}
    }

    private boolean shouldExecute(RuleAction<? super ComponentMetadataDetails> action, ModuleComponentResolveMetadata metadata) {
        List<Class<?>> inputTypes = action.getInputTypes();
        if (!inputTypes.isEmpty()) {
            return inputTypes.stream().anyMatch(input -> MetadataDescriptorFactory.isMatchingMetadata(input, metadata));
        }
        return true;
    }

        def initValue = inputType.contains("Map<") ? "[:]" : "[]"
        def inputDeclaration = "$inputType input = $initValue"
        groovyBuildFile(inputDeclaration, inputValue, operation)

        expect:
        runAndAssert("myTask", expectedResult)

        where:

                            const llvm::SmallVectorImpl<Attribute>& input_types,
                            llvm::SmallVectorImpl<Value>& input_values) const {
    if (input_types.size() <= 1) return;

    Type target_input_type = mlir::cast<TypeAttr>(input_types[0]).getValue();
    auto result_type = UnrankedTensorType::get(target_input_type);
    for (auto i = 1; i < input_types.size(); ++i) {

  if (!module_op) return mlir::failure();

  llvm::SmallVector<XlaArgument, 4> xla_arguments;
  auto args_status = ParseXlaArguments(
      mlir::StringRefToView(input_shapes), mlir::StringRefToView(input_dtypes),
      mlir::StringRefToView(input_types), xla_arguments);
  if (!args_status.ok()) {
    LOG(ERROR) << args_status;
    return mlir::failure();
  }

  XlaCompilationResult compilation_result;
  auto compilation_status =

    const std::vector<std::string>& input_dtypes,
    const std::vector<std::optional<std::vector<int>>>& input_shapes,
    const std::vector<std::string>& output_arrays,
    const std::vector<std::string>& control_output_arrays,
    const GraphdefToMlirOptions& import_options, mlir::MLIRContext* context) {
  auto module_or = GraphdefToMlirImport(
      input, input_arrays, input_dtypes, input_shapes, output_arrays,

        output.text == "${[REMOVED]}"
    }

    def "broken symlink in #inputType.simpleName fails validation"() {
        enableProblemsApiCheck()
        def brokenInputFile = file('brokenInput').createLink("brokenInputFileTarget")
        buildFile << """
            class CustomTask extends DefaultTask {
                @${inputType.simpleName} File brokenInputFile

                @TaskAction execute() {}
            }

          file_prefix_arg_type,
          NameLoc::get(builder.getStringAttr(kTfFilePrefix)));

  SmallVector<Type> input_types(main_func_op.getArgumentTypes());
  input_types.emplace_back(file_prefix_arg_type);

  main_func_op.setType(
      builder.getFunctionType(input_types, main_func_op.getResultTypes()));

  // Add "__tf_file_prefix" to the "tf_saved_model.index_path" attribute for the

                                            tensorflow::Var** maybe_resource) {
  auto* cc_ctx = reinterpret_cast<::tensorflow::OpKernelContext*>(ctx);
  *maybe_resource = nullptr;
  if (cc_ctx->input_dtype(input) == tensorflow::DT_RESOURCE) {
    if (LookupResource(cc_ctx, HandleFromInput(cc_ctx, input), maybe_resource)
            .ok()) {
      return (*maybe_resource)->mu();
    } else {

    if (!init_func) continue;

    auto init_type =
        init_func->getAttrOfType<StringAttr>(kTfSavedModelInitializerTypeAttr);
    if (!init_type) continue;

    if (init_types.contains(init_type)) {
      return init_func->emitError()
             << "Attribute tf_saved_model.initializer_type should not have "
                "duplicate values. Found duplicate: "
             << init_type;
    }

    // Do not create any new function for the operations on the localhost.
    if (IsOnLocalHost(host)) continue;

    llvm::SmallVector<mlir::Type, 4> input_types;
    llvm::SmallVector<mlir::Type, 4> result_types;
    for (Value input : metadata.inputs) {
      input_types.push_back(input.getType());
    }
    for (Value result : metadata.results) {
      result_types.push_back(result.getType());
    }