delegate/opaque/src/Slice.hpp

//
// Copyright © 2023 Arm Ltd and Contributors. All rights reserved.
// SPDX-License-Identifier: MIT
//

#pragma once

#include <OpaqueDelegateUtils.hpp>

namespace armnnOpaqueDelegate
{

TfLiteStatus VisitSliceOperator(DelegateData& delegateData,
                                TfLiteOpaqueContext* tfLiteContext,
                                TfLiteOpaqueNode* tfLiteNode,
                                int nodeIndex,
                                int32_t tfLiteSliceOperatorCode)
{

    TF_LITE_ENSURE_STATUS(ValidateNumInputs(tfLiteContext, tfLiteNode, 3, nodeIndex));
    TF_LITE_ENSURE_STATUS(ValidateNumOutputs(tfLiteContext, tfLiteNode, 1, nodeIndex));

    // Read inputs [input, begin, size]
    // Gather input indices and use to get input tensor.
    const int* inputTensors;
    int numInputs;
    if (TfLiteOpaqueNodeInputs(tfLiteNode, &inputTensors, &numInputs) != kTfLiteOk)
    {
        TF_LITE_OPAQUE_MAYBE_KERNEL_LOG(
                tfLiteContext,
                "TfLiteArmnnOpaqueDelegate: Unable to gather input tensor indices from node #%d: ",
                nodeIndex);
        return kTfLiteError;
    }

    std::vector<const TfLiteOpaqueTensor*> tfLiteInputTensors;
    tfLiteInputTensors.reserve(numInputs);
    for (int i = 0; i < numInputs; i++)
    {
        const TfLiteOpaqueTensor* inputTensor = TfLiteOpaqueContextGetOpaqueTensor(tfLiteContext, inputTensors[i]);
        tfLiteInputTensors.push_back(inputTensor);
        if (!IsValid(tfLiteContext, inputTensor, tfLiteSliceOperatorCode, nodeIndex))
        {
            return kTfLiteError;
        }
    }

    const armnn::TensorInfo& inputTensorInfo  = GetTensorInfoForTfLiteOpaqueTensor(tfLiteInputTensors[0]);

    // We save the begin and size tensors in our descriptor. Therefore we have to read those values from inputs
    unsigned int inputRank = inputTensorInfo.GetNumDimensions();
    auto ReadInt32Input = [&](int inputIndex, std::vector<int32_t>& outputData, const char* name) ->  TfLiteStatus
    {
        if (TfLiteOpaqueTensorType(tfLiteInputTensors[inputIndex]) != kTfLiteInt32)
        {
            TF_LITE_OPAQUE_MAYBE_KERNEL_LOG(
                    tfLiteContext,
                    "TfLiteArmnnOpaqueDelegate: The %s Tensor of the Slice operation needs to "
                    "be of type int32. Operator: #%d node #%d: ",
                    name, tfLiteSliceOperatorCode, nodeIndex);
            return kTfLiteError;
        }
        uint32_t rank = TfLiteOpaqueTensorNumDims(tfLiteInputTensors[inputIndex]);
        if (rank != 1)
        {
            TF_LITE_OPAQUE_MAYBE_KERNEL_LOG(
                    tfLiteContext,
                    "TfLiteArmnnOpaqueDelegate: The %s Tensor of the Slice operation needs to "
                    "be a 1D-Tensor. Operator: #%d node #%d: ",
                    name, tfLiteSliceOperatorCode, nodeIndex);
            return kTfLiteError;
        }
        uint32_t numValues = TfLiteOpaqueTensorDim(tfLiteInputTensors[inputIndex], 0);
        if (numValues != inputRank)
        {
            TF_LITE_OPAQUE_MAYBE_KERNEL_LOG(
                    tfLiteContext,
                    "TfLiteArmnnOpaqueDelegate: The number of values in the %s Tensor of the "
                    "Slice operation needs to be equal to the rank of the Input Tensor. Operator: #%d node #%d: ",
                    name, tfLiteSliceOperatorCode, nodeIndex);
            return kTfLiteError;
        }
        // return tensor data
        auto* tensorDataPtr = static_cast<int32_t*>(TfLiteOpaqueTensorData(tfLiteInputTensors[inputIndex]));
        outputData.assign(tensorDataPtr, tensorDataPtr + numValues);
        return kTfLiteOk;
    };

    std::vector<int32_t> signedBegin;
    if (ReadInt32Input(1, signedBegin, "Begin") != kTfLiteOk)
    {
        return kTfLiteError;
    }

    std::vector<int32_t> signedSize;
    if (ReadInt32Input(2, signedSize, "Size") != kTfLiteOk)
    {
        return kTfLiteError;
    }

    std::vector<uint32_t> begin({ signedBegin.begin(), signedBegin.end() });
    std::vector<uint32_t> size(signedSize.size());

    for (unsigned int i = 0; i < signedSize.size(); ++i)
    {
        int signedValue = signedSize[i];
        if (signedValue < -1 || signedValue > TfLiteOpaqueTensorDim(tfLiteInputTensors[0], i) - signedBegin[i])
        {
            TF_LITE_OPAQUE_MAYBE_KERNEL_LOG(
                    tfLiteContext,
                    "TfLiteArmnnDelegate: Invalid value for Size. Size must be in range [-1, inputDimSize - begin] "
                    "[-1, %d] inclusive but was %d Operator: #%d node #%d: ",
                    TfLiteOpaqueTensorDim(tfLiteInputTensors[0], i) - signedBegin[i], signedValue,
                    tfLiteSliceOperatorCode, nodeIndex);
            return kTfLiteError;
        }
        if (signedValue == -1)
        {
            size[i] = TfLiteOpaqueTensorDim(tfLiteInputTensors[0], i) - signedBegin[i];
        }
        else
        {
            size[i] = static_cast<uint32_t>(signedValue);
        }
    }

    // Write all data to the descriptor
    armnn::SliceDescriptor descriptor(begin, size);

    // Validate output
    // Gather output indices and use to get output tensor.
    const int* outputTensors;
    int numOutputs;
    if (TfLiteOpaqueNodeOutputs(tfLiteNode, &outputTensors, &numOutputs) != kTfLiteOk)
    {
        TF_LITE_OPAQUE_MAYBE_KERNEL_LOG(
                tfLiteContext,
                "TfLiteArmnnOpaqueDelegate: Unable to gather output tensor indices from node #%d: ",
                nodeIndex);
        return kTfLiteError;
    }

    const TfLiteOpaqueTensor* tfLiteOutputTensor = TfLiteOpaqueContextGetOpaqueTensor(tfLiteContext, outputTensors[0]);
    if (!IsValid(tfLiteContext, tfLiteOutputTensor, tfLiteSliceOperatorCode, nodeIndex))
    {
        return kTfLiteError;
    }

    const armnn::TensorInfo& outputTensorInfo = GetTensorInfoForTfLiteOpaqueTensor(tfLiteOutputTensor, true);

    bool isSupported = false;
    armnn::BackendId setBackend;
    auto validateFunc = [&](const armnn::TensorInfo& outInfo, bool& isSupported)
    {
        FORWARD_LAYER_OPAQUE_SUPPORT_FUNC("SLICE",
                                          tfLiteContext,
                                          IsSliceSupported,
                                          delegateData.m_Backends,
                                          isSupported,
                                          setBackend,
                                          inputTensorInfo,
                                          outInfo,
                                          descriptor);
    };

    if (!delegateData.m_Network)
    {
        validateFunc(outputTensorInfo, isSupported);
        return isSupported ? kTfLiteOk : kTfLiteError;
    }

    // Add a Slice layer
    auto layerName = GetName(armnn::LayerType::Slice, nodeIndex);
    armnn::IConnectableLayer* layer = delegateData.m_Network->AddSliceLayer(descriptor, layerName.c_str());
    layer->SetBackendId(setBackend);
    ARMNN_ASSERT(layer != nullptr);

    armnn::IOutputSlot& outputSlot = layer->GetOutputSlot(0);
    outputSlot.SetTensorInfo(outputTensorInfo);

    // try to connect the Constant Inputs if there are any
    if (ProcessInputs(layer, delegateData, tfLiteContext, tfLiteNode, nodeIndex) != kTfLiteOk)
    {
        return kTfLiteError;
    }

    // Connect
    return Connect(layer, tfLiteContext, tfLiteNode, delegateData);
}

} // namespace armnnOpaqueDelegate