponent)
{
    auto socVersion = GetCurrentPlatformInfo().GetSocVersion();
    if (socVersion < SocVersion::ASCEND910B || socVersion > SocVersion::ASCEND910E) {
        return false;
    }
    if (IsRegBase()) {
        return false;
    }
    // self或exponent为bool或bf16类型时,不走AICPU路径
    const auto selfDtype = self->GetDataType();
    const auto expDtype = exponent->GetDataType();
    if (selfDtype == op::DataType::DT_BOOL || expDtype == op::DataType::DT_BOOL || selfDtype == op::DataType::DT_BF16 ||
        expDtype == op::DataType::DT_BF16) {
        return false;
    }
    // 910B AICORE支持的dtype
    static const std::initializer_list<op::DataType> AICORE_DTYPE_LIST = {
        op::DataType::DT_FLOAT, op::DataType::DT_FLOAT16, op::DataType::DT_INT32,
        op::DataType::DT_INT8,  op::DataType::DT_UINT8,   op::DataType::DT_BF16};
    return !CheckType(dtype, AICORE_DTYPE_LIST);
}

static bool CheckDtypeValid(const op::DataType selfDtype, const op::DataType expDtype, const op::DataType outDtype)
{
    if (!CheckSocVersionIsSupportBf16() && (selfDtype == op::DataType::DT_BF16 || expDtype == op::DataType::DT_BF16)) {
        OP_LOGE(ACLNN_ERR_PARAM_INVALID, "Input dtype of pow is not support bfloat16 in current socversion.");
        return false;
    }
    if (!CheckType(selfDtype, DTYPE_SUPPORT_LIST)) {
        OP_LOGE(ACLNN_ERR_PARAM_INVALID, "Self dtype %s should be in dtype support list %s.",
                op::ToString(selfDtype).GetString(), op::ToString(DTYPE_SUPPORT_LIST).GetString());
        return false;
    }
    if (!CheckType(expDtype, DTYPE_SUPPORT_LIST)) {
        OP_LOGE(ACLNN_ERR_PARAM_INVALID, "exp dtype %s should be in dtype support list %s.",
                op::ToString(expDtype).GetString(), op::ToString(DTYPE_SUPPORT_LIST).GetString());
        return false;
    }
    // 检查self和exponent能否做数据类型推导
    op::DataType promoteType = op::PromoteType(selfDtype, expDtype);
    if (promoteType == DataType::DT_UNDEFINED) {
        OP_LOGE(ACLNN_ERR_PARAM_INVALID, "self dtype %s and exponent dtype %s can not promote dtype.",
                op::ToString(selfDtype).GetString(), op::ToString(expDtype).GetString());
        return false;
    }
    if (GetCurrentPlatformInfo().GetCurNpuArch() == NpuArch::DAV_2201 && !CheckType(outDtype, OUT_DTYPE_SUPPORT_LIST)) {
        OP_LOGE(ACLNN_ERR_PARAM_INVALID, "out dtype %s should be in out dtype support list %s.",
                op::ToString(outDtype).GetString(), op::ToString(OUT_DTYPE_SUPPORT_LIST).GetString());
        return false;
    }
    OP_CHECK_RESULT_DTYPE_CAST_FAILED(promoteType, outDtype, return false);
    return true;
}

static inline bool isFloatType(const DataType type)
{
    return type == op::DataType::DT_DOUBLE || type == op::DataType::DT_FLOAT || type == op::DataType::DT_FLOAT16 ||
           type == op::DataType::DT_BF16;
}

static inline op::DataType InferTensorScalarDtype(const aclTensor* self, const aclScalar* exponent,
                                                  const aclTensor* out)
{
    if (IsRegBase()) {
        auto scalarDefaultDtype = GetScalarDefaultDtype(exponent->GetDataType());
        auto promoteType = CombineCategoriesWithComplex(self->GetDataType(), scalarDefaultDtype);
        if (promoteType == DataType::DT_COMPLEX32) {
            promoteType = DataType::DT_COMPLEX64;
        }
        return promoteType;
    }
    if (exponent->GetDataType() == op::DataType::DT_DOUBLE && out->GetDataType() == op::DataType::DT_FLOAT) {
        return op::DataType::DT_FLOAT;
    }

    if (IsComplexType(exponent->GetDataType())) {
        return PromoteType(self->GetDataType(), exponent->GetDataType());
    }
    return isFloatType(self->GetDataType()) ?
               self->GetDataType() :
               ((isFloatType(exponent->GetDataType()) || self->GetDataType() == op::DataType::DT_BOOL) ?
                    PromoteType(self->GetDataType(), exponent->GetDataType()) :
                    self->GetDataType());
}

static inline op::DataType InferScalarTensorDtype(const aclScalar* self, const aclTensor* exponent,
                                                  const aclTensor* out)
{
    if (IsRegBase()) {
        auto scalarDefaultDtype = GetScalarDefaultDtype(self->GetDataType());
        auto promoteType = CombineCategoriesWithComplex(exponent->GetDataType(), scalarDefaultDtype);
        if (promoteType == DataType::DT_COMPLEX32) {
            promoteType = DataType::DT_COMPLEX64;
        }
        return promoteType;
    }
    if (exponent->GetDataType() == op::DataType::DT_DOUBLE && out->GetDataType() == op::DataType::DT_FLOAT) {
        return op::DataType::DT_FLOAT;
    }

    if (IsComplexType(self->GetDataType())) {
        return PromoteType(self->GetDataType(), exponent->GetDataType());
    }
    return isFloatType(exponent->GetDataType()) ?
               exponent->GetDataType() :
               ((isFloatType(self->GetDataType()) || exponent->GetDataType() == op::DataType::DT_BOOL) ?
                    PromoteType(exponent->GetDataType(), self->GetDataType()) :
                    exponent->GetDataType());
}

static bool CheckPromoteType(const op::DataType selfDtype, const op::DataType exponentDtype,
                             const op::DataType outDtype, op::DataType promoteType)
{
    if (promoteType == op::DataType::DT_UNDEFINED) {
        OP_LOGE(ACLNN_ERR_PARAM_INVALID, "Self dtype %s and exponent dtype %s can not promote dtype.",
                op::ToString(selfDtype).GetString(), op::ToString(exponentDtype).GetString());
        return false;
    }
    if ((selfDtype == op::DataType::DT_BOOL) && (exponentDtype == op::DataType::DT_BOOL)) {
        OP_LOGE(ACLNN_ERR_PARAM_INVALID, "Self and exponent dtype are bool is not supported.");
        return false;
    }
    OP_CHECK_RESULT_DTYPE_CAST_FAILED(promoteType, outDtype, return false);
    return true;
}

static bool CheckShape(const aclTensor* self, const aclTensor* out)
{
    OP_CHECK_MAX_DIM(self, MAX_SUPPORT_DIMS_NUMS, return false);
    OP_CHECK_SHAPE_NOT_EQUAL(self, out, return false);
    return true;
}

static aclnnStatus CheckPowTensorScalarParams(const aclTensor* self, const aclScalar* exponent, const aclTensor* out)
{
    // 1. 检查参数是否为空指针
    CHECK_RET(CheckPowTensorScalarNotNull(self, exponent, out), ACLNN_ERR_PARAM_NULLPTR);

    // 2. 检查输入的数据类型是否在API支持的数据类型范围之内,需要根据api定义校验
    CHECK_RET(CheckDtypeValid(self->GetDataType(), exponent->GetDataType(), out->GetDataType()),
              ACLNN_ERR_PARAM_INVALID);

    op::DataType promoteType = InferTensorScalarDtype(self, exponent, out);
    CHECK_RET(CheckPromoteType(self->GetDataType(), exponent->GetDataType(), out->GetDataType(), promoteType),
              ACLNN_ERR_PARAM_INVALID);

    // 3. 检查输入shape
    CHECK_RET(CheckShape(self, out), ACLNN_ERR_PARAM_INVALID);

    return ACLNN_SUCCESS;
}

static bool CheckPowTensorScalarExponet(const DataType inputDtype, const aclScalar* exponent)
{
    // promoteType为整形的情况,exponent需要大于0
    if (IsIntegralType(inputDtype) && (exponent->ToInt64() < 0)) {
        OP_LOGE(ACLNN_ERR_PARAM_INVALID, "Base dtype is integral and exponent negative integers is not allowed.");
        return false;
    }
    return true;
    ;
}

static bool CheckNotOverflow(const DataType pmtType, const aclScalar* exponent)
{
    int8_t overFlowFlag = 1;
    int8_t floatFlag = 1;
    int8_t intFlag = 2;
    int8_t complexFlag = 3;

    switch (pmtType) {
        case op::DataType::DT_FLOAT: {
            overFlowFlag = exponent->CheckOverflows<float>() ? overFlowFlag << floatFlag : overFlowFlag;
            break;
        }
        case op::DataType::DT_FLOAT16: {
            overFlowFlag = exponent->CheckOverflows<op::fp16_t>() ? overFlowFlag << floatFlag : overFlowFlag;
            break;
        }
        case op::DataType::DT_BF16: {
            overFlowFlag = exponent->CheckOverflows<op::bfloat16>() ? overFlowFlag << floatFlag : overFlowFlag;
            break;
        }
        case op::DataType::DT_INT8: {
            overFlowFlag = exponent->CheckOverflows<int8_t>() ? overFlowFlag << intFlag : overFlowFlag;
            break;
        }
        case op::DataType::DT_INT16: {
            overFlowFlag = exponent->CheckOverflows<int16_t>() ? overFlowFlag << intFlag : overFlowFlag;
            break;
        }
        case op::DataType::DT_INT32: {
            overFlowFlag = exponent->CheckOverflows<int32_t>() ? overFlowFlag << intFlag : overFlowFlag;
            break;
        }
        case op::DataType::DT_INT64: {
            overFlowFlag = exponent->CheckOverflows<int64_t>() ? overFlowFlag << intFlag : overFlowFlag;
            break;
        }
        case op::DataType::DT_UINT8: {
            overFlowFlag = exponent->CheckOverflows<uint8_t>() ? overFlowFlag << intFlag : overFlowFlag;
            break;
        }
        case op::DataType::DT_COMPLEX32:
        case op::DataType::DT_COMPLEX64: {
            overFlowFlag = exponent->CheckOverflows<std::complex<float>>() ? overFlowFlag << complexFlag : overFlowFlag;
            break;
        }
        default: {
            return true;
        }
    }

    if ((overFlowFlag >> floatFlag) == 1) {
        OP_LOGE(ACLNN_ERR_PARAM_INVALID, "exponent value cannot be converted to promote type %s without overflow: %lf.",
                op::ToString(pmtType).GetString(), exponent->ToDouble());
    } else if ((overFlowFlag >> intFlag) == 1) {
        OP_LOGE(ACLNN_ERR_PARAM_INVALID, "exponent value cannot be converted to promote type %s without overflow: %ld.",
                op::ToString(pmtType).GetString(), exponent->ToInt64());
    } else if ((overFlowFlag >> complexFlag) == 1) {
        OP_LOGE(ACLNN_ERR_PARAM_INVALID,
                "exponent value cannot be converted to promote type %s without overflow: real is %f, imag is %f.",
                op::ToString(pmtType).GetString(), exponent->ToComplex64().real(), exponent->ToComplex64().imag());
    }
    return overFlowFlag == 1;
}

static aclnnStatus CheckPowScalarTensorParams(const aclScalar* self, const aclTensor* exponent, const aclTensor* out)
{
    // 1. 检查参数是否为空指针
    CHECK_RET(CheckPowScalarTensorNotNull(self, exponent, out), ACLNN_ERR_PARAM_NULLPTR);

    // 2. 检查输入的数据类型是否在API支持的数据类型范围之内,需要根据api定义校验
    CHECK_RET(CheckDtypeValid(self->GetDataType(), exponent->GetDataType(), out->GetDataType()),
              ACLNN_ERR_PARAM_INVALID);

    op::DataType promoteType = InferScalarTensorDtype(self, exponent, out);
    CHECK_RET(CheckPromoteType(self->GetDataType(), exponent->GetDataType(), out->GetDataType(), promoteType),
              ACLNN_ERR_PARAM_INVALID);

    // 3. 检查输入shape
    CHECK_RET(CheckShape(exponent, out), ACLNN_ERR_PARAM_INVALID);

    return ACLNN_SUCCESS;
}

static bool CheckSupportPows(const aclTensor* selfCast, const aclScalar* exponent)
{
    if (std::abs(exponent->ToFloat() - SQRT_EXP) > EXP_EPSILON &&
        std::abs(exponent->ToFloat() - SQUARE_EXP) > EXP_EPSILON &&
        std::abs(exponent->ToFloat() - CUBE_EXP) > EXP_EPSILON &&
        std::abs(exponent->ToFloat() - NEGTIVE_SQRT_EXP) > EXP_EPSILON &&
        std::abs(exponent->ToFloat() - NEGTIVE_ONE_EXP) > EXP_EPSILON &&
        std::abs(exponent->ToFloat() - NEGTIVE_SQUARE_EXP) > EXP_EPSILON) {
        return false;
    }

    if (!CheckType(selfCast->GetDataType(), POWS_DTYPE_SUPPORT_LIST)) {
        return false;
    }

    if (GetCurrentPlatformInfo().GetSocVersion() != SocVersion::ASCEND310P &&
        GetCurrentPlatformInfo().GetSocVersion() != SocVersion::ASCEND910B &&
        GetCurrentPlatformInfo().GetSocVersion() != SocVersion::ASCEND910_93) {
        return false;
    }

    return true;
}

static void CheckFormat(const aclTensor* self)
{
    ge::Format selfStorageFormat = self->GetStorageFormat();
    if (selfStorageFormat != ge::Format::FORMAT_ND) {
        OP_LOGW("aclnnPowTensorScalar/aclnnInplacePowTensorScalar only support format ND.");
    }
}

aclnnStatus aclnnPowTensorScalarGetWorkspaceSize(const aclTensor* self, const aclScalar* exponent, const aclTensor* out,
                                                 uint64_t* workspaceSize, aclOpExecutor** executor)
{
    L2_DFX_PHASE_1(aclnnPowTensorScalar, DFX_IN(self, exponent), DFX_OUT(out));

    // 固定写法,创建OpExecutor
    auto uniqueExecutor = CREATE_EXECUTOR();
    CHECK_RET(uniqueExecutor.get() != nullptr, ACLNN_ERR_INNER_CREATE_EXECUTOR);

    // 固定写法,参数检查
    auto ret = CheckPowTensorScalarParams(self, exponent, out);
    CHECK_RET(ret == ACLNN_SUCCESS, ret);

    CheckFormat(self);

    // pow算子的空tensor在kernel中支持,对标竞品根据算子实际情况补充
    if (self->IsEmpty()) {
        *workspaceSize = 0;
        uniqueExecutor.ReleaseTo(executor);
        return ACLNN_SUCCESS;
    }
    auto promoteType = InferTensorScalarDtype(self, exponent, out);
    if (IsRegBase()) {
        // promoteType为整形的情况,exponent需要大于0
        CHECK_RET(CheckPowTensorScalarExponet(promoteType, exponent), ACLNN_ERR_PARAM_INVALID);

        // 检查exponent是否溢出
        CHECK_RET(CheckNotOverflow(promoteType, exponent), ACLNN_ERR_PARAM_INVALID);
    }
    // 固定写法,将输入self转换成连续的tensor
    auto selfContiguous = l0op::Contiguous(self, uniqueExecutor.get());
    CHECK_RET(selfContiguous != nullptr, ACLNN_ERR_INNER_NULLPTR);

    auto selfCast = l0op::Cast(selfContiguous, promoteType, uniqueExecutor.get());
    CHECK_RET(selfCast != nullptr, ACLNN_ERR_INNER_NULLPTR);

    const aclTensor* powOut = nullptr;
    bool canUseSquare = std::abs(exponent->ToFloat() - SQUARE_EXP) < EXP_EPSILON &&
                        (!IsRegBase() || (IsRegBase() && (selfCast->GetDataType() == op::DataType::DT_FLOAT ||
                                                          selfCast->GetDataType() == op::DataType::DT_BF16 ||
                                                          selfCast->GetDataType() == op::DataType::DT_FLOAT16 ||
                                                          selfCast->GetDataType() == op::DataType::DT_INT64 ||
                                                          selfCast->GetDataType() == op::DataType::DT_COMPLEX64 ||
                                                          selfCast->GetDataType() == op::DataType::DT_COMPLEX128)));
    bool canNoUseOp = std::abs(exponent->ToFloat() - NOOP_EXP) < EXP_EPSILON && !IsRegBase();
    if (CheckSupportPows(selfCast, exponent)) {
        auto expTensor = uniqueExecutor.get()->ConvertToTensor(exponent, promoteType);
        CHECK_RET(expTensor != nullptr, ACLNN_ERR_INNER_NULLPTR);
        // 调用pows进行计算
        powOut = l0op::Pows(selfCast, expTensor, uniqueExecutor.get());
    } else if (canUseSquare) {
        const aclTensor* squareInput = selfCast;
        if (CheckType(selfCast->GetDataType(), SQUARE_NEED_CAST_DTYPE_LIST)) {
            squareInput = l0op::Cast(selfCast, op::DataType::DT_INT32, uniqueExecutor.get());
            CHECK_RET(squareInput != nullptr, ACLNN_ERR_INNER_NULLPTR);
        }
        // 当exponent为2.0时,使用square算子计算
        powOut = l0op::Square(squareInput, uniqueExecutor.get());
    } else if (canNoUseOp) {
        powOut = selfCast;
    } else {
        auto expTensor = uniqueExecutor.get()->ConvertToTensor(exponent, promoteType);
        CHECK_RET(expTensor != nullptr, ACLNN_ERR_INNER_NULLPTR);

        // 调用pow进行计算
        powOut = l0op::Pow(selfCast, expTensor, uniqueExecutor.get());
    }
    CHECK_RET(powOut != nullptr, ACLNN_ERR_INNER_NULLPTR);

    // 固定写法,将计算结果转换成输出out的数据类型
    auto castOut = l0op::Cast(powOut, out->GetDataType(), uniqueExecutor.get());
    CHECK_RET(castOut != nullptr, ACLNN_ERR_INNER_NULLPTR);

    // 固定写法,将计算结果拷贝到输出out上,out可能是非连续的tensor
    auto viewCopyResult = l0op::ViewCopy(castOut, out, uniqueExecutor.get());
    CHECK_RET(viewCopyResult != nullptr, ACLNN_ERR_INNER_NULLPTR);

    *workspaceSize = uniqueExecutor->GetWorkspaceSize();
    uniqueExecutor.ReleaseTo(executor);

    return ACLNN_SUCCESS;
}

aclnnStatus aclnnPowTensorScalar(void* workspace, uint64_t workspaceSize, aclOpExecutor* executor,
                                 const aclrtStream stream)
{
    L2_DFX_PHASE_2(aclnnPowTensorScalar);
    // 固定写法,调用框架能力,完成计算
    return CommonOpExecutorRun(workspace, workspaceSize, executor, stream);
}

aclnnStatus aclnnInplacePowTensorScalarGetWorkspaceSize(const aclTensor* self, const aclScalar* exponent,
                                                        uint64_t* workspaceSize, aclOpExecutor** executor)
{
    auto out = const_cast<aclTensor*>(self);
    return aclnnPowTensorScalarGetWorkspaceSize(self, exponent, out, workspaceSize, executor);
}

aclnnStatus aclnnInplacePowTensorScalar(void* workspace, uint64_t workspaceSize, aclOpExecutor* executor,
                                        aclrtStream stream)
{
    L2_DFX_PHASE_2(aclnnInplacePowTensorScalar);
    // 固定写法,调用框架能力,完成计算
    return CommonOpExecutorRun(workspace, workspaceSize, executor, stream);
}

// self为1时的fill(1)分支
static aclnnStatus BuildPowScalarTensorFillOne(const aclTensor* out, aclOpExecutor* executor)
{
    FVector<int64_t> shape;
    for (size_t idx = 0; idx < out->GetViewShape().GetDimNum(); idx++) {
        int64_t tmpVal = out->GetViewShape().GetDim(idx);
        shape.push_back(tmpVal);
    }
    auto dims = executor->ConvertToTensor(shape.data(), shape.size(), DataType::DT_INT64);
    CHECK_RET(dims != nullptr, ACLNN_ERR_INNER_NULLPTR);
    auto shapeArray = executor->AllocIntArray(shape.data(), shape.size());
    CHECK_RET(shapeArray != nullptr, ACLNN_ERR_INNER_NULLPTR);

    FVector<float> valVector = {1.0};
    auto valTensor = executor->ConvertToTensor(valVector.data(), valVector.size(), out->GetDataType());
    CHECK_RET(valTensor != nullptr, ACLNN_ERR_INNER_NULLPTR);
    auto powOut = l0op::Fill(dims, valTensor, shapeArray, executor);
    CHECK_RET(powOut != nullptr, ACLNN_ERR_INNER_NULLPTR);

    auto viewCopyResult = l0op::ViewCopy(powOut, out, executor);
    CHECK_RET(viewCopyResult != nullptr, ACLNN_ERR_INNER_NULLPTR);

    return ACLNN_SUCCESS;
}

// pow计算分支:contiguous -> cast -> pow -> cast -> viewcopy
static aclnnStatus BuildPowScalarTensorCompute(const aclScalar* self, const aclTensor* exponent, const aclTensor* out,
                                               const op::DataType promoteType, aclOpExecutor* executor)
{
    auto expContiguous = l0op::Contiguous(exponent, executor);
    CHECK_RET(expContiguous != nullptr, ACLNN_ERR_INNER_NULLPTR);

    // 910B且走AICPU时,exponent不cast到promoteType
    bool skipCastForAicpu = IsPowAiCpuOn910B(promoteType, self, exponent);
    const aclTensor* powExp = expContiguous;
    if (!skipCastForAicpu) {
        powExp = l0op::Cast(expContiguous, promoteType, executor);
        CHECK_RET(powExp != nullptr, ACLNN_ERR_INNER_NULLPTR);
    }

    auto selfTensor = executor->ConvertToTensor(self, promoteType);
    CHECK_RET(selfTensor != nullptr, ACLNN_ERR_INNER_NULLPTR);

    // 调用pow进行计算
    auto powOut = l0op::Pow(selfTensor, powExp, executor);
    CHECK_RET(powOut != nullptr, ACLNN_ERR_INNER_NULLPTR);

    // 固定写法,将计算结果转换成输出out的数据类型
    auto castOut = l0op::Cast(powOut, out->GetDataType(), executor);
    CHECK_RET(castOut != nullptr, ACLNN_ERR_INNER_NULLPTR);

    // 固定写法,将计算结果拷贝到输出out上,out可能是非连续的tensor
    auto viewCopyResult = l0op::ViewCopy(castOut, out, executor);
    CHECK_RET(viewCopyResult != nullptr, ACLNN_ERR_INNER_NULLPTR);

    return ACLNN_SUCCESS;
}

aclnnStatus aclnnPowScalarTensorGetWorkspaceSize(const aclScalar* self, const aclTensor* exponent, const aclTensor* out,
                                                 uint64_t* workspaceSize, aclOpExecutor** executor)
{
    L2_DFX_PHASE_1(aclnnPowScalarTensor, DFX_IN(self, exponent), DFX_OUT(out));

    // 固定写法,创建OpExecutor
    auto uniqueExecutor = CREATE_EXECUTOR();
    CHECK_RET(uniqueExecutor.get() != nullptr, ACLNN_ERR_INNER_CREATE_EXECUTOR);

    // 固定写法,参数检查
    auto ret = CheckPowScalarTensorParams(self, exponent, out);
    CHECK_RET(ret == ACLNN_SUCCESS, ret);

    // pow算子的空tensor在kernel中支持,对标竞品根据算子实际情况补充
    if (exponent->IsEmpty()) {
        *workspaceSize = 0;
        uniqueExecutor.ReleaseTo(executor);
        return ACLNN_SUCCESS;
    }

    // fill(1) 分支
    if (IsRegBase() && std::abs(self->ToFloat() - 1.0f) < EXP_EPSILON && !IsComplexType(exponent->GetDataType()) &&
        !IsComplexType(out->GetDataType())) {
        CHECK_RET(BuildPowScalarTensorFillOne(out, uniqueExecutor.get()) == ACLNN_SUCCESS, ACLNN_ERR_INNER);
        *workspaceSize = uniqueExecutor->GetWorkspaceSize();
        uniqueExecutor.ReleaseTo(executor);
        return ACLNN_SUCCESS;
    }

    auto promoteType = InferScalarTensorDtype(self, exponent, out);

    CHECK_RET(BuildPowScalarTensorCompute(self, exponent, out, promoteType, uniqueExecutor.get()) == ACLNN_SUCCESS,
              ACLNN_ERR_INNER);

    *workspaceSize = uniqueExecutor->GetWorkspaceSize();
    uniqueExecutor.ReleaseTo(executor);

    return ACLNN_SUCCESS;
}

aclnnStatus aclnnPowScalarTensor(void* workspace, uint64_t workspaceSize, aclOpExecutor* executor, aclrtStream stream)
{
    L2_DFX_PHASE_2(aclnnPowScalarTensor);
    // 固定写法,调用框架能力,完成计算
    return CommonOpExecutorRun(workspace, workspaceSize, executor, stream);
}
#ifdef __cplusplus
}
#endif