src/Util/ParseUrl.php

   1 <?php
   2 /**
   3  * @file src/Util/ParseUrl.php
   4  * @brief Get informations about a given URL
   5  */
   6 namespace Friendica\Util;
   7
   8 use DOMDocument;
   9 use DOMXPath;
  10 use Friendica\Content\OEmbed;
  11 use Friendica\Core\Hook;
  12 use Friendica\Core\Logger;
  13 use Friendica\Database\DBA;
  14
  15 /**
  16  * @brief Class with methods for extracting certain content from an url
  17  */
  18 class ParseUrl
  19 {
  20         /**
  21          * Maximum number of characters for the description
  22          */
  23         const MAX_DESC_COUNT = 250;
  24
  25         /**
  26          * Minimum number of characters for the description
  27          */
  28         const MIN_DESC_COUNT = 100;
  29
  30         /**
  31          * @brief Search for chached embeddable data of an url otherwise fetch it
  32          *
  33          * @param string $url         The url of the page which should be scraped
  34          * @param bool   $no_guessing If true the parse doens't search for
  35          *                            preview pictures
  36          * @param bool   $do_oembed   The false option is used by the function fetch_oembed()
  37          *                            to avoid endless loops
  38          *
  39          * @return array which contains needed data for embedding
  40          *    string 'url' => The url of the parsed page
  41          *    string 'type' => Content type
  42          *    string 'title' => The title of the content
  43          *    string 'text' => The description for the content
  44          *    string 'image' => A preview image of the content (only available
  45          *                if $no_geuessing = false
  46          *    array'images' = Array of preview pictures
  47          *    string 'keywords' => The tags which belong to the content
  48          *
  49          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  50          * @see   ParseUrl::getSiteinfo() for more information about scraping
  51          * embeddable content
  52          */
  53         public static function getSiteinfoCached($url, $no_guessing = false, $do_oembed = true)
  54         {
  55                 if ($url == "") {
  56                         return false;
  57                 }
  58
  59                 $parsed_url = DBA::selectFirst('parsed_url', ['content'],
  60                         ['url' => Strings::normaliseLink($url), 'guessing' => !$no_guessing, 'oembed' => $do_oembed]
  61                 );
  62                 if (!empty($parsed_url['content'])) {
  63                         $data = unserialize($parsed_url['content']);
  64                         return $data;
  65                 }
  66
  67                 $data = self::getSiteinfo($url, $no_guessing, $do_oembed);
  68
  69                 DBA::insert(
  70                         'parsed_url',
  71                         [
  72                                 'url' => Strings::normaliseLink($url), 'guessing' => !$no_guessing,
  73                                 'oembed' => $do_oembed, 'content' => serialize($data),
  74                                 'created' => DateTimeFormat::utcNow()
  75                         ],
  76                         true
  77                 );
  78
  79                 return $data;
  80         }
  81
  82         /**
  83          * @brief Parse a page for embeddable content information
  84          *
  85          * This method parses to url for meta data which can be used to embed
  86          * the content. If available it prioritizes Open Graph meta tags.
  87          * If this is not available it uses the twitter cards meta tags.
  88          * As fallback it uses standard html elements with meta informations
  89          * like \<title\>Awesome Title\</title\> or
  90          * \<meta name="description" content="An awesome description"\>
  91          *
  92          * @param string $url         The url of the page which should be scraped
  93          * @param bool   $no_guessing If true the parse doens't search for
  94          *                            preview pictures
  95          * @param bool   $do_oembed   The false option is used by the function fetch_oembed()
  96          *                            to avoid endless loops
  97          * @param int    $count       Internal counter to avoid endless loops
  98          *
  99          * @return array which contains needed data for embedding
 100          *    string 'url' => The url of the parsed page
 101          *    string 'type' => Content type
 102          *    string 'title' => The title of the content
 103          *    string 'text' => The description for the content
 104          *    string 'image' => A preview image of the content (only available
 105          *                if $no_geuessing = false
 106          *    array'images' = Array of preview pictures
 107          *    string 'keywords' => The tags which belong to the content
 108          *
 109          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 110          * @todo  https://developers.google.com/+/plugins/snippet/
 111          * @verbatim
 112          * <meta itemprop="name" content="Awesome title">
 113          * <meta itemprop="description" content="An awesome description">
 114          * <meta itemprop="image" content="http://maple.libertreeproject.org/images/tree-icon.png">
 115          *
 116          * <body itemscope itemtype="http://schema.org/Product">
 117          *   <h1 itemprop="name">Shiny Trinket</h1>
 118          *   <img itemprop="image" src="{image-url}" />
 119          *   <p itemprop="description">Shiny trinkets are shiny.</p>
 120          * </body>
 121          * @endverbatim
 122          */
 123         public static function getSiteinfo($url, $no_guessing = false, $do_oembed = true, $count = 1)
 124         {
 125                 $siteinfo = [];
 126
 127                 // Check if the URL does contain a scheme
 128                 $scheme = parse_url($url, PHP_URL_SCHEME);
 129
 130                 if ($scheme == '') {
 131                         $url = 'http://' . trim($url, '/');
 132                 }
 133
 134                 if ($count > 10) {
 135                         Logger::log('Endless loop detected for ' . $url, Logger::DEBUG);
 136                         return $siteinfo;
 137                 }
 138
 139                 $url = trim($url, "'");
 140                 $url = trim($url, '"');
 141
 142                 $url = Network::stripTrackingQueryParams($url);
 143
 144                 $siteinfo['url'] = $url;
 145                 $siteinfo['type'] = 'link';
 146
 147                 $curlResult = Network::curl($url);
 148                 if (!$curlResult->isSuccess()) {
 149                         return $siteinfo;
 150                 }
 151
 152                 // If the file is too large then exit
 153                 if (($curlResult->getInfo()['download_content_length'] ?? 0) > 1000000) {
 154                         return $siteinfo;
 155                 }
 156
 157                 // If it isn't a HTML file then exit
 158                 if (($curlResult->getContentType() != '') && !strstr(strtolower($curlResult->getContentType()), 'html')) {
 159                         return $siteinfo;
 160                 }
 161
 162                 $header = $curlResult->getHeader();
 163                 $body = $curlResult->getBody();
 164
 165                 if ($do_oembed) {
 166                         $oembed_data = OEmbed::fetchURL($url);
 167
 168                         if (!empty($oembed_data->type)) {
 169                                 if (!in_array($oembed_data->type, ['error', 'rich', ''])) {
 170                                         $siteinfo['type'] = $oembed_data->type;
 171                                 }
 172
 173                                 // See https://github.com/friendica/friendica/pull/5763#discussion_r217913178
 174                                 if ($siteinfo['type'] != 'photo') {
 175                                         if (isset($oembed_data->title)) {
 176                                                 $siteinfo['title'] = trim($oembed_data->title);
 177                                         }
 178                                         if (isset($oembed_data->description)) {
 179                                                 $siteinfo['text'] = trim($oembed_data->description);
 180                                         }
 181                                         if (isset($oembed_data->thumbnail_url)) {
 182                                                 $siteinfo['image'] = $oembed_data->thumbnail_url;
 183                                         }
 184                                 }
 185                         }
 186                 }
 187
 188                 // Fetch the first mentioned charset. Can be in body or header
 189                 $charset = '';
 190                 if (preg_match('/charset=(.*?)[\'"\s\n]/', $header, $matches)) {
 191                         $charset = trim(trim(trim(array_pop($matches)), ';,'));
 192                 }
 193
 194                 if ($charset && strtoupper($charset) != 'UTF-8') {
 195                         // See https://github.com/friendica/friendica/issues/5470#issuecomment-418351211
 196                         $charset = str_ireplace('latin-1', 'latin1', $charset);
 197
 198                         Logger::log('detected charset ' . $charset, Logger::DEBUG);
 199                         $body = iconv($charset, 'UTF-8//TRANSLIT', $body);
 200                 }
 201
 202                 $body = mb_convert_encoding($body, 'HTML-ENTITIES', 'UTF-8');
 203
 204                 $doc = new DOMDocument();
 205                 @$doc->loadHTML($body);
 206
 207                 XML::deleteNode($doc, 'style');
 208                 XML::deleteNode($doc, 'script');
 209                 XML::deleteNode($doc, 'option');
 210                 XML::deleteNode($doc, 'h1');
 211                 XML::deleteNode($doc, 'h2');
 212                 XML::deleteNode($doc, 'h3');
 213                 XML::deleteNode($doc, 'h4');
 214                 XML::deleteNode($doc, 'h5');
 215                 XML::deleteNode($doc, 'h6');
 216                 XML::deleteNode($doc, 'ol');
 217                 XML::deleteNode($doc, 'ul');
 218
 219                 $xpath = new DOMXPath($doc);
 220
 221                 $list = $xpath->query('//meta[@content]');
 222                 foreach ($list as $node) {
 223                         $meta_tag = [];
 224                         if ($node->attributes->length) {
 225                                 foreach ($node->attributes as $attribute) {
 226                                         $meta_tag[$attribute->name] = $attribute->value;
 227                                 }
 228                         }
 229
 230                         if (@$meta_tag['http-equiv'] == 'refresh') {
 231                                 $path = $meta_tag['content'];
 232                                 $pathinfo = explode(';', $path);
 233                                 $content = '';
 234                                 foreach ($pathinfo as $value) {
 235                                         if (substr(strtolower($value), 0, 4) == 'url=') {
 236                                                 $content = substr($value, 4);
 237                                         }
 238                                 }
 239                                 if ($content != '') {
 240                                         $siteinfo = self::getSiteinfo($content, $no_guessing, $do_oembed, ++$count);
 241                                         return $siteinfo;
 242                                 }
 243                         }
 244                 }
 245
 246                 $list = $xpath->query('//title');
 247                 if ($list->length > 0) {
 248                         $siteinfo['title'] = trim($list->item(0)->nodeValue);
 249                 }
 250
 251                 $list = $xpath->query('//meta[@name]');
 252                 foreach ($list as $node) {
 253                         $meta_tag = [];
 254                         if ($node->attributes->length) {
 255                                 foreach ($node->attributes as $attribute) {
 256                                         $meta_tag[$attribute->name] = $attribute->value;
 257                                 }
 258                         }
 259
 260                         if (empty($meta_tag['content'])) {
 261                                 continue;
 262                         }
 263
 264                         $meta_tag['content'] = trim(html_entity_decode($meta_tag['content'], ENT_QUOTES, 'UTF-8'));
 265
 266                         switch (strtolower($meta_tag['name'])) {
 267                                 case 'fulltitle':
 268                                         $siteinfo['title'] = trim($meta_tag['content']);
 269                                         break;
 270                                 case 'description':
 271                                         $siteinfo['text'] = trim($meta_tag['content']);
 272                                         break;
 273                                 case 'thumbnail':
 274                                         $siteinfo['image'] = $meta_tag['content'];
 275                                         break;
 276                                 case 'twitter:image':
 277                                         $siteinfo['image'] = $meta_tag['content'];
 278                                         break;
 279                                 case 'twitter:image:src':
 280                                         $siteinfo['image'] = $meta_tag['content'];
 281                                         break;
 282                                 case 'twitter:card':
 283                                         // Detect photo pages
 284                                         if ($meta_tag['content'] == 'summary_large_image') {
 285                                                 $siteinfo['type'] = 'photo';
 286                                         }
 287                                         break;
 288                                 case 'twitter:description':
 289                                         $siteinfo['text'] = trim($meta_tag['content']);
 290                                         break;
 291                                 case 'twitter:title':
 292                                         $siteinfo['title'] = trim($meta_tag['content']);
 293                                         break;
 294                                 case 'dc.title':
 295                                         $siteinfo['title'] = trim($meta_tag['content']);
 296                                         break;
 297                                 case 'dc.description':
 298                                         $siteinfo['text'] = trim($meta_tag['content']);
 299                                         break;
 300                                 case 'keywords':
 301                                         $keywords = explode(',', $meta_tag['content']);
 302                                         break;
 303                                 case 'news_keywords':
 304                                         $keywords = explode(',', $meta_tag['content']);
 305                                         break;
 306                         }
 307                 }
 308
 309                 if (isset($keywords)) {
 310                         $siteinfo['keywords'] = [];
 311                         foreach ($keywords as $keyword) {
 312                                 if (!in_array(trim($keyword), $siteinfo['keywords'])) {
 313                                         $siteinfo['keywords'][] = trim($keyword);
 314                                 }
 315                         }
 316                 }
 317
 318                 $list = $xpath->query('//meta[@property]');
 319                 foreach ($list as $node) {
 320                         $meta_tag = [];
 321                         if ($node->attributes->length) {
 322                                 foreach ($node->attributes as $attribute) {
 323                                         $meta_tag[$attribute->name] = $attribute->value;
 324                                 }
 325                         }
 326
 327                         if (!empty($meta_tag['content'])) {
 328                                 $meta_tag['content'] = trim(html_entity_decode($meta_tag['content'], ENT_QUOTES, 'UTF-8'));
 329
 330                                 switch (strtolower($meta_tag['property'])) {
 331                                         case 'og:image':
 332                                                 $siteinfo['image'] = $meta_tag['content'];
 333                                                 break;
 334                                         case 'og:title':
 335                                                 $siteinfo['title'] = trim($meta_tag['content']);
 336                                                 break;
 337                                         case 'og:description':
 338                                                 $siteinfo['text'] = trim($meta_tag['content']);
 339                                                 break;
 340                                 }
 341                         }
 342                 }
 343
 344                 // Prevent to have a photo type without an image
 345                 if ((empty($siteinfo['image']) || !empty($siteinfo['text'])) && ($siteinfo['type'] == 'photo')) {
 346                         $siteinfo['type'] = 'link';
 347                 }
 348
 349                 if (!empty($siteinfo['image'])) {
 350                         $src = self::completeUrl($siteinfo['image'], $url);
 351
 352                         unset($siteinfo['image']);
 353
 354                         $photodata = Images::getInfoFromURLCached($src);
 355
 356                         if (($photodata) && ($photodata[0] > 10) && ($photodata[1] > 10)) {
 357                                 $siteinfo['images'][] = ['src' => $src,
 358                                         'width' => $photodata[0],
 359                                         'height' => $photodata[1]];
 360                         }
 361                 }
 362
 363                 if (!empty($siteinfo['text']) && mb_strlen($siteinfo['text']) > self::MAX_DESC_COUNT) {
 364                         $siteinfo['text'] = mb_substr($siteinfo['text'], 0, self::MAX_DESC_COUNT) . '…';
 365                         $pos = mb_strrpos($siteinfo['text'], '.');
 366                         if ($pos > self::MIN_DESC_COUNT) {
 367                                 $siteinfo['text'] = mb_substr($siteinfo['text'], 0, $pos + 1);
 368                         }
 369                 }
 370
 371                 Logger::info('Siteinfo fetched', ['url' => $url, 'siteinfo' => $siteinfo]);
 372
 373                 Hook::callAll('getsiteinfo', $siteinfo);
 374
 375                 return $siteinfo;
 376         }
 377
 378         /**
 379          * @brief Convert tags from CSV to an array
 380          *
 381          * @param string $string Tags
 382          * @return array with formatted Hashtags
 383          */
 384         public static function convertTagsToArray($string)
 385         {
 386                 $arr_tags = str_getcsv($string);
 387                 if (count($arr_tags)) {
 388                         // add the # sign to every tag
 389                         array_walk($arr_tags, ["self", "arrAddHashes"]);
 390
 391                         return $arr_tags;
 392                 }
 393         }
 394
 395         /**
 396          * @brief Add a hasht sign to a string
 397          *
 398          *  This method is used as callback function
 399          *
 400          * @param string $tag The pure tag name
 401          * @param int    $k   Counter for internal use
 402          * @return void
 403          */
 404         private static function arrAddHashes(&$tag, $k)
 405         {
 406                 $tag = "#" . $tag;
 407         }
 408
 409         /**
 410          * @brief Add a scheme to an url
 411          *
 412          * The src attribute of some html elements (e.g. images)
 413          * can miss the scheme so we need to add the correct
 414          * scheme
 415          *
 416          * @param string $url    The url which possibly does have
 417          *                       a missing scheme (a link to an image)
 418          * @param string $scheme The url with a correct scheme
 419          *                       (e.g. the url from the webpage which does contain the image)
 420          *
 421          * @return string The url with a scheme
 422          */
 423         private static function completeUrl($url, $scheme)
 424         {
 425                 $urlarr = parse_url($url);
 426
 427                 // If the url does allready have an scheme
 428                 // we can stop the process here
 429                 if (isset($urlarr["scheme"])) {
 430                         return($url);
 431                 }
 432
 433                 $schemearr = parse_url($scheme);
 434
 435                 $complete = $schemearr["scheme"]."://".$schemearr["host"];
 436
 437                 if (!empty($schemearr["port"])) {
 438                         $complete .= ":".$schemearr["port"];
 439                 }
 440
 441                 if (!empty($urlarr["path"])) {
 442                         if (strpos($urlarr["path"], "/") !== 0) {
 443                                 $complete .= "/";
 444                         }
 445
 446                         $complete .= $urlarr["path"];
 447                 }
 448
 449                 if (!empty($urlarr["query"])) {
 450                         $complete .= "?".$urlarr["query"];
 451                 }
 452
 453                 if (!empty($urlarr["fragment"])) {
 454                         $complete .= "#".$urlarr["fragment"];
 455                 }
 456
 457                 return($complete);
 458         }
 459 }