Use native decoding of base64 in Node.js (#269)

* Make decoding of base64 data URIs faster

I saw in microsoft/vscode-js-debug#1911 that
base64 decoding of a data URI was taking a bit of time.

This PR feature-detects the presence of a global `Buffer` to use native
decoding when running in Node.js, which is about 25x faster on a 10MB
data URI than the JS implementation in the library.

I have a bit of a hack in order to test both paths when running tests,
happy to change it if desired :)

* use conditional exports

* fix test import

* fix node test import, and set stringToBuffer encoding

* Create few-adults-rhyme.md

---------

Co-authored-by: Nathan Rajlich <[email protected]>

Author: connor4312

.changeset/few-adults-rhyme.md

@@ -0,0 +1,5 @@
+---
+"data-uri-to-buffer": patch
+---
+
+Use native Buffer decoding in Node.js

packages/data-uri-to-buffer/package.json

@@ -4,6 +4,10 @@
   "description": "Create an ArrayBuffer instance from a Data URI string",
   "main": "./dist/index.js",
   "types": "./dist/index.d.ts",
+  "exports": {
+    "node": "./dist/node.js",
+    "default": "./dist/index.js"
+  },
   "files": [
     "dist"
   ],

packages/data-uri-to-buffer/src/common.ts

@@ -0,0 +1,69 @@
+export interface ParsedDataURI {
+	type: string;
+	typeFull: string;
+	charset: string;
+	buffer: ArrayBuffer;
+}
+
+export interface IBufferConversions {
+	base64ToArrayBuffer(base64: string): ArrayBuffer;
+	stringToBuffer(str: string): ArrayBuffer;
+}
+
+/**
+ * Returns a `Buffer` instance from the given data URI `uri`.
+ *
+ * @param {String} uri Data URI to turn into a Buffer instance
+ */
+export const makeDataUriToBuffer = (convert: IBufferConversions) => (uri: string | URL): ParsedDataURI => {
+	uri = String(uri);
+
+	if (!/^data:/i.test(uri)) {
+		throw new TypeError(
+			'`uri` does not appear to be a Data URI (must begin with "data:")'
+		);
+	}
+
+	// strip newlines
+	uri = uri.replace(/\r?\n/g, '');
+
+	// split the URI up into the "metadata" and the "data" portions
+	const firstComma = uri.indexOf(',');
+	if (firstComma === -1 || firstComma <= 4) {
+		throw new TypeError('malformed data: URI');
+	}
+
+	// remove the "data:" scheme and parse the metadata
+	const meta = uri.substring(5, firstComma).split(';');
+
+	let charset = '';
+	let base64 = false;
+	const type = meta[0] || 'text/plain';
+	let typeFull = type;
+	for (let i = 1; i < meta.length; i++) {
+		if (meta[i] === 'base64') {
+			base64 = true;
+		} else if (meta[i]) {
+			typeFull += `;${meta[i]}`;
+			if (meta[i].indexOf('charset=') === 0) {
+				charset = meta[i].substring(8);
+			}
+		}
+	}
+	// defaults to US-ASCII only if type is not provided
+	if (!meta[0] && !charset.length) {
+		typeFull += ';charset=US-ASCII';
+		charset = 'US-ASCII';
+	}
+
+	// get the encoded data portion and decode URI-encoded chars
+	const data = unescape(uri.substring(firstComma + 1));
+	const buffer = base64 ? convert.base64ToArrayBuffer(data) : convert.stringToBuffer(data);
+
+	return {
+		type,
+		typeFull,
+		charset,
+		buffer,
+	};
+}

packages/data-uri-to-buffer/src/index.ts

@@ -1,9 +1,6 @@
-export interface ParsedDataURI {
-	type: string;
-	typeFull: string;
-	charset: string;
-	buffer: ArrayBuffer;
-}
+import { makeDataUriToBuffer } from './common';
+
+export type { ParsedDataURI } from  './common';
 
 function base64ToArrayBuffer(base64: string) {
 	const chars =
@@ -58,55 +55,4 @@ function stringToBuffer(str: string): ArrayBuffer {
  *
  * @param {String} uri Data URI to turn into a Buffer instance
  */
-export function dataUriToBuffer(uri: string | URL): ParsedDataURI {
-	uri = String(uri);
-
-	if (!/^data:/i.test(uri)) {
-		throw new TypeError(
-			'`uri` does not appear to be a Data URI (must begin with "data:")'
-		);
-	}
-
-	// strip newlines
-	uri = uri.replace(/\r?\n/g, '');
-
-	// split the URI up into the "metadata" and the "data" portions
-	const firstComma = uri.indexOf(',');
-	if (firstComma === -1 || firstComma <= 4) {
-		throw new TypeError('malformed data: URI');
-	}
-
-	// remove the "data:" scheme and parse the metadata
-	const meta = uri.substring(5, firstComma).split(';');
-
-	let charset = '';
-	let base64 = false;
-	const type = meta[0] || 'text/plain';
-	let typeFull = type;
-	for (let i = 1; i < meta.length; i++) {
-		if (meta[i] === 'base64') {
-			base64 = true;
-		} else if (meta[i]) {
-			typeFull += `;${meta[i]}`;
-			if (meta[i].indexOf('charset=') === 0) {
-				charset = meta[i].substring(8);
-			}
-		}
-	}
-	// defaults to US-ASCII only if type is not provided
-	if (!meta[0] && !charset.length) {
-		typeFull += ';charset=US-ASCII';
-		charset = 'US-ASCII';
-	}
-
-	// get the encoded data portion and decode URI-encoded chars
-	const data = unescape(uri.substring(firstComma + 1));
-	const buffer = base64 ? base64ToArrayBuffer(data) : stringToBuffer(data);
-
-	return {
-		type,
-		typeFull,
-		charset,
-		buffer,
-	};
-}
+export const dataUriToBuffer = makeDataUriToBuffer({ stringToBuffer, base64ToArrayBuffer });

packages/data-uri-to-buffer/src/node.ts

@@ -0,0 +1,28 @@
+import { makeDataUriToBuffer } from './common';
+
+export type { ParsedDataURI } from  './common';
+
+function nodeBuffertoArrayBuffer(nodeBuf: Buffer) {
+	if (nodeBuf.byteLength === nodeBuf.buffer.byteLength) {
+		return nodeBuf.buffer; // large strings may get their own memory allocation
+	}
+	const buffer = new ArrayBuffer(nodeBuf.byteLength);
+	const view = new Uint8Array(buffer);
+	view.set(nodeBuf);
+	return buffer;
+}
+
+function base64ToArrayBuffer(base64: string) {
+	return nodeBuffertoArrayBuffer(Buffer.from(base64, 'base64'));
+}
+
+function stringToBuffer(str: string): ArrayBuffer {
+	return nodeBuffertoArrayBuffer(Buffer.from(str, 'ascii'));
+}
+
+/**
+ * Returns a `Buffer` instance from the given data URI `uri`.
+ *
+ * @param {String} uri Data URI to turn into a Buffer instance
+ */
+export const dataUriToBuffer = makeDataUriToBuffer({ stringToBuffer, base64ToArrayBuffer });

packages/data-uri-to-buffer/test/data-uri-to-buffer.test.ts

@@ -1,7 +1,11 @@
 import assert from 'assert';
-import { dataUriToBuffer } from '../src';
+import { dataUriToBuffer as baseline } from '../src/index';
+import { dataUriToBuffer as node } from '../src/node';
 
-describe('data-uri-to-buffer', function () {
+describe('node', () => doTest(node));
+describe('baseline', () => doTest(baseline));
+
+function doTest(dataUriToBuffer: typeof baseline) {
 	it('should decode bare-bones Data URIs', function () {
 		const uri = 'data:,Hello%2C%20World!';
 
@@ -187,4 +191,4 @@ describe('data-uri-to-buffer', function () {
 		assert.equal('UTF-8', parsed.charset);
 		assert.equal('abc', Buffer.from(parsed.buffer).toString());
 	});
-});
+}