187

I got a webSocket comunication, I recieve base64 encoded string, convert it to uint8 and work on it, but now I need to send back, I got the uint8 array, and need to convert it to base64 string, so I can send it. How can I make this conversion?

3

18 Answers 18

161

If your data may contain multi-byte sequences (not a plain ASCII sequence) and your browser has TextDecoder, then you should use that to decode your data (specify the required encoding for the TextDecoder):

var u8 = new Uint8Array([65, 66, 67, 68]);
var decoder = new TextDecoder('utf8');
var b64encoded = btoa(decoder.decode(u8));

If you need to support browsers that do not have TextDecoder (currently just IE and Edge), then the best option is to use a TextDecoder polyfill.

If your data contains plain ASCII (not multibyte Unicode/UTF-8) then there is a simple alternative using String.fromCharCode that should be fairly universally supported:

var ascii = new Uint8Array([65, 66, 67, 68]);
var b64encoded = btoa(String.fromCharCode.apply(null, ascii));

And to decode the base64 string back to a Uint8Array:

var u8_2 = new Uint8Array(atob(b64encoded).split("").map(function(c) {
    return c.charCodeAt(0); }));

If you have very large array buffers then the apply may fail with Maximum call stack size exceeded and you may need to chunk the buffer (based on the one posted by @RohitSengar). Again, note that this is only correct if your buffer only contains non-multibyte ASCII characters:

function Uint8ToString(u8a){
  var CHUNK_SZ = 0x8000;
  var c = [];
  for (var i=0; i < u8a.length; i+=CHUNK_SZ) {
    c.push(String.fromCharCode.apply(null, u8a.subarray(i, i+CHUNK_SZ)));
  }
  return c.join("");
}
// Usage
var u8 = new Uint8Array([65, 66, 67, 68]);
var b64encoded = btoa(Uint8ToString(u8));
Sign up to request clarification or add additional context in comments.

19 Comments

Thanks. All I needed was btoa(String.fromCharCode.apply(null, myArray))
This doesn't work if the byte array is not valid Unicode.
There are no multibyte characters in a base64 string, or in Uint8Array. TextDecoder is absolutely the wrong thing to use here, because if your Uint8Array has bytes in range 128..255, text decoder will erroneously convert them into unicode characters, which will break base64 converter.
Why on earth would you decode the multibyte before going to base64? If it's binary data you shouldn't be converting to a printable string.
I'm still confused as well that this solution is so widely accepted. Perhaps it works well for UTF8-only strings, but one great advantage of base64 is that it works for other data formats such as images, PDFs, or binary blobs. The "accepted" solution must work for both.
|
116

If you are using Node.js then you can use this code to convert Uint8Array to base64

var u8 = new Uint8Array([65, 66, 67, 68]);
var b64 = Buffer.from(u8).toString('base64');

4 Comments

This is a better answer then the hand rolled functions above in terms of performance.
decode: var u8 = new Uint8Array(Buffer.from(b64, 'base64'))
This works 40% slower due to the Buffer.from() part, when you compare it with the manual JS implementation of base64 encoding.
Being an UInt8Array there's a 99% chance that is not a node thing. Still, is way better to create a Buffer from the ArrayBuffer inside the UInt8Array instead of creating a competely new Buffer, which is really slow.
95

Native browser solution (fast!)

To base64-encode a Uint8Array with arbitrary data (not necessarily UTF-8) using native browser functionality:

// note: `buffer` arg can be an ArrayBuffer or a Uint8Array
async function bufferToBase64(buffer) {
  // use a FileReader to generate a base64 data URI:
  const base64url = await new Promise(r => {
    const reader = new FileReader()
    reader.onload = () => r(reader.result)
    reader.readAsDataURL(new Blob([buffer]))
  });
  // remove the `data:...;base64,` part from the start
  return base64url.slice(base64url.indexOf(',') + 1);
}

// example use:
await bufferToBase64(new Uint8Array([1,2,3,100,200]))

Because this is using native browser features, the performance is optimal. It can convert 250 MB per second on my computer (benchmark script), making it about 60x faster than the accepted answer.

12 Comments

FileReader is not available in nodejs natively, but you could use a polyfill. This might mean losing the high performance.
How to decode the result?
if performance is your goal, consider base64url.substring(base64url.indexOf(',')+1)
To answer the question « How to decode the result? »: developer.mozilla.org/en-US/docs/Glossary/Base64
|
50

All solutions already proposed have severe problems. Some solutions fail to work on large arrays, some provide wrong output, some throw an error on btoa call if an intermediate string contains multibyte characters, some consume more memory than needed.

So I implemented a direct conversion function which just works regardless of the input. It converts about 5 million bytes per second on my machine.

https://gist.github.com/enepomnyaschih/72c423f727d395eeaa09697058238727

/*
MIT License
Copyright (c) 2020 Egor Nepomnyaschih
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.
*/

/*
// This constant can also be computed with the following algorithm:
const base64abc = [],
	A = "A".charCodeAt(0),
	a = "a".charCodeAt(0),
	n = "0".charCodeAt(0);
for (let i = 0; i < 26; ++i) {
	base64abc.push(String.fromCharCode(A + i));
}
for (let i = 0; i < 26; ++i) {
	base64abc.push(String.fromCharCode(a + i));
}
for (let i = 0; i < 10; ++i) {
	base64abc.push(String.fromCharCode(n + i));
}
base64abc.push("+");
base64abc.push("/");
*/
const base64abc = [
	"A", "B", "C", "D", "E", "F", "G", "H", "I", "J", "K", "L", "M",
	"N", "O", "P", "Q", "R", "S", "T", "U", "V", "W", "X", "Y", "Z",
	"a", "b", "c", "d", "e", "f", "g", "h", "i", "j", "k", "l", "m",
	"n", "o", "p", "q", "r", "s", "t", "u", "v", "w", "x", "y", "z",
	"0", "1", "2", "3", "4", "5", "6", "7", "8", "9", "+", "/"
];

/*
// This constant can also be computed with the following algorithm:
const l = 256, base64codes = new Uint8Array(l);
for (let i = 0; i < l; ++i) {
	base64codes[i] = 255; // invalid character
}
base64abc.forEach((char, index) => {
	base64codes[char.charCodeAt(0)] = index;
});
base64codes["=".charCodeAt(0)] = 0; // ignored anyway, so we just need to prevent an error
*/
const base64codes = [
	255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255,
	255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255,
	255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 62, 255, 255, 255, 63,
	52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 255, 255, 255, 0, 255, 255,
	255, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14,
	15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 255, 255, 255, 255, 255,
	255, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40,
	41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51
];

function getBase64Code(charCode) {
	if (charCode >= base64codes.length) {
		throw new Error("Unable to parse base64 string.");
	}
	const code = base64codes[charCode];
	if (code === 255) {
		throw new Error("Unable to parse base64 string.");
	}
	return code;
}

export function bytesToBase64(bytes) {
	let result = '', i, l = bytes.length;
	for (i = 2; i < l; i += 3) {
		result += base64abc[bytes[i - 2] >> 2];
		result += base64abc[((bytes[i - 2] & 0x03) << 4) | (bytes[i - 1] >> 4)];
		result += base64abc[((bytes[i - 1] & 0x0F) << 2) | (bytes[i] >> 6)];
		result += base64abc[bytes[i] & 0x3F];
	}
	if (i === l + 1) { // 1 octet yet to write
		result += base64abc[bytes[i - 2] >> 2];
		result += base64abc[(bytes[i - 2] & 0x03) << 4];
		result += "==";
	}
	if (i === l) { // 2 octets yet to write
		result += base64abc[bytes[i - 2] >> 2];
		result += base64abc[((bytes[i - 2] & 0x03) << 4) | (bytes[i - 1] >> 4)];
		result += base64abc[(bytes[i - 1] & 0x0F) << 2];
		result += "=";
	}
	return result;
}

export function base64ToBytes(str) {
	if (str.length % 4 !== 0) {
		throw new Error("Unable to parse base64 string.");
	}
	const index = str.indexOf("=");
	if (index !== -1 && index < str.length - 2) {
		throw new Error("Unable to parse base64 string.");
	}
	let missingOctets = str.endsWith("==") ? 2 : str.endsWith("=") ? 1 : 0,
		n = str.length,
		result = new Uint8Array(3 * (n / 4)),
		buffer;
	for (let i = 0, j = 0; i < n; i += 4, j += 3) {
		buffer =
			getBase64Code(str.charCodeAt(i)) << 18 |
			getBase64Code(str.charCodeAt(i + 1)) << 12 |
			getBase64Code(str.charCodeAt(i + 2)) << 6 |
			getBase64Code(str.charCodeAt(i + 3));
		result[j] = buffer >> 16;
		result[j + 1] = (buffer >> 8) & 0xFF;
		result[j + 2] = buffer & 0xFF;
	}
	return result.subarray(0, result.length - missingOctets);
}

export function base64encode(str, encoder = new TextEncoder()) {
	return bytesToBase64(encoder.encode(str));
}

export function base64decode(str, decoder = new TextDecoder()) {
	return decoder.decode(base64ToBytes(str));
}

6 Comments

Is having base64abc as an array of strings faster than just making it a string? "ABCDEFG..."?
I tried to use that in a Word Web AddIn with Edge and got an error 'TextDecoder' is not defined. Fortunately I needed only needed the bytesToBase64 function and could remove the dependency.
This is beautiful
Brilliant! However, can someone help explain what the purpose of TextDecoder is for here? I too agree with @rominator007, it's not needed for a strict Uint8 -> Base64 conversion in my tests, but I want to make sure I'm not making a mistake by taking it out. Is this just a URL sanitization measure?
I'd say you can chop off the last 6 lines (the last 2 exported functions) for the vast majority of use cases, and thus avoid the TextDecoder/TextEncoder issue entirely. As far as I can tell, base64encode and base64decode are just convenience functions, and the only use I can think of for them is sending ASCII-safe SMTP messages (but why?) or interacting with extremely brittle legacy code that can only tolerate ASCII. Except for very narrow circumstances, and this goes for the currently accepted answer as well, converting from text strings to base64 and back again is a major code smell.
|
44

Very simple solution and test for JavaScript!

ToBase64 = function (u8) {
    return btoa(String.fromCharCode.apply(null, u8));
}

FromBase64 = function (str) {
    return atob(str).split('').map(function (c) { return c.charCodeAt(0); });
}

var u8 = new Uint8Array(256);
for (var i = 0; i < 256; i++)
    u8[i] = i;

var b64 = ToBase64(u8);
console.debug(b64);
console.debug(FromBase64(b64));

6 Comments

it fails on large data (such as images) with RangeError: Maximum call stack size exceeded
It also makes typescript unhappy, but it seems to work.
I am getting the error "InvalidCharacterError: The string contains invalid characters." when attempting to use the function FromBase 64 on the string eJyLjjYy1CEXxeqM6h5Wui1giFzdhngMIEo3BS4fomE qpsWumMB4VPulQ==
I use this code in combination with Pako to compress data sent between Javascript and PHP and vice versa. Avoiding PHP post variables limit while also reducing the data sent between them.
@ChewieTheChorkie, looks like the problem is the space. It probably came from a URL parser replacing + with {space}. This works once you replace it back (..E q.. -> ..E+q..): FromBase64("eJyLjjYy1CEXxeqM6h5Wui1giFzdhngMIEo3BS4fomE+qpsWumMB4VPulQ==")
|
23
function Uint8ToBase64(u8Arr){
  var CHUNK_SIZE = 0x8000; //arbitrary number
  var index = 0;
  var length = u8Arr.length;
  var result = '';
  var slice;
  while (index < length) {
    slice = u8Arr.subarray(index, Math.min(index + CHUNK_SIZE, length)); 
    result += String.fromCharCode.apply(null, slice);
    index += CHUNK_SIZE;
  }
  return btoa(result);
}

You can use this function if you have a very large Uint8Array. This is for Javascript, can be useful in case of FileReader readAsArrayBuffer.

5 Comments

Interestingly, in Chrome I timed this on a 300kb+ buffer and found doing it in chunks like you are to be ever so slightly slower than doing it byte by byte. This surprised me.
@Matt interesting. It's possible that in the meantime, Chrome has now detects this conversion and has a specific optimization for it and chunking the data may reduce its efficiency.
This isn’t safe, is it? If my chunk’s boundary cuts through a multi-byte UTF8 encoded character, then fromCharCode() would not be able to create sensible characters from the bytes on both sides of the boundary, would it?
@Jens String.fromCharCode.apply() methods cannot reproduce UTF-8: UTF-8 characters may vary in length from one byte to four bytes, yet String.fromCharCode.apply() examines a UInt8Array in segments of UInt8, so it erroneously assumes each character to be exactly one byte long and independent of the neighbouring ones. If the characters encoded in the input UInt8Array all happen to be in the ASCII (single-byte) range, it will work by chance, but it cannot reproduce full UTF-8. You need TextDecoder or a similar algorithm for that.
@Jens what multi-byte UTF8 encoded characters in a binary data array? We're not dealing with unicode strings here, but with arbitrary binary data, which should NOT be treated as utf-8 codepoints.
9

Pure JS - no string middlestep (no btoa)

In below solution I omit conversion to string. IDEA is following:

  • join 3 bytes (3 array elements) and you get 24-bits
  • split 24bits to four 6-bit numbers (which take values from 0 to 63)
  • use that numbers as index in base64 alphabet
  • corner case: when input byte array the length is not divided by 3 then add = or == to result

Solution below works on 3-bytes chunks so it is good for large arrays. Similar solution to convert base64 to binary array (without atob) is HERE

function bytesArrToBase64(arr) {
  const abc = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"; // base64 alphabet
  const bin = n => n.toString(2).padStart(8,0); // convert num to 8-bit binary string
  const l = arr.length
  let result = '';

  for(let i=0; i<=(l-1)/3; i++) {
    let c1 = i*3+1>=l; // case when "=" is on end
    let c2 = i*3+2>=l; // case when "=" is on end
    let chunk = bin(arr[3*i]) + bin(c1? 0:arr[3*i+1]) + bin(c2? 0:arr[3*i+2]);
    let r = chunk.match(/.{1,6}/g).map((x,j)=> j==3&&c2 ? '=' :(j==2&&c1 ? '=':abc[+('0b'+x)]));  
    result += r.join('');
  }

  return result;
}


// ----------
// TEST
// ----------

let test = "Alice's Adventure in Wondeland.";
let testBytes = [...test].map(c=> c.charCodeAt(0) );

console.log('test string:', test);
console.log('bytes:', JSON.stringify(testBytes));
console.log('btoa            ', btoa(test));
console.log('bytesArrToBase64', bytesArrToBase64(testBytes));

Caution!

If you want to convert STRING (not bytes array) be aware that btoa in general will fails on utf8 strings like btoa("💩") (one character may be encoded by more than one byte). In this case you must first convert such string to bytes in proper way and then use above solution e.g. :

function bytesArrToBase64(arr) {
  const abc = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"; // base64 alphabet
  const bin = n => n.toString(2).padStart(8,0); // convert num to 8-bit binary string
  const l = arr.length
  let result = '';

  for(let i=0; i<=(l-1)/3; i++) {
    let c1 = i*3+1>=l; // case when "=" is on end
    let c2 = i*3+2>=l; // case when "=" is on end
    let chunk = bin(arr[3*i]) + bin(c1? 0:arr[3*i+1]) + bin(c2? 0:arr[3*i+2]);
    let r = chunk.match(/.{1,6}/g).map((x,j)=> j==3&&c2 ? '=' :(j==2&&c1 ? '=':abc[+('0b'+x)]));  
    result += r.join('');
  }

  return result;
}


// ----------
// TEST
// ----------

let test = "💩";   // base64: 8J+SqQ==
let testBytes = new TextEncoder().encode(test);

console.log('test string      :', test);
console.log('bytes            :', JSON.stringify([...testBytes]));
console.log('bytesArrToBase64 :', bytesArrToBase64(testBytes));


try {
  console.log('test btoa :', btoa(test));
} catch (e) {
  console.error('btoa fails during conversion!', e.message)
}

Snippets tested 2022-08-04 on: chrome 103.0.5060.134 (arm64), safari 15.2, firefox 103.0.1 (64 bit), edge 103.0.1264.77 (arm64), and node-js v12.16.1

2 Comments

I like the compactness but converting to strings representing binary number and then back is much slower than the accepted solution.
This is the only function which worked for me with gz-deflated binary data. Thanks!
6

MDN's docs cover btoa well.

Because you already have binary data, you can convert your Uint8Array into an ASCII string and invoke btoa on that string.

function encodeBase64Bytes(bytes: Uint8Array): string {
  return btoa(
    bytes.reduce((acc, current) => acc + String.fromCharCode(current), "")
  );
}

Complexity with btoa arises when you need to encode arbitrary JS strings, which may occupy more than a single byte, such as "👍". To handle arbitrary JS strings (which are UTF-16), you must first convert the string to a single byte representation. This is not applicable for this use case because you already have binary data.

The linked MDN documentation covers what that conversion looks like for encoding (and the reciprocal steps for decoding).

Comments

5

Native support for arrayBuffer to uint8array conversion has been recently added to the language.

Example usage:

const arr = new Uint8Array([72, 101, 108, 108, 111, 32, 87, 111, 114, 108, 100]);
console.log(arr.toBase64()); // => 'SGVsbG8gV29ybGQ='

See the MDN docs for more details on the available parameters, and for browser support details.

There are polyfils available, such as this polyfill from core-js.

Comments

4

Use the following to convert uint8 array to base64 encoded string

function arrayBufferToBase64(buffer) {
            var binary = '';
            var bytes = [].slice.call(new Uint8Array(buffer));
            bytes.forEach((b) => binary += String.fromCharCode(b));
            return window.btoa(binary);
        };

1 Comment

this works great, also with large u8 arrays or non-utf8 stuff, but it's about 10 times slower than stackoverflow.com/a/66046176/3779853. The slow part seems to be the [].slice.call.
4

Since btoa only works with strings, we can stringify the Uint8Array with String.fromCharCode:

const toBase64 = uInt8Array => btoa(String.fromCharCode(...uInt8Array));

Comments

2

A one liner for the browser:

Uint8Array --> Base64

btoa(String.fromCharCode.apply(null,new Uint8Array([1,2,3,255])))

Base64 --> Uint8Array

new Uint8Array([...atob('AQID/w==')].map(c=>c.charCodeAt()))

Several people were asking how it works, so I thought I'd finally break it down for everyone:

// Uint8Array --> Base64
//
// btoa(String.fromCharCode.apply(null,new Uint8Array([1,2,3,255])))
//
// breaks down to:


input = new Uint8Array([1,2,3,255])
// convert input to string (by applying the function 'fromCharCode' to each element of the input array)
inputString = String.fromCharCode.apply(null, input)
// now we have a string we can use the 'btoa' function to do all the hard work
base64String = btoa(inputString)
// Base64 --> Uint8Array
//
// new Uint8Array([...atob('AQID/w==')].map(c=>c.charCodeAt()))
//
// breaks down to:

// use the built in function to decode the base64
base64String = atob('AQID/w==')
// convert string to array of characters
base64Array = [...base64String]
// convert characters to bytes
bytes = base64Array.map(c=>c.charCodeAt())
// convert array of bytes to an Uint8Array
bytesArray = new Uint8Array(b)

5 Comments

There si a mistake in Base64 --> Uint8Array. c.charCodeAt(0) should be instead c.charCodeAt(c).
Does not work for input [190, 160, 173, 152, 53, 234]
@Marv-CZ I apologise. You are correct! I have updated the code.
@WolfgangKuehn The code has been updated. It works fine now with your input data set.
@Marv-CZ no, c.charCodeAt(0) is correct. You get a single character, say, "6", and get the charcode of its zeroth character, again 6, which gives 0x36. If you use c.charCodeAt(c), it will seem to work because "a".charCodeAt("a") is "a".charCodeAt(0), as "a" is not a number. But with 53 - that is, "5" - as Wolfgang Kuehn noticed, you will get "5".charCodeAt("5"), which is the fifth character of "5", which does not exist, and you will get NaN and break the code. charCodeAt() works because the missing argument is considered 0, but is still not correct.
1

Here's a solution that works for strings longer than 255 elements, because it doesn't use the "splat operator" (Spread syntax):

function uint8ArrayFromBase64(s) {
  // 1. Call atob()
  var b = atob(s);
  // 2. Construct Uint8Array from String
  return Uint8Array.from(b, c => c.charCodeAt(0));
}

function uint8ArrayToBase64(arr) {
  // 1. Preprocess Uint8Array into String
  // (TODO: fix RAM usage from intermediate array creation)
  var b = arr.map(c => String.fromCharCode(c)).join(String());
  // 2. Call btoa()
  return btoa(b);
}
Demo:

<form action="javascript:" onsubmit="(({target:form,submitter:{value:action}})=>{eval(action)(form)})(event)">
<input name="b64" value="AAAAB3NzaC1yc2E=">
<button type="submit" value="({b64,u8a})=>{u8a.value=`[${uint8ArrayFromBase64(b64.value)}]`;}">Convert to Uint8Array</button>
<br />
<input name="u8a" value="">
<button type="submit" value="({u8a,b64})=>{b64.value=(uint8ArrayToBase64(u8a.value.replace(/(?:^\[|\]$)/g, '').split(',')));}">Convert to Base64</button>
</form>

Comments

0

As of July 2025 Firefox and Safari support .toBase64() on Uint8Array.

The toBase64() method of Uint8Array instances returns a base64-encoded string based on the data in this Uint8Array object.

const myBase64String = myUint8Array.toBase64();

To decode from base64 to a Uint8Array use Uint8Array.fromBase64().

The Uint8Array.fromBase64() static method creates a new Uint8Array object from a base64-encoded string.

const myUint8Array = Uint8Array.fromBase64(myBase64String);

// This creates a Uint8Array from a string.
const bytes = new TextEncoder().encode("Hello, unicode world 😋");

// Encode to base64.
const base64 = bytes.toBase64();
console.log(base64);

// To decode from base64. First get a Uint8Array.
const decodedBytes = Uint8Array.fromBase64(base64);

// And now convert it back to a string.
const decoded = new TextDecoder().decode(decodedBytes);
console.log(decoded);

Polyfills are available at https://www.npmjs.com/package/es-arraybuffer-base64

Comments

0

For Node.js

const bytes = new Uint8Array([0, 1, 2, 3]);
const base64 = Buffer.from(bytes.buffer).toString("base64");

For Browsers

Nowadays, Uint8Array.prototype.toBase64() is implemented in modern browsers.

const bytes = new Uint8Array([0, 1, 2, 3]);
const base64 = bytes.toBase64();

If you want to support older browsers, you may use FileReader.readAsDataURL and btoa.

FileReader.readAsDataURL

function toBase64(bytes) {
    return new Promise(r => {
        const reader = new FileReader()
        reader.onload = () => r(reader.result)
        reader.readAsDataURL(new Blob([bytes]))
    }).then(url => url.slice(url.indexOf(",") + 1));
}

const bytes = new Uint8Array([0, 1, 2, 3]);
const base64 = await toBase64(bytes);

btoa

const bytes = new Uint8Array([0, 1, 2, 3]);
const base64 = btoa(bytes.reduce((data, byte) => data + String.fromCharCode(byte), ""));

Please note that String.fromCharCode should be used instead of TextDecoder to support non-printable characters. Also, bytes.reduce should be used instead of String.fromCharCode(...bytes) to support long data.

Write Your Own

Of course you can write your own Base64 encoding function.

function toBase64(bytes) {
    const dict = [
        "A", "B", "C", "D", "E", "F", "G", "H", "I", "J", "K", "L", "M", "N", "O", "P", "Q", "R", "S", "T", "U", "V", "W", "X", "Y", "Z",
        "a", "b", "c", "d", "e", "f", "g", "h", "i", "j", "k", "l", "m", "n", "o", "p", "q", "r", "s", "t", "u", "v", "w", "x", "y", "z",
        "0", "1", "2", "3", "4", "5", "6", "7", "8", "9", "+", "/"
    ];
    let base64 = "";
    let i;
    for (i = 0; i < bytes.length - 3; i += 3) {
        const byte1 = bytes[i];
        const byte2 = bytes[i + 1];
        const byte3 = bytes[i + 2];
        base64 += dict[byte1 >>> 2] + dict[(byte1 & 3) << 4 | byte2 >>> 4] + dict[(byte2 & 15) << 2 | byte3 >>> 6] + dict[byte3 & 63];
    }
    const has2 = i + 1 < bytes.length;
    const byte1 = bytes[i];
    const byte2 = has2 ? bytes[i + 1] : 0;
    base64 += dict[byte1 >>> 2] + dict[(byte1 & 3) << 4 | byte2 >>> 4];
    if (has2) {
        const has3 = i + 2 < bytes.length;
        const byte3 = has3 ? bytes[i + 2] : 0;
        return base64 + dict[(byte2 & 15) << 2 | byte3 >>> 6] + (has3 ? dict[byte3 & 63] : "=");
    }
    return base64 + "==";
}

const bytes = new Uint8Array([0, 1, 2, 3]);
const base64 = toBase64(bytes);

Performance

If you want to know the performance differences between different methods, you can see it here.

enter image description here

Comments

-1

Here is a JS Function to this:

This function is needed because Chrome doesn't accept a base64 encoded string as value for applicationServerKey in pushManager.subscribe yet https://bugs.chromium.org/p/chromium/issues/detail?id=802280

function urlBase64ToUint8Array(base64String) {
  var padding = '='.repeat((4 - base64String.length % 4) % 4);
  var base64 = (base64String + padding)
    .replace(/\-/g, '+')
    .replace(/_/g, '/');

  var rawData = window.atob(base64);
  var outputArray = new Uint8Array(rawData.length);

  for (var i = 0; i < rawData.length; ++i) {
    outputArray[i] = rawData.charCodeAt(i);
  }
  return outputArray;
}

1 Comment

This converts base64 to Uint8Array. But the question asks how to convert Uint8Array to base64
-5

If all you want is a JS implementation of a base64-encoder, so that you can send data back, you can try the btoa function.

b64enc = btoa(uint);

A couple of quick notes on btoa - it's non-standard, so browsers aren't forced to support it. However, most browsers do. The big ones, at least. atob is the opposite conversion.

If you need a different implementation, or you find an edge-case where the browser has no idea what you're talking about, searching for a base64 encoder for JS wouldn't be too hard.

I think there are 3 of them hanging around on my company's website, for some reason...

3 Comments

Thanks, i didnt try that out before.
Couple of notes. btoa and atob are actually part of the HTML5 standardization process and most browsers do support them in mostly the same way already. Secondly, btoa and atob work with strings only. Running btoa on the Uint8Array will first convert the buffer to a string using toString(). This results in the string "[object Uint8Array]". That's probably not what is intended.
@CaioKeto you might want to consider changing your selected answer. This answer is not correct.
-6

npm install google-closure-library --save

require("google-closure-library");
goog.require('goog.crypt.base64');

var result =goog.crypt.base64.encodeByteArray(Uint8Array.of(1,83,27,99,102,66));
console.log(result);

$node index.js would write AVMbY2Y= to the console.

1 Comment

It's funny that a -ve voted answer is accepted rather than a highly +ve one.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.