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 $arr['text'];
  46                 killme();
  47         }
  48
  49
  50         if($url && $title && $text) {
  51
  52                 $text = '<br /><br /><blockquote>' . $text . '</blockquote><br />';
  53                 $title = str_replace(array("\r","\n"),array('',''),$title);
  54
  55                 $result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
  56
  57                 logger('parse_url (unparsed): returns: ' . $result);
  58
  59                 echo $result;
  60                 killme();
  61         }
  62
  63
  64         if($url) {
  65                 $s = fetch_url($url);
  66         } else {
  67                 echo '';
  68                 killme();
  69         }
  70
  71 //      logger('parse_url: data: ' . $s, LOGGER_DATA);
  72
  73         if(! $s) {
  74                 echo sprintf($template,$url,$url,'') . $str_tags;
  75                 killme();
  76         }
  77
  78         $matches = '';
  79         $c = preg_match('/\<head(.*?)\>(.*?)\<\/head\>/ism',$s,$matches);
  80         if($c) {
  81 //              logger('parse_url: header: ' . $matches[2], LOGGER_DATA);
  82                 try {
  83                         $domhead = HTML5_Parser::parse($matches[2]);
  84                 } catch (DOMException $e) {
  85                         logger('scrape_dfrn: parse error: ' . $e);
  86                 }
  87                 if($domhead)
  88                         logger('parsed header');
  89         }
  90
  91         if(! $title) {
  92                 if(strpos($s,'<title>')) {
  93                         $title = substr($s,strpos($s,'<title>')+7,64);
  94                         if(strpos($title,'<') !== false)
  95                                 $title = strip_tags(substr($title,0,strpos($title,'<')));
  96                 }
  97         }
  98
  99         $config = HTMLPurifier_Config::createDefault();
 100         $config->set('Cache.DefinitionImpl', null);
 101         $purifier = new HTMLPurifier($config);
 102         $s = $purifier->purify($s);
 103
 104 //      logger('purify_output: ' . $s);
 105
 106         try {
 107                 $dom = HTML5_Parser::parse($s);
 108         } catch (DOMException $e) {
 109                 logger('scrape_dfrn: parse error: ' . $e);
 110         }
 111
 112         if(! $dom) {
 113                 echo sprintf($template,$url,$url,'') . $str_tags;
 114                 killme();
 115         }
 116
 117         $items = $dom->getElementsByTagName('title');
 118
 119         if($items) {
 120                 foreach($items as $item) {
 121                         $title = trim($item->textContent);
 122                         break;
 123                 }
 124         }
 125
 126
 127         if(! $text) {
 128                 $divs = $dom->getElementsByTagName('div');
 129                 if($divs) {
 130                         foreach($divs as $div) {
 131                                 $class = $div->getAttribute('class');
 132                                 if($class && (stristr($class,'article') || stristr($class,'content'))) {
 133                                         $items = $div->getElementsByTagName('p');
 134                                         if($items) {
 135                                                 foreach($items as $item) {
 136                                                         $text = $item->textContent;
 137                                                         if(stristr($text,'<script')) {
 138                                                                 $text = '';
 139                                                                 continue;
 140                                                         }
 141                                                         $text = strip_tags($text);
 142                                                         if(strlen($text) < 100) {
 143                                                                 $text = '';
 144                                                                 continue;
 145                                                         }
 146                                                         $text = substr($text,0,250) . '...' ;
 147                                                         break;
 148                                                 }
 149                                         }
 150                                 }
 151                                 if($text)
 152                                         break;
 153                         }
 154                 }
 155
 156                 if(! $text) {
 157                         $items = $dom->getElementsByTagName('p');
 158                         if($items) {
 159                                 foreach($items as $item) {
 160                                         $text = $item->textContent;
 161                                         if(stristr($text,'<script'))
 162                                                 continue;
 163                                         $text = strip_tags($text);
 164                                         if(strlen($text) < 100) {
 165                                                 $text = '';
 166                                                 continue;
 167                                         }
 168                                         $text = substr($text,0,250) . '...' ;
 169                                         break;
 170                                 }
 171                         }
 172                 }
 173         }
 174
 175         if(! $text) {
 176                 logger('parsing meta');
 177                 $items = $domhead->getElementsByTagName('meta');
 178                 if($items) {
 179                         foreach($items as $item) {
 180                                 $property = $item->getAttribute('property');
 181                                 if($property && (stristr($property,':description'))) {
 182
 183                                         $text = $item->getAttribute('content');
 184                                         if(stristr($text,'<script')) {
 185                                                 $text = '';
 186                                                 continue;
 187                                         }
 188                                         $text = strip_tags($text);
 189
 190
 191                                         $text = substr($text,0,250) . '...' ;
 192                                 }
 193                                 if($property && (stristr($property,':image'))) {
 194
 195                                         $image = $item->getAttribute('content');
 196                                         if(stristr($text,'<script')) {
 197                                                 $image = '';
 198                                                 continue;
 199                                         }
 200                                         $image = strip_tags($image);
 201
 202                                         $i = fetch_url($image);
 203                                         if($i) {
 204                                                 require_once('include/Photo.php');
 205                                                 $ph = new Photo($i);
 206                                                 if($ph->is_valid()) {
 207                                                         if($ph->getWidth() > 300 || $ph->getHeight() > 300) {
 208                                                                 $ph->scaleImage(300);
 209                                                                 $new_width = $ph->getWidth();
 210                                                                 $new_height = $ph->getHeight();
 211                                                                 $image = '<br /><br /><img height="' . $new_height . '" width="' . $new_width . '" src="' .$image . '" alt="photo" />';
 212                                                         }
 213                                                         else
 214                                                                 $image = '<br /><br /><img src="' . $image . '" alt="photo" />';
 215                                                 }
 216                                                 else
 217                                                         $image = '';
 218
 219                                         }
 220                                 }
 221                         }
 222                 }
 223         }
 224
 225         if(strlen($text)) {
 226                 $text = '<br /><br /><blockquote>' . $text . '</blockquote><br />';
 227         }
 228
 229         if($image) {
 230                 $text = $image . '<br />' . $text;
 231         }
 232         $title = str_replace(array("\r","\n"),array('',''),$title);
 233
 234         $result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
 235
 236         logger('parse_url: returns: ' . $result);
 237
 238         echo $result;
 239         killme();
 240 }