normalize-url 3.0.0 → 3.3.0

package/index.js

@@ -1,15 +1,18 @@
 'use strict';
-const {URL} = require('url');
+// TODO: Use the `URL` global when targeting Node.js 10
+const URLParser = typeof URL === 'undefined' ? require('url').URL : URL;
 
-function testParameter(name, filters) {
+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,
-		normalizeHttps: false,
-		stripFragment: true,
+		forceHttp: false,
+		forceHttps: false,
+		stripHash: true,
 		stripWWW: true,
 		removeQueryParameters: [/^utm_\w+/i],
 		removeTrailingSlash: true,
@@ -17,6 +20,19 @@ module.exports = (urlString, opts) => {
 		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('//');
@@ -24,23 +40,38 @@ module.exports = (urlString, opts) => {
 
 	// Prepend protocol
 	if (!isRelativeUrl) {
-		urlString = urlString.replace(/^(?!(?:\w+:)?\/\/)|^\/\//, 'http://');
+		urlString = urlString.replace(/^(?!(?:\w+:)?\/\/)|^\/\//, opts.defaultProtocol);
 	}
 
-	const urlObj = new URL(urlString);
+	const urlObj = new URLParser(urlString);
+
+	if (opts.forceHttp && opts.forceHttps) {
+		throw new Error('The `forceHttp` and `forceHttps` options cannot be used together');
+	}
 
-	if (opts.normalizeHttps && urlObj.protocol === 'https:') {
+	if (opts.forceHttp && urlObj.protocol === 'https:') {
 		urlObj.protocol = 'http:';
 	}
 
-	// Remove fragment
-	if (opts.stripFragment) {
+	if (opts.forceHttps && urlObj.protocol === 'http:') {
+		urlObj.protocol = 'https:';
+	}
+
+	// Remove hash
+	if (opts.stripHash) {
 		urlObj.hash = '';
 	}
 
-	// Remove duplicate slashes
+	// Remove duplicate slashes if not preceded by a protocol
 	if (urlObj.pathname) {
-		urlObj.pathname = urlObj.pathname.replace(/\/{2,}/g, '/');
+		// 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
@@ -68,7 +99,11 @@ module.exports = (urlString, opts) => {
 		urlObj.hostname = urlObj.hostname.replace(/\.$/, '');
 
 		// Remove `www.`
-		if (opts.stripWWW) {
+		// 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\./, '');
 		}
 	}

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "normalize-url",
-	"version": "3.0.0",
+	"version": "3.3.0",
 	"description": "Normalize a URL",
 	"license": "MIT",
 	"repository": "sindresorhus/normalize-url",
@@ -13,7 +13,7 @@
 		"node": ">=6"
 	},
 	"scripts": {
-		"test": "xo && ava"
+		"test": "xo && nyc ava"
 	},
 	"files": [
 		"index.js"
@@ -35,6 +35,8 @@
 	],
 	"devDependencies": {
 		"ava": "*",
+		"coveralls": "^3.0.0",
+		"nyc": "^12.0.2",
 		"xo": "*"
 	}
 }

package/readme.md

@@ -1,4 +1,4 @@
-# normalize-url [![Build Status](https://travis-ci.org/sindresorhus/normalize-url.svg?branch=master)](https://travis-ci.org/sindresorhus/normalize-url)
+# normalize-url [![Build Status](https://travis-ci.org/sindresorhus/normalize-url.svg?branch=master)](https://travis-ci.org/sindresorhus/normalize-url) [![Coverage Status](https://coveralls.io/repos/github/sindresorhus/normalize-url/badge.svg?branch=master)](https://coveralls.io/github/sindresorhus/normalize-url?branch=master)
 
 > [Normalize](https://en.wikipedia.org/wiki/URL_normalization) a URL
 
@@ -39,12 +39,17 @@ URL to normalize.
 
 Type: `Object`
 
+##### defaultProtocol
+
+Type: `string`<br>
+Default: `http:`
+
 ##### normalizeProtocol
 
 Type: `boolean`<br>
 Default: `true`
 
-Prepend `http:` to the URL if it's protocol-relative.
+Prepends `defaultProtocol` to the URL if it's protocol-relative.
 
 ```js
 normalizeUrl('//sindresorhus.com:80/');
@@ -54,12 +59,12 @@ normalizeUrl('//sindresorhus.com:80/', {normalizeProtocol: false});
 //=> '//sindresorhus.com'
 ```
 
-##### normalizeHttps
+##### forceHttp
 
 Type: `boolean`<br>
 Default: `false`
 
-Normalize `https:` URLs to `http:`.
+Normalizes `https:` URLs to `http:`.
 
 ```js
 normalizeUrl('https://sindresorhus.com:80/');
@@ -69,18 +74,35 @@ normalizeUrl('https://sindresorhus.com:80/', {normalizeHttps: true});
 //=> 'http://sindresorhus.com'
 ```
 
-##### stripFragment
+##### forceHttps
+
+Type: `boolean`<br>
+Default: `false`
+
+Normalizes `http:` URLs to `https:`.
+
+```js
+normalizeUrl('https://sindresorhus.com:80/');
+//=> 'https://sindresorhus.com'
+
+normalizeUrl('http://sindresorhus.com:80/', {normalizeHttp: true});
+//=> 'https://sindresorhus.com'
+```
+
+This option can't be used with the `forceHttp` option at the same time.
+
+##### stripHash
 
 Type: `boolean`<br>
 Default: `true`
 
-Remove the fragment at the end of the URL.
+Removes hash from the URL.
 
 ```js
 normalizeUrl('sindresorhus.com/about.html#contact');
 //=> 'http://sindresorhus.com/about.html'
 
-normalizeUrl('sindresorhus.com/about.html#contact', {stripFragment: false});
+normalizeUrl('sindresorhus.com/about.html#contact', {stripHash: false});
 //=> 'http://sindresorhus.com/about.html#contact'
 ```
 
@@ -89,7 +111,7 @@ normalizeUrl('sindresorhus.com/about.html#contact', {stripFragment: false});
 Type: `boolean`<br>
 Default: `true`
 
-Remove `www.` from the URL.
+Removes `www.` from the URL.
 
 ```js
 normalizeUrl('http://www.sindresorhus.com/about.html#contact');
@@ -104,7 +126,7 @@ normalizeUrl('http://www.sindresorhus.com/about.html#contact', {stripWWW: false}
 Type: `Array<RegExp|string>`<br>
 Default: `[/^utm_\w+/i]`
 
-Remove query parameters that matches any of the provided strings or regexes.
+Removes query parameters that matches any of the provided strings or regexes.
 
 ```js
 normalizeUrl('www.sindresorhus.com?foo=bar&ref=test_ref', {
@@ -118,7 +140,7 @@ normalizeUrl('www.sindresorhus.com?foo=bar&ref=test_ref', {
 Type: `boolean`<br>
 Default: `true`
 
-Remove trailing slash.
+Removes trailing slash.
 
 **Note:** Trailing slash is always removed if the URL doesn't have a pathname.
 
@@ -138,7 +160,7 @@ normalizeUrl('http://sindresorhus.com/', {removeTrailingSlash: false});
 Type: `boolean` `Array<RegExp|string>`<br>
 Default: `false`
 
-Remove the default directory index file from path that matches any of the provided strings or regexes. When `true`, the regex `/^index\.[a-z]+$/` is used.
+Removes the default directory index file from path that matches any of the provided strings or regexes. When `true`, the regex `/^index\.[a-z]+$/` is used.
 
 ```js
 normalizeUrl('www.sindresorhus.com/foo/default.php', {
@@ -152,7 +174,7 @@ normalizeUrl('www.sindresorhus.com/foo/default.php', {
 Type: `boolean`<br>
 Default: `true`
 
-Sort the query parameters alphabetically by key.
+Sorts the query parameters alphabetically by key.
 
 ```js
 normalizeUrl('www.sindresorhus.com?b=two&a=one&c=three', {