DCOMMENT_OP(op.getOperation(), "Infer shape for if ");
  bool changed = false;
  auto then_results =
      op.ResolveThenFunction(&symbol_table_).getFunctionType().getResults();
  auto else_results =
      op.ResolveElseFunction(&symbol_table_).getFunctionType().getResults();
  for (auto it : llvm::zip(op.getResults(), then_results, else_results)) {
    // If then and else types do not match, skip refinement for that result.

  }
  if (!body_fn) {
    return emitOpError("body refers to an undefined function : ") << getBody();
  }

  auto cond_fn_type = cond_fn.getFunctionType();
  auto body_fn_type = body_fn.getFunctionType();

  // Verify that the cond function has exactly one result.
  if (cond_fn_type.getNumResults() != 1)
    return emitOpError("requires cond function to have exactly one result");

    auto* node = graph_->FindNodeId(ret->id());
    TF_ASSIGN_OR_RETURN(auto type, InferInputType(*node, /*idx=*/0, builder));
    ret_types.push_back(type);
  }

  return builder.getFunctionType(arg_types, ret_types);
}

// Stateful helper class to import a TensorFlow model expressed in GraphDef into
// an MLIR Module.
//
// The nodes defined in the graph are converted to a function called

      return op->emitOpError()
             << "expects " << branch_name(branch.index()) << " ("
             << branch.value() << ") to point to a defined function";

    FunctionType branch_type = branch_func.getFunctionType();
    std::string desc = branch_name(branch.index()) + " input";
    TypeRangeWithDesc branch_input{branch_type.getInputs(), desc};
    if (failed(VerifyTypeRangesAreCompatible(op, branch_input, input)))

    }
    mlir::func::FuncOp callee = module_op_->lookupSymbol<mlir::func::FuncOp>(
        mlir::cast<FlatSymbolRefAttr>(op.getCalledComputations()[0]));
    mlir::FunctionType callee_type = callee.getFunctionType();
    SmallVector<Type, 4> expected_callee_input_types;
    auto num_inputs = op.getInputs().size() / 2;
    for (unsigned i = 0; i < num_inputs; ++i) {
      auto input_type =