mod/parse_url.php

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