Serialize and deserialize TOON (Token-Oriented Object Notation) - an LLM-optimized format that uses ~40-50% fewer tokens than JSON.
Uses the json-io library
Download the .lex file from releases and install via Lucee Admin, or drop it into your server's deploy directory.
result = SerializeTOON( var [, options] )Arguments:
- var - CFML data to serialize (struct, array, or any value) (required)
- options - Optional struct with serialization options:
prettyPrint- boolean (default false). Format output for readability
result = DeserializeTOON( source [, options] )Arguments:
- source - TOON string content or file path to read from (required)
- options - Optional struct with deserialization options:
charset- string (default UTF-8). Character encoding when reading from file
data = { name: "Zac", age: 42, tags: [ "lucee", "cfml" ] };
// Serialize to TOON
toon = SerializeTOON( data );
// With pretty print
toon = SerializeTOON( data, { prettyPrint: true } );// From string
toonString = '(name:"Zac" age:42 tags:["lucee" "cfml"])';
data = DeserializeTOON( toonString );
// From file
data = DeserializeTOON( "/path/to/data.toon" );
// With charset option
data = DeserializeTOON( "/path/to/data.toon", { charset: "UTF-16" } );TOON is designed for LLM (Large Language Model) communication:
- 40-50% fewer tokens than JSON
- No quotes around keys
- No commas between elements
- Compact representation
Example comparison:
{"name":"Zac","age":42,"tags":["lucee","cfml"]}(name:"Zac" age:42 tags:["lucee" "cfml"])
- Circular references not supported - Data structures with circular references will cause a StackOverflowError. This is due to a bug in json-io's ToonWriter where cycle detection is not applied to Map/Collection types.
- Currently only for Lucee 7.1.0.21+ due to OSGI tech debt, only fixed in 7.1
Uses json-io (bundled, Apache 2.0 license).
Apache License 2.0
