What is typed-function?
The typed-function npm package allows you to define functions with typed arguments in JavaScript. It provides a way to enforce type checking at runtime, making your code more robust and easier to debug.
What are typed-function's main functionalities?
Define a typed function
This feature allows you to define a function with multiple signatures, each with different types of arguments. The function will execute the appropriate implementation based on the types of the provided arguments.
const typed = require('typed-function');
const add = typed({
'number, number': function (a, b) {
return a + b;
},
'string, string': function (a, b) {
return a + b;
}
});
console.log(add(2, 3)); // 5
console.log(add('Hello, ', 'world!')); // 'Hello, world!'
Type checking
This feature enforces type checking at runtime, throwing an error if the provided arguments do not match any of the defined signatures. This helps catch type-related bugs early in the development process.
const typed = require('typed-function');
const multiply = typed({
'number, number': function (a, b) {
return a * b;
}
});
try {
console.log(multiply(2, '3')); // Throws an error
} catch (err) {
console.error(err.message); // 'TypeError: Unexpected type of argument in function multiply (expected: number, actual: string, index: 1)'
}
Default types
This feature allows you to define default types for your function arguments. If no arguments are provided or if they do not match any specific type, the function will fall back to the default implementation.
const typed = require('typed-function');
const greet = typed({
'string': function (name) {
return 'Hello, ' + name + '!';
},
'any': function () {
return 'Hello, world!';
}
});
console.log(greet('Alice')); // 'Hello, Alice!'
console.log(greet()); // 'Hello, world!'
Other packages similar to typed-function
io-ts
io-ts is a runtime type system for IO decoding/encoding in TypeScript. It allows you to define types and validate data at runtime. Compared to typed-function, io-ts is more focused on data validation and transformation rather than function overloading.
runtypes
Runtypes provides a way to define and validate types at runtime in TypeScript. It offers a similar type-checking functionality but is more geared towards defining and validating data structures rather than function signatures.
ts-runtime
ts-runtime is a TypeScript transformer that adds runtime type checks to your TypeScript code. It provides a more integrated approach to type checking in TypeScript, whereas typed-function is a standalone library for JavaScript.
typed-function
Move type checking logic and type conversions outside of your function in a
flexible, organized way. Automatically throw informative errors in case of
wrong input arguments.
Features
typed-function has the following features:
- Runtime type-checking of input arguments.
- Automatic type conversion of arguments.
- Compose typed functions with multiple signatures.
- Supports union types, any type, and variable arguments.
- Detailed error messaging.
Supported environments: node.js, Chrome, Firefox, Safari, Opera, IE9+.
Why?
In JavaScript, functions can be called with any number and any type of arguments.
When writing a function, the easiest way is to just assume that the function
will be called with the correct input. This leaves the function's behavior on
invalid input undefined. The function may throw some error, or worse,
it may silently fail or return wrong results. Typical errors are
TypeError: undefined is not a function or TypeError: Cannot call method
'request' of undefined. These error messages are not very helpful. It can be
hard to debug them, as they can be the result of a series of nested function
calls manipulating and propagating invalid or incomplete data.
Often, JavaScript developers add some basic type checking where it is important,
using checks like typeof fn === 'function'
, date instanceof Date
, and
Array.isArray(arr)
. For functions supporting multiple signatures,
the type checking logic can grow quite a bit, and distract from the actual
logic of the function.
For functions dealing with a considerable amount of type checking and conversion
logic, or functions facing a public API, it can be very useful to use the
typed-function
module to handle the type-checking logic. This way:
- Users of the function get useful and consistent error messages when using
the function wrongly.
- The function cannot silently fail or silently give wrong results due to
invalid input.
- Correct type of input is assured inside the function. The function's code
becomes easier to understand as it only contains the actual function logic.
Lower level utility functions called by the type-checked function can
possibly be kept simpler as they don't need to do additional type checking.
It's important however not to overuse type checking:
- Locking down the type of input that a function accepts can unnecessary limit
it's flexibility. Keep functions as flexible and forgiving as possible,
follow the
robustness principle
here: "be liberal in what you accept and conservative in what you send"
(Postel's law).
- There is no need to apply type checking to all functions. It may be
enough to apply type checking to one tier of public facing functions.
- There is a performance penalty involved for all type checking, so applying
it everywhere can unnecessarily worsen the performance.
Load
Install via npm:
npm install typed-function
Usage
Here some usage examples. More examples are available in the
/examples folder.
var typed = require('typed-function');
var fn1 = typed('number, string', function (a, b) {
return 'a is a number, b is a string';
});
var fn2 = typed('string, number | boolean', function (a, b) {
return 'a is a string, b is a number or a boolean';
});
var fn3 = typed('string, any', function (a, b) {
return 'a is a string, b can be anything';
});
var fn4 = typed({
'number': function (a) {
return 'a is a number';
},
'number, boolean': function (a, b) {
return 'a is a number, b is a boolean';
},
'number, number': function (a, b) {
return 'a is a number, b is a number';
}
});
console.log(fn1(2, 'foo'));
console.log(fn4(2));
try {
fn2('hello', 'world');
}
catch (err) {
console.log(err.toString());
}
Types
typed-function has the following built-in types:
null
boolean
number
string
function
Array
Date
RegExp
Object
The following type expressions are supported:
- Multiple arguments:
string, number, function
- Union types:
number | string
- Variable arguments:
...number
- Any type:
any
API
Construction
A typed function can be constructed in three ways:
-
With a single signature:
typed(signature: string, fn: function) : function
typed(name: string, signature: string, fn: function) : function
-
With multiple signatures:
typed(signatures: Object.<string, function>) : function
typed(name: string, signatures: Object.<string, function>) : function
-
Merge multiple typed functions into a new typed function
typed(functions: ...function) : function
Properties
- `typed.conversions: Array`
An Array with built-in conversions. Empty by default. Can be used for example
to defined conversions from `boolean` to `number`. For example:
```js
typed.conversions.push({
from: 'boolean',
to: 'number',
convert: function (x) {
return +x;
});
```
- `typed.config: Object`
An object with configuration options for typed-function:
- `minify: boolean`
If true (default), the functions are generated from minified code.
If false the typed-functions have a nicely readable .toString() source.
### Output
The functions generated with `typed({...})` have:
- A function `toString`. When `typed.config.minify` is set to `true` (is `false`
by default), the `toString` function will return well readable code which can
be used to see what the function exactly does. For debugging purposes.
- A property `signatures`, which holds a map with the (normalized)
signatures as key and the original sub-functions as value.
- A property `name` containing name of the typed function or an empty string.
## Roadmap
### Version 1
- Be able to turn off exception throwing.
- Extend function signatures:
- Optional arguments like `'[number], array'` or like `number=, array`
- Nullable arguments like `'?Object'`
- Create a good benchmark, to get insight in the overhead.
- Allow conversions to fail (for example string to number is not always
possible). Call this `fallible` or `optional`?
### Version 2
- Extend function signatures:
- Constants like `'"linear" | "cubic"'`, `'0..10'`, etc.
- Object definitions like `'{name: string, age: number}'`
- Object definitions like `'Object.<string, Person>'`
- Array definitions like `'Array.<Person>'`
- Improve performance of both generating a typed function as well as
the performance and memory footprint of a typed function.
## Test
To test the library, run:
npm test
## Minify
To generate the minified version of the library, run:
npm run minify