Add Daikatana pak file format description to README

README

@@ -36,7 +36,54 @@ written one after the other without any space between
 them. A directory entry is 64 bytes long has entries
 in the following order:
  - A 56 byte file name (in ASCII)
- - A 4 byte value defining the position if the file
+ - A 4 byte integer value defining the position if the file
    as an offset in bytes to the start of the pak file.
  - A 4 byte integer giving the length of the file
    in bytes.
+
+-------------------------------------------------------
+
+            The Daikatana Pak File Format
+            -----------------------------
+
+The Daikatana pak file format is similar to the Quake II one, but
+additionally supports compressing files within the pak.
+Daikatana only did that for .tga .bmp .wal .pcx and .bsp files and might
+expect other files to be uncompressed (in case you want to write a compressor).
+
+In Daikatana directory entries are 72 bytes long, because they have to
+additional 4byte integer fields at the end:
+ - A 4 byte integer giving the compressed length of the file in bytes,
+   if it is compressed.
+ - A 4 byte integer indicating whether the file is compressed
+   (0: it's not compressed, else it's compressed)
+
+If the file is compressed, the file length field (the one also defined in
+Quake II pak) indicates the uncompressed length of the file, while the
+additional "compressed length" field indicates the number of bytes the
+compressed data of that file takes in the .pak.
+
+Compressed files are decompressed like this:
+	while not done:
+		read a byte (unsigned char) _x_.
+		// x is never written to output, only used to determine what to do
+		if      x < 64:
+			x+1 bytes of uncompressed data follow (just read+write them as they are)
+		else if x < 128:
+			// run-length encoded zeros
+			write (x - 62) zero-bytes to output
+		else if x < 192:
+			// run-length encoded data
+			read one byte, write it (x-126) times to output
+		else if x < 254:
+			// this references previously uncompressed data
+			read one byte to get _offset_
+			read (x-190) bytes from the already uncompressed and written output data,
+				starting at (offset+2) bytes before the current write position
+				(and add them to output, of course)
+		else if x == 255:
+			you're done decompressing (used as terminator)
+			// but I'd also abort once compressed_length bytes are read, to be sure
+
+See also https://gist.github.com/DanielGibson/8bde6241c93e5efe8b75e5e00d0b9858
+("Description of Daikatana .pak format")