trimmer.js

'use strict';

/**
 * https://github.com/SalesforceEng/secure-filters
 * 
 * @description
 * Frequently strings are dirty and need to 
 * have some HTML removed. Often times you may
 * want to control the filtering a bit more, for 
 * example removing all elements except anchors. 
 * Maybe you want to remove all tags except 
 * anchors that have [data-keep]. 
 * 
 * This module supports removing HTML with
 * options for finer-grained control
 * 
 */

function TrimHtml() {

	/**
	 * Regexs to find element tags
	 * @type {Object}
	 */
	var regex = {
		// @<[\/\!]*?[^<>]*?>@si -alternative regex
		// tags : /<\/?([a-z]+)([a-z0-9\=\"\'\.\?\:\s\/]+)?\>/g,
		tags : /<\/?([a-z]+)([a-z0-9\=\"\'\.\?\:\s\/]+)?\>/g,
		tagAndContent : function(element) {
			// Original /\<([a-z]+)([a-z0-9\=\"\'\.\?\:\s\/]+)?\>(.*)<\/[a-z]+\>/g
			return new RegExp('\\<' + element + '([a-z0-9\\=\"\'\\.\\?\\:\\s\/]+)?\\>(.*)<\/'+element+'\\>', 'g');
		} 
	};

	/**
	 * Hash map of functions for custom replace rules for an element
	 * @type {Object}
	 */
	var replaceRules = {};

	/**
	 * Tags that we also want to remove the inner content for
	 * @type {Array}
	 */
	var removeInner = [
		'style',
		'script'
	];

	var HTML_CONTROL = /[\x00-\x08\x0B\x0C\x0E-\x1F\x7F-\x9F]/g;

	

	this.escape = function(string) {
		return string.replace(/\&/g, '&amp;')
					 .replace(/\"/g, '&quot;')
					 .replace(/</g, '&lt;')
					 .replace(/\>/g, '&gt;');
	}
	/**
	 * Strip the html tags from a given string
	 * @param  {String} string 
	 * @return {String}       
	 */
	this.trim = function(string) {

		string = string.replace(HTML_CONTROL, ' ');

		/**
		 * Handles the initial string replace
		 * which removes elements but maintains
		 * their content
		 * @param  {String} string 
		 * @param  {String} el     
		 * @return {String}       
		 */
		function replaceTags(string, el) { 

			/**
			 * If the tag does not need to be 
			 * completely removed with it's 
			 * content .. handle here 
			 */
			if (removeInner.indexOf(el) !== -1) {
				return string;
			}

			/**
			 * If there a custom rule for the replacing
			 * the element ..
			 */
			if (replaceRules[el] !== undefined) {
				return replaceRules[el](string);
			}

			return '';
		}

		/**
		 * Replace matches for element and contents
		 * @param  {String} string [description]
		 * @param  {String} el     [description]
		 * @return {String}        [description]
		 */
		function replaceTagsContent(string, el) {
			return '';
		}

		// Remove tags but not content
		var trimmed = string.replace(regex.tags, replaceTags);


		// Remove tags AND content
		for (var idx in removeInner) {
			var _regex;

			_regex  = regex.tagAndContent(removeInner[idx]);
			trimmed = trimmed.replace(_regex, replaceTagsContent);
		}
		
		return trimmed;
	};

	/**
	 * Add an element to the list of elements
	 * that we remove the element and content
	 * @param {String} elementName
	 */
	this.addRemoveInner = function(elementName) {
		removeInner.push(elementName);
	};

	/**
	 * Add a custom replace rule
	 * @param {String} elementName
	 * @param {Function} func     
	 */
	this.addReplaceRule = function(elementName, func) {
		replaceRules[ elementName ] = func;
	};

	return this;
}

module.exports = TrimHtml;