Merge pull request #24480 from BoloniniD/yaml-fuzz

Add fuzzing tests for YAMLParser

src/Common/Config/YAMLParser.cpp

@@ -57,7 +57,7 @@ void processNode(const YAML::Node & node, Poco::XML::Element & parent_xml_elemen
     {
         case YAML::NodeType::Scalar:
         {
-            auto value = node.as<std::string>();
+            std::string value = node.as<std::string>();
             Poco::AutoPtr<Poco::XML::Text> xml_value = xml_document->createTextNode(value);
             parent_xml_element.appendChild(xml_value);
             break;
@@ -110,13 +110,13 @@ void processNode(const YAML::Node & node, Poco::XML::Element & parent_xml_elemen
             {
                 const auto & key_node = key_value_pair.first;
                 const auto & value_node = key_value_pair.second;
-                auto key = key_node.as<std::string>();
+                std::string key = key_node.as<std::string>();
                 bool is_attribute = (key.starts_with(YAML_ATTRIBUTE_PREFIX) && value_node.IsScalar());
                 if (is_attribute)
                 {
                     /// we use substr(1) here to remove YAML_ATTRIBUTE_PREFIX from key
                     auto attribute_name = key.substr(1);
-                    auto value = value_node.as<std::string>();
+                    std::string value = value_node.as<std::string>();
                     parent_xml_element.setAttribute(attribute_name, value);
                 }
                 else
@@ -148,7 +148,7 @@ Poco::AutoPtr<Poco::XML::Document> YAMLParser::parse(const String& path)
     catch (const YAML::ParserException& e)
     {
         /// yaml-cpp cannot parse the file because its contents are incorrect
-        throw Exception(ErrorCodes::CANNOT_PARSE_YAML, "Unable to parse YAML configuration file {}", path, e.what());
+        throw Exception(ErrorCodes::CANNOT_PARSE_YAML, "Unable to parse YAML configuration file {}, {}", path, e.what());
     }
     catch (const YAML::BadFile&)
     {
@@ -158,7 +158,14 @@ Poco::AutoPtr<Poco::XML::Document> YAMLParser::parse(const String& path)
     Poco::AutoPtr<Poco::XML::Document> xml = new Document;
     Poco::AutoPtr<Poco::XML::Element> root_node = xml->createElement("yandex");
     xml->appendChild(root_node);
+    try
+    {
         processNode(node_yml, *root_node);
+    }
+    catch (const YAML::TypedBadConversion<std::string>&)
+    {
+        throw Exception(ErrorCodes::CANNOT_PARSE_YAML, "YAMLParser has encountered node with key or value which cannot be represented as string and cannot continue parsing of the file");
+    }
     return xml;
 }
 

src/Common/examples/CMakeLists.txt

@@ -80,3 +80,8 @@ target_link_libraries (average PRIVATE clickhouse_common_io)
 
 add_executable (shell_command_inout shell_command_inout.cpp)
 target_link_libraries (shell_command_inout PRIVATE clickhouse_common_io)
+
+if (ENABLE_FUZZING)
+    add_executable(YAML_fuzzer YAML_fuzzer.cpp ${SRCS})
+    target_link_libraries(YAML_fuzzer PRIVATE clickhouse_parsers ${LIB_FUZZING_ENGINE})
+endif ()

src/Common/examples/YAML_fuzzer.cpp

@@ -0,0 +1,39 @@
+#include <iostream>
+#include <fstream>
+#include <string>
+#include <cstdio>
+#include <time.h>
+#include <filesystem>
+
+extern "C" int LLVMFuzzerTestOneInput(const uint8_t * data, size_t size)
+{
+    /// How to test:
+    /// build ClickHouse with YAML_fuzzer.cpp
+    /// ./YAML_fuzzer YAML_CORPUS
+    /// where YAML_CORPUS is a directory with different YAML configs for libfuzzer
+    char file_name[L_tmpnam];
+    if (!std::tmpnam(file_name))
+    {
+        std::cerr << "Cannot create temp file!\n";
+        return 1;
+    }
+    std::string input = std::string(reinterpret_cast<const char*>(data), size);
+    DB::YAMLParser parser;
+
+    {
+        std::ofstream temp_file(file_name);
+        temp_file << input;
+    }
+
+    try
+    {
+        DB::YAMLParser::parse(std::string(file_name));
+    }
+    catch (...)
+    {
+        std::cerr << "YAML_fuzzer failed: " << getCurrentExceptionMessage() << std::endl;
+        return 1;
+    }
+    return 0;
+}
+