entities/lib/decode.js

var entityMap = require("../maps/entities.json"),
    legacyMap = require("../maps/legacy.json"),
    xmlMap    = require("../maps/xml.json"),
    decodeCodePoint = require("./decode_codepoint.js");

var decodeXMLStrict  = getStrictDecoder(xmlMap),
    decodeHTMLStrict = getStrictDecoder(entityMap);

function getStrictDecoder(map){
	var keys = Object.keys(map).join("|"),
	    replace = getReplacer(map);

	keys += "|#[xX][\\da-fA-F]+|#\\d+";

	var re = new RegExp("&(?:" + keys + ");", "g");

	return function(str){
		return String(str).replace(re, replace);
	};
}

var decodeHTML = (function(){
	var legacy = Object.keys(legacyMap)
		.sort(sorter);

	var keys = Object.keys(entityMap)
		.sort(sorter);

	for(var i = 0, j = 0; i < keys.length; i++){
		if(legacy[j] === keys[i]){
			keys[i] += ";?";
			j++;
		} else {
			keys[i] += ";";
		}
	}

	var re = new RegExp("&(?:" + keys.join("|") + "|#[xX][\\da-fA-F]+;?|#\\d+;?)", "g"),
	    replace = getReplacer(entityMap);

	function replacer(str){
		if(str.substr(-1) !== ";") str += ";";
		return replace(str);
	}

	//TODO consider creating a merged map
	return function(str){
		return String(str).replace(re, replacer);
	};
}());

function sorter(a, b){
	return a < b ? 1 : -1;
}

function getReplacer(map){
	return function replace(str){
		if(str.charAt(1) === "#"){
			if(str.charAt(2) === "X" || str.charAt(2) === "x"){
				return decodeCodePoint(parseInt(str.substr(3), 16));
			}
			return decodeCodePoint(parseInt(str.substr(2), 10));
		}
		return map[str.slice(1, -1)];
	};
}

module.exports = {
	XML: decodeXMLStrict,
	HTML: decodeHTML,
	HTMLStrict: decodeHTMLStrict
};