lib/AkConverters/AkHtmlToText.php

   1 <?php
   2 /* vim: set expandtab tabstop=4 shiftwidth=4 softtabstop=4: */
   3
   4 // +----------------------------------------------------------------------+
   5 // | Akelos Framework - http://www.akelos.org                             |
   6 // +----------------------------------------------------------------------+
   7 // | Copyright (c) 2002-2006, Akelos Media, S.L.  & Bermi Ferrer Martinez |
   8 // | Released under the GNU Lesser General Public License, see LICENSE.txt|
   9 // +----------------------------------------------------------------------+
  10
  11 /**
  12  * @package ActiveSupport
  13  * @subpackage Converters
  14  * @author Bermi Ferrer <bermi a.t akelos c.om>
  15  * @copyright Copyright (c) 2002-2006, Akelos Media, S.L. http://www.akelos.org
  16  * @license GNU Lesser General Public License <http://www.gnu.org/copyleft/lesser.html>
  17  */
  18
  19 class AkHtmlToText
  20 {
  21     function convert()
  22     {
  23         require_once(AK_VENDOR_DIR.DS.'TextParsers'.DS.'html2text.php');
  24         $Converter = new html2text(true, 0, false);
  25         $markdown = str_replace('__AK:AMP__','&', $Converter->load_string(str_replace('&','__AK:AMP__', $this->source)));
  26
  27         require_once(AK_VENDOR_DIR.DS.'TextParsers'.DS.'smartypants.php');
  28         $Smartypants = new SmartyPantsTypographer_Parser();
  29         $markdown = Ak::html_entity_decode(strip_tags($Smartypants->transform($markdown)));
  30
  31         return trim($this->_simplifyMarkdown($markdown));
  32     }
  33
  34
  35
  36     function _simplifyMarkdown($markdown)
  37     {
  38         $markdown = trim($markdown);
  39         if(strstr($markdown,"\n")){
  40             if(preg_match_all('/([ \t]*[#]{1,2})(.*)(\{#[a-z0-9_]+}|$)+/i',$markdown,$matches)){
  41                 foreach ($matches[0] as $k=>$match){
  42                     $simple_markdown = trim($matches[2][$k]);
  43                     if($match[0] == '#'){
  44                         if($simple_markdown[0] == '#'){ // h3, h4, h5
  45                             $markdown = str_replace($match,'##'.$simple_markdown,$markdown);
  46                         }else{
  47                             $separator = strlen($matches[1][$k]) === 1 ? '=' : '-';
  48                             $simple_markdown .= "\n".str_repeat($separator,strlen($simple_markdown));
  49                             $markdown = str_replace($match,$simple_markdown,$markdown);
  50                         }
  51                     }
  52                 }
  53             }
  54         }
  55         return $markdown;
  56     }
  57 }
  58
  59 ?>