Fix error generation for very large inputs in TextFormat::Parser.

PiperOrigin-RevId: 952895478

Author: sbenzaquen

src/google/protobuf/text_format.cc

@@ -337,6 +337,24 @@ const Descriptor* DefaultFinderFindAnyType(const Message& message,
                                            const std::string& name) {
   return message.GetDescriptor()->file()->pool()->FindMessageTypeByName(name);
 }
+
+void ReportErrorImpl(int line, int col, absl::string_view message,
+                     const Descriptor* root_message_type,
+                     io::ErrorCollector* error_collector) {
+  if (error_collector == nullptr) {
+    if (line >= 0) {
+      ABSL_LOG(ERROR) << "Error parsing text-format "
+                      << root_message_type->full_name() << ": " << (line + 1)
+                      << ":" << (col + 1) << ": " << message;
+    } else {
+      ABSL_LOG(ERROR) << "Error parsing text-format "
+                      << root_message_type->full_name() << ": " << message;
+    }
+  } else {
+    error_collector->RecordError(line, col, message);
+  }
+}
+
 }  // namespace
 
 auto TextFormat::Parser::UnsetFieldsMetadata::GetUnsetFieldId(
@@ -454,18 +472,7 @@ class TextFormat::Parser::ParserImpl {
 
   void ReportError(int line, int col, absl::string_view message) {
     had_errors_ = true;
-    if (error_collector_ == nullptr) {
-      if (line >= 0) {
-        ABSL_LOG(ERROR) << "Error parsing text-format "
-                        << root_message_type_->full_name() << ": " << (line + 1)
-                        << ":" << (col + 1) << ": " << message;
-      } else {
-        ABSL_LOG(ERROR) << "Error parsing text-format "
-                        << root_message_type_->full_name() << ": " << message;
-      }
-    } else {
-      error_collector_->RecordError(line, col, message);
-    }
+    ReportErrorImpl(line, col, message, root_message_type_, error_collector_);
   }
 
   void ReportWarning(int line, int col, const absl::string_view message) {
@@ -1958,23 +1965,19 @@ TextFormat::Parser::Parser()
       allow_singular_overwrites_(false),
       recursion_limit_(kDefaultRecursionLimit) {}
 
-namespace {
-
 template <typename T>
-bool CheckParseInputSize(T& input, io::ErrorCollector* error_collector) {
+bool TextFormat::Parser::CheckParseInputSize(T& input, Message* output) const {
   if (input.size() > INT_MAX) {
-    error_collector->RecordError(
-        -1, 0,
-        absl::StrCat(
-            "Input size too large: ", static_cast<int64_t>(input.size()),
-            " bytes", " > ", INT_MAX, " bytes."));
+    ReportErrorImpl(-1, 0,
+                    absl::StrCat("Input size too large: ",
+                                 static_cast<int64_t>(input.size()), " bytes",
+                                 " > ", INT_MAX, " bytes."),
+                    output->GetDescriptor(), error_collector_);
     return false;
   }
   return true;
 }
 
-}  // namespace
-
 bool TextFormat::Parser::Parse(io::ZeroCopyInputStream* input,
                                Message* output) {
   output->Clear();
@@ -1994,14 +1997,14 @@ bool TextFormat::Parser::Parse(io::ZeroCopyInputStream* input,
 
 bool TextFormat::Parser::ParseFromString(absl::string_view input,
                                          Message* output) {
-  DO(CheckParseInputSize(input, error_collector_));
+  DO(CheckParseInputSize(input, output));
   io::ArrayInputStream input_stream(input.data(), input.size());
   return Parse(&input_stream, output);
 }
 
 bool TextFormat::Parser::ParseFromCord(const absl::Cord& input,
                                        Message* output) {
-  DO(CheckParseInputSize(input, error_collector_));
+  DO(CheckParseInputSize(input, output));
   io::CordInputStream input_stream(&input);
   return Parse(&input_stream, output);
 }
@@ -2019,7 +2022,7 @@ bool TextFormat::Parser::Merge(io::ZeroCopyInputStream* input,
 
 bool TextFormat::Parser::MergeFromString(absl::string_view input,
                                          Message* output) {
-  DO(CheckParseInputSize(input, error_collector_));
+  DO(CheckParseInputSize(input, output));
   io::ArrayInputStream input_stream(input.data(), input.size());
   return Merge(&input_stream, output);
 }

src/google/protobuf/text_format.h

@@ -863,6 +863,9 @@ class PROTOBUF_EXPORT TextFormat {
     // representations (see text_format.cc for implementation).
     class ParserImpl;
 
+    template <typename T>
+    bool CheckParseInputSize(T& input, Message* output) const;
+
     // Like TextFormat::Merge().  The provided implementation is used
     // to do the parsing.
     bool MergeUsingImpl(io::ZeroCopyInputStream* input, Message* output,

src/google/protobuf/text_format_unittest.cc

@@ -82,6 +82,7 @@ namespace text_format_unittest {
 using ::absl_testing::IsOk;
 using ::absl_testing::StatusIs;
 using ::google::protobuf::internal::UnsetFieldsMetadataTextFormatTestUtil;
+using ::testing::_;
 using ::testing::AllOf;
 using ::testing::HasSubstr;
 using ::testing::UnorderedElementsAre;
@@ -295,6 +296,64 @@ TEST_F(TextFormatTest, ShortFormat) {
                   value_replacement, kTextMarkerRegex)));
 }
 
+TEST_F(TextFormatTest, InputTooLarge) {
+  if (sizeof(size_t) <= sizeof(int)) {
+    GTEST_SKIP() << "Not supported in platform.";
+  }
+  unittest::TestAllTypes msg;
+
+  constexpr absl::string_view kError = "Input size too large";
+
+  const auto expect_error = [&](auto f) {
+    {
+      // First without a collector.
+      TextFormat::Parser parser;
+      absl::ScopedMockLog log(absl::MockLogDefault::kDisallowUnexpected);
+      EXPECT_CALL(log, Log(absl::LogSeverity::kError, _, HasSubstr(kError)))
+          .Times(1);
+      log.StartCapturingLogs();
+      EXPECT_FALSE(f(parser));
+    }
+
+    // Then with a collector.
+    TextFormat::Parser parser;
+    class MockErrorCollector : public io::ErrorCollector {
+     public:
+      MockErrorCollector() = default;
+      ~MockErrorCollector() override = default;
+
+      std::string text_;
+
+      // implements ErrorCollector -------------------------------------
+      void RecordError(int line, int column,
+                       absl::string_view message) override {
+        text_ = absl::StrCat(message);
+      }
+
+      void RecordWarning(int line, int column,
+                         absl::string_view message) override {}
+    };
+
+    MockErrorCollector error_collector;
+    parser.RecordErrorsTo(&error_collector);
+    EXPECT_FALSE(f(parser));
+    EXPECT_THAT(error_collector.text_, HasSubstr(kError));
+  };
+
+  // Use a fake string_view with very large size.
+  // The contents don't matter because it should fail just by the size.
+  const absl::string_view too_large_sv(
+      "asdf", size_t{std::numeric_limits<int>::max()} + 1);
+  expect_error([&](auto& p) { return p.ParseFromString(too_large_sv, &msg); });
+  expect_error([&](auto& p) { return p.MergeFromString(too_large_sv, &msg); });
+
+  absl::Cord too_large_cord("sdaf");
+  while (too_large_cord.size() < std::numeric_limits<int>::max()) {
+    too_large_cord.Append(too_large_cord);
+  }
+  expect_error([&](auto& p) { return p.ParseFromCord(too_large_cord, &msg); });
+}
+
 TEST_F(TextFormatTest, Utf8Format) {
   unittest::RedactedFields proto;
   unittest::TestNestedMessageRedaction redacted_nested_proto;