normalize-url/index.js

'use strict';
// TODO: Use the `URL` global when targeting Node.js 10
const URLParser = typeof URL === 'undefined' ? require('url').URL : URL;

const testParameter = (name, filters) => {
	return filters.some(filter => filter instanceof RegExp ? filter.test(name) : filter === name);
};

module.exports = (urlString, opts) => {
	opts = Object.assign({
		defaultProtocol: 'http:',
		normalizeProtocol: true,
		forceHttp: false,
		forceHttps: false,
		stripHash: true,
		stripWWW: true,
		removeQueryParameters: [/^utm_\w+/i],
		removeTrailingSlash: true,
		removeDirectoryIndex: false,
		sortQueryParameters: true
	}, opts);

	// Backwards compatibility
	if (Reflect.has(opts, 'normalizeHttps')) {
		opts.forceHttp = opts.normalizeHttps;
	}

	if (Reflect.has(opts, 'normalizeHttp')) {
		opts.forceHttps = opts.normalizeHttp;
	}

	if (Reflect.has(opts, 'stripFragment')) {
		opts.stripHash = opts.stripFragment;
	}

	urlString = urlString.trim();

	const hasRelativeProtocol = urlString.startsWith('//');
	const isRelativeUrl = !hasRelativeProtocol && /^\.*\//.test(urlString);

	// Prepend protocol
	if (!isRelativeUrl) {
		urlString = urlString.replace(/^(?!(?:\w+:)?\/\/)|^\/\//, opts.defaultProtocol);
	}

	const urlObj = new URLParser(urlString);

	if (opts.forceHttp && opts.forceHttps) {
		throw new Error('The `forceHttp` and `forceHttps` options cannot be used together');
	}

	if (opts.forceHttp && urlObj.protocol === 'https:') {
		urlObj.protocol = 'http:';
	}

	if (opts.forceHttps && urlObj.protocol === 'http:') {
		urlObj.protocol = 'https:';
	}

	// Remove hash
	if (opts.stripHash) {
		urlObj.hash = '';
	}

	// Remove duplicate slashes if not preceded by a protocol
	if (urlObj.pathname) {
		// TODO: Use the following instead when targeting Node.js 10
		// `urlObj.pathname = urlObj.pathname.replace(/(?<!https?:)\/{2,}/g, '/');`
		urlObj.pathname = urlObj.pathname.replace(/((?![https?:]).)\/{2,}/g, (_, p1) => {
			if (/^(?!\/)/g.test(p1)) {
				return `${p1}/`;
			}
			return '/';
		});
	}

	// Decode URI octets
	if (urlObj.pathname) {
		urlObj.pathname = decodeURI(urlObj.pathname);
	}

	// Remove directory index
	if (opts.removeDirectoryIndex === true) {
		opts.removeDirectoryIndex = [/^index\.[a-z]+$/];
	}

	if (Array.isArray(opts.removeDirectoryIndex) && opts.removeDirectoryIndex.length > 0) {
		let pathComponents = urlObj.pathname.split('/');
		const lastComponent = pathComponents[pathComponents.length - 1];

		if (testParameter(lastComponent, opts.removeDirectoryIndex)) {
			pathComponents = pathComponents.slice(0, pathComponents.length - 1);
			urlObj.pathname = pathComponents.slice(1).join('/') + '/';
		}
	}

	if (urlObj.hostname) {
		// Remove trailing dot
		urlObj.hostname = urlObj.hostname.replace(/\.$/, '');

		// Remove `www.`
		// eslint-disable-next-line no-useless-escape
		if (opts.stripWWW && /^www\.([a-z\-\d]{2,63})\.([a-z\.]{2,5})$/.test(urlObj.hostname)) {
			// Each label should be max 63 at length (min: 2).
			// The extension should be max 5 at length (min: 2).
			// Source: https://en.wikipedia.org/wiki/Hostname#Restrictions_on_valid_host_names
			urlObj.hostname = urlObj.hostname.replace(/^www\./, '');
		}
	}

	// Remove query unwanted parameters
	if (Array.isArray(opts.removeQueryParameters)) {
		for (const key of [...urlObj.searchParams.keys()]) {
			if (testParameter(key, opts.removeQueryParameters)) {
				urlObj.searchParams.delete(key);
			}
		}
	}

	// Sort query parameters
	if (opts.sortQueryParameters) {
		urlObj.searchParams.sort();
	}

	// Take advantage of many of the Node `url` normalizations
	urlString = urlObj.toString();

	// Remove ending `/`
	if (opts.removeTrailingSlash || urlObj.pathname === '/') {
		urlString = urlString.replace(/\/$/, '');
	}

	// Restore relative protocol, if applicable
	if (hasRelativeProtocol && !opts.normalizeProtocol) {
		urlString = urlString.replace(/^http:\/\//, '//');
	}

	return urlString;
};