Safen

Build Coverage License Language Typescript
deno.land/x/safen Version Downloads

Safen is a high-performance validation and sanitization library with easy type inference. Its syntax is similar to TypeScript interface, making it easy to create validation rules.

https://user-images.githubusercontent.com/4086535/203831205-8b3481cb-bb8d-4f3c-9876-e41adb6855fd.mp4

Installation

Node

npm install safen

Deno

import {
  s, // sanitize,
  v, // validate,
} from "https://deno.land/x/safen/mod.ts";

Basic Usage

Create Validate Fn

import { v } from "https://deno.land/x/safen/mod.ts";

const validate = v(String); // now, validate: (data: unknown) => data is string

const input = {} as unknown;
if (validate(input)) {
  // now input is string!
}

Create Sanitize Fn

import { s } from "https://deno.land/x/safen/mod.ts";

const sanitize = s(String); // now, sanitize: (data: unknown) => string

const input = {} as unknown; // some unknown value

sanitize("something" as unknown); // return "something"
sanitize(null as unknown); // throw InvalidValueError

Types

// Primitive Types
const validate = v(String); // (data: unknown) => data is string
const validate = v(Number); // (data: unknown) => data is number
const validate = v(Boolean); // (data: unknown) => data is boolean
const validate = v(BigInt); // (data: unknown) => data is bigint
const validate = v(Symbol); // (data: unknown) => data is symbol

// Literal Types
const validate = v("foo"); // (data: unknown) => data is "foo"
const validate = v(1024); // (data: unknown) => data is 1024
const validate = v(true); // (data: unknown) => data is true
const validate = v(2048n); // (data: unknown) => data is 2048n
const validate = v(null); // (data: unknown) => data is null
const validate = v(undefined); // (data: unknown) => data is undefined

// Special
const validate = v(v.any()); // (data: unknown) => data is any
const validate = v(Array); // (data: unknown) => data is any[]

// Object
const Point = { x: Number, y: Number };
const validate = v({ p1: Point, p2: Point }); // (data: unknown) => data is { p1: { x: number, y: number }, p2: { x: number, y: number } }

// Union
const validate = v(v.union([String, Number])); // (data: unknown) => data is string | number

// Array
const validate = v([String]); // (data: unknown) => data is string[]
const validate = v([v.union([String, Number])]); // (data: unknown) => data is (string | number)[]

Decorator

Decorators do not affect type inference, but do affect additional validation and data transformation.

Step1. Basic Sanitize

const sanitize = s(s.union([
  String,
  null,
]));

sanitize("hello world!"); // return "hello world!"
sanitize("  hello world!  "); // return "  hello world!  "
sanitize("    "); // return "    "
sanitize(null); // return null

Step2. Add trim decorator

const sanitize = s(s.union([
  s.decorate(String, (d) => d.trim()),
  null,
]));

sanitize("hello world!"); // return "hello world!"
sanitize("  hello world!  "); // return "hello world!"
sanitize("    "); // return ""
sanitize(null); // return null

Step3. Add emptyToNull decorator

const sanitize = s(
  s.decorate(
    s.union([
      s.decorate(String, (d) => d.trim()),
      null,
    ]),
    (d) => d.emptyToNull(),
  ),
);

sanitize("hello world!"); // return "hello world!"
sanitize("  hello world!  "); // return "hello world!"
sanitize("    "); // return null
sanitize(null); // return null

Defined Decorators

Decorator Validate Transform Type Description
alpha string contains only letters([a-zA-Z]).
alphanum string contains only letters and numbers([a-zA-Z0-9])
ascii string contains only ascii characters.
base64 string Base64.
between(min, max) string, number value is between {min} and {max}. (ex) between("aaa","zzz"), between(1,100)
ceil number Math.ceil. (ref. floor, round)
creditcard string valid Credit Card number. cf. 0000-0000-0000-0000
dateformat string valid Date string(RFC2822, ISO8601). cf. 2018-12-25, 12/25/2018, Dec 25, 2018
email string valid E-mail string.
emptyToNull string or null empty string("") to null
floor number Math.floor. (ref. ceil, round)
hexcolor string valid Hex Color string. cf. #ffffff
ip(version = null) string valid UUID.
version is one of null(both, default), v4, and v6.
json string valid JSON.
length(size) string, any[] length is {size}.
lengthBetween(min, max) string, any[] length is between {min} and {max}.
lengthMax(max) string, any[] length is less than {max}.
lengthMin(min) string, any[] length is greater than {min}.
lowercase string lowercase.
macaddress string valid Mac Address.
max(max) string, number value is less than {min}.
min(min) string, number value is greater than {max}.
port number valid PORT(0-65535).
re string match RegExp.
round number Math.round. (ref. ceil, floor)
stringify string cast to string
toLower string change to lower case.
toUpper string change to upper case.
trim string trim.
uppercase string uppercase.
url string valid URL.
uuid(version = null) string valid UUID.
version is one of null(default), v3, v4, and v5.

Custom Decorator

graph LR;
  A[input] -->|type = unknown| B{cast};
  B -->|type = T| C{validate};
  C -->|true| D{transform};
  C -->|false| E[error];
  D --> F[output];
interface Decorator<T> {
  name: string;
  cast?(v: unknown): T;
  validate?(v: T): boolean;
  transform?(v: T): T;
}

The cast function is invoked at the beginning of the data processing pipeline, before the validate and transform functions. The purpose of the cast function is to ensure that the data is in the right type before being processed further.

This is an example of a cast-only function:

const decorator: Decorator<string> = {
  name: "json_string",
  cast: (v) => JSON.stringify(v),
};

Once the data has been casted, the validate function is called to verify the content and format of the data. This function ensures that the data is valid and meets the specified criteria before being processed further.

The transform function, on the other hand, is invoked only after the validation function returns a true result. The transform function then processes the data according to the specified rules and criteria.

Therefore, the cast, validate, and transform functions work together to ensure that the data is in the right format, is valid, and is properly processed.

Benchmark

Please see benchmark results.

Old Version Docs