sanitizeUrl.js

/**
 * Sanitize URL — Validate it looks like a URL, then make it less dirty.
 * 
 * Oddnut Software
 * Copyright (c) 2010 Eric Ferraiuolo - http://eric.ferraiuolo.name
 * MIT License - http://www.opensource.org/licenses/mit-license.php
 * 
 * Examples:
 * 
 * 'Http://WWW.example.com/'        » 'http://www.example.com/'
 * 'example.com'                    » 'http://example.com'
 * 'example.c'                      » null
 * 'https://foo:[email protected]'    » 'https://foo:[email protected]'
 * ' WWW.EXAMPLE.COM '              » 'http://www.example.com'
 */

var sanitizeUrl = (function(){

    var URL_REGEXP      = /^(?:(https?:)\/\/)?(?:([^:@\s]+:?[^:@\s]+?)@)?((?:[^;:@=\/\?\.\s]+\.)+[A-Za-z0-9\-]{2,})(?::(\d+))?(?=\/|$)(\/.*)?/i,
        TRIM_REGEXP     = /^\s+|\s+$/g,
        HTTP_PROTOCOL   = 'http:',
        SLASH_SLASH     = '//',
        AT              = '@',
        COLON           = ':',
        EMPTY_STRING    = '';
        
    function trim (s) {
        
        try {
            return s.replace(TRIM_REGEXP, EMPTY_STRING);
        } catch(e) {
            return s;
        }
    }
    
    return function (url) {
        
        url = trim(url);
        
        if (url.length < 1) { return null; }
        
        var urlParts = url.match(URL_REGEXP),
            protocol, authorization, host, port, path;
            
        if ( ! urlParts) { return null; }
        
        protocol        = (urlParts[1] ? urlParts[1].toLowerCase() : HTTP_PROTOCOL) + SLASH_SLASH;
        authorization   = (urlParts[2] ? (urlParts[2] + AT) : EMPTY_STRING);
        host            = (urlParts[3] ? urlParts[3].toLowerCase() : EMPTY_STRING);
        port            = (urlParts[4] ? (COLON + urlParts[4]) : EMPTY_STRING);
        path            = (urlParts[5] || EMPTY_STRING);
        
        return ( protocol + authorization + host + port + path );
    }

}());