Add data URL support

index.js

@@ -6,6 +6,52 @@ const testParameter = (name, filters) => {
 	return filters.some(filter => filter instanceof RegExp ? filter.test(name) : filter === name);
 };
 
+// From https://github.com/killmenot/valid-data-url/blob/8d56cab866469a4608001f0e8c5d73f65789ba13/index.js#L24
+// eslint-disable-next-line no-useless-escape
+const dataURLRegex = /^data:([a-z]+\/[a-z0-9-+.]+(;[a-z0-9-.!#$%*+.{}|~`]+=[a-z0-9-.!#$%*+.{}|~`]+)*)?(;base64)?,([a-z0-9!$&',()*+;=\-._~:@\/?%\s]*?)$/i;
+
+const isValidDataURL = urlString => {
+	return dataURLRegex.test(urlString);
+};
+
+const parseDataURL = urlString => {
+	if (!isValidDataURL(urlString)) {
+		throw new Error(`Invalid URL: ${urlString}`);
+	}
+
+	const parts = urlString.trim().match(dataURLRegex);
+	const parsed = {};
+
+	if (parts[1]) {
+		const mimeType = parts[1].toLowerCase();
+		const [contentType, ...attributes] = mimeType.split(';');
+		// TODO: use `Object.fromEntries`
+		// Object.assign(parsed, Object.fromEntries(attributes.map(attribute => attribute.split('='))))
+		attributes.reduce((parsed, attribute) => {
+			const [key, value] = attribute.split('=');
+			parsed[key] = value;
+			return parsed;
+		}, parsed);
+
+		parsed.mimeType = mimeType;
+		parsed.contentType = contentType;
+	}
+
+	parsed.base64 = Boolean(parts[parts.length - 2]);
+	parsed.body = parts[parts.length - 1] || '';
+
+	return parsed;
+};
+
+const normalizeDataURL = urlString => {
+	const data = parseDataURL(urlString);
+
+	const body = data.base64 ? data.body.trim() : data.body;
+	const {mimeType = ''} = data;
+
+	return `data:${mimeType}${data.base64 ? ';base64' : ''},${body}`;
+};
+
 const normalizeUrl = (urlString, options) => {
 	options = {
 		defaultProtocol: 'http:',
@@ -41,7 +87,7 @@ const normalizeUrl = (urlString, options) => {
 	const isRelativeUrl = !hasRelativeProtocol && /^\.*\//.test(urlString);
 
 	// Prepend protocol
-	if (!isRelativeUrl) {
+	if (!isRelativeUrl && !/^data?:/i.test(urlString)) {
 		urlString = urlString.replace(/^(?!(?:\w+:)?\/\/)|^\/\//, options.defaultProtocol);
 	}
 
@@ -130,6 +176,12 @@ const normalizeUrl = (urlString, options) => {
 		urlObj.searchParams.sort();
 	}
 
+	// DataUrls
+	if (urlObj.protocol === 'data:') {
+		const url = normalizeDataURL(`${urlObj.protocol}${urlObj.pathname}`);
+		return `${url}${urlObj.search}${urlObj.hash}`;
+	}
+
 	if (options.removeTrailingSlash) {
 		urlObj.pathname = urlObj.pathname.replace(/\/$/, '');
 	}

test.js

@@ -204,3 +204,38 @@ test('remove duplicate pathname slashes', t => {
 	t.is(normalizeUrl('http://sindresorhus.com:5000//foo'), 'http://sindresorhus.com:5000/foo');
 	t.is(normalizeUrl('http://sindresorhus.com//foo'), 'http://sindresorhus.com/foo');
 });
+
+test('data URL', t => {
+	t.throws(() => {
+		normalizeUrl('data:text/plain;charset=UTF-8;,foo');
+	}, 'Invalid URL: data:text/plain;charset=UTF-8;,foo');
+
+	// Empty MIME type.
+	t.is(normalizeUrl('data:,'), 'data:,');
+
+	// Lowercase the MIME type.
+	t.is(normalizeUrl('data:TEXT/plain;charset=UTF-8,foo'), 'data:text/plain;charset=utf-8,foo');
+
+	// Remove spaces after the comma when it's base64.
+	t.is(normalizeUrl('data:image/gif;base64, R0lGODlhAQABAAAAACw= ?foo=bar'), 'data:image/gif;base64,R0lGODlhAQABAAAAACw=?foo=bar');
+
+	// Keep spaces when it's not base64.
+	t.is(normalizeUrl('data:text/plain;charset=utf-8, foo ?foo=bar'), 'data:text/plain;charset=utf-8, foo?foo=bar');
+
+	// Data URL with query and hash.
+	t.is(normalizeUrl('data:image/gif;base64,R0lGODlhAQABAAAAACw=?foo=bar#baz'), 'data:image/gif;base64,R0lGODlhAQABAAAAACw=?foo=bar#baz');
+
+	// Options.
+	t.is(normalizeUrl('data:text/plain;charset=utf-8,www.foo/index.html?foo=bar&a=a&utm_medium=test#baz', {
+		defaultProtocol: 'http:',
+		normalizeProtocol: true,
+		forceHttp: true,
+		stripHash: true,
+		stripWWW: true,
+		stripProtocol: true,
+		removeQueryParameters: [/^utm_\w+/i, 'ref'],
+		sortQueryParameters: true,
+		removeTrailingSlash: true,
+		removeDirectoryIndex: true
+	}), 'data:text/plain;charset=utf-8,www.foo/index.html?a=a&foo=bar');
+});