mlir::DenseIntElementsAttr shape_attr;
    if (matchPattern(op_state.operands[1], m_Constant(&shape_attr))) {
      auto shape_ty =
          op_state.operands[1].getType().dyn_cast<RankedTensorType>();
      if (shape_ty != nullptr && shape_ty.hasRank() && shape_ty.getRank() > 1) {
        llvm::SmallVector<mlir::Attribute, 4> shape;
        int32_t dim_size = 0;

TEST_F(CApiAttributesTest, ShapeList) {
  const int64_t shape_1[] = {1, 3};
  const int64_t shape_2[] = {2, 4, 6};
  const int64_t* list[] = {&shape_1[0], &shape_2[0]};
  const size_t list_size = TF_ARRAYSIZE(list);
  const int ndims[] = {TF_ARRAYSIZE(shape_1), TF_ARRAYSIZE(shape_2)};
  const int total_ndims = 5;  // ndims[0] + ndims[1]

  auto desc = init("list(shape)");
  TF_SetAttrShapeList(desc, "v", list, ndims, list_size);

                   ranked_ty.getDimSize(idx)));
    }
  }

  auto shape = builder->create<TF::ShapeOp>(loc, input, use_32bit);
  return builder->create<TF::StridedSliceOp>(
      loc, mlir::RankedTensorType::get({}, getElementTypeOrSelf(shape)), shape,
      /*begin=*/
      builder->create<TF::ConstOp>(loc, builder->getI32TensorAttr({idx})),
      /*end=*/

  return
}
```
### `-tf-tpu-annotate-dynamic-shape-inputs`

_Annotate the inputs returned by TPUCopyWithDynamicShapeOp with dynamic shape_

This pass looks for the usage of the result of TPUCopyWithDynamicShapeOp
and sets the shape of these inputs to be dynamic shaped. This will ensure
that the generated HLO program is correctly reflecting the dynamic shape.
### `-tf-tpu-cleanup-cluster-attributes`