# HG changeset patch
# User Laman
# Date 2019-06-29 14:48:23
# Node ID 63e17c1c419e60b399652d968b2f1c769428b12c
# Parent  d27d9e2dd943a3f3288c5a65f13d8aef7c79403f

String <--> UTF-8 conversion

diff --git a/util.js b/util.js
--- a/util.js
+++ b/util.js
@@ -32,7 +32,7 @@ function bytes2hex(arr){
 function str2utf8(s){
 	let res=[];
 	let c=s.codePointAt(0);
-	for(let i=0;c!==undefined;i++,c=s.codePointAt(i)){
+	for(let i=0; c!==undefined; i++,c=s.codePointAt(i)){
 		if(c<0x80){res.push(c);}
 		else if(c<0x800){
 			res.push(0b11000000|(c>>>6));
@@ -49,11 +49,44 @@ function str2utf8(s){
 			res.push(0b10000000|((c>>>6)&0b111111));
 			res.push(0b10000000|(c&0b111111));
 		}
+		if(c>0xffff){i++;} // skip surrogate
 	}
 	return res;
 }
 
-/*console.log(str2utf8("$").map(x=>x.toString(16)));
-console.log(str2utf8("¢").map(x=>x.toString(16)));
-console.log(str2utf8("€").map(x=>x.toString(16)));
-console.log(str2utf8("𐍈").map(x=>x.toString(16)));*/
+function utf82str(arr){
+	let res=[];
+	for(let i=0;i<arr.length;i++){
+		let x=arr[i];
+		if(x<=0b1111111){res.push(x);}
+		else if(x<=0b11011111){
+			let a=x&0b11111;
+			let b=arr[++i]&0b111111;
+			res.push(a<<6|b);
+		}
+		else if(x<=0b11101111){
+			let a=x&0b1111;
+			let b=arr[++i]&0b111111;
+			let c=arr[++i]&0b111111;
+			res.push(a<<12|b<<6|c);
+		}
+		else{
+			let a=x&0b111;
+			let b=arr[++i]&0b111111;
+			let c=arr[++i]&0b111111;
+			let d=arr[++i]&0b111111;
+			res.push(a<<18|b<<12|c<<6|d);
+		}
+	}
+	return res.map(x=>String.fromCodePoint(x)).join("");
+}
+
+if(typeof module!=='undefined'&&module.hasOwnProperty('exports')){
+	module.exports.bytes2int32=bytes2int32;
+	module.exports.bytes2int32s=bytes2int32s;
+	module.exports.int322bytes=int322bytes;
+	module.exports.int32s2bytes=int32s2bytes;
+	module.exports.bytes2hex=bytes2hex;
+	module.exports.str2utf8=str2utf8;
+	module.exports.utf82str=utf82str;
+}