validate-value

validate-value

validate-value validates values against JSON schemas.

Status

Category	Status
Version
Build
License

Installation

$ npm install validate-value

Philosophy

The rewrite of validate-value in version 9.0.0 follows the infamous quote "parse, don't validate"¹. The quote asserts that parsing data is more valuable than validating it. This is very obvious when comparing the two approaches in TypeScript:

const doThingsWithValidation = async function (options: unknown): Promise<void> {
  // This throws, if the options are not valid.
  validateOptions(options);

  const typedOptions = options as Options;

  doSomethingWithAnOption(typedOptions.someOption);
};

const doThingsWithParsing = async function (options: unknown): Promise<Result<void, Error>> {
  // This parses the options and unwraps them, throwing if they were invalid.
  // If invalid options were something our program could handle, we would not
  // want to throw here and instead handle the error appropriately.
  // In this example we want the function to throw, if the options are invalid.
  const typedOptions = parseOptions(options).unwrapOrThrow();

  doSomethingWithAnOption(typedOptions.someOption);
};

In the second example, the typedOptions contained in the Result returned from the parseOptions call already have the type that we expect them to have and we don't have to assert them or assign them to a new variable in any way. This combines better support from the TypeScript compiler with better error handling from defekt.

Quick start

First you need to integrate validate-value into your application:

const { parse, Parser } = require('validate-value');

If you use TypeScript, use the following code instead:

import { parse, Parser } from 'validate-value';

Then, create a new instance and provide a JSON schema that you would like to use for parsing:

const parser = new Parser({
  type: 'object',
  properties: {
    username: { type: 'string' },
    password: { type: 'string' }
  },
  additionalProperties: false,
  required: [ 'username', 'password' ]
});

If you are using TypeScript, you will want to provide a type for the parsed value:

interface User {
  username: string;
  password: string;
}

const parser = new Parser<User>({
  type: 'object',
  properties: {
    username: { type: 'string' },
    password: { type: 'string' }
  },
  additionalProperties: false,
  required: [ 'username', 'password' ]
});

Afterwards, you may use the parse function to parse a value:

const user = {
  username: 'Jane Doe',
  password: 'secret'
};

const parseResult = parser.parse(user);

if (parseResult.hasError()) {
  // The user object did not match the parser's schema. This should be handled.
} else {
  // The user object was parsed successfully and the result's value can now be used.
  doSomethingWithUser(parseResult.value);
}

After parsing, the value in the returned Result will have the type User, since it was passed to the parser upon construction.

For details on the error handling in the example above, check the documentation of the Result type in defekt.

Configuring the parser

If you want to, you can configure the parser according to your needs.

Configuring the error messages

By default, the error message uses value as identifier and . as the separator for the object that is parsed, but sometimes you may want to change this. Therefore, provide the desired identifier and separator as second parameter to the parse function:

const user = {
  username: 'Jane Doe',
  password: 'secret'
};

const parseResult = parser.parse(user, { valueName: 'person', separator: '/' });

Using a custom validator

Sometimes it is desired to use a custom validator instance, e.g. if you want to add custom formats. For that, you can hand over a custom Ajv instance to the parser:

const schema = // ...
const ajvInstance = new Ajv();

// Adjust Ajv instance according to your needs...

const parser = new Parser<User>(schema, {
  ajvInstance
});

Optionally, you can build upon the default instance that is used internally. For that, call the getDefaultAjvInstance function:

const schema = // ...
const ajvInstance = getDefaultAjvInstance();

// Adjust Ajv instance according to your needs...

const parser = new Parser<User>(schema, {
  ajvInstance
});

Parsing without a parser instance

For convenience, there is also the parse function, which skips the creation of a parser instance. You can use this if you're only going to use a schema for validation once. Otherwise, it is recommended to first create a parser instance, since then the JSON schema is only compiled once:

import { parse } from 'validate-value';

const parseResult = parse<User>(user, {
  type: 'object',
  properties: {
    username: { type: 'string' },
    password: { type: 'string' }
  },
  additionalProperties: false,
  required: [ 'username', 'password' ]
});

Verifying that a variable is of a specific type

To verify that a variable is of a specific type, use the isOfType function. Hand over the value you would like to verify, and a JSON schema describing that type. The function returns true if the given variable matches the schema, and false if it doesn't:

const { isOfType } = require('validate-value');

const user = {
  username: 'Jane Doe',
  password: 'secret'
};

const schema = {
  type: 'object',
  properties: {
    username: { type: 'string' },
    password: { type: 'string' }
  },
  additionalProperties: false,
  required: [ 'username', 'password' ]
};

if (isOfType(user, schema)) {
  // ...
}

When using TypeScript, you may even specify a generic type parameter, and use the function as a type guard:

import { isOfType, JsonSchema } from 'validate-value';

interface User {
  username: string;
  password: string;
}

const user = {
  username: 'Jane Doe',
  password: 'secret'
};

const schema: JsonSchema = {
  type: 'object',
  properties: {
    username: { type: 'string' },
    password: { type: 'string' }
  },
  additionalProperties: false,
  required: [ 'username', 'password' ]
};

if (isOfType<User>(user, schema)) {
  // ...
}

Resources

1: "Parse, don't validate", Alexis King, 2019.

Running quality assurance

To run quality assurance for this module use roboter:

$ npx roboter