levn

levn

Light ECMAScript (JavaScript) Value Notation Levn is a library which allows you to parse a string into a JavaScript value based on an expected type. It is meant for short amounts of human entered data (eg. config files, command line arguments).

Levn aims to concisely describe JavaScript values in text, and allow for the extraction and validation of those values. Levn uses type-check for its type format, and to validate the results. MIT license. Version 0.4.1.

How is this different than JSON? levn is meant to be written by humans only, is (due to the previous point) much more concise, can be validated against supplied types, has regex and date literals, and can easily be extended with custom types. On the other hand, it is probably slower and thus less efficient at transporting large amounts of data, which is fine since this is not its purpose.

npm install levn

For updates on levn, follow me on twitter.

Quick Examples

var parse = require('levn').parse;
parse('Number', '2');      // 2
parse('String', '2');      // '2'
parse('String', 'levn');   // 'levn'
parse('String', 'a b');    // 'a b'
parse('Boolean', 'true');  // true

parse('Date', '#2011-11-11#'); // (Date object)
parse('Date', '2011-11-11');   // (Date object)
parse('RegExp', '/[a-z]/gi');  // /[a-z]/gi
parse('RegExp', 're');         // /re/
parse('Int', '2');             // 2

parse('Number | String', 'str'); // 'str'
parse('Number | String', '2');   // 2

parse('[Number]', '[1,2,3]');                      // [1,2,3]
parse('(String, Boolean)', '(hi, false)');         // ['hi', false]
parse('{a: String, b: Number}', '{a: str, b: 2}'); // {a: 'str', b: 2}

// at the top level, you can ommit surrounding delimiters
parse('[Number]', '1,2,3');                      // [1,2,3]
parse('(String, Boolean)', 'hi, false');         // ['hi', false]
parse('{a: String, b: Number}', 'a: str, b: 2'); // {a: 'str', b: 2}

// wildcard - auto choose type
parse('*', '[hi,(null,[42]),{k: true}]'); // ['hi', [null, [42]], {k: true}]

Usage

require('levn'); returns an object that exposes three properties. VERSION is the current version of the library as a string. parse and parsedTypeParse are functions.

// parse(type, input, options);
parse('[Number]', '1,2,3'); // [1, 2, 3]

// parsedTypeParse(parsedType, input, options);
var parsedType = require('type-check').parseType('[Number]');
parsedTypeParse(parsedType, '1,2,3'); // [1, 2, 3]

parse(type, input, options)

parse casts the string input into a JavaScript value according to the specified type in the type format (and taking account the optional options) and returns the resulting JavaScript value.

arguments

• type - String - the type written in the type format which to check against
• input - String - the value written in the levn format
• options - Maybe Object - an optional parameter specifying additional options

returns

* - the resulting JavaScript value

example

parse('[Number]', '1,2,3'); // [1, 2, 3]

parsedTypeParse(parsedType, input, options)

parsedTypeParse casts the string input into a JavaScript value according to the specified type which has already been parsed (and taking account the optional options) and returns the resulting JavaScript value. You can parse a type using the type-check library's parseType function.

arguments

• type - Object - the type in the parsed type format which to check against
• input - String - the value written in the levn format
• options - Maybe Object - an optional parameter specifying additional options

returns

* - the resulting JavaScript value

example

var parsedType = require('type-check').parseType('[Number]');
parsedTypeParse(parsedType, '1,2,3'); // [1, 2, 3]

Levn Format

Levn can use the type information you provide to choose the appropriate value to produce from the input. For the same input, it will choose a different output value depending on the type provided. For example, parse('Number', '2') will produce the number 2, but parse('String', '2') will produce the string "2".

If you do not provide type information, and simply use *, levn will parse the input according the unambiguous "explicit" mode, which we will now detail - you can also set the explicit option to true manually in the options.

• "string", 'string' are parsed as a String, eg. "a msg" is "a msg"
• #date# is parsed as a Date, eg. #2011-11-11# is new Date('2011-11-11')
• /regexp/flags is parsed as a RegExp, eg. /re/gi is /re/gi
• undefined, null, NaN, true, and false are all their JavaScript equivalents
• [element1, element2, etc] is an Array, and the casting procedure is recursively applied to each element. Eg. [1,2,3] is [1,2,3].
• (element1, element2, etc) is an tuple, and the casting procedure is recursively applied to each element. Eg. (1, a) is (1, a) (is [1, 'a']).
• {key1: val1, key2: val2, ...} is an Object, and the casting procedure is recursively applied to each property. Eg. {a: 1, b: 2} is {a: 1, b: 2}.
• Any test which does not fall under the above, and which does not contain special characters ([``]``(``)``{``}``:``,) is a string, eg. $12- blah is "$12- blah".

If you do provide type information, you can make your input more concise as the program already has some information about what it expects. Please see the type format section of type-check for more information about how to specify types. There are some rules about what levn can do with the information:

• If a String is expected, and only a String, all characters of the input (including any special ones) will become part of the output. Eg. [({})] is "[({})]", and "hi" is '"hi"'.
• If a Date is expected, the surrounding # can be omitted from date literals. Eg. 2011-11-11 is new Date('2011-11-11').
• If a RegExp is expected, no flags need to be specified, and the regex is not using any of the special characters,the opening and closing / can be omitted - this will have the affect of setting the source of the regex to the input. Eg. regex is /regex/.
• If an Array is expected, and it is the root node (at the top level), the opening [ and closing ] can be omitted. Eg. 1,2,3 is [1,2,3].
• If a tuple is expected, and it is the root node (at the top level), the opening ( and closing ) can be omitted. Eg. 1, a is (1, a) (is [1, 'a']).
• If an Object is expected, and it is the root node (at the top level), the opening { and closing } can be omitted. Eg a: 1, b: 2 is {a: 1, b: 2}.

If you list multiple types (eg. Number | String), it will first attempt to cast to the first type and then validate - if the validation fails it will move on to the next type and so forth, left to right. You must be careful as some types will succeed with any input, such as String. Thus put String at the end of your list. In non-explicit mode, Date and RegExp will succeed with a large variety of input - also be careful with these and list them near the end if not last in your list.

Whitespace between special characters and elements is inconsequential.

Options

Options is an object. It is an optional parameter to the parse and parsedTypeParse functions.

Explicit

A Boolean. By default it is false.

Example:

parse('RegExp', 're', {explicit: false});          // /re/
parse('RegExp', 're', {explicit: true});           // Error: ... does not type check...
parse('RegExp | String', 're', {explicit: true});  // 're'

explicit sets whether to be in explicit mode or not. Using * automatically activates explicit mode. For more information, read the levn format section.

customTypes

An Object. Empty {} by default.

Example:

var options = {
  customTypes: {
    Even: {
      typeOf: 'Number',
      validate: function (x) {
        return x % 2 === 0;
      },
      cast: function (x) {
        return {type: 'Just', value: parseInt(x)};
      }
    }
  }
}
parse('Even', '2', options); // 2
parse('Even', '3', options); // Error: Value: "3" does not type check...

Another Example:

function Person(name, age){
  this.name = name;
  this.age = age;
}
var options = {
  customTypes: {
    Person: {
      typeOf: 'Object',
      validate: function (x) {
        x instanceof Person;
      },
      cast: function (value, options, typesCast) {
        var name, age;
        if ({}.toString.call(value).slice(8, -1) !== 'Object') {
          return {type: 'Nothing'};
        }
        name = typesCast(value.name, [{type: 'String'}], options);
        age = typesCast(value.age, [{type: 'Numger'}], options);
        return {type: 'Just', value: new Person(name, age)};
    }
  }
}
parse('Person', '{name: Laura, age: 25}', options); // Person {name: 'Laura', age: 25}

customTypes is an object whose keys are the name of the types, and whose values are an object with three properties, typeOf, validate, and cast. For more information about typeOf and validate, please see the custom types section of type-check.

cast is a function which receives three arguments, the value under question, options, and the typesCast function. In cast, attempt to cast the value into the specified type. If you are successful, return an object in the format {type: 'Just', value: CAST-VALUE}, if you know it won't work, return {type: 'Nothing'}. You can use the typesCast function to cast any child values. Remember to pass options to it. In your function you can also check for options.explicit and act accordingly.

Technical About

levn is written in LiveScript - a language that compiles to JavaScript. It uses type-check to both parse types and validate values. It also uses the prelude.ls library.

Similar packages

Other packages similar to levn

yargs-parser

Yargs-parser is a package that parses command line arguments. It provides similar functionality to levn in terms of parsing and type coercion but is more focused on CLI arguments processing.

convict

Convict is a configuration management library that includes schema-based validation. It is similar to levn in that it can coerce environment variables and JSON files into the correct types based on a predefined schema.

joi

Joi is an object schema validation library which is often used to validate data types and structures. While it doesn't directly parse strings like levn, it provides a robust set of features for ensuring data types and structures are as expected.