/**
 * Copyright (c) 2026 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.
 */

/*!
 * \file aclnn_hypot.cpp
 * \brief hypot L2 aclnn two-phase interface
 */

#include "aclnn_hypot.h"
#include "hypot.h"
#include "aclnn_kernels/cast.h"
#include "aclnn_kernels/contiguous.h"
#include "aclnn_kernels/common/op_error_check.h"
#include "op_api/aclnn_check.h"
#include "opdev/common_types.h"
#include "opdev/data_type_utils.h"
#include "op_api/data_type_utils.h"
#include "opdev/format_utils.h"
#include "opdev/op_dfx.h"
#include "opdev/op_executor.h"
#include "opdev/op_log.h"
#include "opdev/shape_utils.h"
#include "opdev/tensor_view_utils.h"
#include "opdev/platform.h"

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

constexpr size_t MAX_DIM_LEN = 8;

// FLOAT / FLOAT16 / BFLOAT16(异 dtype 组合由 kernel 内 DTYPE_SELF/DTYPE_OTHER 直读,
// 计算在 FP32 寄存器完成,不再插入 Cast 拼接算子)
static const std::initializer_list<op::DataType> DTYPE_SUPPORT_LIST = {op::DataType::DT_FLOAT, op::DataType::DT_FLOAT16,
                                                                       op::DataType::DT_BF16};

static bool CheckNotNull(const aclTensor* self, const aclTensor* other, const aclTensor* out)
{
    OP_CHECK_NULL(self, return false);
    OP_CHECK_NULL(other, return false);
    OP_CHECK_NULL(out, return false);
    return true;
}

static bool CheckDtypeValid(const aclTensor* self, const aclTensor* other, const aclTensor* out)
{
    OP_CHECK_DTYPE_NOT_SUPPORT(self, DTYPE_SUPPORT_LIST, return false);
    OP_CHECK_DTYPE_NOT_SUPPORT(other, DTYPE_SUPPORT_LIST, return false);
    OP_CHECK_DTYPE_NOT_SUPPORT(out, DTYPE_SUPPORT_LIST, return false);
    return true;
}

static bool CheckComputeDtype(const aclTensor* self, const aclTensor* other, const aclTensor* out)
{
    // dtype 互推导(np.promote_types 语义)
    op::DataType promoteType = op::PromoteType(self->GetDataType(), other->GetDataType());
    if (promoteType == DataType::DT_UNDEFINED) {
        OP_LOGE(ACLNN_ERR_PARAM_INVALID, "self dtype %s and other dtype %s can not promote dtype.",
                op::ToString(self->GetDataType()).GetString(), op::ToString(other->GetDataType()).GetString());
        return false;
    }
    // 推导结果须为 FLOAT / FLOAT16 / BFLOAT16 之一
    if (!CheckType(promoteType, DTYPE_SUPPORT_LIST)) {
        OP_LOGE(ACLNN_ERR_PARAM_INVALID, "Hypot promote dtype %s is not supported.",
                op::ToString(promoteType).GetString());
        return false;
    }
    // 检查推导后的数据类型能否转换为输出的数据类型
    OP_CHECK_RESULT_DTYPE_CAST_FAILED(promoteType, out->GetDataType(), return false);
    return true;
}

static bool CheckShape(const aclTensor* self, const aclTensor* other, const aclTensor* out)
{
    OP_CHECK_MAX_DIM(self, MAX_DIM_LEN, return false);
    OP_CHECK_MAX_DIM(other, MAX_DIM_LEN, return false);

    op::Shape broadcastShape;
    OP_CHECK_BROADCAST_AND_INFER_SHAPE(self, other, broadcastShape, return false);
    OP_CHECK_SHAPE_NOT_EQUAL_WITH_EXPECTED_SIZE(out, broadcastShape, return false);
    return true;
}

static aclnnStatus CheckParams(const aclTensor* self, const aclTensor* other, const aclTensor* out)
{
    // null pointer check failure reports ACLNN_ERR_PARAM_INVALID (161002)
    CHECK_RET(CheckNotNull(self, other, out), ACLNN_ERR_PARAM_INVALID);
    CHECK_RET(CheckDtypeValid(self, other, out), ACLNN_ERR_PARAM_INVALID);
    CHECK_RET(CheckComputeDtype(self, other, out), ACLNN_ERR_PARAM_INVALID);
    CHECK_RET(CheckShape(self, other, out), ACLNN_ERR_PARAM_INVALID);
    return ACLNN_SUCCESS;
}

aclnnStatus aclnnHypotGetWorkspaceSize(const aclTensor* self, const aclTensor* other, aclTensor* out,
                                       uint64_t* workspaceSize, aclOpExecutor** executor)
{
    L2_DFX_PHASE_1(aclnnHypot, DFX_IN(self, other), DFX_OUT(out));
    auto uniqueExecutor = CREATE_EXECUTOR();
    CHECK_RET(uniqueExecutor.get() != nullptr, ACLNN_ERR_INNER_CREATE_EXECUTOR);
    auto ret = CheckParams(self, other, out);
    CHECK_RET(ret == ACLNN_SUCCESS, ret);
    if (self->GetStorageFormat() != Format::FORMAT_ND) {
        OP_LOGW("Format only supports ND");
    }
    if (self->IsEmpty() || other->IsEmpty()) {
        *workspaceSize = 0;
        uniqueExecutor.ReleaseTo(executor);
        return ACLNN_SUCCESS;
    }
    DataType computeType = op::PromoteType(self->GetDataType(), other->GetDataType());
    CHECK_RET(computeType != DataType::DT_UNDEFINED, ACLNN_ERR_INNER);
    bool needOutputCast = (computeType != out->GetDataType());
    auto selfContiguous = l0op::Contiguous(self, uniqueExecutor.get());
    CHECK_RET(selfContiguous != nullptr, ACLNN_ERR_INNER_NULLPTR);
    auto otherContiguous = l0op::Contiguous(other, uniqueExecutor.get());
    CHECK_RET(otherContiguous != nullptr, ACLNN_ERR_INNER_NULLPTR);
    const aclTensor* hypotOpOut = l0op::Hypot(selfContiguous, otherContiguous, uniqueExecutor.get());
    CHECK_RET(hypotOpOut != nullptr, ACLNN_ERR_INNER_NULLPTR);
    const aclTensor* finalOut = hypotOpOut;
    if (needOutputCast) {
        finalOut = l0op::Cast(hypotOpOut, out->GetDataType(), uniqueExecutor.get());
        CHECK_RET(finalOut != nullptr, ACLNN_ERR_INNER_NULLPTR);
    }
    auto viewCopyResult = l0op::ViewCopy(finalOut, out, uniqueExecutor.get());
    CHECK_RET(viewCopyResult != nullptr, ACLNN_ERR_INNER_NULLPTR);
    *workspaceSize = uniqueExecutor->GetWorkspaceSize();
    uniqueExecutor.ReleaseTo(executor);
    return ACLNN_SUCCESS;
}

aclnnStatus aclnnHypot(void* workspace, uint64_t workspaceSize, aclOpExecutor* executor, aclrtStream stream)
{
    L2_DFX_PHASE_2(aclnnHypot);
    return CommonOpExecutorRun(workspace, workspaceSize, executor, stream);
}

#ifdef __cplusplus
}
#endif