[zimpenfish]

If I remember correctly, it took a 128GB AWS EC2 to parse that file without OOMing. Go is not that efficient at deep multi-level size- and type-unknown data structures.