| 
<?php /* Created on: 12.07.2007 */ ?>
<html>
 <head>
 <title>HTMLCleaner :: Clean MS Word HTML quickly</title>
 <style type="text/css">
 body,table    {font-family:tahoma,verdana,arial;font-size:9pt}
 </style>
 </head>
 <body>
 <h2>CRIOS<i>WEB</i>'s HTMLCleaner</h2>
 <form enctype="multipart/form-data" method="POST" action="<?$_SERVER['PHP_SELF']?>">
 Upload HTML Document to be cleaned:
 <input type="file" name="doc">
 or paste your code here:
 <br><br>
 <textarea name="html" rows="20" cols="80"></textarea>
 <br><br>
 <input type="submit" value="Process" name="process">
 </form>
 <p>HTML source code cleaner (great help for cleaning MS Word content)</p>
 <p>PHP class: <a href="http://luci.criosweb.ro/blog/2007/08/04/html-cleaner/" target="_blank">http://luci.criosweb.ro/blog/2007/08/04/html-cleaner/</a></p>
 
 <p>Licenced under Creative Commons LGPL (creativecommons.org/licenses/LGPL/2.1/)<br>
 for personal, non-commercial use</p>
 
 <p>For commercial use one developer licence costs 30 EUROs</p>
 <?php
 $word=$_POST['html'];
 if(isset($_POST["process"])){
 
 require("HTMLCleaner.php");
 
 if(!$word){
 $filename=$_FILES["doc"]["tmp_name"];
 
 $fp=fopen($filename,"r");
 $word=fread($fp,filesize($filename));
 fclose($fp);
 unlink($filename);
 }
 
 $cleaner=new HTMLCleaner();
 
 $cleaner->Options['UseTidy']=false;
 $cleaner->Options['OutputXHTML']=false;
 $cleaner->Options['Optimize']=true;
 
 $cleaner->html=$word;
 $cleanHTML=$cleaner->cleanUp('latin1');
 
 echo '<textarea style="width:100%;height:300px">'.htmlspecialchars($cleanHTML,ENT_COMPAT,"ISO-8859-1")."</textarea>";
 echo $cleanHTML;
 }
 ?>
 </body>
 </html>
 
 |