mod/parse_url.php

   1 <?php
   2
   3 require_once('library/HTML5/Parser.php');
   4
   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
  36         if(! $s) {
  37                 echo sprintf($template,$url,$url,'');
  38                 killme();
  39         }
  40
  41         $dom = @HTML5_Parser::parse($s);
  42
  43         if(! $dom)
  44                 return $ret;
  45
  46         $items = $dom->getElementsByTagName('title');
  47
  48         if($items) {
  49                 foreach($items as $item) {
  50                         $title = trim($item->textContent);
  51                         break;
  52                 }
  53         }
  54
  55
  56         $divs = $dom->getElementsByTagName('div');
  57         if($divs) {
  58                 foreach($divs as $div) {
  59                         $class = $div->getAttribute('class');
  60                         if($class && stristr($class,'article')) {
  61                                 $items = $div->getElementsByTagName('p');
  62                                 if($items) {
  63                                         foreach($items as $item) {
  64                                                 if($item->getElementsByTagName('script'))
  65                                                         continue;
  66                                                 $text = $item->textContent;
  67                                                 $text = strip_tags($text);
  68                                                 if(strlen($text) < 100)
  69                                                         continue;
  70                                                 $text = substr($text,0,250) . '...' ;
  71                                                 break;
  72                                         }
  73                                 }
  74                         }
  75                 }
  76         }
  77
  78         if(! $text) {
  79                 $items = $dom->getElementsByTagName('p');
  80                 if($items) {
  81                         foreach($items as $item) {
  82                                 if($item->getElementsByTagName('script'))
  83                                         continue;
  84                                 $text = $item->textContent;
  85                                 $text = strip_tags($text);
  86                                 if(strlen($text) < 100)
  87                                         continue;
  88                                 $text = substr($text,0,250) . '...' ;
  89                                 break;
  90                         }
  91                 }
  92         }
  93
  94         if(strlen($text)) {
  95                 $text = '<br />' . $text;
  96         }
  97
  98         echo sprintf($template,$url,$title,$text);
  99         killme();
 100 }