LEX (File Format)

From Custom Mario Kart
Revision as of 12:23, 13 June 2019 by Wiimm (talk | contribs)
Jump to navigation Jump to search

LEX (LE-CODE Extension) is a file format defined by Wiimm. A LEX file is usually stored as part of a track SZS under the name course.lex. It's like KMP's course.kmp, but it provides additional settings and modifications outside of a KMP. The main idea is, that global game settings can be overridden by track dependent settings in cases, where it is not possible yet.

LEX is designed to be open for future extension by third persons. Anyway, each new section identifier (a 4 byte ASCII magic) should be requested to Wiimm. He will guarantee unique identifiers.

At least LE-CODE will support LEX. Maybe CTGP will do it too.


Design Notes

There were some prerequisites before the definition of the file format:

  • The format should be open for future extensions.
  • Extensions are managed by sections. Each section is identified by an unique ID (magic).
  • ID (magic) should be 4 ASCII characters to make viewing by a hex-dumper easier.
  • Old implementations must work, event if new unknown subtypes were added. Therefore, the code needs an easy skip mechanism.
  • Tools to edit LEX files must be able to handle unknown subtypes as well (not deleting them when saving).
  • The order must be stable to support sections, that depend on other prior defined sections.
  • Fast and easy scanning support.
  • The data of an element should be easy to use for the game extender like LE-CODE.

File Format

Each file starts with a LEX header, followed by a sequence of LEX elements. The chain of elements is closed by a special termination element.

All values are stored in network byte order (big endian). Each element is aligned to 4 bytes.

File header of LEX files
Offset Type Description
0x00 char[4] File magic, always »LE-X«
0x04 uint16 The major version number. Usually 1. It is incremented, if the data structure changes in an incompatible way. This is not expected.
0x06 uint16 The minor version number. It starts with 0 and is incremented, if this header is expanded in a compatible way.
0x08 uint32 Total size of file. Use it for sanity cheeks.
0x0c uint32 Offset of first element of the chain relative to the beginning of this structure. Usually the value is 0x10, but this may change in future versions.
0x10 End of file header
Notes
  • The version number is printed as »%u.%u«. At the moment it is 1.0.
  • Don't expect the first element of the chain at offset 0x10. Use Offset of first element instead.

The list of elements is managed as chain. The header of each element contains a magic to identify it, and the size of the embedded data:

Element header of LEX files
Offset Type Description
0x00 uint32 A magic to identify the element (type of element). An alternative datatype is char[4].
0x04 uint32 Size of the element data. It is an offset to the next element relative to the data member too. The size is always a multiple of 4.
0x08 u8[SIZE] Binary data of the element.
0x08+SIZE The next element
Notes
  • Duplicates of the same element are forbidden.
  • The chain of elements is well ordered. Tools should usually not change this order, except they know exactly what they do.
  • The offset of the next element is: current_offset + 8 + element_size
  • Magic and size of the final element (terminator) are 0.

Example Implementation

For example code in C, see »LEX (File Format)/C example«.

Elements

Each element has its own identity (magic). This magic is an integer of 4 bytes. It can be interpreted as 4 characters. So it created by printable ASCII characters. At the moment, the identities are manged by Wiimm to guarantee unique values.

List of defined elements
Identity / Magic Date of
definition
Description
ASCII uint32
0 2019-02-28 Termintor of chain.
---- 0x2d2d2d2d 2019-02-28 Invalidated element. Ignore it.
CANN 0x43414e4e 2019-02-28 Cannon settings.

CANN : Cannon settings

Data of element "CANN" (ID 0x43414e4e)
Offset Type Description
0x00 uint32 N = Number of defined cannon types
0x04 float[4] 4 settings for cannon type #0
0x14 float[4] 4 settings for cannon type #1
0x24+ ...
4+N*16 End of data

See »Cannon Characteristics« for details about the settings. This format can be used to define more than three cannon types, but doing so makes the track incompatible with distributions not supporting the LEX file format.

Wiimms SZS Tools

As of v2.01a, Wiimms SZS Tools will support LEX files. The implementation is similar to KMP:

  • Command wlect cat[1] will print the content of a LEX file to stdout.
  • Command wlect decode[2] will decode a binary LEX file into a text form, that can be modified by tools and editors.
  • Command wlect encode[3] will create a text file into a binary LEX file. There a 2 pass compiler is used.
  • Command wszst lex[4] will search file course.lex and will print the content of it to stdout.
  • Encoding and decoding is integrated in tool wszst for automated decoding and encoding (like KMP before).
  • Commands filetype and fileattrib of all tools detect LEX files (binary and text).