mod/parse_url.php

   1 <?php
   2 /* To-Do
   3 https://developers.google.com/+/plugins/snippet/
   4
   5 <meta itemprop="name" content="Toller Titel">
   6 <meta itemprop="description" content="Eine tolle Beschreibung">
   7 <meta itemprop="image" content="http://maple.libertreeproject.org/images/tree-icon.png">
   8
   9 <body itemscope itemtype="http://schema.org/Product">
  10   <h1 itemprop="name">Shiny Trinket</h1>
  11   <img itemprop="image" src="{image-url}" />
  12   <p itemprop="description">Shiny trinkets are shiny.</p>
  13 </body>
  14 */
  15
  16 if(!function_exists('deletenode')) {
  17         function deletenode(&$doc, $node)
  18         {
  19                 $xpath = new DomXPath($doc);
  20                 $list = $xpath->query("//".$node);
  21                 foreach ($list as $child)
  22                         $child->parentNode->removeChild($child);
  23         }
  24 }
  25
  26 function completeurl($url, $scheme) {
  27         $urlarr = parse_url($url);
  28
  29         if (isset($urlarr["scheme"]))
  30                 return($url);
  31
  32         $schemearr = parse_url($scheme);
  33
  34         $complete = $schemearr["scheme"]."://".$schemearr["host"];
  35
  36         if (@$schemearr["port"] != "")
  37                 $complete .= ":".$schemearr["port"];
  38
  39                 if(strpos($urlarr['path'],'/') !== 0)
  40                         $complete .= '/';
  41
  42         $complete .= $urlarr["path"];
  43
  44         if (@$urlarr["query"] != "")
  45                 $complete .= "?".$urlarr["query"];
  46
  47         if (@$urlarr["fragment"] != "")
  48                 $complete .= "#".$urlarr["fragment"];
  49
  50         return($complete);
  51 }
  52
  53 function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $count = 1) {
  54         require_once("include/network.php");
  55
  56         $a = get_app();
  57
  58         $siteinfo = array();
  59
  60         if ($count > 10) {
  61                 logger("parseurl_getsiteinfo: Endless loop detected for ".$url, LOGGER_DEBUG);
  62                 return($siteinfo);
  63         }
  64
  65         $url = trim($url, "'");
  66         $url = trim($url, '"');
  67
  68         $url = original_url($url);
  69
  70         $siteinfo["url"] = $url;
  71         $siteinfo["type"] = "link";
  72
  73         $ch = curl_init();
  74         curl_setopt($ch, CURLOPT_URL, $url);
  75         curl_setopt($ch, CURLOPT_HEADER, 1);
  76         curl_setopt($ch, CURLOPT_NOBODY, 0);
  77         curl_setopt($ch, CURLOPT_TIMEOUT, 3);
  78         curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
  79         //curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 1);
  80         curl_setopt($ch, CURLOPT_USERAGENT, $a->get_useragent());
  81
  82         $header = curl_exec($ch);
  83         $curl_info = @curl_getinfo($ch);
  84         $http_code = $curl_info['http_code'];
  85         curl_close($ch);
  86
  87         if ((($curl_info['http_code'] == "301") OR ($curl_info['http_code'] == "302") OR ($curl_info['http_code'] == "303") OR ($curl_info['http_code'] == "307"))
  88                 AND (($curl_info['redirect_url'] != "") OR ($curl_info['location'] != ""))) {
  89                 if ($curl_info['redirect_url'] != "")
  90                         $siteinfo = parseurl_getsiteinfo($curl_info['redirect_url'], $no_guessing, $do_oembed, ++$count);
  91                 else
  92                         $siteinfo = parseurl_getsiteinfo($curl_info['location'], $no_guessing, $do_oembed, ++$count);
  93                 return($siteinfo);
  94         }
  95
  96         if ($do_oembed) {
  97                 require_once("include/oembed.php");
  98
  99                 $oembed_data = oembed_fetch_url($url);
 100
 101                 if ($oembed_data->type != "error")
 102                         $siteinfo["type"] = $oembed_data->type;
 103         }
 104
 105         // Fetch the first mentioned charset. Can be in body or header
 106         $charset = "";
 107         if (preg_match('/charset=(.*?)['."'".'"\s\n]/', $header, $matches))
 108                 $charset = trim(trim(trim(array_pop($matches)), ';,'));
 109
 110         if ($charset == "")
 111                 $charset = "utf-8";
 112
 113         $pos = strpos($header, "\r\n\r\n");
 114
 115         if ($pos)
 116                 $body = trim(substr($header, $pos));
 117         else
 118                 $body = $header;
 119
 120         if (($charset != '') AND (strtoupper($charset) != "UTF-8")) {
 121                 logger("parseurl_getsiteinfo: detected charset ".$charset, LOGGER_DEBUG);
 122                 //$body = mb_convert_encoding($body, "UTF-8", $charset);
 123                 $body = iconv($charset, "UTF-8//TRANSLIT", $body);
 124         }
 125
 126         $body = mb_convert_encoding($body, 'HTML-ENTITIES', "UTF-8");
 127
 128         $doc = new DOMDocument();
 129         @$doc->loadHTML($body);
 130
 131         deletenode($doc, 'style');
 132         deletenode($doc, 'script');
 133         deletenode($doc, 'option');
 134         deletenode($doc, 'h1');
 135         deletenode($doc, 'h2');
 136         deletenode($doc, 'h3');
 137         deletenode($doc, 'h4');
 138         deletenode($doc, 'h5');
 139         deletenode($doc, 'h6');
 140         deletenode($doc, 'ol');
 141         deletenode($doc, 'ul');
 142
 143         $xpath = new DomXPath($doc);
 144
 145         $list = $xpath->query("//meta[@content]");
 146         foreach ($list as $node) {
 147                 $attr = array();
 148                 if ($node->attributes->length)
 149                         foreach ($node->attributes as $attribute)
 150                                 $attr[$attribute->name] = $attribute->value;
 151
 152                 if (@$attr["http-equiv"] == 'refresh') {
 153                         $path = $attr["content"];
 154                         $pathinfo = explode(";", $path);
 155                         $content = "";
 156                         foreach ($pathinfo AS $value) {
 157                                 if (substr(strtolower($value), 0, 4) == "url=")
 158                                         $content = substr($value, 4);
 159                         }
 160                         if ($content != "") {
 161                                 $siteinfo = parseurl_getsiteinfo($content, $no_guessing, $do_oembed, ++$count);
 162                                 return($siteinfo);
 163                         }
 164                 }
 165         }
 166
 167         //$list = $xpath->query("head/title");
 168         $list = $xpath->query("//title");
 169         foreach ($list as $node)
 170                 $siteinfo["title"] =  html_entity_decode($node->nodeValue, ENT_QUOTES, "UTF-8");
 171
 172         //$list = $xpath->query("head/meta[@name]");
 173         $list = $xpath->query("//meta[@name]");
 174         foreach ($list as $node) {
 175                 $attr = array();
 176                 if ($node->attributes->length)
 177                         foreach ($node->attributes as $attribute)
 178                                 $attr[$attribute->name] = $attribute->value;
 179
 180                 $attr["content"] = trim(html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8"));
 181
 182                 if ($attr["content"] != "")
 183                         switch (strtolower($attr["name"])) {
 184                                 case "fulltitle":
 185                                         $siteinfo["title"] = $attr["content"];
 186                                         break;
 187                                 case "description":
 188                                         $siteinfo["text"] = $attr["content"];
 189                                         break;
 190                                 case "twitter:image":
 191                                         $siteinfo["image"] = $attr["content"];
 192                                         break;
 193                                 case "twitter:image:src":
 194                                         $siteinfo["image"] = $attr["content"];
 195                                         break;
 196                                 case "twitter:card":
 197                                         if (($siteinfo["type"] == "") OR ($attr["content"] == "photo"))
 198                                                 $siteinfo["type"] = $attr["content"];
 199                                         break;
 200                                 case "twitter:description":
 201                                         $siteinfo["text"] = $attr["content"];
 202                                         break;
 203                                 case "twitter:title":
 204                                         $siteinfo["title"] = $attr["content"];
 205                                         break;
 206                                 case "dc.title":
 207                                         $siteinfo["title"] = $attr["content"];
 208                                         break;
 209                                 case "dc.description":
 210                                         $siteinfo["text"] = $attr["content"];
 211                                         break;
 212                                 case "keywords":
 213                                         $keywords = explode(",", $attr["content"]);
 214                                         break;
 215                                 case "news_keywords":
 216                                         $keywords = explode(",", $attr["content"]);
 217                                         break;
 218                         }
 219                 if ($siteinfo["type"] == "summary")
 220                         $siteinfo["type"] = "link";
 221         }
 222
 223         if (isset($keywords)) {
 224                 $siteinfo["keywords"] = array();
 225                 foreach ($keywords as $keyword)
 226                         $siteinfo["keywords"][] = trim($keyword);
 227         }
 228
 229         //$list = $xpath->query("head/meta[@property]");
 230         $list = $xpath->query("//meta[@property]");
 231         foreach ($list as $node) {
 232                 $attr = array();
 233                 if ($node->attributes->length)
 234                         foreach ($node->attributes as $attribute)
 235                                 $attr[$attribute->name] = $attribute->value;
 236
 237                 $attr["content"] = trim(html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8"));
 238
 239                 if ($attr["content"] != "")
 240                         switch (strtolower($attr["property"])) {
 241                                 case "og:image":
 242                                         $siteinfo["image"] = $attr["content"];
 243                                         break;
 244                                 case "og:title":
 245                                         $siteinfo["title"] = $attr["content"];
 246                                         break;
 247                                 case "og:description":
 248                                         $siteinfo["text"] = $attr["content"];
 249                                         break;
 250                         }
 251         }
 252
 253         if (isset($oembed_data) AND ($oembed_data->type == "link") AND ($siteinfo["type"] != "photo")) {
 254                 if (isset($oembed_data->title) AND (trim($oembed_data->title) != ""))
 255                         $siteinfo["title"] = $oembed_data->title;
 256                 if (isset($oembed_data->description) AND (trim($oembed_data->description) != ""))
 257                         $siteinfo["text"] = trim($oembed_data->description);
 258                 if (isset($oembed_data->thumbnail_url) AND (trim($oembed_data->thumbnail_url) != ""))
 259                         $siteinfo["image"] = $oembed_data->thumbnail_url;
 260         }
 261
 262         if ((@$siteinfo["image"] == "") AND !$no_guessing) {
 263             $list = $xpath->query("//img[@src]");
 264             foreach ($list as $node) {
 265                 $attr = array();
 266                 if ($node->attributes->length)
 267                     foreach ($node->attributes as $attribute)
 268                         $attr[$attribute->name] = $attribute->value;
 269
 270                         $src = completeurl($attr["src"], $url);
 271                         $photodata = @getimagesize($src);
 272
 273                         if (($photodata) && ($photodata[0] > 150) and ($photodata[1] > 150)) {
 274                                 if ($photodata[0] > 300) {
 275                                         $photodata[1] = round($photodata[1] * (300 / $photodata[0]));
 276                                         $photodata[0] = 300;
 277                                 }
 278                                 if ($photodata[1] > 300) {
 279                                         $photodata[0] = round($photodata[0] * (300 / $photodata[1]));
 280                                         $photodata[1] = 300;
 281                                 }
 282                                 $siteinfo["images"][] = array("src"=>$src,
 283                                                                 "width"=>$photodata[0],
 284                                                                 "height"=>$photodata[1]);
 285                         }
 286
 287                 }
 288     } else {
 289                 $src = completeurl($siteinfo["image"], $url);
 290
 291                 unset($siteinfo["image"]);
 292
 293                 $photodata = @getimagesize($src);
 294
 295                 if (($photodata) && ($photodata[0] > 10) and ($photodata[1] > 10))
 296                         $siteinfo["images"][] = array("src"=>$src,
 297                                                         "width"=>$photodata[0],
 298                                                         "height"=>$photodata[1]);
 299         }
 300
 301         if ((@$siteinfo["text"] == "") AND (@$siteinfo["title"] != "") AND !$no_guessing) {
 302                 $text = "";
 303
 304                 $list = $xpath->query("//div[@class='article']");
 305                 foreach ($list as $node)
 306                         if (strlen($node->nodeValue) > 40)
 307                                 $text .= " ".trim($node->nodeValue);
 308
 309                 if ($text == "") {
 310                         $list = $xpath->query("//div[@class='content']");
 311                         foreach ($list as $node)
 312                                 if (strlen($node->nodeValue) > 40)
 313                                         $text .= " ".trim($node->nodeValue);
 314                 }
 315
 316                 // If none text was found then take the paragraph content
 317                 if ($text == "") {
 318                         $list = $xpath->query("//p");
 319                         foreach ($list as $node)
 320                                 if (strlen($node->nodeValue) > 40)
 321                                         $text .= " ".trim($node->nodeValue);
 322                 }
 323
 324                 if ($text != "") {
 325                         $text = trim(str_replace(array("\n", "\r"), array(" ", " "), $text));
 326
 327                         while (strpos($text, "  "))
 328                                 $text = trim(str_replace("  ", " ", $text));
 329
 330                         $siteinfo["text"] = trim(html_entity_decode(substr($text,0,350), ENT_QUOTES, "UTF-8").'...');
 331                 }
 332         }
 333
 334         logger("parseurl_getsiteinfo: Siteinfo for ".$url." ".print_r($siteinfo, true), LOGGER_DEBUG);
 335
 336         return($siteinfo);
 337 }
 338
 339 function arr_add_hashes(&$item,$k) {
 340         $item = '#' . $item;
 341 }
 342
 343 function parse_url_content(&$a) {
 344
 345         $text = null;
 346         $str_tags = '';
 347
 348         $textmode = false;
 349
 350         if(local_user() && (! feature_enabled(local_user(),'richtext')))
 351                 $textmode = true;
 352
 353         //if($textmode)
 354         $br = (($textmode) ? "\n" : '<br />');
 355
 356         if(x($_GET,'binurl'))
 357                 $url = trim(hex2bin($_GET['binurl']));
 358         else
 359                 $url = trim($_GET['url']);
 360
 361         if($_GET['title'])
 362                 $title = strip_tags(trim($_GET['title']));
 363
 364         if($_GET['description'])
 365                 $text = strip_tags(trim($_GET['description']));
 366
 367         if($_GET['tags']) {
 368                 $arr_tags = str_getcsv($_GET['tags']);
 369                 if(count($arr_tags)) {
 370                         array_walk($arr_tags,'arr_add_hashes');
 371                         $str_tags = $br . implode(' ',$arr_tags) . $br;
 372                 }
 373         }
 374
 375         logger('parse_url: ' . $url);
 376
 377         if($textmode)
 378                 $template = '[bookmark=%s]%s[/bookmark]%s';
 379         else
 380                 $template = "<a class=\"bookmark\" href=\"%s\" >%s</a>%s";
 381
 382         $arr = array('url' => $url, 'text' => '');
 383
 384         call_hooks('parse_link', $arr);
 385
 386         if(strlen($arr['text'])) {
 387                 echo $arr['text'];
 388                 killme();
 389         }
 390
 391
 392         if($url && $title && $text) {
 393
 394                 $title = str_replace(array("\r","\n"),array('',''),$title);
 395
 396                 if($textmode)
 397                         $text = '[quote]' . trim($text) . '[/quote]' . $br;
 398                 else {
 399                         $text = '<blockquote>' . htmlspecialchars(trim($text)) . '</blockquote><br />';
 400                         $title = htmlspecialchars($title);
 401                 }
 402
 403                 $result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
 404
 405                 logger('parse_url (unparsed): returns: ' . $result);
 406
 407                 echo $result;
 408                 killme();
 409         }
 410
 411         $siteinfo = parseurl_getsiteinfo($url);
 412
 413         $url= $siteinfo["url"];
 414
 415         $sitedata = "";
 416
 417         if($siteinfo["title"] == "") {
 418                 $sitedata .= sprintf($template,$url,$url,'') . $str_tags;
 419                 killme();
 420         } else {
 421                 $text = $siteinfo["text"];
 422                 $title = $siteinfo["title"];
 423         }
 424
 425         $image = "";
 426
 427         if (($siteinfo["type"] != "video") AND (sizeof($siteinfo["images"]) > 0)){
 428                 /* Execute below code only if image is present in siteinfo */
 429
 430                 $total_images = 0;
 431                 $max_images = get_config('system','max_bookmark_images');
 432                 if($max_images === false)
 433                         $max_images = 2;
 434                 else
 435                         $max_images = intval($max_images);
 436
 437                 foreach ($siteinfo["images"] as $imagedata) {
 438                         if($textmode)
 439                                 $image .= '[img='.$imagedata["width"].'x'.$imagedata["height"].']'.$imagedata["src"].'[/img]' . "\n";
 440                         else
 441                                 $image .= '<img height="'.$imagedata["height"].'" width="'.$imagedata["width"].'" src="'.$imagedata["src"].'" alt="photo" /><br />';
 442                         $total_images ++;
 443                         if($max_images && $max_images >= $total_images)
 444                                 break;
 445                 }
 446         }
 447
 448         if(strlen($text)) {
 449                 if($textmode)
 450                         $text = '[quote]'.trim($text).'[/quote]';
 451                 else
 452                         $text = '<blockquote>'.htmlspecialchars(trim($text)).'</blockquote>';
 453         }
 454
 455         if($image)
 456                 $text = $br.$br.$image.$text;
 457         else
 458                 $text = $br.$text;
 459
 460         $title = str_replace(array("\r","\n"),array('',''),$title);
 461
 462         $result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
 463
 464         logger('parse_url: returns: ' . $result);
 465
 466         $sitedata .=  trim($result);
 467
 468         if (($siteinfo["type"] == "video") AND ($url != ""))
 469                 echo "[class=type-video]".$sitedata."[/class]";
 470         elseif (($siteinfo["type"] != "photo"))
 471                 echo "[class=type-link]".$sitedata."[/class]";
 472         else
 473                 echo "[class=type-photo]".$title.$br.$image."[/class]";
 474
 475         killme();
 476 }
 477 ?>