RST0 (File Format)

Revision as of 03:13, 10 September 2015 by >Aruki (Created page with "The '''RST0 format''' is an archive format found in ''Donkey Kong Country Returns 3D''. It appears under a few different file extensions: '''.lvl''', '''.res''', and '''.t...")
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

The RST0 format is an archive format found in Donkey Kong Country Returns 3D. It appears under a few different file extensions: .lvl, .res, and .trk; all are the same format though, with the extension just describing the contents of the file rather than the format. Like all DKCR3D formats, RST0 is a little-endian format; all fourCCs and magic values that appear in the file appear reversed.


This file format needs more research
There are a lot of unknowns left


Format

RST0 files contain a header, which is a static size of 0x44 bytes, then are padded until 0x80, where a table of contents is present, listing every resource in the file. Some RST0 files are compressed; on compressed files, the header will be uncompressed, and most of the rest of the file (including the ToC and most of the resources in the file) will be compressed. There are some RST0 files that contain uncompressed resource data following the end of the compressed data, though.

Header

Offset Type Name Notes
0x0 char[4] Magic RST0
0x4 u32 Header Size Always 0x44
0x8 u32 Unknown Always 0x12?
0xC u32 Unknown Compression-related. Always 0 on uncompressed files and 2 on compressed files. No idea what the number means though; maybe it's a "compression type" enum?
0x10 u32 File Size Includes the header, but doesn't include the padding at the end of the file.
0x14 u32 Unknown Possibly a hash?
0x18 u32 Table of Contents Offset Always 0x80
0x1C u32 Resource Count
0x20 u32 Also resource count? Always matches the first resource count
0x24 u32 Decompressed Size Matches the decompressed sizes in the compressed section header. Equal to the file size (minus the header) on uncompressed files.
0x28 u32 Compressed Size Size of the entire compressed section. Equal to the file size (minus the header) on uncompressed files.
0x2C u32 Unknown
0x30 u32 Unknown
0x34 u32 String Table Size Size of the string table that contains the resource file names
0x38 u32 Resource ID Table Size Size of the table of resource IDs after the string table
0x3C u32 Unknown Some sort of hash?
0x40 u32 Unknown Some sort of hash?
0x44 End of header; padding to 0x80

Table of Contents

The ToC contains three arrays:

Type Name
Resources Resource Definitions
string Filename Table
Padding; the size of the name table is always a multiple of 0x10
Resource IDs Resource ID Table
End of ToC; pad to 0x80 then file data starts

Resources

Offset Type Name Notes
0x0 u32 Filename Offset Relative to the start of the name table
0x4 char[4] Resource Type Since the file is little-endian, the fourCC is backwards; eg, LDOM is actually MODL, XET is actually TEX , etc
0x8 u32 Unknown Flags? 0x10000000 seems to be set on uncompressed resources in compressed archives.
0xC u32 Resource Size
0x10 u32 Resource Offset Relative to the start of the ToC
0x14 u32 Unknown Some sort of hash?
0x18 u32 Resource ID Likely a hash of the resource name. This ID appears again later in the ID table and is used by other files to reference this resource.
0x1C u32 Unknown
0x20 u32 Unknown
0x24 u32 Unknown
0x28 u32 Unknown
0x2C End of resource definition

Resource IDs

Offset Type Name
0x0 u32 Resource ID
0x4 u32 Resource Index
0x8 End of resource ID

Compression

RST0 files use a custom run-length encoding compression algorithm. The compressed data begins with a 0x19-byte header before the actual compressed data begins. The header is actually two separate headers - one for the entire compressed section, comprising the first 0x10 bytes, and then one for the compressed data, comprising the last 0x9 bytes, which is why there's some redundant values included.

Offset Type Name Notes
0x0 char[4] Magic rCMP
0x4 u32 Unknown Possibly a hash of the compressed data?
0x8 u32 Compressed data size This is 0x10 smaller than the compressed size from the RST0 header because it doesn't include the initial compressed section header.
0xC u32 Decompressed data size This will match the decompressed size from the RST0 header.
0x10 u8 Unknown Always 0x4F?
0x11 u32 Compressed data size Matches the other compressed data size value.
0x15 u32 Decompressed data size Matches the other decompressed data size value as well as the one from the RST0 header.
0x19 End of header; compressed data begins

The compressed data is made up of multiple chunks. Each chunk has a 32-bit value as a header, where each bit tells you how to read the next part of the data. After reading in the header value, check each bit separately, starting at the least significant bit (0x1) and going up to the 31st bit (0x40000000). Note that although the header value is 32-bit, the top bit (0x80000000) is always set and doesn't actually correspond to anything in the file, so it should be ignored. After checking all 31 bits, read the next header byte and repeat.

For each bit, do the following depending on the value of the bit:

  • 0: Read a single byte directly out of the compressed data stream
  • 1: Seek backwards in the decompressed data buffer and copy bytes out of it back to the front

0 is straightforward and doesn't need much further explanation. 1 is where things get a little bit complicated. The next bytes in the compressed stream are an indicator value that will tell you how far back to seek in the decompressed data buffer and how much data to read out of them. The value will be a different size depending on how much data is needed at any given moment. Read one byte, and check the bottom two bits (0x1 and 0x2); they will tell you how large the indicator value is, and how to use the data in it. It's also important to note that the indicator value is little-endian, even though it has to be read one byte at a time; so the first byte you read is the bottom byte, and the last one is the top byte.

Using these definitions:

A Number of bytes to seek back in the decompressed buffer
B Number of bytes to read back from the decompressed buffer
C Indicator size/type flags
D These bits will be 0 to indicate a 32-bit indicator value. If non-zero, then the indicator is 24-bit.

Then depending on the value of the bottom two bits, this is the structure of the indicator value:

Value Size Structure
00 8-bit AAAAAACC

The byte count is always 3.

01 16-bit AAAAAAAAAAAAAACC

The byte count is always 3.

10 16-bit AAAAAAAAAABBBBCC

Add 3 to the byte count.

11 24- or 32-bit This one is different depending on the contents of the bottom byte. Check 5 of the middle bits (the D bits):

BDDDDDCC

If any of the D bits are non-zero, then the value is 24-bit: (add 2 to the byte count)

AAAAAAAAAAAAAAAAABBBBBCC

Otherwise, the value is 32-bit: (add 3 to the byte count)

AAAAAAAAAAAAAAAAABBBBBBBBDDDDDCC