mod/parse_url.php

   1 <?php
   2
   3 require_once('library/HTML5/Parser.php');
   4 require_once('library/HTMLPurifier.auto.php');
   5
   6 function arr_add_hashes(&$item,$k) {
   7         $item = '#' . $item;
   8 }
   9
  10 function parse_url_content(&$a) {
  11
  12         $text = null;
  13         $str_tags = '';
  14
  15         if(x($_GET,'binurl'))
  16                 $url = trim(hex2bin($_GET['binurl']));
  17         else
  18                 $url = trim($_GET['url']);
  19
  20         if($_GET['title'])
  21                 $title = strip_tags(trim($_GET['title']));
  22
  23         if($_GET['description'])
  24                 $text = strip_tags(trim($_GET['description']));
  25
  26         if($_GET['tags']) {
  27                 $arr_tags = str_getcsv($_GET['tags']);
  28                 if(count($arr_tags)) {
  29                         array_walk($arr_tags,'arr_add_hashes');
  30                         $str_tags = '<br />' . implode(' ',$arr_tags) . '<br />';
  31                 }
  32         }
  33
  34         logger('parse_url: ' . $url);
  35
  36
  37         $template = "<br /><a class=\"bookmark\" href=\"%s\" >%s</a>%s<br />";
  38
  39
  40         $arr = array('url' => $url, 'text' => '');
  41
  42         call_hooks('parse_link', $arr);
  43
  44         if(strlen($arr['text'])) {
  45                 echo $arr['text'];
  46                 killme();
  47         }
  48
  49         if($url && $title && $text) {
  50
  51                 $text = '<br /><br /><blockquote>' . $text . '</blockquote><br />';
  52                 $title = str_replace(array("\r","\n"),array('',''),$title);
  53
  54                 $result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
  55
  56                 logger('parse_url (unparsed): returns: ' . $result);
  57
  58                 echo $result;
  59                 killme();
  60         }
  61
  62
  63         if($url) {
  64                 $s = fetch_url($url);
  65         } else {
  66                 echo '';
  67                 killme();
  68         }
  69
  70         logger('parse_url: data: ' . $s, LOGGER_DATA);
  71
  72         if(! $s) {
  73                 echo sprintf($template,$url,$url,'') . $str_tags;
  74                 killme();
  75         }
  76
  77         if(! $title) {
  78                 if(strpos($s,'<title>')) {
  79                         $title = substr($s,strpos($s,'<title>')+7,64);
  80                         if(strpos($title,'<') !== false)
  81                                 $title = strip_tags(substr($title,0,strpos($title,'<')));
  82                 }
  83         }
  84
  85         $config = HTMLPurifier_Config::createDefault();
  86         $config->set('Cache.DefinitionImpl', null);
  87
  88         $purifier = new HTMLPurifier($config);
  89         $s = $purifier->purify($s);
  90
  91         try {
  92                 $dom = HTML5_Parser::parse($s);
  93         } catch (DOMException $e) {
  94                 logger('scrape_dfrn: parse error: ' . $e);
  95         }
  96
  97         if(! $dom) {
  98                 echo sprintf($template,$url,$url,'') . $str_tags;
  99                 killme();
 100         }
 101
 102         $items = $dom->getElementsByTagName('title');
 103
 104         if($items) {
 105                 foreach($items as $item) {
 106                         $title = trim($item->textContent);
 107                         break;
 108                 }
 109         }
 110
 111
 112         if(! $text) {
 113                 $divs = $dom->getElementsByTagName('div');
 114                 if($divs) {
 115                         foreach($divs as $div) {
 116                                 $class = $div->getAttribute('class');
 117                                 if($class && (stristr($class,'article') || stristr($class,'content'))) {
 118                                         $items = $div->getElementsByTagName('p');
 119                                         if($items) {
 120                                                 foreach($items as $item) {
 121                                                         $text = $item->textContent;
 122                                                         if(stristr($text,'<script')) {
 123                                                                 $text = '';
 124                                                                 continue;
 125                                                         }
 126                                                         $text = strip_tags($text);
 127                                                         if(strlen($text) < 100) {
 128                                                                 $text = '';
 129                                                                 continue;
 130                                                         }
 131                                                         $text = substr($text,0,250) . '...' ;
 132                                                         break;
 133                                                 }
 134                                         }
 135                                 }
 136                                 if($text)
 137                                         break;
 138                         }
 139                 }
 140
 141                 if(! $text) {
 142                         $items = $dom->getElementsByTagName('p');
 143                         if($items) {
 144                                 foreach($items as $item) {
 145                                         $text = $item->textContent;
 146                                         if(stristr($text,'<script'))
 147                                                 continue;
 148                                         $text = strip_tags($text);
 149                                         if(strlen($text) < 100) {
 150                                                 $text = '';
 151                                                 continue;
 152                                         }
 153                                         $text = substr($text,0,250) . '...' ;
 154                                         break;
 155                                 }
 156                         }
 157                 }
 158         }
 159
 160         if(strlen($text)) {
 161                 $text = '<br /><br /><blockquote>' . $text . '</blockquote><br />';
 162         }
 163
 164         $title = str_replace(array("\r","\n"),array('',''),$title);
 165
 166         $result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
 167
 168         logger('parse_url: returns: ' . $result);
 169
 170         echo $result;
 171         killme();
 172 }