Export a Unicode class as replacement for string.

[utf16js.git] / utf16.js

if (typeof define !== 'function') { var define = require('amdefine')(module) }

define(function(require) {

var 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)
            points.push(unit);
        else if (++i == e)
            return null;
        else {
            var next = string.charCodeAt(i);
            if ((next & 0xfc00) != 0xdc00)
                return null;
            points.push(0x10000 | (unit & 0x03ff) << 10 | next & 0x03ff);
        }
    }

    return points;
};

var encode = function(points) {
    var units = [];
    for (var i = 0, e = points.length; i != e; ++i) {
        var point = points[i];
        if (point < 0x10000)
            units.push(point);
        else {
            point -= 0x10000;
            units.push(0xd800 | (0xffc00 & point) >> 10, 0xdc00 | 0x03ff & point);
        }
    } return String.fromCharCode.apply(String, units);
};

var 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;
};

var point2unit = function(string, offsets) {
    return adjust(false, string, offsets);
};

var unit2point = function(string, offsets) {
    return adjust(true, string, offsets);
};

var charAt = function(string, offset) {
    return substring(string, offset, offset + 1);
};

var length = function(string) {
    return unit2point(string, [string.length])[0];
};

var substr = function(string, start, length) {
    var stop;
    if (typeof length !== "undefined")
        stop = start + length;
    return substring(string, start, stop);
};

var substring = function(string, start, stop) {
    var range = point2unit(string, [start, stop]);
    return string.substring(range[0], range[1]);
};

var Unicode = function(string) {
    this.string = string;
    this.length = length(string);
};

Unicode.prototype.charAt = function(index) {
    return new Unicode(charAt(this.string, index));
};

Unicode.prototype.substr = function(start, length) {
    return new Unicode(substr(this.string, start, length));
};

Unicode.prototype.substring = function(start, end) {
    return new Unicode(substring(this.string, start, end));
};

Unicode.prototype.toString = function() {
    return this.string;
};

return {
    decode: decode,
    encode: encode,

    point2unit: point2unit,
    unit2point: unit2point,

    charAt: charAt,
    length: length,
    substr: substr,
    substring: substring,

    Unicode: Unicode,
};

});