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 = "<br /><a class=\"bookmark\" href=\"%s\" >%s</a>%s<br />";
  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                         $title = strip_tags(substr($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 //      logger('parse_url: purified: ' . $s, LOGGER_DATA);
  55
  56         $dom = @HTML5_Parser::parse($s);
  57
  58         if(! $dom) {
  59                 echo sprintf($template,$url,$url,'');
  60                 killme();
  61         }
  62
  63         $items = $dom->getElementsByTagName('title');
  64
  65         if($items) {
  66                 foreach($items as $item) {
  67                         $title = trim($item->textContent);
  68                         break;
  69                 }
  70         }
  71
  72         $divs = $dom->getElementsByTagName('div');
  73         if($divs) {
  74                 foreach($divs as $div) {
  75                         $class = $div->getAttribute('class');
  76                         if($class && (stristr($class,'article') || stristr($class,'content'))) {
  77                                 $items = $div->getElementsByTagName('p');
  78                                 if($items) {
  79                                         foreach($items as $item) {
  80                                                 $text = $item->textContent;
  81                                                 if(stristr($text,'<script')) {
  82                                                         $text = '';
  83                                                         continue;
  84                                                 }
  85                                                 $text = strip_tags($text);
  86                                                 if(strlen($text) < 100) {
  87                                                         $text = '';
  88                                                         continue;
  89                                                 }
  90                                                 $text = substr($text,0,250) . '...' ;
  91                                                 break;
  92                                         }
  93                                 }
  94                         }
  95                         if($text)
  96                                 break;
  97                 }
  98         }
  99
 100         if(! $text) {
 101                 $items = $dom->getElementsByTagName('p');
 102                 if($items) {
 103                         foreach($items as $item) {
 104                                 $text = $item->textContent;
 105                                 if(stristr($text,'<script'))
 106                                         continue;
 107                                 $text = strip_tags($text);
 108                                 if(strlen($text) < 100) {
 109                                         $text = '';
 110                                         continue;
 111                                 }
 112                                 $text = substr($text,0,250) . '...' ;
 113                                 break;
 114                         }
 115                 }
 116         }
 117
 118         if(strlen($text)) {
 119                 $text = '<br /><br /><blockquote>' . $text . '</blockquote><br />';
 120         }
 121
 122         $title = str_replace("\n",'',$title);
 123
 124         $result = sprintf($template,$url,($title) ? $title : $url,$text);
 125
 126         logger('parse_url: returns: ' . $result);
 127
 128         echo $result;
 129         killme();
 130 }