ALE — Abstract List Expression

Specification v0.0.2

ALE is a human-readable, token-efficient data notation inspired by JSON, TOON, and YAML. It is designed to represent structured data deterministically with minimal syntax overhead, making it especially suitable for configs, datasets, and LLM prompts.

---

$3

- High human readability
- Low syntactic noise (token-efficient)
- Deterministic parsing (no heuristics)
- Native support for:
- implicit objects
- heterogeneous arrays
- homogeneous tables (arrays of objects)
- complex nested blocks (slots)

ALE is data-only (no execution, no expressions).

---

$3

- Hierarchy is defined only by indentation.
- Tabs or spaces are allowed but must be consistent.
- Dedentation closes the current block.
- Empty lines are ignored.

---

$3

ALE can represent:
- Object
- Array
- String
- Number (int / float)
- Boolean
- Null

---

$3

3.1 Key–Value Pair

``
key: value
`

3.2 Object Continuity Rule

Consecutive key: value lines belong to the same object unless:
- a > appears (new sibling), or
- indentation level decreases.

Example:

`
id: 1
name: Booker
tags: [shark, friend]
`
`json
{ "id": 1, "name": "Booker", "tags": ["shark", "friend"] }
`

---

$3

4.1 Block-Inferred Array

`
items:
1
"hello"
[1,2]
`

If a key ends with : and the indented block does not start with key: value, the key is an array.

4.2 Inline Array

`json
[1, 2, 3, 4.5]
`

4.3 Empty Array

`
tags: []
`

---

$3

5.1 Table Header

`
key{a, b, c}:
v1, v2, v3
v1, v2, v3
`

Each row produces an object by column position.

Example:

`
characters{id, name}:
1, Eleanor
2, Swifty
`
`json
[
{ "id": 1, "name": "Eleanor" },
{ "id": 2, "name": "Swifty" }
]
`

5.2 Null Values in Tables
- - → null
- Empty cell (Booker,) → null

---

$3

6.1 [{a,b}] — Array as a Single Element

`
mixed:
[{name, age}]:
Alexa, 28
Booker, -
`
`json
[
[
{ "name": "Alexa", "age": 28 },
{ "name": "Booker", "age": null }
]
]
`

6.2 ...{a,b} — Spread / Unpack

`
mixed:
...{name, age}:
Alexa, 28
Booker, -
`
`json
[
{ "name": "Alexa", "age": 28 },
{ "name": "Booker", "age": null }
]
`

---

$3

Inside arrays, > starts a new sibling element.

`
mixed:
id: 1
name: Booker
>id: 0
name: Alexa
`
`json
[
{ "id": 1, "name": "Booker" },
{ "id": 0, "name": "Alexa" }
]
`

---

$3

Syntax | Result
--------|------------
!flag | flag: false
¡flag | flag: true

Explicit booleans are also valid:

active: true

---

$3

value: -

---

$3

10.1 Bare Strings

name: Alexa

10.2 Escaping
- \n → newline
- \, → literal comma
- \" → quote

`
text: Hola\, soy Ale\nEncantada
`

10.3 Quoted Strings

Required when containing commas, structural symbols, or leading spaces:

`
text: "Hola, soy \"Ale\" (Alexa)"
`

---

$3

11.1 Definition

[:] defines a slot, not a normal empty array.
- Slots always resolve to arrays
- Empty slot → []
- Slots accept only free elements

Conceptually:

let arr = [];
arr.push(...)

Example:

`
notes: [:]
"text"
word: konnichiwa
`

---

$3

When multiple slots exist:
- < → move to next slot
-

`
german, 0, [:], {aspect, difficulty, attempts}
word: -
meaning: -
>word: fuchs
¡favAnimal

declination, 80, 4
pronunciation, 90, 2
`

---

$3

Indented lines that are not:
- key: value
- flags
- \> or <
- table rows

are treated as string literals and appended to the active slot.

---

$3

Keys may contain spaces if quoted:

`
"full name": Alexa
"user tags": [fox, artist]
`

---

$3

1. Indentation
2. > / <
3. Flags (!, ¡)
4. Table headers {}
5. Spread ...{}
6. Array-as-element [{ } ]
7. Inline arrays []
8. Scalars (null → bool → number → string)

---

Token Efficiency (Measured)

Using OpenAI Tokenizer (model: GPT-4o / GPT-4o mini):

Format | Tokens
-------|--------
JSON | 544
TOON | 385
ALE | 278

ALE achieves ~49% token reduction vs JSON and ~28% vs TOON.

Examples used:

$3

`json
{

"name": "Alexa",
"age": 28,
"characters": [
{ "id": 1, "name": "Eleanor" },
{ "id": 2, "name": "Swifty" }
],
"mixedArray": [
1,
[1, 2, 3, 4.5],
{
"id": 1,
"name": "Booker",
"tags": ["shark", "friend"]
},
{
"id": 0,
"name": "Alexa",
"active": false,
"isFox": true,
"languages": [
{
"name": "spanish",
"level": 100,
"notes": "lingua franca",
"easy": true
},
{
"name": "english",
"level": 70,
"notes": ["Oskar", "Kilo"],
"easy": true
},
{
"name": "japanese",
"level": 20,
"notes": [
{
"word": "konnichiwa",
"easy": true
},
{
"katakana": "キツネ",
"meaning": "fox"
},
"same",
"neko",
"shika"
],
"easy": false
},{
"name": "german",
"level": 0,
"notes": [
{
"word": null,
"meaning": null
},
{
"word": "fuchs",
"favAnimal": true
}
],
"easy": [
{
"aspect": "declination",
"difficulty": 80,
"attempts": 4
},
{
"aspect": "pronunciation",
"difficulty": 90,
"attempts": 2
},
{
"aspect": "vocabulary",
"difficulty": 40,
"attempts": 5
}
]
}
]
},
{ "name": "Alexa", "age": 28 },
{ "name": "Booker", "age": null },
{ "name": "Eleanor", "age": 30 },
[
{ "name": "Alexa", "age": 28 },
{ "name": "Booker", "age": null },
{ "name": "Eleanor", "age": 30 }
]
]
}
`

$3

`
name: Alexa
age: 28
characters: items[2]{id,name}:
1,Eleanor
2,Swifty
mixedArray: items[8]:
- 1
- items[4]:
- 1
- 2
- 3
- 4.5
- id: 1
name: Booker
tags: items[2]:
- shark
- friend
- id: 0
name: Alexa
active: false
isFox: true
languages: items[4]:
- name: spanish
level: 100
notes: lingua franca
easy: true
- name: english
level: 70
notes: items[2]:
- Oskar
- Kilo
easy: true
- name: japanese
level: 20
notes: items[5]:
- word: konnichiwa
easy: true
- katakana: キツネ
meaning: fox
- same
- neko
- shika
easy: false
- name: german
level: 0
notes: items[2]:
- word:
meaning:
- word: fuchs
favAnimal: true
easy: items[3]{aspect,difficulty,attempts}:
declination,80,4
pronunciation,90,2
vocabulary,40,5
- name: Alexa
age: 28
- name: Booker
age:
- name: Eleanor
age: 30
- items[3]{name,age}:
Alexa,28
Booker,
Eleanor,30
`

$3

`
name: Alexa
age: 28
characters{id, name}:
1, Eleanor
2, Swifty
mixedArray:
1
[1, 2, 3, 4.5]
id: 1
name: Booker
tags: [shark, friend]
>id: 0
name: Alexa
!active
¡isFox
languages{name, level, notes, easy}:
spanish, 100, lingua franca, ¡
english, 70, [Oskar, Kilo], ¡
japanese, 20, [:], !
word: konnichiwa
¡easy
>katakana: キツネ
meaning: fox
same
neko
shika
german, 0, [:], {aspect, difficulty, attempts}
word:
meaning:
>word: fuchs
¡favAnimal
pronunciation, 90, 2
vocabulary, 40, 5
...{name, age}:
Alexa, 28
Booker,
Eleanor, 30
{name, age}:
Alexa, 28
Booker,
Eleanor, 30
`

---

Status

- v0.2 — Stable specification
- Deterministic, human-friendly, LLM-efficient
- Suitable for bidirectional ALE ⇄ JSON parsing

Usage

$3

`bash
npm i @alexa.kitsune/ale
`

$3


`js
import { parseALE } from "@alexa.kitsune/ale";

const src =
name: Alexa
age: 28
characters{id, name}:
1, Eleanor
2, Swifty
;

const data = parseALE(src);
console.log(data);
`

$3


`js
import { stringifyALE } from "@alexa.kitsune/ale";

const obj = {
name: "Alexa",
age: 28,
characters: [
{ id: 1, name: "Eleanor" },
{ id: 2, name: "Swifty" }
]
};

console.log(stringifyALE(obj));
``

> NOTE
> The package is published as ESM (type: "module"). Use import ... (Node >= 16).

---

End of specification.