我有一些UTF-8编码的数据生活在Javascript Uint8Array元素的范围内。是否有一种有效的方法来解码这些到一个常规的javascript字符串(我相信javascript使用16位Unicode)?我不想一次添加一个字符,因为字符串连接会变得CPU密集。
当前回答
Albert给出的解决方案,只要不经常调用所提供的函数,并且只用于大小适中的数组,就能很好地工作,否则效率非常低。下面是一个增强的普通JavaScript解决方案,它适用于Node和浏览器,具有以下优点:
•有效工作于所有八字节数组大小
•不生成中间丢弃字符串
•在现代JS引擎上支持4字节字符(否则“?”将被替换)
var utf8ArrayToStr = (function () {
var charCache = new Array(128); // Preallocate the cache for the common single byte chars
var charFromCodePt = String.fromCodePoint || String.fromCharCode;
var result = [];
return function (array) {
var codePt, byte1;
var buffLen = array.length;
result.length = 0;
for (var i = 0; i < buffLen;) {
byte1 = array[i++];
if (byte1 <= 0x7F) {
codePt = byte1;
} else if (byte1 <= 0xDF) {
codePt = ((byte1 & 0x1F) << 6) | (array[i++] & 0x3F);
} else if (byte1 <= 0xEF) {
codePt = ((byte1 & 0x0F) << 12) | ((array[i++] & 0x3F) << 6) | (array[i++] & 0x3F);
} else if (String.fromCodePoint) {
codePt = ((byte1 & 0x07) << 18) | ((array[i++] & 0x3F) << 12) | ((array[i++] & 0x3F) << 6) | (array[i++] & 0x3F);
} else {
codePt = 63; // Cannot convert four byte code points, so use "?" instead
i += 3;
}
result.push(charCache[codePt] || (charCache[codePt] = charFromCodePt(codePt)));
}
return result.join('');
};
})();
其他回答
在NodeJS中,我们有缓冲区可用,使用它们进行字符串转换非常容易。更好的是,它很容易将Uint8Array转换为Buffer。试试这段代码,它为我在节点基本上任何转换涉及Uint8Arrays:
let str = Buffer.from(uint8arr.buffer).toString();
我们只是从Uint8Array中提取ArrayBuffer,然后将其转换为适当的NodeJS Buffer。然后我们将Buffer转换为一个字符串(如果你愿意,你可以使用十六进制或base64编码)。
如果我们想从一个字符串转换回Uint8Array,那么我们会这样做:
let uint8arr = new Uint8Array(Buffer.from(str));
注意,如果你在转换为字符串时声明了一个像base64这样的编码,那么如果你使用base64或任何其他你使用的编码,你就必须使用Buffer.from(str, "base64")。
这将在没有模块的浏览器中不起作用!NodeJS的Buffer在浏览器中是不存在的,所以除非你在浏览器中添加Buffer功能,否则这个方法是行不通的。这其实很容易做到,只要使用像这样的模块,它既小又快!
来自Encoding标准的TextEncoder和TextDecoder,由stringencoding库填充,在字符串和ArrayBuffers之间转换:
var uint8array = new TextEncoder().encode("someString");
var string = new TextDecoder().decode(uint8array);
在节点“缓冲区实例也是Uint8Array实例”,所以buf.toString()在这种情况下工作。
我正在使用这个函数,它对我有用:
function uint8ArrayToBase64(data) {
return btoa(Array.from(data).map((c) => String.fromCharCode(c)).join(''));
}
class UTF8{
static encode(str:string){return new UTF8().encode(str)}
static decode(data:Uint8Array){return new UTF8().decode(data)}
private EOF_byte:number = -1;
private EOF_code_point:number = -1;
private encoderError(code_point) {
console.error("UTF8 encoderError",code_point)
}
private decoderError(fatal, opt_code_point?):number {
if (fatal) console.error("UTF8 decoderError",opt_code_point)
return opt_code_point || 0xFFFD;
}
private inRange(a:number, min:number, max:number) {
return min <= a && a <= max;
}
private div(n:number, d:number) {
return Math.floor(n / d);
}
private stringToCodePoints(string:string) {
/** @type {Array.<number>} */
let cps = [];
// Based on http://www.w3.org/TR/WebIDL/#idl-DOMString
let i = 0, n = string.length;
while (i < string.length) {
let c = string.charCodeAt(i);
if (!this.inRange(c, 0xD800, 0xDFFF)) {
cps.push(c);
} else if (this.inRange(c, 0xDC00, 0xDFFF)) {
cps.push(0xFFFD);
} else { // (inRange(c, 0xD800, 0xDBFF))
if (i == n - 1) {
cps.push(0xFFFD);
} else {
let d = string.charCodeAt(i + 1);
if (this.inRange(d, 0xDC00, 0xDFFF)) {
let a = c & 0x3FF;
let b = d & 0x3FF;
i += 1;
cps.push(0x10000 + (a << 10) + b);
} else {
cps.push(0xFFFD);
}
}
}
i += 1;
}
return cps;
}
private encode(str:string):Uint8Array {
let pos:number = 0;
let codePoints = this.stringToCodePoints(str);
let outputBytes = [];
while (codePoints.length > pos) {
let code_point:number = codePoints[pos++];
if (this.inRange(code_point, 0xD800, 0xDFFF)) {
this.encoderError(code_point);
}
else if (this.inRange(code_point, 0x0000, 0x007f)) {
outputBytes.push(code_point);
} else {
let count = 0, offset = 0;
if (this.inRange(code_point, 0x0080, 0x07FF)) {
count = 1;
offset = 0xC0;
} else if (this.inRange(code_point, 0x0800, 0xFFFF)) {
count = 2;
offset = 0xE0;
} else if (this.inRange(code_point, 0x10000, 0x10FFFF)) {
count = 3;
offset = 0xF0;
}
outputBytes.push(this.div(code_point, Math.pow(64, count)) + offset);
while (count > 0) {
let temp = this.div(code_point, Math.pow(64, count - 1));
outputBytes.push(0x80 + (temp % 64));
count -= 1;
}
}
}
return new Uint8Array(outputBytes);
}
private decode(data:Uint8Array):string {
let fatal:boolean = false;
let pos:number = 0;
let result:string = "";
let code_point:number;
let utf8_code_point = 0;
let utf8_bytes_needed = 0;
let utf8_bytes_seen = 0;
let utf8_lower_boundary = 0;
while (data.length > pos) {
let _byte = data[pos++];
if (_byte == this.EOF_byte) {
if (utf8_bytes_needed != 0) {
code_point = this.decoderError(fatal);
} else {
code_point = this.EOF_code_point;
}
} else {
if (utf8_bytes_needed == 0) {
if (this.inRange(_byte, 0x00, 0x7F)) {
code_point = _byte;
} else {
if (this.inRange(_byte, 0xC2, 0xDF)) {
utf8_bytes_needed = 1;
utf8_lower_boundary = 0x80;
utf8_code_point = _byte - 0xC0;
} else if (this.inRange(_byte, 0xE0, 0xEF)) {
utf8_bytes_needed = 2;
utf8_lower_boundary = 0x800;
utf8_code_point = _byte - 0xE0;
} else if (this.inRange(_byte, 0xF0, 0xF4)) {
utf8_bytes_needed = 3;
utf8_lower_boundary = 0x10000;
utf8_code_point = _byte - 0xF0;
} else {
this.decoderError(fatal);
}
utf8_code_point = utf8_code_point * Math.pow(64, utf8_bytes_needed);
code_point = null;
}
} else if (!this.inRange(_byte, 0x80, 0xBF)) {
utf8_code_point = 0;
utf8_bytes_needed = 0;
utf8_bytes_seen = 0;
utf8_lower_boundary = 0;
pos--;
code_point = this.decoderError(fatal, _byte);
} else {
utf8_bytes_seen += 1;
utf8_code_point = utf8_code_point + (_byte - 0x80) * Math.pow(64, utf8_bytes_needed - utf8_bytes_seen);
if (utf8_bytes_seen !== utf8_bytes_needed) {
code_point = null;
} else {
let cp = utf8_code_point;
let lower_boundary = utf8_lower_boundary;
utf8_code_point = 0;
utf8_bytes_needed = 0;
utf8_bytes_seen = 0;
utf8_lower_boundary = 0;
if (this.inRange(cp, lower_boundary, 0x10FFFF) && !this.inRange(cp, 0xD800, 0xDFFF)) {
code_point = cp;
} else {
code_point = this.decoderError(fatal, _byte);
}
}
}
}
//Decode string
if (code_point !== null && code_point !== this.EOF_code_point) {
if (code_point <= 0xFFFF) {
if (code_point > 0)result += String.fromCharCode(code_point);
} else {
code_point -= 0x10000;
result += String.fromCharCode(0xD800 + ((code_point >> 10) & 0x3ff));
result += String.fromCharCode(0xDC00 + (code_point & 0x3ff));
}
}
}
return result;
}
`