#include "mlir/Dialect/Tensor/IR/Tensor.h"
#include "mlir/Dialect/Tosa/IR/TosaOps.h"
#include "mlir/IR/BuiltinAttributes.h"
#include "mlir/IR/BuiltinTypes.h"
#include "mlir/IR/Location.h"
#include "mlir/IR/PatternMatch.h"
#include "mlir/IR/Value.h"
#include "mlir/Transforms/DialectConversion.h"

#include "llvm/ADT/SmallVector.h"
#include "llvm/ADT/Twine.h"
#include "llvm/Support/Casting.h"

#include <cassert>
#include <optional>
#include <utility>

#include "Common.hpp"
#include "src/Accelerators/PIM/Compiler/PimCompilerOptions.hpp"
#include "src/Accelerators/PIM/Dialect/Spatial/SpatialOps.hpp"
#include "src/Dialect/ONNX/ONNXOps.hpp"

using namespace mlir;

namespace onnx_mlir {

SmallVector<Value> sliceTensor(
  const Value& tensorToSlice, size_t axis, int64_t sliceSize, ConversionPatternRewriter& rewriter, Location loc) {
  ArrayRef<long> shape = getTensorShape(tensorToSlice);
  assert("Invalid axis" && axis < shape.size());

  SmallVector<OpFoldResult> strides(shape.size(), rewriter.getIndexAttr(1));
  SmallVector<OpFoldResult> offsets(shape.size(), rewriter.getIndexAttr(0));
  SmallVector<OpFoldResult> sizes;
  sizes.reserve(shape.size());
  for (const auto size : shape)
    sizes.push_back(rewriter.getIndexAttr(size));
  sizes[axis] = rewriter.getIndexAttr(sliceSize);

  long length = shape[axis];
  auto [numSlices, lastSliceSize] = ceilIntegerDivideWithRemainder(length, sliceSize);
  SmallVector<Value> slices;
  slices.reserve(numSlices);

  for (int64_t i = 0; i < numSlices; i++) {
    offsets[axis] = rewriter.getIndexAttr(i * sliceSize);
    if (i == numSlices - 1 && lastSliceSize != 0)
      sizes[axis] = rewriter.getIndexAttr(lastSliceSize);

    Value slice = tensor::ExtractSliceOp::create(rewriter, loc, tensorToSlice, offsets, sizes, strides);
    slices.push_back(slice);
  }

  return slices;
}

SmallVector<Value>
sliceVector(const Value& vectorToSlice, int64_t sliceSize, ConversionPatternRewriter& rewriter, Location loc) {
  ArrayRef<long> shape = getTensorShape(vectorToSlice);
  assert("Not a vector" && isVectorShape(shape));
  size_t axis = shape[0] != 1 ? 0 : 1;
  return sliceTensor(vectorToSlice, axis, sliceSize, rewriter, loc);
}

DenseMap<CoreId, SmallVector<Value>>
sliceVectorPerCrossbarPerCore(const Value& vectorToSlice, ConversionPatternRewriter& rewriter, Location loc) {
  SmallVector<Value> slices = sliceVector(vectorToSlice, crossbarSize, rewriter, loc);
  DenseMap<CoreId, SmallVector<Value>> slicesPerCore;
  for (size_t sliceId = 0; sliceId < slices.size(); sliceId++) {
    size_t coreId = sliceId / crossbarCountInCore;
    slicesPerCore[coreId].push_back(slices[sliceId]);
  }
  return slicesPerCore;
}

DenseMap<HSliceId, DenseMap<CoreId, SmallVector<Value>>> tileMatrix(
  Value& matrixToTile, int64_t hSliceSize, int64_t vSliceSize, ConversionPatternRewriter& rewriter, Location& loc) {
  assert("Not a matrix" && isMatrixShape(getTensorShape(matrixToTile)));

  DenseMap<HSliceId, DenseMap<CoreId, SmallVector<Value>>> tiles;

  SmallVector<Value> hSlices = sliceTensor(matrixToTile, 1, hSliceSize, rewriter, loc);
  size_t numHSlices = hSlices.size();
  for (size_t hSliceId = 0; hSliceId < numHSlices; hSliceId++) {
    Value hSlice = hSlices[hSliceId];
    SmallVector<Value> vSlices = sliceTensor(hSlice, 0, vSliceSize, rewriter, loc);
    for (size_t vSliceId = 0; vSliceId < vSlices.size(); vSliceId++) {
      size_t coreId = vSliceId / crossbarCountInCore;
      Value vSlice = vSlices[vSliceId];
      tiles[hSliceId][coreId].push_back(vSlice);
    }
  }
  return tiles;
}

tensor::SplatOp
broadcastToVector(Value scalarToBroadcast, int64_t length, ConversionPatternRewriter& rewriter, Location loc) {
  auto oldType = cast<RankedTensorType>(scalarToBroadcast.getType());
  Type elementType = oldType.getElementType();
  int64_t shape[2] = {1, length};
  Type type = oldType.cloneWith(ArrayRef(shape), elementType);

  auto zero = arith::ConstantIndexOp::create(rewriter, loc, 0).getResult();
  SmallVector<Value> index(oldType.getRank(), zero);
  auto elementValue = tensor::ExtractOp::create(rewriter, loc, scalarToBroadcast, index).getResult();

  return tensor::SplatOp::create(rewriter, loc, type, elementValue);
}

Value sumTensors(ArrayRef<Value> tensors, ConversionPatternRewriter& rewriter) {
  if (tensors.size() == 1)
    return tensors[0];

  SmallVector<Value> tensors1 = {tensors.begin(), tensors.end()};
  SmallVector<Value> tensors2;
  tensors2.reserve(tensors.size() / 2);

  auto* currTensors = &tensors1;
  auto* nextTensors = &tensors2;
  while (currTensors->size() > 1) {
    for (size_t i = 0; i < currTensors->size() - 1; i += 2) {
      Value a = (*currTensors)[i];
      Value b = (*currTensors)[i + 1];
      rewriter.setInsertionPointAfterValue(b);
      auto addedValue = spatial::SpatVAddOp::create(rewriter, a.getLoc(), a.getType(), a, b);
      nextTensors->push_back(addedValue);
    }
    if (currTensors->size() % 2 == 1)
      nextTensors->push_back(currTensors->back());
    std::swap(currTensors, nextTensors);
    nextTensors->clear();
  }
  assert(currTensors->size() == 1 && "Expected a single input at this point.");
  return (*currTensors)[0];
}

}; // namespace onnx_mlir