mod/parse_url.php

   1 <?php
   2
   3 require_once('library/HTML5/Parser.php');
   4 require_once('library/HTMLPurifier.auto.php');
   5
   6 function parse_url_content(&$a) {
   7
   8         logger('parse_url: ' . $_GET['url']);
   9
  10         $url = trim(hex2bin($_GET['url']));
  11
  12         logger('parse_url: ' . $url);
  13
  14         $text = null;
  15
  16         $template = "<a href=\"%s\" >%s</a>\n%s";
  17
  18
  19         $arr = array('url' => $url, 'text' => '');
  20
  21         call_hooks('parse_link', $arr);
  22
  23         if(strlen($arr['text'])) {
  24                 echo $arr['text'];
  25                 killme();
  26         }
  27
  28         if($url) {
  29                 $s = fetch_url($url);
  30         } else {
  31                 echo '';
  32                 killme();
  33         }
  34
  35         logger('parse_url: data: ' . $s, LOGGER_DATA);
  36
  37         if(! $s) {
  38                 echo sprintf($template,$url,$url,'');
  39                 killme();
  40         }
  41
  42         if(strpos($s,'<title>')) {
  43                 $title = substr($s,strpos($s,'<title>')+7,64);
  44                 if(strpos($title,'<') !== false)
  45                         $title = substr($title,0,strpos($title,'<'));
  46         }
  47
  48         $config = HTMLPurifier_Config::createDefault();
  49         $config->set('Cache.DefinitionImpl', null);
  50
  51         $purifier = new HTMLPurifier($config);
  52         $s = $purifier->purify($s);
  53
  54         $dom = @HTML5_Parser::parse($s);
  55
  56         if(! $dom) {
  57                 echo sprintf($template,$url,$url,'');
  58                 killme();
  59         }
  60
  61         $items = $dom->getElementsByTagName('title');
  62
  63         if($items) {
  64                 foreach($items as $item) {
  65                         $title = trim($item->textContent);
  66                         break;
  67                 }
  68         }
  69
  70         $divs = $dom->getElementsByTagName('div');
  71         if($divs) {
  72                 foreach($divs as $div) {
  73                         $class = $div->getAttribute('class');
  74                         if($class && stristr($class,'article')) {
  75                                 $items = $div->getElementsByTagName('p');
  76                                 if($items) {
  77                                         foreach($items as $item) {
  78                                                 if($item->getElementsByTagName('script'))
  79                                                         continue;
  80                                                 $text = $item->textContent;
  81                                                 $text = strip_tags($text);
  82                                                 if(strlen($text) < 100)
  83                                                         continue;
  84                                                 $text = substr($text,0,250) . '...' ;
  85                                                 break;
  86                                         }
  87                                 }
  88                         }
  89                 }
  90         }
  91
  92         if(! $text) {
  93                 $items = $dom->getElementsByTagName('p');
  94                 if($items) {
  95                         foreach($items as $item) {
  96                                 if($item->getElementsByTagName('script'))
  97                                         continue;
  98                                 $text = $item->textContent;
  99                                 $text = strip_tags($text);
 100                                 if(strlen($text) < 100)
 101                                         continue;
 102                                 $text = substr($text,0,250) . '...' ;
 103                                 break;
 104                         }
 105                 }
 106         }
 107
 108         if(strlen($text)) {
 109                 $text = '<br />' . $text;
 110         }
 111
 112         echo sprintf($template,$url,($title) ? $title : $url,$text);
 113         killme();
 114 }