🍠 trnp

A tiny data markup language for non-technical humans that's Hard To Mess Upᵗᵐ

features:

• Minimal syntax
• Reduce opening/closing syntax whenever possible (eg. quotes, braces, etc.)
• Minimal data types: Booleans, numbers, strings; Lists; Dictionaries
• Javascript-style Multiline and single-line comments
• Multiline strings without additional syntax
• No weird type coercion. Eg. 'yes' -> true
• Indentation matters. Nesting behavior is defined by indentation, regardless of tabs or spaces.
• Never errors
• Only 36 lines of code

Example

test: -34_000.00
yes:

/* This is a test */

// alert: 'danger!'

foo:
  bar : Oh hello
        this is a multiline string
  tags : [user, false, -1_234.05]

users :
  - name : scott
    id   : foo
  - name : mark
    id   : boop

more_test : "This is a string: with a colon in it"

this converts to:

{
	test: -34000.00,
	yes: undefined,
	foo: {
		bar : "Oh hello\nthis is a multiline string",
		tags: ["user", false, -1234.05]
	},
	users : [
		{ name : 'scott', id : 'foo' },
		{ name : 'mark', id  : 'boop' },
	],
	more_test : 'This is a string: with a colon in it'
}

Usage

npm install pico-trnp

const trnp = require('pico-trnp');

const groceries = trnp(`
- eggs
- bacon
- pancakes
`);

Types

• Booleans: must be either true or false exactly
• Numbers: Handles negatives, decimals, and underscores anywhere within the number (eg. -45_000.3_4_5 is valid)
• Strings: Any raw text. trnp does not use any quotes for parsing. Multiline strings must be the same indentation across lines.
• Arrays: Simple single line arrays, wrapped in [], and comma separated
• Lists: Line must start with - . Contiguous lines with the same indentation will be joined into one list
• Dictionaries: Key-val separated by a : . Must have a space after if it's value is on the same line. Subsequent lines indented deeper than the key are processed as the value.

Why not use YAML/TOML/JSON ?

I have found that with many of my projects having stake-holders, project managers, designers, etc. be able to modify and update some configuration data has lead to reducing development bottlenecks and allowed these project members to have more control over minor changes without going through developers each time.

Unfortunately most data markup languages are quite strict, and non-technical humans have a hard time interacting and updating these types of files.

YAML is close to what I want, but I have found that it's spec has too many weird edge cases, and the parsers for it are clunky and hard to use when dealing with server-side/client-side bundlers.

How it works

trnp runs in three steps:

1. First remove anything that looks like a comment
2. Use regex to split the input into tokens. Each token has an indentation quantity, and is either a value, list, or dictionary.
3. Steps through each token to create a final value. If it encounters a list or dictionary, it plucks out all tokens with a indentation greater than that token and runs the parse step on just those tokens. That result becomes the value of that group.

To demonstrate, these are the tokens from the above example:

[
  { key: 'test', idt: 3 },
  { val: -34000, idt: 9 },
  { key: 'yes', idt: 3 },
  { key: 'foo', idt: 3 },
  { key: 'bar', idt: 5 },
  { val: 'Oh hello', idt: 11 },
  { val: 'this is a multiline string', idt: 11 },
  { key: 'tags', idt: 5 },
  { val: ["user", false, -1234.05], idt: 9},
  { key: 'goop', idt: 5 },
  { val: false, idt: 12 },
  { key: 'users', idt: 3 },
  { arr: true, idt: 5 },
  { key: 'name', idt: 7 },
  { val: 'scott', idt: 14 },
  { key: 'id', idt: 7 },
  { val: 'foo', idt: 14 },
  { arr: true, idt: 5 },
  { key: 'name', idt: 7 },
  { val: 'mark', idt: 14 },
  { key: 'id', idt: 7 },
  { val: 'boop', idt: 14 },
  { key: 'more_test', idt: 3 },
  { val: '"This is a string: with a colon in it"', idt: 15 }
]