I'm trying to base64 encode a utf8 string containing Thai characters. I'm using the browser's built in btoa function. It works for ascii text, however Thai is causing it to throw a INVALID_CHARACTER_ERR: DOM Exception 5 exception.
Here's a sample that fails (the character that looks like an "n" is Thai)
btoa('aก')
What do I need to do to base64 encode non-ascii strings?
var Base64 = {
encode: function(s) {
return btoa(unescape(encodeURIComponent(s)));
},
decode: function(s) {
return decodeURIComponent(escape(atob(s)));
}
};
Unfortunately btoa/atob aren't specified in any standard, but the implementations in firefox and webkit both fail on multibyte characters so even if they were now specified those builtin functions would not be able to support multibyte characters (as the input and output strings would necessarily change).
It would seem your only option would be to roll your own base64 encode+decode routines
check this workaround http://ecmanaut.blogspot.com/2006/07/encoding-decoding-utf8-in-javascript.html
I know this is old, but I was recently looking for a UTF8-to-Base64 encoder as well. I found a handy little script at http://www.webtoolkit.info/javascript-base64.html, and a performance improved version at http://jsbase64.codeplex.com/.
Here is the script:
var B64 = {
alphabet: 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/=',
lookup: null,
ie: /MSIE /.test(navigator.userAgent),
ieo: /MSIE [67]/.test(navigator.userAgent),
encode: function (s) {
var buffer = B64.toUtf8(s),
position = -1,
len = buffer.length,
nan0, nan1, nan2, enc = [, , , ];
if (B64.ie) {
var result = [];
while (++position < len) {
nan0 = buffer[position];
nan1 = buffer[++position];
enc[0] = nan0 >> 2;
enc[1] = ((nan0 & 3) << 4) | (nan1 >> 4);
if (isNaN(nan1))
enc[2] = enc[3] = 64;
else {
nan2 = buffer[++position];
enc[2] = ((nan1 & 15) << 2) | (nan2 >> 6);
enc[3] = (isNaN(nan2)) ? 64 : nan2 & 63;
}
result.push(B64.alphabet.charAt(enc[0]), B64.alphabet.charAt(enc[1]), B64.alphabet.charAt(enc[2]), B64.alphabet.charAt(enc[3]));
}
return result.join('');
} else {
var result = '';
while (++position < len) {
nan0 = buffer[position];
nan1 = buffer[++position];
enc[0] = nan0 >> 2;
enc[1] = ((nan0 & 3) << 4) | (nan1 >> 4);
if (isNaN(nan1))
enc[2] = enc[3] = 64;
else {
nan2 = buffer[++position];
enc[2] = ((nan1 & 15) << 2) | (nan2 >> 6);
enc[3] = (isNaN(nan2)) ? 64 : nan2 & 63;
}
result += B64.alphabet[enc[0]] + B64.alphabet[enc[1]] + B64.alphabet[enc[2]] + B64.alphabet[enc[3]];
}
return result;
}
},
decode: function (s) {
if (s.length % 4)
throw new Error("InvalidCharacterError: 'B64.decode' failed: The string to be decoded is not correctly encoded.");
var buffer = B64.fromUtf8(s),
position = 0,
len = buffer.length;
if (B64.ieo) {
var result = [];
while (position < len) {
if (buffer[position] < 128)
result.push(String.fromCharCode(buffer[position++]));
else if (buffer[position] > 191 && buffer[position] < 224)
result.push(String.fromCharCode(((buffer[position++] & 31) << 6) | (buffer[position++] & 63)));
else
result.push(String.fromCharCode(((buffer[position++] & 15) << 12) | ((buffer[position++] & 63) << 6) | (buffer[position++] & 63)));
}
return result.join('');
} else {
var result = '';
while (position < len) {
if (buffer[position] < 128)
result += String.fromCharCode(buffer[position++]);
else if (buffer[position] > 191 && buffer[position] < 224)
result += String.fromCharCode(((buffer[position++] & 31) << 6) | (buffer[position++] & 63));
else
result += String.fromCharCode(((buffer[position++] & 15) << 12) | ((buffer[position++] & 63) << 6) | (buffer[position++] & 63));
}
return result;
}
},
toUtf8: function (s) {
var position = -1,
len = s.length,
chr, buffer = [];
if (/^[\x00-\x7f]*$/.test(s)) while (++position < len)
buffer.push(s.charCodeAt(position));
else while (++position < len) {
chr = s.charCodeAt(position);
if (chr < 128)
buffer.push(chr);
else if (chr < 2048)
buffer.push((chr >> 6) | 192, (chr & 63) | 128);
else
buffer.push((chr >> 12) | 224, ((chr >> 6) & 63) | 128, (chr & 63) | 128);
}
return buffer;
},
fromUtf8: function (s) {
var position = -1,
len, buffer = [],
enc = [, , , ];
if (!B64.lookup) {
len = B64.alphabet.length;
B64.lookup = {};
while (++position < len)
B64.lookup[B64.alphabet.charAt(position)] = position;
position = -1;
}
len = s.length;
while (++position < len) {
enc[0] = B64.lookup[s.charAt(position)];
enc[1] = B64.lookup[s.charAt(++position)];
buffer.push((enc[0] << 2) | (enc[1] >> 4));
enc[2] = B64.lookup[s.charAt(++position)];
if (enc[2] == 64)
break;
buffer.push(((enc[1] & 15) << 4) | (enc[2] >> 2));
enc[3] = B64.lookup[s.charAt(++position)];
if (enc[3] == 64)
break;
buffer.push(((enc[2] & 3) << 6) | enc[3]);
}
return buffer;
}
};
Disclaimer: I haven't tested this with Thai characters specifically, but assume it will work.
Sav
