Binary Encoding

All token-based formats in LEGO Racers share a common serialization layer implemented in LRBinaryReader/LRBinaryWriter and BinaryFileHelper.

Token Enum

Every typed value in the binary stream is prefixed by a single-byte token:

Token	Hex	Size	Description
String	0x02	Variable	Null-terminated ASCII string
Float	0x03	4 bytes	IEEE 754 single-precision float
Int32	0x04	4 bytes	32-bit signed integer
LeftCurly	0x05	0	Block open {
RightCurly	0x06	0	Block close }
LeftBracket	0x07	0	Array header open [
RightBracket	0x08	0	Array header close ]
Fract8 / SByte	0x0B	1 byte	8-bit fixed-point (÷16) or signed byte
Byte	0x0C	1 byte	Unsigned byte
Fract16 / Short	0x0D	2 bytes	16-bit fixed-point (÷256) or signed short
UShort	0x0E	2 bytes	Unsigned 16-bit integer
Array	0x14	Variable	Compression pass only — not in decompressed stream
Struct	0x16	Variable	Compression pass only — struct definition

Note: Fract8/SByte share token 0x0B and Fract16/Short share 0x0D. Context determines interpretation.

Byte Order

All multi-byte values are little-endian.

File Structure Pattern

Every format class follows the same pattern:

1. Read a block ID byte (format-specific, typically starting at 0x27)
2. Switch on the block ID to read the appropriate data structure
3. Repeat until end of stream

Within blocks, property ID bytes identify individual fields. Properties are read until a RightCurly token is encountered.

Compression / Decompression

Binary files on disk may be compressed using a struct-based compression scheme. BinaryFileHelper.Decompress() handles this transparently:

• 0x16 (Struct) — Defines a reusable struct template: [struct_id] [field_count] [field_token_types...]
• 0x14 (Array) — Inline array: [array_length as int16] [element_token_type] [elements...]
• Struct IDs encountered later in the stream are expanded using the stored template

After decompression, the stream contains only the basic tokens (0x02–0x0E) and format-specific block/property IDs (0x27+).

Common Data Structures

Arrays

[array_length_token]    // 0x07 (LeftBracket)
  [Int32] count         // 0x04 + 4-byte length
[end_length]            // 0x08 (RightBracket)
[begin_data]            // 0x05 (LeftCurly)
  element[0]
  element[1]
  ...
[end_data]              // 0x06 (RightCurly)

Struct Arrays

Same as arrays, but each element is wrapped: [type_byte] { element_data }.

Dictionaries

Like struct arrays, but each element has an optional string key:

[type_byte]
"key_name"              // optional — if next token isn't String, uses index as key
{ element_data }

Integral Type Selection

ReadIntegralWithHeader() accepts any of: SByte, Byte, Int32, UShort, Short. The file format picks the smallest type that fits the value. This is why the same field may use different token types in different files.

See Primitive Types for the composite types built on these tokens.