add unit tests for known-problematic fuzz inputs

test/CMakeLists.txt

@@ -265,3 +265,62 @@ if(RYML_TEST_SUITE)
         ryml_add_test_from_suite(${case})
     endforeach()
 endif(RYML_TEST_SUITE)
+
+
+#------------------------------------------------------------------------------
+#------------------------------------------------------------------------------
+#------------------------------------------------------------------------------
+
+# run every known fuzz crash
+set(fuzzdefault ON)
+if(EMSCRIPTEN)
+    set(fuzzdefault OFF)
+endif()
+option(RYML_TEST_FUZZ "Enable tests for known problematic fuzz cases." ${fuzzdefault})
+
+if(RYML_TEST_FUZZ)
+    c4_download_remote_proj(rapidyaml-data rapidyaml_data_dir
+        GIT_REPOSITORY https://github.com/biojppm/rapidyaml-data
+        GIT_TAG master)
+    if(NOT EXISTS ${rapidyaml_data_dir}/fuzz/yaml.dict)
+        c4_err("cannot find rapidyaml-data at ${rapidyaml_data_dir} -- was there an error downloading the project?")
+    endif()
+
+    set(corpus_suite_dir ${rapidyaml_data_dir}/fuzz/yaml_test_suite)
+    set(corpus_generated_dir ${rapidyaml_data_dir}/fuzz/yaml_generated)
+    set(corpus_artifacts_dir ${rapidyaml_data_dir}/fuzz/yaml_artifacts)
+    set(corpus_merged_dir ${rapidyaml_data_dir}/fuzz/yaml_merged)
+    set(yaml_dict ${rapidyaml_data_dir}/fuzz/yaml.dict)
+    file(GLOB_RECURSE fuzz_files RELATIVE "${corpus_artifacts_dir}" "${corpus_artifacts_dir}/*")
+    file(GLOB_RECURSE suite_files RELATIVE "${corpus_suite_dir}" "${corpus_suite_dir}/*")
+
+    # add individual tests for problematic fuzz files
+    function(ryml_add_fuzz_test name)
+        c4_add_executable(ryml-test-fuzz-${name}
+            SOURCES
+            test_fuzz/test_fuzz_common.hpp
+            test_fuzz/test_fuzz_${name}.cpp
+            test_fuzz/test_fuzz_main.cpp
+            ${ARGN}
+            INC_DIRS ${CMAKE_CURRENT_LIST_DIR}
+            LIBS ryml c4fs
+            FOLDER test/fuzz)
+        if(RYML_DBG)
+            target_compile_definitions(ryml-test-fuzz-${name} PUBLIC RYML_DBG)
+        endif()
+        add_dependencies(ryml-test-build ryml-test-fuzz-${name})
+        ryml_get_target_exe(ryml-test-fuzz-${name} tgtexe)
+        function(ryml_add_fuzz_test_file name_ dir file)
+            string(REPLACE "/" "_" fuzz_name "${file}")
+            add_test(NAME ryml-test-fuzz-${name_}-${fuzz_name}
+                COMMAND ${tgtexe} ${dir}/${file})
+        endfunction()
+        foreach(fuzz_file ${fuzz_files})
+            ryml_add_fuzz_test_file(${name} ${corpus_artifacts_dir} ${fuzz_file})
+        endforeach()
+    endfunction()
+    ryml_add_fuzz_test(parse_emit)
+    ryml_add_fuzz_test(events
+        ../test/test_suite/test_suite_event_handler.hpp
+        ../test/test_suite/test_suite_event_handler.cpp)
+endif()

test/test_fuzz/test_fuzz_common.hpp

@@ -0,0 +1,135 @@
+#pragma once
+#ifndef TEST_FUZZ_COMMON_H
+#define TEST_FUZZ_COMMON_H
+
+#ifdef RYML_SINGLE_HEADER
+#include <ryml_all.hpp>
+#else
+#include <c4/yml/std/std.hpp>
+#include <c4/yml/parse.hpp>
+#include <c4/yml/emit.hpp>
+#include <c4/yml/event_handler_tree.hpp>
+#include <c4/yml/parse_engine.def.hpp>
+#endif
+#include <test_suite/test_suite_events.hpp>
+#include <test_suite/test_suite_event_handler.hpp>
+#include <cstdio>
+
+#ifdef C4_EXCEPTIONS
+#include <stdexcept>
+#else
+#include <csetjmp>
+std::jmp_buf jmp_env = {};
+c4::csubstr jmp_msg = {};
+#endif
+
+
+#ifdef RYML_DBG
+#define _if_dbg(...) __VA_ARGS__
+bool report_errors = true;
+#else
+#define _if_dbg(...)
+bool report_errors = false;
+#endif
+
+inline void report_error(const char* msg, size_t length, c4::yml::Location loc, FILE *f)
+{
+    if(!report_errors)
+        return;
+    if(!loc.name.empty())
+    {
+        fwrite(loc.name.str, 1, loc.name.len, f);
+        fputc(':', f);
+    }
+    fprintf(f, "%zu:", loc.line);
+    if(loc.col)
+        fprintf(f, "%zu:", loc.col);
+    if(loc.offset)
+        fprintf(f, " (%zuB):", loc.offset);
+    fputc(' ', f);
+    fprintf(f, "%.*s\n", static_cast<int>(length), msg);
+    fflush(f);
+}
+
+// watchout: VS2022 requires C4_NORETURN to come before inline
+C4_NORETURN inline void errcallback(const char *msg, size_t msg_len, c4::yml::Location location, void *)
+{
+    report_error(msg, msg_len, location, stderr);
+    C4_IF_EXCEPTIONS(
+        throw std::runtime_error({msg, msg_len});
+        ,
+        jmp_msg.assign(msg, msg_len);
+        std::longjmp(jmp_env, 1);
+    );
+}
+
+inline c4::yml::Callbacks create_custom_callbacks()
+{
+    c4::set_error_flags(c4::ON_ERROR_CALLBACK);
+    c4::set_error_callback([](const char *msg, size_t msg_len){
+        errcallback(msg, msg_len, {}, nullptr);
+    });
+    c4::yml::Callbacks callbacks = {};
+    callbacks.m_error = errcallback;
+    return callbacks;
+}
+
+namespace c4 {
+namespace yml {
+
+inline int fuzztest_parse_emit(uint32_t case_number, csubstr src)
+{
+    C4_UNUSED(case_number);
+    set_callbacks(create_custom_callbacks());
+    Tree tree(create_custom_callbacks());
+    bool parse_success = false;
+    C4_IF_EXCEPTIONS_(try, if(setjmp(jmp_env) == 0))
+    {
+        RYML_ASSERT(tree.empty());
+        _if_dbg(_dbg_printf("in[{}]: [{}]~~~\n{}\n~~~\n", case_number, src.len, src); fflush(NULL));
+        parse_in_arena(src, &tree);
+        parse_success = true;
+        _if_dbg(print_tree("parsed tree", tree));
+        _if_dbg(_dbg_printf("in[{}]: [{}]~~~\n{}\n~~~\n", case_number, src.len, src); fflush(NULL));
+        std::string dst = emitrs_yaml<std::string>(tree);
+        _if_dbg(_dbg_printf("emitted[{}]: [{}]~~~\n{}\n~~~\n", case_number, dst.size(), to_csubstr(dst)); fflush(NULL));
+        C4_DONT_OPTIMIZE(dst);
+        C4_DONT_OPTIMIZE(parse_success);
+    }
+    C4_IF_EXCEPTIONS_(catch(std::exception const&), else)
+    {
+        // if an exception leaks from here, it is likely because of a greedy noexcept
+        _if_dbg(if(parse_success) print_tree("parsed tree", tree));
+        return 1;
+    }
+    return 0;
+}
+
+inline int fuzztest_yaml_events(uint32_t case_number, csubstr src)
+{
+    C4_UNUSED(case_number);
+    set_callbacks(create_custom_callbacks());
+    EventHandlerYamlStd::EventSink sink = {};
+    EventHandlerYamlStd handler(&sink, create_custom_callbacks());
+    ParseEngine<EventHandlerYamlStd> parser(&handler);
+    std::string str(src.begin(), src.end());
+    C4_IF_EXCEPTIONS_(try, if(setjmp(jmp_env) == 0))
+    {
+        _if_dbg(_dbg_printf("in[{}]: [{}]~~~\n{}\n~~~\n", case_number, src.len, src); fflush(NULL));
+        parser.parse_in_place_ev("input", c4::to_substr(str));
+        _if_dbg(_dbg_printf("evts[{}]: ~~~\n{}\n~~~\n", case_number, sink); fflush(NULL));
+        C4_DONT_OPTIMIZE(sink);
+    }
+    C4_IF_EXCEPTIONS_(catch(std::exception const&), else)
+    {
+        // if an exception leaks from here, it is likely because of a greedy noexcept
+        _if_dbg(fprintf(stdout, "err\n"); fflush(NULL));
+        return 1;
+    }
+    return 0;
+}
+
+} // namespace yml
+} // namespace c4
+
+#endif /* TEST_FUZZ_COMMON_H */

test/test_fuzz/test_fuzz_events.cpp

@@ -0,0 +1,9 @@
+#include "./test_fuzz_common.hpp"
+#include <atomic>
+
+extern "C" int LLVMFuzzerTestOneInput(const uint8_t *str, size_t len)
+{
+    static std::atomic<uint32_t> case_number{0};
+    c4::csubstr src = {reinterpret_cast<const char*>(str), len};
+    return c4::yml::fuzztest_yaml_events(case_number++, src);
+}

test/test_fuzz/test_fuzz_main.cpp

@@ -0,0 +1,16 @@
+#include <stdio.h>
+#include <c4/fs/fs.hpp>
+
+extern "C" int LLVMFuzzerTestOneInput(const uint8_t *str, size_t len);
+
+int main(int argc, const char *argv[])
+{
+    if(argc < 2)
+        return 1;
+    const char *filename = argv[1];
+    if(!c4::fs::file_exists(filename))
+        return 1;
+    std::string file = c4::fs::file_get_contents<std::string>(filename);
+    (void)LLVMFuzzerTestOneInput(reinterpret_cast<const uint8_t*>(&file[0]), file.size());
+    return 0;
+}

test/test_fuzz/test_fuzz_parse_emit.cpp

@@ -0,0 +1,9 @@
+#include "./test_fuzz_common.hpp"
+#include <atomic>
+
+extern "C" int LLVMFuzzerTestOneInput(const uint8_t *str, size_t len)
+{
+    static std::atomic<uint32_t> case_number{0};
+    c4::csubstr src = {reinterpret_cast<const char*>(str), len};
+    return c4::yml::fuzztest_parse_emit(case_number++, src);
+}

tools/parse_emit.cpp

@@ -180,7 +180,6 @@ yml::Callbacks create_custom_callbacks()
     };
     return callbacks;
 }
-// LCOV_EXCL_STOP
 
 
 //-----------------------------------------------------------------------------
@@ -241,6 +240,7 @@ void process_file(csubstr file)
             emit_json_docs(tree);
     }
 }
+// LCOV_EXCL_STOP
 
 int main(int argc, const char *argv[])
 {