add Slice operator

include/aidge/operator/Slice.hpp

+/********************************************************************************
+ * Copyright (c) 2023 CEA-List
+ *
+ * This program and the accompanying materials are made available under the
+ * terms of the Eclipse Public License 2.0 which is available at
+ * http://www.eclipse.org/legal/epl-2.0.
+ *
+ * SPDX-License-Identifier: EPL-2.0
+ *
+ ********************************************************************************/
+
+#ifndef AIDGE_CORE_OPERATOR_SLICE_H_
+#define AIDGE_CORE_OPERATOR_SLICE_H_
+
+#include <cassert>
+#include <memory>
+#include <vector>
+
+#include "aidge/utils/Registrar.hpp"
+#include "aidge/operator/Operator.hpp"
+#include "aidge/backend/OperatorImpl.hpp"
+#include "aidge/data/Tensor.hpp"
+#include "aidge/data/Data.hpp"
+#include "aidge/graph/Node.hpp"
+#include "aidge/utils/Types.h"
+
+namespace Aidge {
+
+class Slice_Op : public Operator,
+    public Registrable<Slice_Op, std::string, std::unique_ptr<OperatorImpl>(const Slice_Op&)> {
+public:
+    // FIXME: change accessibility
+    std::array<std::shared_ptr<Tensor>, 4> mInputs = {std::make_shared<Tensor>(),
+                                                      std::make_shared<Tensor>(),
+                                                      std::make_shared<Tensor>(),
+                                                      std::make_shared<Tensor>()};
+    const std::shared_ptr<Tensor> mOutput = std::make_shared<Tensor>();
+
+public:
+    static constexpr const char* Type = "Slice";
+
+    Slice_Op()
+            : Operator(Type)
+    {
+        setDatatype(DataType::Float32);
+    }
+
+    /**
+     * @brief Copy-constructor. Copy the operator attributes and its output tensor(s), but not its input tensors (the new operator has no input associated).
+     * @param op Operator to copy.
+     */
+    Slice_Op(const Slice_Op& op)
+        : Operator(Type),
+          mOutput(std::make_shared<Tensor>(*op.mOutput))
+    {
+        // cpy-ctor
+        setDatatype(op.mOutput->dataType());
+        mImpl = op.mImpl ? Registrar<Slice_Op>::create(mOutput->getImpl()->backend())(*this) : nullptr;
+    }
+
+    /**
+     * @brief Clone the operator using its copy-constructor.
+     * @see Operator::Slice_Op
+     */
+    std::shared_ptr<Operator> clone() const override {
+        return std::make_shared<Slice_Op>(*this);
+    }
+
+    void associateInput(const IOIndex_t inputIdx, std::shared_ptr<Data> data) override final {
+        assert(inputIdx < 4 && "operator Slice supports only 4 inputs");
+        assert(strcmp(data->type(), Tensor::Type)==0 && "input data must be of Tensor type");
+        mInputs[inputIdx] = std::dynamic_pointer_cast<Tensor>(data);
+    }
+
+    void computeOutputDims() override final {
+        if (!mInputs[0]->empty() && !mInputs[1]->empty() && !mInputs[2]->empty()&& !mInputs[3]->empty())
+        {
+            const int* axes = static_cast<const int*>(mInputs[1]->getImpl()->rawPtr());
+            const int* starts = static_cast<const int*>(mInputs[2]->getImpl()->rawPtr());
+            const int* ends = static_cast<const int*>(mInputs[3]->getImpl()->rawPtr());
+            DimSize_t nbAxes = mInputs[1]->dims()[0];
+            std::vector<DimSize_t> outDims;
+            for(std::size_t i=0; i<mInputs[0]->dims().size();++i)
+            {
+
+                const int* idxPos = std::find(axes, axes + nbAxes, static_cast<int>(i));
+                if(idxPos != (axes + nbAxes))
+                {
+                    // TODO make sure all indxes are positive before this
+                    size_t idx = static_cast<size_t>(*idxPos);
+                    int startVal = starts[idx];
+                    int endVal = ends[idx];
+                    outDims.push_back(endVal - startVal);
+                }
+                else
+                {
+                    outDims.push_back(mInputs[0]->dims()[i]);
+                }
+            }
+            mOutput->resize(outDims);
+        }
+    }
+
+    bool outputDimsForwarded() const override final {
+        return !(mOutput->empty());
+    }
+
+
+    inline Tensor& input(const IOIndex_t inputIdx) const override final {
+        assert(static_cast<std::size_t>(inputIdx) < 4 && "wrong inputIdx for Slice operator.");
+        return *(mInputs[inputIdx].get());
+    }
+    inline Tensor& output(const IOIndex_t /*outputIdx*/) const override final { return *(mOutput.get()); }
+
+
+    inline std::shared_ptr<Tensor> getInput(const IOIndex_t inputIdx) const override final {
+        assert((inputIdx < 4) && "Slice Operator has 4 inputs");
+        return mInputs[inputIdx];
+    }
+    inline std::shared_ptr<Tensor> getOutput(const IOIndex_t outputIdx) const override final {
+        assert((outputIdx == 0) && "Slice Operator has only 1 output");
+        (void) outputIdx; // avoid unused warning
+        return mOutput;
+    }
+
+
+    std::shared_ptr<Data> getRawInput(const IOIndex_t inputIdx) const override final {
+        assert(inputIdx < 4 && "operator supports only 4 inputs");
+        return std::static_pointer_cast<Data>(mInputs[inputIdx]);
+    }
+    std::shared_ptr<Data> getRawOutput(const IOIndex_t outputIdx) const override final {
+        assert(outputIdx == 0 && "operator supports only 1 output");
+        (void) outputIdx; // avoid unused warning
+        return std::static_pointer_cast<Data>(mOutput);
+    }
+
+
+    void setBackend(const std::string& name) override {
+        mImpl = Registrar<Slice_Op>::create(name)(*this);
+        mOutput->setBackend(name);
+
+        // FIXME: temporary workaround
+        mInputs[0]->setBackend(name);
+        mInputs[1]->setBackend(name);
+        mInputs[2]->setBackend(name);
+        mInputs[3]->setBackend(name);
+    }
+    void setDatatype(const DataType& datatype) override {
+        mOutput->setDatatype(datatype);
+
+        // FIXME: temporary workaround
+        mInputs[0]->setDatatype(datatype);
+        mInputs[1]->setDatatype(DataType::Int32);
+        mInputs[2]->setDatatype(DataType::Int32);
+        mInputs[3]->setDatatype(DataType::Int32);
+    }
+
+    inline IOIndex_t nbInputs() const noexcept override final { return 4; }
+    inline IOIndex_t nbDataInputs() const noexcept override final { return 4; }
+    inline IOIndex_t nbOutputs() const noexcept override final { return 1; }
+    static const std::vector<std::string> getInputsName(){
+        return {"data_input", "starts", "ends", "axes"};
+    }
+    static const std::vector<std::string> getOutputsName(){
+        return {"data_output"};
+    }
+};
+
+inline std::shared_ptr<Node> Slice(const std::string& name = "") {
+    return std::make_shared<Node>(std::make_shared<Slice_Op>(), name);
+}
+}
+
+#endif /* AIDGE_CORE_OPERATOR_SLICE_H_ */

python_binding/operator/pybind_Slice.cpp

+/********************************************************************************
+ * Copyright (c) 2023 CEA-List
+ *
+ * This program and the accompanying materials are made available under the
+ * terms of the Eclipse Public License 2.0 which is available at
+ * http://www.eclipse.org/legal/epl-2.0.
+ *
+ * SPDX-License-Identifier: EPL-2.0
+ *
+ ********************************************************************************/
+
+#include <pybind11/pybind11.h>
+
+#include "aidge/operator/Slice.hpp"
+#include "aidge/operator/Operator.hpp"
+
+namespace py = pybind11;
+namespace Aidge {
+
+void init_Slice(py::module& m) {
+    py::class_<Slice_Op, std::shared_ptr<Slice_Op>, Operator>(m, "SliceOp", py::multiple_inheritance())
+    .def("get_inputs_name", &Slice_Op::getInputsName)
+    .def("get_outputs_name", &Slice_Op::getOutputsName);
+
+    m.def("Slice", &Slice, py::arg("name") = "");
+}
+}  // namespace Aidge