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         $textmode = false;
  16         if(local_user() && intval(get_pconfig(local_user(),'system','plaintext')))
  17                 $textmode = true;
  18
  19         if($textmode)
  20         $br = (($textmode) ? "\n" : '<br /?');
  21
  22         if(x($_GET,'binurl'))
  23                 $url = trim(hex2bin($_GET['binurl']));
  24         else
  25                 $url = trim($_GET['url']);
  26
  27         if($_GET['title'])
  28                 $title = strip_tags(trim($_GET['title']));
  29
  30         if($_GET['description'])
  31                 $text = strip_tags(trim($_GET['description']));
  32
  33         if($_GET['tags']) {
  34                 $arr_tags = str_getcsv($_GET['tags']);
  35                 if(count($arr_tags)) {
  36                         array_walk($arr_tags,'arr_add_hashes');
  37                         $str_tags = $br . implode(' ',$arr_tags) . $br;
  38                 }
  39         }
  40
  41         logger('parse_url: ' . $url);
  42
  43
  44         if($textmode)
  45                 $template = $br . '[bookmark=%s]%s[/bookmark]%s' . $br;
  46         else
  47                 $template = "<br /><a class=\"bookmark\" href=\"%s\" >%s</a>%s<br />";
  48
  49
  50         $arr = array('url' => $url, 'text' => '');
  51
  52         call_hooks('parse_link', $arr);
  53
  54         if(strlen($arr['text'])) {
  55                 echo $arr['text'];
  56                 killme();
  57         }
  58
  59
  60         if($url && $title && $text) {
  61
  62                 if($textmode)
  63                         $text = $br . $br . '[quote]' . $text . '[/quote]' . $br;
  64                 else
  65                         $text = '<br /><br /><blockquote>' . $text . '</blockquote><br />';
  66
  67                 $title = str_replace(array("\r","\n"),array('',''),$title);
  68
  69                 $result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
  70
  71                 logger('parse_url (unparsed): returns: ' . $result);
  72
  73                 echo $result;
  74                 killme();
  75         }
  76
  77
  78         if($url) {
  79                 $s = fetch_url($url);
  80         } else {
  81                 echo '';
  82                 killme();
  83         }
  84
  85 //      logger('parse_url: data: ' . $s, LOGGER_DATA);
  86
  87         if(! $s) {
  88                 echo sprintf($template,$url,$url,'') . $str_tags;
  89                 killme();
  90         }
  91
  92         $matches = '';
  93         $c = preg_match('/\<head(.*?)\>(.*?)\<\/head\>/ism',$s,$matches);
  94         if($c) {
  95 //              logger('parse_url: header: ' . $matches[2], LOGGER_DATA);
  96                 try {
  97                         $domhead = HTML5_Parser::parse($matches[2]);
  98                 } catch (DOMException $e) {
  99                         logger('scrape_dfrn: parse error: ' . $e);
 100                 }
 101                 if($domhead)
 102                         logger('parsed header');
 103         }
 104
 105         if(! $title) {
 106                 if(strpos($s,'<title>')) {
 107                         $title = substr($s,strpos($s,'<title>')+7,64);
 108                         if(strpos($title,'<') !== false)
 109                                 $title = strip_tags(substr($title,0,strpos($title,'<')));
 110                 }
 111         }
 112
 113         $config = HTMLPurifier_Config::createDefault();
 114         $config->set('Cache.DefinitionImpl', null);
 115         $purifier = new HTMLPurifier($config);
 116         $s = $purifier->purify($s);
 117
 118 //      logger('purify_output: ' . $s);
 119
 120         try {
 121                 $dom = HTML5_Parser::parse($s);
 122         } catch (DOMException $e) {
 123                 logger('scrape_dfrn: parse error: ' . $e);
 124         }
 125
 126         if(! $dom) {
 127                 echo sprintf($template,$url,$url,'') . $str_tags;
 128                 killme();
 129         }
 130
 131         $items = $dom->getElementsByTagName('title');
 132
 133         if($items) {
 134                 foreach($items as $item) {
 135                         $title = trim($item->textContent);
 136                         break;
 137                 }
 138         }
 139
 140
 141         if(! $text) {
 142                 $divs = $dom->getElementsByTagName('div');
 143                 if($divs) {
 144                         foreach($divs as $div) {
 145                                 $class = $div->getAttribute('class');
 146                                 if($class && (stristr($class,'article') || stristr($class,'content'))) {
 147                                         $items = $div->getElementsByTagName('p');
 148                                         if($items) {
 149                                                 foreach($items as $item) {
 150                                                         $text = $item->textContent;
 151                                                         if(stristr($text,'<script')) {
 152                                                                 $text = '';
 153                                                                 continue;
 154                                                         }
 155                                                         $text = strip_tags($text);
 156                                                         if(strlen($text) < 100) {
 157                                                                 $text = '';
 158                                                                 continue;
 159                                                         }
 160                                                         $text = substr($text,0,250) . '...' ;
 161                                                         break;
 162                                                 }
 163                                         }
 164                                 }
 165                                 if($text)
 166                                         break;
 167                         }
 168                 }
 169
 170                 if(! $text) {
 171                         $items = $dom->getElementsByTagName('p');
 172                         if($items) {
 173                                 foreach($items as $item) {
 174                                         $text = $item->textContent;
 175                                         if(stristr($text,'<script'))
 176                                                 continue;
 177                                         $text = strip_tags($text);
 178                                         if(strlen($text) < 100) {
 179                                                 $text = '';
 180                                                 continue;
 181                                         }
 182                                         $text = substr($text,0,250) . '...' ;
 183                                         break;
 184                                 }
 185                         }
 186                 }
 187         }
 188
 189         if(! $text) {
 190                 logger('parsing meta');
 191                 $items = (isset($domhead) && is_object($domhead) ? $domhead->getElementsByTagName('meta') : null);
 192                 if($items) {
 193                         foreach($items as $item) {
 194                                 $property = $item->getAttribute('property');
 195                                 if($property && (stristr($property,':description'))) {
 196
 197                                         $text = $item->getAttribute('content');
 198                                         if(stristr($text,'<script')) {
 199                                                 $text = '';
 200                                                 continue;
 201                                         }
 202                                         $text = strip_tags($text);
 203
 204
 205                                         $text = substr($text,0,250) . '...' ;
 206                                 }
 207                                 if($property && (stristr($property,':image'))) {
 208
 209                                         $image = $item->getAttribute('content');
 210                                         if(stristr($text,'<script')) {
 211                                                 $image = '';
 212                                                 continue;
 213                                         }
 214                                         $image = strip_tags($image);
 215
 216                                         $i = fetch_url($image);
 217                                         if($i) {
 218                                                 require_once('include/Photo.php');
 219                                                 // guess mimetype from headers or filename
 220                                                 $type = guess_image_type($image,true);
 221
 222                                                 $ph = new Photo($i, $type);
 223                                                 if($ph->is_valid()) {
 224                                                         if($ph->getWidth() > 300 || $ph->getHeight() > 300) {
 225                                                                 $ph->scaleImage(300);
 226                                                                 $new_width = $ph->getWidth();
 227                                                                 $new_height = $ph->getHeight();
 228                                                                 if($textmode)
 229                                                                         $image = $br . $br . '[img=' . $new_width . 'x' . $new_height . ']' . $image . '[/img]';
 230                                                                 else
 231                                                                         $image = '<br /><br /><img height="' . $new_height . '" width="' . $new_width . '" src="' .$image . '" alt="photo" />';
 232                                                         }
 233                                                         else {
 234                                                                 if($textmode)
 235                                                                         $image = $br . $br . '[img]' . $image . '[/img]';
 236                                                                 else
 237                                                                         $image = '<br /><br /><img src="' . $image . '" alt="photo" />';
 238                                                         }
 239                                                 }
 240                                                 else
 241                                                         $image = '';
 242
 243                                         }
 244                                 }
 245                         }
 246                 }
 247         }
 248
 249         if(strlen($text)) {
 250                 if($textmode)
 251                         $text = $br .$br . '[quote]' . $text . '[/quote]' . $br ;
 252                 else
 253                         $text = '<br /><br /><blockquote>' . $text . '</blockquote><br />';
 254         }
 255
 256         if($image) {
 257                 $text = $image . $br . $text;
 258         }
 259         $title = str_replace(array("\r","\n"),array('',''),$title);
 260
 261         $result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
 262
 263         logger('parse_url: returns: ' . $result);
 264
 265         echo $result;
 266         killme();
 267 }