Skip to main content

Module parser

serializer::llm

Module parser 

Source
Expand description

DX Serializer LLM Format Parser

Parses the token-optimized LLM format into DxDocument. 52-73% more token-efficient than JSON.

§LLM Format Syntax (Wrapped Dataframe)

# Key-Value Pairs
name=MyApp
port=8080
description="Multi word string"

# Arrays (square brackets)
tags=[rust performance serialization]
editors=[neovim zed "firebase studio"]

# Objects (parentheses)
config(host=localhost port=5432 debug=true)
server(url="https://api.example.com" timeout=30)

# Tables (wrapped dataframes - deterministic parsing)
users[id name email](
1 Alice alice@ex.com
2 Bob bob@ex.com
3 Carol carol@ex.com
)

# Multi-word values use quotes
employees[id name dept](
1 "James Smith" Engineering
2 "Mary Johnson" "Research and Development"
)

§Removed Features (for TypeScript compatibility)

The following features were removed to maintain compatibility with the TypeScript implementation:

  • Boolean shorthand: + for true, - for false (use true/false instead)
  • Null shorthand: ~ for null (use null instead)

Structs§

LlmParser
Dx Serializer format parser

Enums§

ParseError
Parse errors for Dx Serializer format