Anyhow despite hex workshop being my choice of almost always I have very little experience in defining file types with it.
Anyhow with thanks to deufeufeu:
First few bytes are ALAR (a common theme in the nitroSDK)
id
offset (normally just the two bytes unless a massive file is happening)
size (not always the same as offset difference as padding is sometimes used so as to end up word (or higher) aligned.
pad (00's but if you are dealing with a massive aar file then they may be filled with offset data)
filename ASCII, the foward slashes indicate directory. If you just want extraction then I suggest counting back 12 bytes from the name and then flip the previous 2 bytes (can be slightly confusing as an ASCII letter can appear before the name: i not J in the example below)
I've so far found a lot of the ALAR header is related to the file number. This can be seen within all the /chr/*.* files. But otherwise I'm good to go with extraction. Though it looks like each file contains the picture, the bounding box, and some node data.
QUOTE(FAST6191 @ Sep 19 2007, 04:47 AM)Followed by standard LZ, a decompression app here:
http://gbadev.org/tools.php?showinfo=56