[NaaS] Register CEL extension provided string functions in evaluator.

PiperOrigin-RevId: 773278978

Author: CEL Dev Team

extensions/BUILD

@@ -175,6 +175,9 @@ cc_library(
     srcs = ["regex_functions.cc"],
     hdrs = ["regex_functions.h"],
     deps = [
+        "//checker:type_checker_builder",
+        "//common:decl",
+        "//common:type",
         "//common:value",
         "//eval/public:cel_function_registry",
         "//eval/public:cel_options",
@@ -215,8 +218,12 @@ cc_test(
     ],
     deps = [
         ":regex_functions",
+        "//checker:standard_library",
+        "//checker:validation_result",
         "//common:value",
         "//common:value_testing",
+        "//compiler",
+        "//compiler:compiler_factory",
         "//extensions/protobuf:runtime_adapter",
         "//internal:status_macros",
         "//internal:testing",

extensions/regex_functions.cc

@@ -23,6 +23,9 @@
 #include "absl/status/status.h"
 #include "absl/status/statusor.h"
 #include "absl/strings/string_view.h"
+#include "checker/type_checker_builder.h"
+#include "common/decl.h"
+#include "common/type.h"
 #include "common/value.h"
 #include "eval/public/cel_function_registry.h"
 #include "eval/public/cel_options.h"
@@ -164,6 +167,30 @@ absl::Status RegisterRegexFunctions(FunctionRegistry& registry) {
   return absl::OkStatus();
 }
 
+absl::Status RegisterRegexDecls(TypeCheckerBuilder& builder) {
+  CEL_ASSIGN_OR_RETURN(
+      FunctionDecl regex_extract_decl,
+      MakeFunctionDecl(
+          std::string(kRegexExtract),
+          MakeOverloadDecl("regex_extract", StringType(), StringType(),
+                           StringType(), StringType())));
+  CEL_RETURN_IF_ERROR(builder.AddFunction(regex_extract_decl));
+
+  CEL_ASSIGN_OR_RETURN(
+      FunctionDecl regex_capture_decl,
+      MakeFunctionDecl(std::string(kRegexCapture),
+                       MakeOverloadDecl("regex_capture", StringType(),
+                                        StringType(), StringType())));
+  CEL_RETURN_IF_ERROR(builder.AddFunction(regex_capture_decl));
+
+  CEL_ASSIGN_OR_RETURN(
+      FunctionDecl regex_capture_n_decl,
+      MakeFunctionDecl(std::string(kRegexCaptureN),
+                       MakeOverloadDecl("regex_capture", MapType(),
+                                        StringType(), StringType())));
+  return builder.AddFunction(regex_capture_n_decl);
+}
+
 }  // namespace
 
 absl::Status RegisterRegexFunctions(FunctionRegistry& registry,
@@ -182,4 +209,8 @@ absl::Status RegisterRegexFunctions(CelFunctionRegistry* registry,
   return absl::OkStatus();
 }
 
+CheckerLibrary RegexCheckerLibrary() {
+  return {.id = "regex", .configure = RegisterRegexDecls};
+}
+
 }  // namespace cel::extensions

extensions/regex_functions.h

@@ -17,16 +17,17 @@
 
 #include "absl/status/status.h"
 #include "absl/strings/string_view.h"
+#include "checker/type_checker_builder.h"
 #include "eval/public/cel_function_registry.h"
 #include "eval/public/cel_options.h"
 #include "runtime/function_registry.h"
 #include "runtime/runtime_options.h"
 
 namespace cel::extensions {
 
-constexpr absl::string_view kRegexExtract = "re.extract";
-constexpr absl::string_view kRegexCapture = "re.capture";
-constexpr absl::string_view kRegexCaptureN = "re.captureN";
+inline constexpr absl::string_view kRegexExtract = "re.extract";
+inline constexpr absl::string_view kRegexCapture = "re.capture";
+inline constexpr absl::string_view kRegexCaptureN = "re.captureN";
 
 // Register Extract and Capture Functions for RE2
 // Requires options.enable_regex to be true
@@ -36,5 +37,8 @@ absl::Status RegisterRegexFunctions(
 absl::Status RegisterRegexFunctions(FunctionRegistry& registry,
                                     const RuntimeOptions& options);
 
+// Declarations for the regex extension library.
+CheckerLibrary RegexCheckerLibrary();
+
 }  // namespace cel::extensions
 #endif  // THIRD_PARTY_CEL_CPP_EXTENSIONS_REGEX_FUNCTIONS_H_

extensions/regex_functions_test.cc

@@ -23,8 +23,12 @@
 #include "absl/status/status.h"
 #include "absl/status/status_matchers.h"
 #include "absl/status/statusor.h"
+#include "checker/standard_library.h"
+#include "checker/validation_result.h"
 #include "common/value.h"
 #include "common/value_testing.h"
+#include "compiler/compiler.h"
+#include "compiler/compiler_factory.h"
 #include "extensions/protobuf/runtime_adapter.h"
 #include "internal/status_macros.h"
 #include "internal/testing.h"
@@ -226,6 +230,60 @@ TEST_P(RegexFunctionsTest, RegexFunctionsTests) {
 INSTANTIATE_TEST_SUITE_P(RegexFunctionsTest, RegexFunctionsTest,
                          ValuesIn(createParams()));
 
+struct RegexCheckerTestCase {
+  const std::string expr_string;
+  bool is_valid;
+};
+
+class RegexCheckerLibraryTest
+    : public ::testing::TestWithParam<RegexCheckerTestCase> {
+ public:
+  void SetUp() override {
+    // Arrange: Configure the compiler.
+    // Add the regex checker library to the compiler builder.
+    ASSERT_OK_AND_ASSIGN(std::unique_ptr<CompilerBuilder> compiler_builder,
+                         NewCompilerBuilder(descriptor_pool_));
+    ASSERT_THAT(compiler_builder->AddLibrary(StandardCheckerLibrary()), IsOk());
+    ASSERT_THAT(compiler_builder->AddLibrary(RegexCheckerLibrary()), IsOk());
+    ASSERT_OK_AND_ASSIGN(compiler_, std::move(*compiler_builder).Build());
+  }
+
+  const google::protobuf::DescriptorPool* descriptor_pool_ =
+      internal::GetTestingDescriptorPool();
+  std::unique_ptr<Compiler> compiler_;
+};
+
+TEST_P(RegexCheckerLibraryTest, RegexFunctionsTypeCheckerSuccess) {
+  // Act & Assert: Compile the expression and validate the result.
+  ASSERT_OK_AND_ASSIGN(ValidationResult result,
+                       compiler_->Compile(GetParam().expr_string));
+  EXPECT_EQ(result.IsValid(), GetParam().is_valid);
+}
+
+// Returns a vector of test cases for the RegexCheckerLibraryTest.
+// Returns both positive and negative test cases for the regex functions.
+std::vector<RegexCheckerTestCase> createRegexCheckerParams() {
+  return {
+      {R"(re.extract('testuser@google.com', '(.*)@([^.]*)', '\\2!\\1'))", true},
+      {R"(re.extract('testuser@google.com', '(.*)@([^.]*)', 2))", false},
+      {R"(re.extract('testuser@google.com', false, '\\2!\\1'))", false},
+      {R"(re.extract(['foo', 'bar'], '(.*)@([^.]*)', '\\2!\\1'))", false},
+      {R"(re.captureN(
+          'The user testuser belongs to testdomain',
+          'The (user|domain) (?P<Username>.*) belongs to (?P<Domain>.*)'
+        ))",
+       true},
+      {R"(re.captureN('testuser@', '(?P<username>.*)@'))", true},
+      {R"(re.captureN('testuser@', 2))", false},
+      {R"(re.captureN(['foo', 'bar'], '(?P<username>.*)@'))", false},
+      {R"(re.capture('foo', 'fo(o)'))", true},
+      {R"(re.capture('foo', 2))", false},
+      {R"(re.capture(true, 'fo(o)'))", false}};
+}
+
+INSTANTIATE_TEST_SUITE_P(RegexCheckerLibraryTest, RegexCheckerLibraryTest,
+                         ValuesIn(createRegexCheckerParams()));
+
 }  // namespace
 
 }  // namespace cel::extensions