I’ve spent two or three days working on the Birthright Resource Utility (BRUt) and have managed to implement displaying the file index and exporting uncompressed files. Unfortunately, I’ve run into a bit of a problem with extracting compressed files.

For context, the resource files on the Birthright CD contain a total of 9427 files, of which 125 files (1.3%) are uncompressed. This leaves 98.7% of files unextractable. This makes the tool effectively useless.

After extracting the uncompressed .WAV files in WAVE.RES, I got to hear some sound effects I had never heard outside of the game before, so that was fun. If I implement adding and removing files, this might still be somewhat useful for modding. I’m not happy with that as an alternative, though. If I absolutely can’t get the compression algorith to work, I’ll use that as a fallback.

As I mentioned in my previous post, the resource files use a compression algorithm called LZSS. This algorithm has been around for decades, so I was hoping I could use existing tools to decompress the files.

Sadly, I’ve been having trouble finding good quality libraries that support LZSS. I decided to try AuroraLib, but it requires setting some configuration variables that I was unable to get to work properly. WinRAR appears to use LZSS in version 3 and provides a DLL and licensing for decompression, but requires paying for compression.

I could probably try several less well-known libraries, but there’s a chance I’ll spend a lot of time and effort only to find out Synergistic Software had a bug in their algorithm or did something non-standard. I already ran into unexpected behavior in their filename hashing algorithm1.

With all that said, I’m currently leaning towards rewriting the compression and decompression algorithms in C# by hand. Since these were written in Assembly Language, which I haven’t used in a decade, this might take me longer than I was hoping to spend on the basic functionality.

In the process, I may discover the right configuration for AuroraLib to work properly. Either way, I think it’s a worthwhile endeavor if I ended up with working code in the end.

  1. The HashCRC function hashes the filename to use as an index. It adds a NULL byte to the end of the filename because of an index overflow bug. I haven’t done thorough testing, so I can only assume it’s consistently NULL. I’ll do more thorough testing later on. ↩︎