mod/parse_url.php

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