textscan.js

;
(function () {

    /**
     * The TextScan object constructor.
     *
     * @constructor
     */
    function TextScan() {

        /**
         * The string to replaces found matches with. Defaults to ***
         *
         * @type {String}
         */
        var replacementString = '***';

        /**
         * The color used for highlighting
         *
         * @type {string}
         */
        var highlightColor = 'F2B8B8';

        /**
         * Holds the currently matched text.
         *
         * @type {{replace: string, hasMatches: boolean}}
         */
        var currentMatch = {
            replace : '',
            hasMatches : false
        };

        /**
         * The available patterns. These are as follows:
         *  [name]              [description]
         *  - long_number       ; Matches long, consecutive numbers
         *  - phone_number      ; Matches phone numbers.
         *  - email_address     ; Matches email addresses in many formats.
         *  - url               ; Matches URL patterns/
         *  - words             ; Finds words, even when in disguise.
         *
         * @type {{long_number: {pattern: RegExp, enabled: boolean}, phone_number: {pattern: RegExp, enabled: boolean}, email_address: {pattern: RegExp, enabled: boolean}, url: {pattern: RegExp, enabled: boolean}, words: {enabled: boolean, pattern: Array}}}
         */
        var patterns = {
            long_number : {
                pattern : /\d{8,}/,
                enabled : true
            },
            phone_number : {
                pattern : /([+-]?[\d]{1,}[\d\s]+|\([\d]+\))[-\d.\s]{8,}/gi,
                enabled : true
            },
            email_address : {
                pattern : /[\w._%+-]+(@|\[at\]|\(at\))[\w.-]+(\.|\[dot\]|\(dot\)|\(punt\)|\[punt\])[a-zA-Z]{2,4}/gi,
                enabled : true
            },
            url : {
                pattern : /((https?:\/{1,2})?[-\w.]{2,}(\.|\[dot\]|\(dot\)|\(punt\)|\[punt\])([a-zA-Z]{2}\.[a-zA-Z]{2,3}|[a-zA-Z]{2,4}).*?(?=$|[^\w\/-]))/gi,
                enabled : true
            },
            words : {
                pattern : [],
                enabled : true
            }
        };

        /**
         * A mapping that maps regular characters to 1337 characters.
         *
         * @type {{o: string, g: string, b: Array, t: string, s: string, a: string, e: string, z: string, i: string, l: string}}
         */
        var map1337 = {
            o : '0',
            g : '9',
            b : ['8', '6'],
            t : '7',
            s : '5',
            a : '4',
            e : '3',
            z : '2',
            i : '1',
            l : '1'
        };

        /**
         * Check if given subject is an array.
         *
         * @param   subject
         * @returns {boolean}
         */
        var isArray = function (subject) {
            return Object.prototype.toString.call(subject) === '[object Array]';
        };

        /**
         * Enable a filter by name.
         *
         * @param   {String}    filter
         * @returns {TextScan}
         */
        this.enableFilter = function (filter) {
            if (typeof patterns[filter] === 'undefined') {
                throw 'Invalid filter supplied.';
            }

            patterns[filter].enabled = true;

            return this;
        };

        /**
         * Add multiple filterWords.
         *
         * @param   {{}}        words
         * @returns {TextScan}
         */
        this.addFilterWords = function (words) {
            if (!isArray(words)) {
                throw 'Invalid type supplied for addFilterWords. Expected array.';
            }

            for (var i = 0; i < words.length; i++) {
                this.addFilterWord(words[i]);
            }

            return this;
        };

        /**
         * Add a word to filter out.
         *
         * @param   {String}    word
         * @returns {TextScan}
         */
        this.addFilterWord = function (word) {
            var pattern = '',
                any = '[^a-z0-9]?',
                last = false,
                character;

            for (var i = 0; i < word.length; i++) {
                last = i === (word.length - 1);
                character = word.charAt(i);

                if (typeof map1337[character] === 'undefined') {
                    pattern += (character + (!last ? any : ''));

                    continue;
                }

                if (typeof map1337[character] === 'string') {
                    pattern += ('((' + character + '|' + map1337[character] + ')' + (!last ? any : '') + ')');

                    continue;
                }

                pattern += '((' + character;

                for (var m = 0; m < map1337[character].length; m++) {
                    pattern += '|' + map1337[character][m];
                }

                pattern += ')' + (!last ? any : '') + ')';
            }

            patterns.words.pattern.push(new RegExp(pattern, 'ig'));

            return this;
        };

        /**
         * Disable a filter by name.
         *
         * @param   {String}    filter
         * @returns {TextScan}
         */
        this.disableFilter = function (filter) {
            if (typeof patterns[filter] === 'undefined') {
                throw 'Invalid filter supplied.';
            }

            patterns[filter].enabled = false;

            return this;
        };

        /**
         * Enable multiple filters at once.
         *
         * @param   {Array}     filters
         * @returns {TextScan}
         * @see     TextScan.enableFilter
         */
        this.enableFilters = function (filters) {
            if (!isArray(filters)) {
                throw 'Invalid filters type supplied. Expected Array.';
            }

            for (var i = 0; i < filters.length; i++) {
                this.enableFilter(filters[i]);
            }

            return this;
        };

        /**
         * Set the string to replace matches in the filterString() method.
         *
         * @param   {String}    str
         * @returns {TextScan}
         */
        this.setReplacementString = function (str) {
            if (typeof str !== 'string') {
                throw 'Invalid replacementString type supplied. Expected string.';
            }

            replacementString = str;

            return this;
        };

        /**
         * @returns {String}
         */
        this.getReplacementString = function () {
            return replacementString;
        };

        /**
         * Returns if text matched.
         *
         * @returns {Boolean}
         */
        this.test = function () {
            return currentMatch.hasMatches;
        };

        /**
         * Prepare some text to be matched against.
         *
         * @param   {String}    str
         * @param   {Boolean}   highlight
         * @returns {TextScan}
         */
        this.prepare = function (str, highlight) {
            currentMatch.replace = this.filterString(str, highlight);
            currentMatch.hasMatches = str !== currentMatch.replace;

            return this;
        };

        /**
         * Get the filtered text.
         *
         * @returns {TextScan}
         */
        this.replace = function () {
            return currentMatch.replace;
        };

        /**
         * Filter the string.
         *
         * @param   {String}    str
         * @param   {Boolean}   highlight
         * @returns {String}}
         */
        this.filterString = function (str, highlight) {
            highlight = highlight || false;

            var replace = function (str, pattern) {
                if (!highlight) {
                    return str.replace(pattern, replacementString);
                }

                return str.replace(pattern, function (match) {
                    return '<span style="background: #' + highlightColor + ';">' + match + '</span>';
                });
            };

            if (typeof str !== 'string') {
                throw 'Invalid "str" type supplied in filterString. Expected string.';
            }

            for (var p in patterns) {
                if (!patterns[p].enabled) {
                    continue;
                }

                if (patterns[p].pattern instanceof RegExp) {
                    str = replace(str, patterns[p].pattern);

                    continue;
                }

                if (!isArray(patterns[p].pattern)) {
                    throw 'Invalid pattern type supplied. Expected Array.';
                }

                for (var i = 0; i < patterns[p].pattern.length; i++) {
                    if (!patterns[p].pattern[i] instanceof RegExp) {
                        throw 'Expected valid RegExp.';
                    }

                    str = replace(str, patterns[p].pattern[i]);
                }
            }

            return str;
        };
    }

    /*
     * Make sure TextScan is loadable through amd.
     */
    if (typeof define === "function" && define.amd) {
        define([], function () {
            return TextScan;
        });

        return;
    }

    /*
     * Make sure TextScan is accessible beyond our scope.
     */
    window.TextScan = TextScan;
})();