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) {
  54         $siteinfo = array();
  55
  56         $url = trim($url, "'");
  57         $url = trim($url, '"');
  58         $siteinfo["url"] = $url;
  59         $siteinfo["type"] = "link";
  60
  61         $ch = curl_init();
  62         curl_setopt($ch, CURLOPT_URL, $url);
  63         curl_setopt($ch, CURLOPT_HEADER, 1);
  64         curl_setopt($ch, CURLOPT_NOBODY, 0);
  65         curl_setopt($ch, CURLOPT_TIMEOUT, 3);
  66         curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
  67         //curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 1);
  68         curl_setopt($ch,CURLOPT_USERAGENT,' Mozilla/5.0 (Macintosh; Intel Mac OS X 10.8; rv:24.0) Gecko/20100101 Firefox/24.0');
  69
  70         $header = curl_exec($ch);
  71         $curl_info = @curl_getinfo($ch);
  72         $http_code = $curl_info['http_code'];
  73         curl_close($ch);
  74
  75         if ((($curl_info['http_code'] == "301") OR ($curl_info['http_code'] == "302") OR ($curl_info['http_code'] == "303") OR ($curl_info['http_code'] == "307"))
  76                 AND (($curl_info['redirect_url'] != "") OR ($curl_info['location'] != ""))) {
  77                 if ($curl_info['redirect_url'] != "")
  78                         $siteinfo = parseurl_getsiteinfo($curl_info['redirect_url']);
  79                 else
  80                         $siteinfo = parseurl_getsiteinfo($curl_info['location']);
  81                 return($siteinfo);
  82         }
  83
  84         require_once("include/oembed.php");
  85
  86         $oembed_data = oembed_fetch_url($url);
  87
  88         if ($oembed_data->type != "error")
  89                 $siteinfo["type"] = $oembed_data->type;
  90
  91         // Fetch the first mentioned charset. Can be in body or header
  92         $charset = "";
  93         if (preg_match('/charset=(.*?)['."'".'"\s\n]/', $header, $matches))
  94                 $charset = trim(array_pop($matches));
  95
  96         if ($charset == "")
  97                 $charset = "utf-8";
  98
  99         $pos = strpos($header, "\r\n\r\n");
 100
 101         if ($pos)
 102                 $body = trim(substr($header, $pos));
 103         else
 104                 $body = $header;
 105
 106         $body = mb_convert_encoding($body, "UTF-8", $charset);
 107         $body = mb_convert_encoding($body, 'HTML-ENTITIES', "UTF-8");
 108
 109         $doc = new DOMDocument();
 110         @$doc->loadHTML($body);
 111
 112         deletenode($doc, 'style');
 113         deletenode($doc, 'script');
 114         deletenode($doc, 'option');
 115         deletenode($doc, 'h1');
 116         deletenode($doc, 'h2');
 117         deletenode($doc, 'h3');
 118         deletenode($doc, 'h4');
 119         deletenode($doc, 'h5');
 120         deletenode($doc, 'h6');
 121         deletenode($doc, 'ol');
 122         deletenode($doc, 'ul');
 123
 124         $xpath = new DomXPath($doc);
 125
 126         $list = $xpath->query("//meta[@content]");
 127         foreach ($list as $node) {
 128                 $attr = array();
 129                 if ($node->attributes->length)
 130                         foreach ($node->attributes as $attribute)
 131                                 $attr[$attribute->name] = $attribute->value;
 132
 133                 if (@$attr["http-equiv"] == 'refresh') {
 134                         $path = $attr["content"];
 135                         $pathinfo = explode(";", $path);
 136                         $content = "";
 137                         foreach ($pathinfo AS $value) {
 138                                 if (substr(strtolower($value), 0, 4) == "url=")
 139                                         $content = substr($value, 4);
 140                         }
 141                         if ($content != "") {
 142                                 $siteinfo = parseurl_getsiteinfo($content);
 143                                 return($siteinfo);
 144                         }
 145                 }
 146         }
 147
 148         //$list = $xpath->query("head/title");
 149         $list = $xpath->query("//title");
 150         foreach ($list as $node)
 151                 $siteinfo["title"] =  html_entity_decode($node->nodeValue, ENT_QUOTES, "UTF-8");
 152
 153         //$list = $xpath->query("head/meta[@name]");
 154         $list = $xpath->query("//meta[@name]");
 155         foreach ($list as $node) {
 156                 $attr = array();
 157                 if ($node->attributes->length)
 158                         foreach ($node->attributes as $attribute)
 159                                 $attr[$attribute->name] = $attribute->value;
 160
 161                 $attr["content"] = trim(html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8"));
 162
 163                 if ($attr["content"] != "")
 164                         switch (strtolower($attr["name"])) {
 165                                 case "fulltitle":
 166                                         $siteinfo["title"] = $attr["content"];
 167                                         break;
 168                                 case "description":
 169                                         $siteinfo["text"] = $attr["content"];
 170                                         break;
 171                                 case "twitter:image":
 172                                         $siteinfo["image"] = $attr["content"];
 173                                         break;
 174                                 case "twitter:card":
 175                                         if ($siteinfo["type"] == "")
 176                                                 $siteinfo["type"] = $attr["content"];
 177                                         break;
 178                                 case "twitter:description":
 179                                         $siteinfo["text"] = $attr["content"];
 180                                         break;
 181                                 case "twitter:title":
 182                                         $siteinfo["title"] = $attr["content"];
 183                                         break;
 184                                 case "dc.title":
 185                                         $siteinfo["title"] = $attr["content"];
 186                                         break;
 187                                 case "dc.description":
 188                                         $siteinfo["text"] = $attr["content"];
 189                                         break;
 190                         }
 191                 if ($siteinfo["type"] == "summary")
 192                         $siteinfo["type"] = "link";
 193         }
 194
 195         //$list = $xpath->query("head/meta[@property]");
 196         $list = $xpath->query("//meta[@property]");
 197         foreach ($list as $node) {
 198                 $attr = array();
 199                 if ($node->attributes->length)
 200                         foreach ($node->attributes as $attribute)
 201                                 $attr[$attribute->name] = $attribute->value;
 202
 203                 $attr["content"] = trim(html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8"));
 204
 205                 if ($attr["content"] != "")
 206                         switch (strtolower($attr["property"])) {
 207                                 case "og:image":
 208                                         $siteinfo["image"] = $attr["content"];
 209                                         break;
 210                                 case "og:title":
 211                                         $siteinfo["title"] = $attr["content"];
 212                                         break;
 213                                 case "og:description":
 214                                         $siteinfo["text"] = $attr["content"];
 215                                         break;
 216                         }
 217         }
 218
 219         if ($oembed_data->type == "link") {
 220                 if (isset($oembed_data->title) AND (trim($oembed_data->title) != ""))
 221                         $siteinfo["title"] = $oembed_data->title;
 222                 if (isset($oembed_data->description) AND (trim($oembed_data->description) != ""))
 223                         $siteinfo["text"] = trim($oembed_data->description);
 224                 if (isset($oembed_data->thumbnail_url) AND (trim($oembed_data->thumbnail_url) != ""))
 225                         $siteinfo["image"] = $oembed_data->thumbnail_url;
 226         }
 227
 228         if ((@$siteinfo["image"] == "") AND !$no_guessing) {
 229             $list = $xpath->query("//img[@src]");
 230             foreach ($list as $node) {
 231                 $attr = array();
 232                 if ($node->attributes->length)
 233                     foreach ($node->attributes as $attribute)
 234                         $attr[$attribute->name] = $attribute->value;
 235
 236                         $src = completeurl($attr["src"], $url);
 237                         $photodata = @getimagesize($src);
 238
 239                         if (($photodata) && ($photodata[0] > 150) and ($photodata[1] > 150)) {
 240                                 if ($photodata[0] > 300) {
 241                                         $photodata[1] = round($photodata[1] * (300 / $photodata[0]));
 242                                         $photodata[0] = 300;
 243                                 }
 244                                 if ($photodata[1] > 300) {
 245                                         $photodata[0] = round($photodata[0] * (300 / $photodata[1]));
 246                                         $photodata[1] = 300;
 247                                 }
 248                                 $siteinfo["images"][] = array("src"=>$src,
 249                                                                 "width"=>$photodata[0],
 250                                                                 "height"=>$photodata[1]);
 251                         }
 252
 253                 }
 254     } else {
 255                 $src = completeurl($siteinfo["image"], $url);
 256
 257                 unset($siteinfo["image"]);
 258
 259                 $photodata = @getimagesize($src);
 260
 261                 if (($photodata) && ($photodata[0] > 10) and ($photodata[1] > 10))
 262                         $siteinfo["images"][] = array("src"=>$src,
 263                                                         "width"=>$photodata[0],
 264                                                         "height"=>$photodata[1]);
 265         }
 266
 267         if ((@$siteinfo["text"] == "") AND (@$siteinfo["title"] != "") AND !$no_guessing) {
 268                 $text = "";
 269
 270                 $list = $xpath->query("//div[@class='article']");
 271                 foreach ($list as $node)
 272                         if (strlen($node->nodeValue) > 40)
 273                                 $text .= " ".trim($node->nodeValue);
 274
 275                 if ($text == "") {
 276                         $list = $xpath->query("//div[@class='content']");
 277                         foreach ($list as $node)
 278                                 if (strlen($node->nodeValue) > 40)
 279                                         $text .= " ".trim($node->nodeValue);
 280                 }
 281
 282                 // If none text was found then take the paragraph content
 283                 if ($text == "") {
 284                         $list = $xpath->query("//p");
 285                         foreach ($list as $node)
 286                                 if (strlen($node->nodeValue) > 40)
 287                                         $text .= " ".trim($node->nodeValue);
 288                 }
 289
 290                 if ($text != "") {
 291                         $text = trim(str_replace(array("\n", "\r"), array(" ", " "), $text));
 292
 293                         while (strpos($text, "  "))
 294                                 $text = trim(str_replace("  ", " ", $text));
 295
 296                         $siteinfo["text"] = trim(html_entity_decode(substr($text,0,350), ENT_QUOTES, "UTF-8").'...');
 297                 }
 298         }
 299
 300         logger("parseurl_getsiteinfo: Siteinfo for ".$url." ".print_r($siteinfo, true), LOGGER_DEBUG);
 301
 302         return($siteinfo);
 303 }
 304
 305 function arr_add_hashes(&$item,$k) {
 306         $item = '#' . $item;
 307 }
 308
 309 function parse_url_content(&$a) {
 310
 311         $text = null;
 312         $str_tags = '';
 313
 314         $textmode = false;
 315
 316         if(local_user() && (! feature_enabled(local_user(),'richtext')))
 317                 $textmode = true;
 318
 319         //if($textmode)
 320         $br = (($textmode) ? "\n" : '<br />');
 321
 322         if(x($_GET,'binurl'))
 323                 $url = trim(hex2bin($_GET['binurl']));
 324         else
 325                 $url = trim($_GET['url']);
 326
 327         if($_GET['title'])
 328                 $title = strip_tags(trim($_GET['title']));
 329
 330         if($_GET['description'])
 331                 $text = strip_tags(trim($_GET['description']));
 332
 333         if($_GET['tags']) {
 334                 $arr_tags = str_getcsv($_GET['tags']);
 335                 if(count($arr_tags)) {
 336                         array_walk($arr_tags,'arr_add_hashes');
 337                         $str_tags = $br . implode(' ',$arr_tags) . $br;
 338                 }
 339         }
 340
 341         logger('parse_url: ' . $url);
 342
 343         if($textmode)
 344                 $template = '[bookmark=%s]%s[/bookmark]%s';
 345         else
 346                 $template = "<a class=\"bookmark\" href=\"%s\" >%s</a>%s";
 347
 348         $arr = array('url' => $url, 'text' => '');
 349
 350         call_hooks('parse_link', $arr);
 351
 352         if(strlen($arr['text'])) {
 353                 echo $arr['text'];
 354                 killme();
 355         }
 356
 357
 358         if($url && $title && $text) {
 359
 360                 $title = str_replace(array("\r","\n"),array('',''),$title);
 361
 362                 if($textmode)
 363                         $text = '[quote]' . trim($text) . '[/quote]' . $br;
 364                 else {
 365                         $text = '<blockquote>' . htmlspecialchars(trim($text)) . '</blockquote><br />';
 366                         $title = htmlspecialchars($title);
 367                 }
 368
 369                 $result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
 370
 371                 logger('parse_url (unparsed): returns: ' . $result);
 372
 373                 echo $result;
 374                 killme();
 375         }
 376
 377         $siteinfo = parseurl_getsiteinfo($url);
 378
 379         $url= $siteinfo["url"];
 380
 381         $sitedata = "";
 382
 383         if($siteinfo["title"] == "") {
 384                 $sitedata .= sprintf($template,$url,$url,'') . $str_tags;
 385                 killme();
 386         } else {
 387                 $text = $siteinfo["text"];
 388                 $title = $siteinfo["title"];
 389         }
 390
 391         $image = "";
 392
 393         if (($siteinfo["type"] != "video") AND (sizeof($siteinfo["images"]) > 0)){
 394                 /* Execute below code only if image is present in siteinfo */
 395
 396                 $total_images = 0;
 397                 $max_images = get_config('system','max_bookmark_images');
 398                 if($max_images === false)
 399                         $max_images = 2;
 400                 else
 401                         $max_images = intval($max_images);
 402
 403                 foreach ($siteinfo["images"] as $imagedata) {
 404                         if($textmode)
 405                                 $image .= '[img='.$imagedata["width"].'x'.$imagedata["height"].']'.$imagedata["src"].'[/img]' . "\n";
 406                         else
 407                                 $image .= '<img height="'.$imagedata["height"].'" width="'.$imagedata["width"].'" src="'.$imagedata["src"].'" alt="photo" /><br />';
 408                         $total_images ++;
 409                         if($max_images && $max_images >= $total_images)
 410                                 break;
 411                 }
 412         }
 413
 414         if(strlen($text)) {
 415                 if($textmode)
 416                         $text = '[quote]'.trim($text).'[/quote]';
 417                 else
 418                         $text = '<blockquote>'.htmlspecialchars(trim($text)).'</blockquote>';
 419         }
 420
 421         if($image)
 422                 $text = $br.$br.$image.$text;
 423         else
 424                 $text = $br.$text;
 425
 426         $title = str_replace(array("\r","\n"),array('',''),$title);
 427
 428         $result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
 429
 430         logger('parse_url: returns: ' . $result);
 431
 432         $sitedata .=  trim($result);
 433
 434         if (($siteinfo["type"] == "video") AND ($url != ""))
 435                 echo "[class=type-video]".$sitedata."[/class]";
 436         elseif (($siteinfo["type"] != "photo"))
 437                 echo "[class=type-link]".$sitedata."[/class]";
 438         else
 439                 echo "[class=type-photo]".$title.$br.$image."[/class]";
 440
 441         killme();
 442 }