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) {
  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($ch,CURLOPT_USERAGENT,' Mozilla/5.0 (Macintosh; Intel Mac OS X 10.8; rv:24.0) Gecko/20100101 Firefox/24.0');
  70         curl_setopt($ch,CURLOPT_USERAGENT, "Mozilla/5.0 (compatible; Friendica)");
  71
  72         $header = curl_exec($ch);
  73         $curl_info = @curl_getinfo($ch);
  74         $http_code = $curl_info['http_code'];
  75         curl_close($ch);
  76
  77         if ((($curl_info['http_code'] == "301") OR ($curl_info['http_code'] == "302") OR ($curl_info['http_code'] == "303") OR ($curl_info['http_code'] == "307"))
  78                 AND (($curl_info['redirect_url'] != "") OR ($curl_info['location'] != ""))) {
  79                 if ($curl_info['redirect_url'] != "")
  80                         $siteinfo = parseurl_getsiteinfo($curl_info['redirect_url']);
  81                 else
  82                         $siteinfo = parseurl_getsiteinfo($curl_info['location']);
  83                 return($siteinfo);
  84         }
  85
  86         if ($do_oembed) {
  87                 require_once("include/oembed.php");
  88
  89                 $oembed_data = oembed_fetch_url($url);
  90
  91                 if ($oembed_data->type != "error")
  92                         $siteinfo["type"] = $oembed_data->type;
  93         }
  94
  95         // Fetch the first mentioned charset. Can be in body or header
  96         $charset = "";
  97         if (preg_match('/charset=(.*?)['."'".'"\s\n]/', $header, $matches))
  98                 $charset = trim(array_pop($matches));
  99
 100         if ($charset == "")
 101                 $charset = "utf-8";
 102
 103         $pos = strpos($header, "\r\n\r\n");
 104
 105         if ($pos)
 106                 $body = trim(substr($header, $pos));
 107         else
 108                 $body = $header;
 109
 110         $body = mb_convert_encoding($body, "UTF-8", $charset);
 111         $body = mb_convert_encoding($body, 'HTML-ENTITIES', "UTF-8");
 112
 113         $doc = new DOMDocument();
 114         @$doc->loadHTML($body);
 115
 116         deletenode($doc, 'style');
 117         deletenode($doc, 'script');
 118         deletenode($doc, 'option');
 119         deletenode($doc, 'h1');
 120         deletenode($doc, 'h2');
 121         deletenode($doc, 'h3');
 122         deletenode($doc, 'h4');
 123         deletenode($doc, 'h5');
 124         deletenode($doc, 'h6');
 125         deletenode($doc, 'ol');
 126         deletenode($doc, 'ul');
 127
 128         $xpath = new DomXPath($doc);
 129
 130         $list = $xpath->query("//meta[@content]");
 131         foreach ($list as $node) {
 132                 $attr = array();
 133                 if ($node->attributes->length)
 134                         foreach ($node->attributes as $attribute)
 135                                 $attr[$attribute->name] = $attribute->value;
 136
 137                 if (@$attr["http-equiv"] == 'refresh') {
 138                         $path = $attr["content"];
 139                         $pathinfo = explode(";", $path);
 140                         $content = "";
 141                         foreach ($pathinfo AS $value) {
 142                                 if (substr(strtolower($value), 0, 4) == "url=")
 143                                         $content = substr($value, 4);
 144                         }
 145                         if ($content != "") {
 146                                 $siteinfo = parseurl_getsiteinfo($content);
 147                                 return($siteinfo);
 148                         }
 149                 }
 150         }
 151
 152         //$list = $xpath->query("head/title");
 153         $list = $xpath->query("//title");
 154         foreach ($list as $node)
 155                 $siteinfo["title"] =  html_entity_decode($node->nodeValue, ENT_QUOTES, "UTF-8");
 156
 157         //$list = $xpath->query("head/meta[@name]");
 158         $list = $xpath->query("//meta[@name]");
 159         foreach ($list as $node) {
 160                 $attr = array();
 161                 if ($node->attributes->length)
 162                         foreach ($node->attributes as $attribute)
 163                                 $attr[$attribute->name] = $attribute->value;
 164
 165                 $attr["content"] = trim(html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8"));
 166
 167                 if ($attr["content"] != "")
 168                         switch (strtolower($attr["name"])) {
 169                                 case "fulltitle":
 170                                         $siteinfo["title"] = $attr["content"];
 171                                         break;
 172                                 case "description":
 173                                         $siteinfo["text"] = $attr["content"];
 174                                         break;
 175                                 case "twitter:image":
 176                                         $siteinfo["image"] = $attr["content"];
 177                                         break;
 178                                 case "twitter:card":
 179                                         if (($siteinfo["type"] == "") OR ($attr["content"] == "photo"))
 180                                                 $siteinfo["type"] = $attr["content"];
 181                                         break;
 182                                 case "twitter:description":
 183                                         $siteinfo["text"] = $attr["content"];
 184                                         break;
 185                                 case "twitter:title":
 186                                         $siteinfo["title"] = $attr["content"];
 187                                         break;
 188                                 case "dc.title":
 189                                         $siteinfo["title"] = $attr["content"];
 190                                         break;
 191                                 case "dc.description":
 192                                         $siteinfo["text"] = $attr["content"];
 193                                         break;
 194                         }
 195                 if ($siteinfo["type"] == "summary")
 196                         $siteinfo["type"] = "link";
 197         }
 198
 199         //$list = $xpath->query("head/meta[@property]");
 200         $list = $xpath->query("//meta[@property]");
 201         foreach ($list as $node) {
 202                 $attr = array();
 203                 if ($node->attributes->length)
 204                         foreach ($node->attributes as $attribute)
 205                                 $attr[$attribute->name] = $attribute->value;
 206
 207                 $attr["content"] = trim(html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8"));
 208
 209                 if ($attr["content"] != "")
 210                         switch (strtolower($attr["property"])) {
 211                                 case "og:image":
 212                                         $siteinfo["image"] = $attr["content"];
 213                                         break;
 214                                 case "og:title":
 215                                         $siteinfo["title"] = $attr["content"];
 216                                         break;
 217                                 case "og:description":
 218                                         $siteinfo["text"] = $attr["content"];
 219                                         break;
 220                         }
 221         }
 222
 223         if (isset($oembed_data) AND ($oembed_data->type == "link")) {
 224                 if (isset($oembed_data->title) AND (trim($oembed_data->title) != ""))
 225                         $siteinfo["title"] = $oembed_data->title;
 226                 if (isset($oembed_data->description) AND (trim($oembed_data->description) != ""))
 227                         $siteinfo["text"] = trim($oembed_data->description);
 228                 if (isset($oembed_data->thumbnail_url) AND (trim($oembed_data->thumbnail_url) != ""))
 229                         $siteinfo["image"] = $oembed_data->thumbnail_url;
 230         }
 231
 232         if ((@$siteinfo["image"] == "") AND !$no_guessing) {
 233             $list = $xpath->query("//img[@src]");
 234             foreach ($list as $node) {
 235                 $attr = array();
 236                 if ($node->attributes->length)
 237                     foreach ($node->attributes as $attribute)
 238                         $attr[$attribute->name] = $attribute->value;
 239
 240                         $src = completeurl($attr["src"], $url);
 241                         $photodata = @getimagesize($src);
 242
 243                         if (($photodata) && ($photodata[0] > 150) and ($photodata[1] > 150)) {
 244                                 if ($photodata[0] > 300) {
 245                                         $photodata[1] = round($photodata[1] * (300 / $photodata[0]));
 246                                         $photodata[0] = 300;
 247                                 }
 248                                 if ($photodata[1] > 300) {
 249                                         $photodata[0] = round($photodata[0] * (300 / $photodata[1]));
 250                                         $photodata[1] = 300;
 251                                 }
 252                                 $siteinfo["images"][] = array("src"=>$src,
 253                                                                 "width"=>$photodata[0],
 254                                                                 "height"=>$photodata[1]);
 255                         }
 256
 257                 }
 258     } else {
 259                 $src = completeurl($siteinfo["image"], $url);
 260
 261                 unset($siteinfo["image"]);
 262
 263                 $photodata = @getimagesize($src);
 264
 265                 if (($photodata) && ($photodata[0] > 10) and ($photodata[1] > 10))
 266                         $siteinfo["images"][] = array("src"=>$src,
 267                                                         "width"=>$photodata[0],
 268                                                         "height"=>$photodata[1]);
 269         }
 270
 271         if ((@$siteinfo["text"] == "") AND (@$siteinfo["title"] != "") AND !$no_guessing) {
 272                 $text = "";
 273
 274                 $list = $xpath->query("//div[@class='article']");
 275                 foreach ($list as $node)
 276                         if (strlen($node->nodeValue) > 40)
 277                                 $text .= " ".trim($node->nodeValue);
 278
 279                 if ($text == "") {
 280                         $list = $xpath->query("//div[@class='content']");
 281                         foreach ($list as $node)
 282                                 if (strlen($node->nodeValue) > 40)
 283                                         $text .= " ".trim($node->nodeValue);
 284                 }
 285
 286                 // If none text was found then take the paragraph content
 287                 if ($text == "") {
 288                         $list = $xpath->query("//p");
 289                         foreach ($list as $node)
 290                                 if (strlen($node->nodeValue) > 40)
 291                                         $text .= " ".trim($node->nodeValue);
 292                 }
 293
 294                 if ($text != "") {
 295                         $text = trim(str_replace(array("\n", "\r"), array(" ", " "), $text));
 296
 297                         while (strpos($text, "  "))
 298                                 $text = trim(str_replace("  ", " ", $text));
 299
 300                         $siteinfo["text"] = trim(html_entity_decode(substr($text,0,350), ENT_QUOTES, "UTF-8").'...');
 301                 }
 302         }
 303
 304         logger("parseurl_getsiteinfo: Siteinfo for ".$url." ".print_r($siteinfo, true), LOGGER_DEBUG);
 305
 306         return($siteinfo);
 307 }
 308
 309 function arr_add_hashes(&$item,$k) {
 310         $item = '#' . $item;
 311 }
 312
 313 function parse_url_content(&$a) {
 314
 315         $text = null;
 316         $str_tags = '';
 317
 318         $textmode = false;
 319
 320         if(local_user() && (! feature_enabled(local_user(),'richtext')))
 321                 $textmode = true;
 322
 323         //if($textmode)
 324         $br = (($textmode) ? "\n" : '<br />');
 325
 326         if(x($_GET,'binurl'))
 327                 $url = trim(hex2bin($_GET['binurl']));
 328         else
 329                 $url = trim($_GET['url']);
 330
 331         if($_GET['title'])
 332                 $title = strip_tags(trim($_GET['title']));
 333
 334         if($_GET['description'])
 335                 $text = strip_tags(trim($_GET['description']));
 336
 337         if($_GET['tags']) {
 338                 $arr_tags = str_getcsv($_GET['tags']);
 339                 if(count($arr_tags)) {
 340                         array_walk($arr_tags,'arr_add_hashes');
 341                         $str_tags = $br . implode(' ',$arr_tags) . $br;
 342                 }
 343         }
 344
 345         logger('parse_url: ' . $url);
 346
 347         if($textmode)
 348                 $template = '[bookmark=%s]%s[/bookmark]%s';
 349         else
 350                 $template = "<a class=\"bookmark\" href=\"%s\" >%s</a>%s";
 351
 352         $arr = array('url' => $url, 'text' => '');
 353
 354         call_hooks('parse_link', $arr);
 355
 356         if(strlen($arr['text'])) {
 357                 echo $arr['text'];
 358                 killme();
 359         }
 360
 361
 362         if($url && $title && $text) {
 363
 364                 $title = str_replace(array("\r","\n"),array('',''),$title);
 365
 366                 if($textmode)
 367                         $text = '[quote]' . trim($text) . '[/quote]' . $br;
 368                 else {
 369                         $text = '<blockquote>' . htmlspecialchars(trim($text)) . '</blockquote><br />';
 370                         $title = htmlspecialchars($title);
 371                 }
 372
 373                 $result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
 374
 375                 logger('parse_url (unparsed): returns: ' . $result);
 376
 377                 echo $result;
 378                 killme();
 379         }
 380
 381         $siteinfo = parseurl_getsiteinfo($url);
 382
 383         $url= $siteinfo["url"];
 384
 385         $sitedata = "";
 386
 387         if($siteinfo["title"] == "") {
 388                 $sitedata .= sprintf($template,$url,$url,'') . $str_tags;
 389                 killme();
 390         } else {
 391                 $text = $siteinfo["text"];
 392                 $title = $siteinfo["title"];
 393         }
 394
 395         $image = "";
 396
 397         if (($siteinfo["type"] != "video") AND (sizeof($siteinfo["images"]) > 0)){
 398                 /* Execute below code only if image is present in siteinfo */
 399
 400                 $total_images = 0;
 401                 $max_images = get_config('system','max_bookmark_images');
 402                 if($max_images === false)
 403                         $max_images = 2;
 404                 else
 405                         $max_images = intval($max_images);
 406
 407                 foreach ($siteinfo["images"] as $imagedata) {
 408                         if($textmode)
 409                                 $image .= '[img='.$imagedata["width"].'x'.$imagedata["height"].']'.$imagedata["src"].'[/img]' . "\n";
 410                         else
 411                                 $image .= '<img height="'.$imagedata["height"].'" width="'.$imagedata["width"].'" src="'.$imagedata["src"].'" alt="photo" /><br />';
 412                         $total_images ++;
 413                         if($max_images && $max_images >= $total_images)
 414                                 break;
 415                 }
 416         }
 417
 418         if(strlen($text)) {
 419                 if($textmode)
 420                         $text = '[quote]'.trim($text).'[/quote]';
 421                 else
 422                         $text = '<blockquote>'.htmlspecialchars(trim($text)).'</blockquote>';
 423         }
 424
 425         if($image)
 426                 $text = $br.$br.$image.$text;
 427         else
 428                 $text = $br.$text;
 429
 430         $title = str_replace(array("\r","\n"),array('',''),$title);
 431
 432         $result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
 433
 434         logger('parse_url: returns: ' . $result);
 435
 436         $sitedata .=  trim($result);
 437
 438         if (($siteinfo["type"] == "video") AND ($url != ""))
 439                 echo "[class=type-video]".$sitedata."[/class]";
 440         elseif (($siteinfo["type"] != "photo"))
 441                 echo "[class=type-link]".$sitedata."[/class]";
 442         else
 443                 echo "[class=type-photo]".$title.$br.$image."[/class]";
 444
 445         killme();
 446 }
 447 ?>