Unable to parse files with abnormal encoding causing program crashes

[HuluMan]

Dear Contributors

Environment

• branches: [master]
• Build System: [CMake]
• Operating System: [Windows]

Problem Description

As the title states，When I parse an encoding exception file, an exception 0xC0000005 is thrown.

Steps to reproduce

1. Add vdf_parser.hpp file in my test project
2. Writing my test code

int main(int argc, char *argv[])
{
    std::list<std::string> list{
        "C:/Users/Administrator/Desktop/steamapps/appmanifest_a.acf",
        "C:/Users/Administrator/Desktop/steamapps/appmanifest_b.acf",
        "C:/Users/Administrator/Desktop/steamapps/appmanifest_b.acf",
    };
    for(auto path : list) {
        try{
            std::ifstream file(path);
            auto root = tyti::vdf::read(file);
            std::cout << path  << std::endl;
            for(auto [l, r] : root.attribs) {
                std::cout << l << " : " << r << std::endl;
            }
            std::cout << "--------------" <<std::endl;
        }catch(std::exception& e) {
            std::cout << path << e.what() << std::endl;
        }
    }
    return 0;
}

4. The exception is then thrown

Debug Logs

Exception thrown at 0x00C8D2E6 in TestSketch.exe: 0xC0000005: Access violation reading location 0x0000003C.

[HuluMan]

appmanifest_a.zip
Here I provide an exception file.

[Matthias Möller]

Hi,
in what encoding is the file? Neither ANSI with randomly codepgaes, nor utf8/16 show me any results.

Is this a binary file? Binary files are not supported by the parser as I cannot find any real specification to it. So it must be a text file. See https://developer.valvesoftware.com/wiki/KeyValues#About_KeyValues_Text_File_Format

if you know the encoding, try to change your locale or try to use wchar_t.

[HuluMan]

if you know the encoding, try to change your locale or try to use wchar_t.

Hi, TinyTinni, In fact, I can't be sure of the encoding of this file, which is generated by the file Steam.

---

My idea is, do we need to make some security treatment to avoid the file encoding non-conformity to crash the program when we can't be sure if the encoding of the file is canonical or not?

[Matthias Möller]

My idea is, do we need to make some security treatment to avoid the file encoding non-conformity to crash the program when we can't be sure if the encoding of the file is canonical or not?

I totally agree, it should be fixed. I just wanted to clarify, that the file cannot be parsed by this parser, even when the crash is fixed.