type-is

This version switches the package to ESM and node >= 22.

Changed:

• Publish the package as ESM and bump to node 22 (#106)

• Publishes built-in types using TypeScript (#94)

• Refactors the API shape to a class (#103)

  • 2-4x faster by initializing the class once (instead of every request)

  • 20% faster by initializing inline (e.g. every request)

  • Added support for matching parameters too (e.g. text/html; charset=utf-8)

  • Returns the original input string not the matched MIME type

    • Previously text/* would return e.g. text/html on match, now it returns the original text/*

  • Returns undefined instead of false on no match

  • Returns undefined when no input types are defined

  • Throws on invalid MIME type used to construct the matcher (e.g. new TypeIs(["invalid"]), it does not error if the header is invalid)

• Matches return the original input string (#109)

Removed:

• Removes mime-types from the package, only has 3 built in "shortcuts" (e.g. json -> application/json)

  • Pass mime-types.lookup as an option to override

Before:

const isText = typeis(req, ['text/*'])

After:

const isText = new TypeIs(['text/*']).request(req);

If you are upgrading an upstream dependency (e.g. express), and want maximum compatibility, something like this can help with compatibility for the extension aliases:

import * as mimeTypes from 'mime-types';
import { TypeIs, DEFAULT_LOOKUP } from 'type-is';

function lookup (value) {
  return mimeTypes.lookup(value) || DEFAULT_LOOKUP(value);
}

req.is = function (types) {
  const matcher = new TypeIs(types, { lookup });
  return matcher.request(this);
};

However the returned string has changed. Backward compatibility with the old string is achievable by mimicking the old behavior -- parse the MIME type yourself (example) and use the matcher.contentType method. This is not recommended though.