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
  55         $siteinfo = array();
  56
  57         $url = trim($url, "'");
  58         $url = trim($url, '"');
  59         $siteinfo["url"] = $url;
  60         $siteinfo["type"] = "link";
  61
  62         $ch = curl_init();
  63         curl_setopt($ch, CURLOPT_URL, $url);
  64         curl_setopt($ch, CURLOPT_HEADER, 1);
  65         curl_setopt($ch, CURLOPT_NOBODY, 0);
  66         curl_setopt($ch, CURLOPT_TIMEOUT, 3);
  67         curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
  68         //curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 1);
  69         curl_setopt($ch,CURLOPT_USERAGENT,' Mozilla/5.0 (Macintosh; Intel Mac OS X 10.8; rv:24.0) Gecko/20100101 Firefox/24.0');
  70
  71         $header = curl_exec($ch);
  72         $curl_info = @curl_getinfo($ch);
  73         $http_code = $curl_info['http_code'];
  74         curl_close($ch);
  75
  76         if ((($curl_info['http_code'] == "301") OR ($curl_info['http_code'] == "302") OR ($curl_info['http_code'] == "303") OR ($curl_info['http_code'] == "307"))
  77                 AND (($curl_info['redirect_url'] != "") OR ($curl_info['location'] != ""))) {
  78                 if ($curl_info['redirect_url'] != "")
  79                         $siteinfo = parseurl_getsiteinfo($curl_info['redirect_url']);
  80                 else
  81                         $siteinfo = parseurl_getsiteinfo($curl_info['location']);
  82                 return($siteinfo);
  83         }
  84
  85         require_once("include/oembed.php");
  86
  87         $oembed_data = oembed_fetch_url($url);
  88
  89         if ($oembed_data->type != "error")
  90                 $siteinfo["type"] = $oembed_data->type;
  91
  92         // Fetch the first mentioned charset. Can be in body or header
  93         $charset = "";
  94         if (preg_match('/charset=(.*?)['."'".'"\s\n]/', $header, $matches))
  95                 $charset = trim(array_pop($matches));
  96
  97         if ($charset == "")
  98                 $charset = "utf-8";
  99
 100         $pos = strpos($header, "\r\n\r\n");
 101
 102         if ($pos)
 103                 $body = trim(substr($header, $pos));
 104         else
 105                 $body = $header;
 106
 107         $body = mb_convert_encoding($body, "UTF-8", $charset);
 108         $body = mb_convert_encoding($body, 'HTML-ENTITIES', "UTF-8");
 109
 110         $doc = new DOMDocument();
 111         @$doc->loadHTML($body);
 112
 113         deletenode($doc, 'style');
 114         deletenode($doc, 'script');
 115         deletenode($doc, 'option');
 116         deletenode($doc, 'h1');
 117         deletenode($doc, 'h2');
 118         deletenode($doc, 'h3');
 119         deletenode($doc, 'h4');
 120         deletenode($doc, 'h5');
 121         deletenode($doc, 'h6');
 122         deletenode($doc, 'ol');
 123         deletenode($doc, 'ul');
 124
 125         $xpath = new DomXPath($doc);
 126
 127         $list = $xpath->query("//meta[@content]");
 128         foreach ($list as $node) {
 129                 $attr = array();
 130                 if ($node->attributes->length)
 131                         foreach ($node->attributes as $attribute)
 132                                 $attr[$attribute->name] = $attribute->value;
 133
 134                 if (@$attr["http-equiv"] == 'refresh') {
 135                         $path = $attr["content"];
 136                         $pathinfo = explode(";", $path);
 137                         $content = "";
 138                         foreach ($pathinfo AS $value) {
 139                                 if (substr(strtolower($value), 0, 4) == "url=")
 140                                         $content = substr($value, 4);
 141                         }
 142                         if ($content != "") {
 143                                 $siteinfo = parseurl_getsiteinfo($content);
 144                                 return($siteinfo);
 145                         }
 146                 }
 147         }
 148
 149         //$list = $xpath->query("head/title");
 150         $list = $xpath->query("//title");
 151         foreach ($list as $node)
 152                 $siteinfo["title"] =  html_entity_decode($node->nodeValue, ENT_QUOTES, "UTF-8");
 153
 154         //$list = $xpath->query("head/meta[@name]");
 155         $list = $xpath->query("//meta[@name]");
 156         foreach ($list as $node) {
 157                 $attr = array();
 158                 if ($node->attributes->length)
 159                         foreach ($node->attributes as $attribute)
 160                                 $attr[$attribute->name] = $attribute->value;
 161
 162                 $attr["content"] = trim(html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8"));
 163
 164                 if ($attr["content"] != "")
 165                         switch (strtolower($attr["name"])) {
 166                                 case "fulltitle":
 167                                         $siteinfo["title"] = $attr["content"];
 168                                         break;
 169                                 case "description":
 170                                         $siteinfo["text"] = $attr["content"];
 171                                         break;
 172                                 case "twitter:image":
 173                                         $siteinfo["image"] = $attr["content"];
 174                                         break;
 175                                 case "twitter:card":
 176                                         if (($siteinfo["type"] == "") OR ($attr["content"] == "photo"))
 177                                                 $siteinfo["type"] = $attr["content"];
 178                                         break;
 179                                 case "twitter:description":
 180                                         $siteinfo["text"] = $attr["content"];
 181                                         break;
 182                                 case "twitter:title":
 183                                         $siteinfo["title"] = $attr["content"];
 184                                         break;
 185                                 case "dc.title":
 186                                         $siteinfo["title"] = $attr["content"];
 187                                         break;
 188                                 case "dc.description":
 189                                         $siteinfo["text"] = $attr["content"];
 190                                         break;
 191                         }
 192                 if ($siteinfo["type"] == "summary")
 193                         $siteinfo["type"] = "link";
 194         }
 195
 196         //$list = $xpath->query("head/meta[@property]");
 197         $list = $xpath->query("//meta[@property]");
 198         foreach ($list as $node) {
 199                 $attr = array();
 200                 if ($node->attributes->length)
 201                         foreach ($node->attributes as $attribute)
 202                                 $attr[$attribute->name] = $attribute->value;
 203
 204                 $attr["content"] = trim(html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8"));
 205
 206                 if ($attr["content"] != "")
 207                         switch (strtolower($attr["property"])) {
 208                                 case "og:image":
 209                                         $siteinfo["image"] = $attr["content"];
 210                                         break;
 211                                 case "og:title":
 212                                         $siteinfo["title"] = $attr["content"];
 213                                         break;
 214                                 case "og:description":
 215                                         $siteinfo["text"] = $attr["content"];
 216                                         break;
 217                         }
 218         }
 219
 220         if ($oembed_data->type == "link") {
 221                 if (isset($oembed_data->title) AND (trim($oembed_data->title) != ""))
 222                         $siteinfo["title"] = $oembed_data->title;
 223                 if (isset($oembed_data->description) AND (trim($oembed_data->description) != ""))
 224                         $siteinfo["text"] = trim($oembed_data->description);
 225                 if (isset($oembed_data->thumbnail_url) AND (trim($oembed_data->thumbnail_url) != ""))
 226                         $siteinfo["image"] = $oembed_data->thumbnail_url;
 227         }
 228
 229         if ((@$siteinfo["image"] == "") AND !$no_guessing) {
 230             $list = $xpath->query("//img[@src]");
 231             foreach ($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                         $src = completeurl($attr["src"], $url);
 238                         $photodata = @getimagesize($src);
 239
 240                         if (($photodata) && ($photodata[0] > 150) and ($photodata[1] > 150)) {
 241                                 if ($photodata[0] > 300) {
 242                                         $photodata[1] = round($photodata[1] * (300 / $photodata[0]));
 243                                         $photodata[0] = 300;
 244                                 }
 245                                 if ($photodata[1] > 300) {
 246                                         $photodata[0] = round($photodata[0] * (300 / $photodata[1]));
 247                                         $photodata[1] = 300;
 248                                 }
 249                                 $siteinfo["images"][] = array("src"=>$src,
 250                                                                 "width"=>$photodata[0],
 251                                                                 "height"=>$photodata[1]);
 252                         }
 253
 254                 }
 255     } else {
 256                 $src = completeurl($siteinfo["image"], $url);
 257
 258                 unset($siteinfo["image"]);
 259
 260                 $photodata = @getimagesize($src);
 261
 262                 if (($photodata) && ($photodata[0] > 10) and ($photodata[1] > 10))
 263                         $siteinfo["images"][] = array("src"=>$src,
 264                                                         "width"=>$photodata[0],
 265                                                         "height"=>$photodata[1]);
 266         }
 267
 268         if ((@$siteinfo["text"] == "") AND (@$siteinfo["title"] != "") AND !$no_guessing) {
 269                 $text = "";
 270
 271                 $list = $xpath->query("//div[@class='article']");
 272                 foreach ($list as $node)
 273                         if (strlen($node->nodeValue) > 40)
 274                                 $text .= " ".trim($node->nodeValue);
 275
 276                 if ($text == "") {
 277                         $list = $xpath->query("//div[@class='content']");
 278                         foreach ($list as $node)
 279                                 if (strlen($node->nodeValue) > 40)
 280                                         $text .= " ".trim($node->nodeValue);
 281                 }
 282
 283                 // If none text was found then take the paragraph content
 284                 if ($text == "") {
 285                         $list = $xpath->query("//p");
 286                         foreach ($list as $node)
 287                                 if (strlen($node->nodeValue) > 40)
 288                                         $text .= " ".trim($node->nodeValue);
 289                 }
 290
 291                 if ($text != "") {
 292                         $text = trim(str_replace(array("\n", "\r"), array(" ", " "), $text));
 293
 294                         while (strpos($text, "  "))
 295                                 $text = trim(str_replace("  ", " ", $text));
 296
 297                         $siteinfo["text"] = trim(html_entity_decode(substr($text,0,350), ENT_QUOTES, "UTF-8").'...');
 298                 }
 299         }
 300
 301         logger("parseurl_getsiteinfo: Siteinfo for ".$url." ".print_r($siteinfo, true), LOGGER_DEBUG);
 302
 303         return($siteinfo);
 304 }
 305
 306 function arr_add_hashes(&$item,$k) {
 307         $item = '#' . $item;
 308 }
 309
 310 function parse_url_content(&$a) {
 311
 312         $text = null;
 313         $str_tags = '';
 314
 315         $textmode = false;
 316
 317         if(local_user() && (! feature_enabled(local_user(),'richtext')))
 318                 $textmode = true;
 319
 320         //if($textmode)
 321         $br = (($textmode) ? "\n" : '<br />');
 322
 323         if(x($_GET,'binurl'))
 324                 $url = trim(hex2bin($_GET['binurl']));
 325         else
 326                 $url = trim($_GET['url']);
 327
 328         if($_GET['title'])
 329                 $title = strip_tags(trim($_GET['title']));
 330
 331         if($_GET['description'])
 332                 $text = strip_tags(trim($_GET['description']));
 333
 334         if($_GET['tags']) {
 335                 $arr_tags = str_getcsv($_GET['tags']);
 336                 if(count($arr_tags)) {
 337                         array_walk($arr_tags,'arr_add_hashes');
 338                         $str_tags = $br . implode(' ',$arr_tags) . $br;
 339                 }
 340         }
 341
 342         logger('parse_url: ' . $url);
 343
 344         if($textmode)
 345                 $template = '[bookmark=%s]%s[/bookmark]%s';
 346         else
 347                 $template = "<a class=\"bookmark\" href=\"%s\" >%s</a>%s";
 348
 349         $arr = array('url' => $url, 'text' => '');
 350
 351         call_hooks('parse_link', $arr);
 352
 353         if(strlen($arr['text'])) {
 354                 echo $arr['text'];
 355                 killme();
 356         }
 357
 358
 359         if($url && $title && $text) {
 360
 361                 $title = str_replace(array("\r","\n"),array('',''),$title);
 362
 363                 if($textmode)
 364                         $text = '[quote]' . trim($text) . '[/quote]' . $br;
 365                 else {
 366                         $text = '<blockquote>' . htmlspecialchars(trim($text)) . '</blockquote><br />';
 367                         $title = htmlspecialchars($title);
 368                 }
 369
 370                 $result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
 371
 372                 logger('parse_url (unparsed): returns: ' . $result);
 373
 374                 echo $result;
 375                 killme();
 376         }
 377
 378         $siteinfo = parseurl_getsiteinfo($url);
 379
 380         $url= $siteinfo["url"];
 381
 382         $sitedata = "";
 383
 384         if($siteinfo["title"] == "") {
 385                 $sitedata .= sprintf($template,$url,$url,'') . $str_tags;
 386                 killme();
 387         } else {
 388                 $text = $siteinfo["text"];
 389                 $title = $siteinfo["title"];
 390         }
 391
 392         $image = "";
 393
 394         if (($siteinfo["type"] != "video") AND (sizeof($siteinfo["images"]) > 0)){
 395                 /* Execute below code only if image is present in siteinfo */
 396
 397                 $total_images = 0;
 398                 $max_images = get_config('system','max_bookmark_images');
 399                 if($max_images === false)
 400                         $max_images = 2;
 401                 else
 402                         $max_images = intval($max_images);
 403
 404                 foreach ($siteinfo["images"] as $imagedata) {
 405                         if($textmode)
 406                                 $image .= '[img='.$imagedata["width"].'x'.$imagedata["height"].']'.$imagedata["src"].'[/img]' . "\n";
 407                         else
 408                                 $image .= '<img height="'.$imagedata["height"].'" width="'.$imagedata["width"].'" src="'.$imagedata["src"].'" alt="photo" /><br />';
 409                         $total_images ++;
 410                         if($max_images && $max_images >= $total_images)
 411                                 break;
 412                 }
 413         }
 414
 415         if(strlen($text)) {
 416                 if($textmode)
 417                         $text = '[quote]'.trim($text).'[/quote]';
 418                 else
 419                         $text = '<blockquote>'.htmlspecialchars(trim($text)).'</blockquote>';
 420         }
 421
 422         if($image)
 423                 $text = $br.$br.$image.$text;
 424         else
 425                 $text = $br.$text;
 426
 427         $title = str_replace(array("\r","\n"),array('',''),$title);
 428
 429         $result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
 430
 431         logger('parse_url: returns: ' . $result);
 432
 433         $sitedata .=  trim($result);
 434
 435         if (($siteinfo["type"] == "video") AND ($url != ""))
 436                 echo "[class=type-video]".$sitedata."[/class]";
 437         elseif (($siteinfo["type"] != "photo"))
 438                 echo "[class=type-link]".$sitedata."[/class]";
 439         else
 440                 echo "[class=type-photo]".$title.$br.$image."[/class]";
 441
 442         killme();
 443 }
 444 ?>