Export Unicode class as the module for easier use.

[utf16js.git] / utf16.js

diff --git a/utf16.js b/utf16.js
index 89322ec7fc86ea264063ec9761beebb4f75e67cb..b76fc714864106a53a8b96fc1e7ec75c281a14a6 100644 (file)
--- a/utf16.js
+++ b/utf16.js
@@ -2,31 +2,39 @@ if (typeof define !== 'function') { var define = require('amdefine')(module) }
 
 define(function(require) {
 
-var decode = function(array, string) {
+var utf16 = function(string) {
+    this.string = string;
+    this.length = utf16.strlen(string);
+};
+
+utf16.decode = function(string, points) {
+    if (typeof points === "undefined")
+        points = [];
+
     for (var i = 0, e = string.length; i != e; ++i) {
         var unit = string.charCodeAt(i);
         var part = unit & 0xfc00;
         if (part == 0xdc00)
             return null;
         else if (part != 0xd800)
-            array.push(unit);
+            points.push(unit);
         else if (++i == e)
             return null;
         else {
             var next = string.charCodeAt(i);
             if ((next & 0xfc00) != 0xdc00)
                 return null;
-            array.push(0x10000 | (unit & 0x03ff) << 10 | next & 0x03ff);
+            points.push(0x10000 | (unit & 0x03ff) << 10 | next & 0x03ff);
         }
     }
 
-    return array;
+    return points;
 };
 
-var encode = function(array) {
+utf16.encode = function(points) {
     var units = [];
-    for (var i = 0, e = array.length; i != e; ++i) {
-        var point = array[i];
+    for (var i = 0, e = points.length; i != e; ++i) {
+        var point = points[i];
         if (point < 0x10000)
             units.push(point);
         else {
@@ -36,9 +44,104 @@ var encode = function(array) {
     } return String.fromCharCode.apply(String, units);
 };
 
-return {
-    decode: decode,
-    encode: encode,
+utf16.fromCharCode = function() {
+    return new utf16(utf16.encode(arguments), arguments.length);
+};
+
+utf16.adjust = function(mode, string, after) {
+    var before = after.slice();
+    var offset = 0, index = 0;
+
+    var check = function() {
+        var from = mode ? index : offset;
+        var to = mode ? offset : index;
+
+        for (var o = 0, l = after.length; o != l; ++o)
+            if (before[o] == from) {
+                before[o] = null;
+                after[o] = to;
+            }
+    };
+
+    for (var e = string.length; index != e; ++index) {
+        check();
+
+        var unit = string.charCodeAt(index);
+        var part = unit & 0xfc00;
+
+        if (part == 0xdc00)
+            return null;
+        else if (part != 0xd800);
+        else if (++index == e)
+            return null;
+        else {
+            var next = string.charCodeAt(index);
+            if ((next & 0xfc00) != 0xdc00)
+                return null;
+        }
+
+        ++offset;
+    }
+
+    check();
+    return after;
+};
+
+utf16.point2unit = function(string, offsets) {
+    return utf16.adjust(false, string, offsets);
 };
 
+utf16.unit2point = function(string, offsets) {
+    return utf16.adjust(true, string, offsets);
+};
+
+utf16.charAt = function(string, offset) {
+    return utf16.substring(string, offset, offset + 1);
+};
+
+utf16.charCodeAt = function(string, offset) {
+    var char = utf16.charAt(string, offset);
+    if (char.length != 2)
+        return char.charCodeAt(0);
+    return 0x10000 | (char.charCodeAt(0) & 0x03ff) << 10 | char.charCodeAt(1) & 0x03ff
+};
+
+utf16.strlen = function(string) {
+    return utf16.unit2point(string, [string.length])[0];
+};
+
+utf16.substr = function(string, start, length) {
+    var stop;
+    if (typeof length !== "undefined")
+        stop = start + length;
+    return utf16.substring(string, start, stop);
+};
+
+utf16.substring = function(string, start, stop) {
+    var range = utf16.point2unit(string, [start, stop]);
+    return string.substring(range[0], range[1]);
+};
+
+utf16.prototype.charAt = function(index) {
+    return new utf16(utf16.charAt(this.string, index));
+};
+
+utf16.prototype.charCodeAt = function(index) {
+    return utf16.charCodeAt(this.string, index);
+};
+
+utf16.prototype.substr = function(start, length) {
+    return new utf16(utf16.substr(this.string, start, length));
+};
+
+utf16.prototype.substring = function(start, end) {
+    return new utf16(utf16.substring(this.string, start, end));
+};
+
+utf16.prototype.toString = function() {
+    return this.string;
+};
+
+return utf16;
+
 });