src/Protocol/Feed.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2020, Friendica
   4  *
   5  * @license GNU AGPL version 3 or any later version
   6  *
   7  * This program is free software: you can redistribute it and/or modify
   8  * it under the terms of the GNU Affero General Public License as
   9  * published by the Free Software Foundation, either version 3 of the
  10  * License, or (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU Affero General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Affero General Public License
  18  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  19  *
  20  */
  21
  22 namespace Friendica\Protocol;
  23
  24 use DOMDocument;
  25 use DOMXPath;
  26 use Friendica\Content\PageInfo;
  27 use Friendica\Content\Text\BBCode;
  28 use Friendica\Content\Text\HTML;
  29 use Friendica\Core\Cache\Duration;
  30 use Friendica\Core\Logger;
  31 use Friendica\Core\Protocol;
  32 use Friendica\Database\DBA;
  33 use Friendica\DI;
  34 use Friendica\Model\Contact;
  35 use Friendica\Model\Item;
  36 use Friendica\Model\Post;
  37 use Friendica\Model\Tag;
  38 use Friendica\Model\User;
  39 use Friendica\Util\DateTimeFormat;
  40 use Friendica\Util\Network;
  41 use Friendica\Util\ParseUrl;
  42 use Friendica\Util\Strings;
  43 use Friendica\Util\XML;
  44
  45 /**
  46  * This class contain functions to import feeds (RSS/RDF/Atom)
  47  */
  48 class Feed
  49 {
  50         /**
  51          * consume - process atom feed and update anything/everything we might need to update
  52          *
  53          * $xml = the (atom) feed to consume - RSS isn't as fully supported but may work for simple feeds.
  54          *
  55          * $importer = the contact_record (joined to user_record) of the local user who owns this relationship.
  56          *             It is this person's stuff that is going to be updated.
  57          * $contact =  the person who is sending us stuff. If not set, we MAY be processing a "follow" activity
  58          *             from an external network and MAY create an appropriate contact record. Otherwise, we MUST
  59          *             have a contact record.
  60          * $hub = should we find a hub declation in the feed, pass it back to our calling process, who might (or
  61          *        might not) try and subscribe to it.
  62          * $datedir sorts in reverse order
  63          * $pass - by default ($pass = 0) we cannot guarantee that a parent item has been
  64          *      imported prior to its children being seen in the stream unless we are certain
  65          *      of how the feed is arranged/ordered.
  66          * With $pass = 1, we only pull parent items out of the stream.
  67          * With $pass = 2, we only pull children (comments/likes).
  68          *
  69          * So running this twice, first with pass 1 and then with pass 2 will do the right
  70          * thing regardless of feed ordering. This won't be adequate in a fully-threaded
  71          * model where comments can have sub-threads. That would require some massive sorting
  72          * to get all the feed items into a mostly linear ordering, and might still require
  73          * recursion.
  74          *
  75          * @param       $xml
  76          * @param array $importer
  77          * @param array $contact
  78          * @param       $hub
  79          * @throws ImagickException
  80          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  81          */
  82         public static function consume($xml, array $importer, array $contact, &$hub)
  83         {
  84                 if ($contact['network'] === Protocol::OSTATUS) {
  85                         Logger::info('Consume OStatus messages');
  86                         OStatus::import($xml, $importer, $contact, $hub);
  87
  88                         return;
  89                 }
  90
  91                 if ($contact['network'] === Protocol::FEED) {
  92                         Logger::info('Consume feeds');
  93                         self::import($xml, $importer, $contact);
  94
  95                         return;
  96                 }
  97
  98                 if ($contact['network'] === Protocol::DFRN) {
  99                         Logger::info('Consume DFRN messages');
 100                         $dfrn_importer = DFRN::getImporter($contact['id'], $importer['uid']);
 101                         if (!empty($dfrn_importer)) {
 102                                 Logger::info('Now import the DFRN feed');
 103                                 DFRN::import($xml, $dfrn_importer, true);
 104                                 return;
 105                         }
 106                 }
 107         }
 108
 109         /**
 110          * Read a RSS/RDF/Atom feed and create an item entry for it
 111          *
 112          * @param string $xml      The feed data
 113          * @param array  $importer The user record of the importer
 114          * @param array  $contact  The contact record of the feed
 115          *
 116          * @return array Returns the header and the first item in dry run mode
 117          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 118          */
 119         public static function import($xml, array $importer = [], array $contact = [])
 120         {
 121                 $dryRun = empty($importer) && empty($contact);
 122
 123                 if ($dryRun) {
 124                         Logger::info("Test Atom/RSS feed");
 125                 } else {
 126                         Logger::info("Import Atom/RSS feed '" . $contact["name"] . "' (Contact " . $contact["id"] . ") for user " . $importer["uid"]);
 127                 }
 128
 129                 if (empty($xml)) {
 130                         Logger::info('XML is empty.');
 131                         return [];
 132                 }
 133
 134                 if (!empty($contact['poll'])) {
 135                         $basepath = $contact['poll'];
 136                 } elseif (!empty($contact['url'])) {
 137                         $basepath = $contact['url'];
 138                 } else {
 139                         $basepath = '';
 140                 }
 141
 142                 $doc = new DOMDocument();
 143                 @$doc->loadXML(trim($xml));
 144                 $xpath = new DOMXPath($doc);
 145                 $xpath->registerNamespace('atom', ActivityNamespace::ATOM1);
 146                 $xpath->registerNamespace('dc', "http://purl.org/dc/elements/1.1/");
 147                 $xpath->registerNamespace('content', "http://purl.org/rss/1.0/modules/content/");
 148                 $xpath->registerNamespace('rdf', "http://www.w3.org/1999/02/22-rdf-syntax-ns#");
 149                 $xpath->registerNamespace('rss', "http://purl.org/rss/1.0/");
 150                 $xpath->registerNamespace('media', "http://search.yahoo.com/mrss/");
 151                 $xpath->registerNamespace('poco', ActivityNamespace::POCO);
 152
 153                 $author = [];
 154                 $entries = null;
 155
 156                 // Is it RDF?
 157                 if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
 158                         $author["author-link"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:link/text()');
 159                         $author["author-name"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:title/text()');
 160
 161                         if (empty($author["author-name"])) {
 162                                 $author["author-name"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:description/text()');
 163                         }
 164                         $entries = $xpath->query('/rdf:RDF/rss:item');
 165                 }
 166
 167                 // Is it Atom?
 168                 if ($xpath->query('/atom:feed')->length > 0) {
 169                         $alternate = XML::getFirstAttributes($xpath, "atom:link[@rel='alternate']");
 170                         if (is_object($alternate)) {
 171                                 foreach ($alternate AS $attribute) {
 172                                         if ($attribute->name == "href") {
 173                                                 $author["author-link"] = $attribute->textContent;
 174                                         }
 175                                 }
 176                         }
 177
 178                         if (empty($author["author-link"])) {
 179                                 $self = XML::getFirstAttributes($xpath, "atom:link[@rel='self']");
 180                                 if (is_object($self)) {
 181                                         foreach ($self AS $attribute) {
 182                                                 if ($attribute->name == "href") {
 183                                                         $author["author-link"] = $attribute->textContent;
 184                                                 }
 185                                         }
 186                                 }
 187                         }
 188
 189                         if (empty($author["author-link"])) {
 190                                 $author["author-link"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:id/text()');
 191                         }
 192                         $author["author-avatar"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:logo/text()');
 193
 194                         $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:title/text()');
 195
 196                         if (empty($author["author-name"])) {
 197                                 $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:subtitle/text()');
 198                         }
 199
 200                         if (empty($author["author-name"])) {
 201                                 $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:name/text()');
 202                         }
 203
 204                         $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:displayName/text()');
 205                         if ($value != "") {
 206                                 $author["author-name"] = $value;
 207                         }
 208
 209                         if ($dryRun) {
 210                                 $author["author-id"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:id/text()');
 211
 212                                 // See https://tools.ietf.org/html/rfc4287#section-3.2.2
 213                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/atom:uri/text()');
 214                                 if ($value != "") {
 215                                         $author["author-link"] = $value;
 216                                 }
 217
 218                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:preferredUsername/text()');
 219                                 if ($value != "") {
 220                                         $author["author-nick"] = $value;
 221                                 }
 222
 223                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:address/poco:formatted/text()');
 224                                 if ($value != "") {
 225                                         $author["author-location"] = $value;
 226                                 }
 227
 228                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:note/text()');
 229                                 if ($value != "") {
 230                                         $author["author-about"] = $value;
 231                                 }
 232
 233                                 $avatar = XML::getFirstAttributes($xpath, "atom:author/atom:link[@rel='avatar']");
 234                                 if (is_object($avatar)) {
 235                                         foreach ($avatar AS $attribute) {
 236                                                 if ($attribute->name == "href") {
 237                                                         $author["author-avatar"] = $attribute->textContent;
 238                                                 }
 239                                         }
 240                                 }
 241                         }
 242
 243                         $author["edited"] = $author["created"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:updated/text()');
 244
 245                         $author["app"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:generator/text()');
 246
 247                         $entries = $xpath->query('/atom:feed/atom:entry');
 248                 }
 249
 250                 // Is it RSS?
 251                 if ($xpath->query('/rss/channel')->length > 0) {
 252                         $author["author-link"] = XML::getFirstNodeValue($xpath, '/rss/channel/link/text()');
 253
 254                         $author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/title/text()');
 255                         $author["author-avatar"] = XML::getFirstNodeValue($xpath, '/rss/channel/image/url/text()');
 256
 257                         if (empty($author["author-name"])) {
 258                                 $author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/copyright/text()');
 259                         }
 260
 261                         if (empty($author["author-name"])) {
 262                                 $author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/description/text()');
 263                         }
 264
 265                         $author["edited"] = $author["created"] = XML::getFirstNodeValue($xpath, '/rss/channel/pubDate/text()');
 266
 267                         $author["app"] = XML::getFirstNodeValue($xpath, '/rss/channel/generator/text()');
 268
 269                         $entries = $xpath->query('/rss/channel/item');
 270                 }
 271
 272                 if (!$dryRun) {
 273                         $author["author-link"] = $contact["url"];
 274
 275                         if (empty($author["author-name"])) {
 276                                 $author["author-name"] = $contact["name"];
 277                         }
 278
 279                         $author["author-avatar"] = $contact["thumb"];
 280
 281                         $author["owner-link"] = $contact["url"];
 282                         $author["owner-name"] = $contact["name"];
 283                         $author["owner-avatar"] = $contact["thumb"];
 284                 }
 285
 286                 $header = [];
 287                 $header["uid"] = $importer["uid"] ?? 0;
 288                 $header["network"] = Protocol::FEED;
 289                 $header["wall"] = 0;
 290                 $header["origin"] = 0;
 291                 $header["gravity"] = GRAVITY_PARENT;
 292                 $header["private"] = Item::PUBLIC;
 293                 $header["verb"] = Activity::POST;
 294                 $header["object-type"] = Activity\ObjectType::NOTE;
 295
 296                 $header["contact-id"] = $contact["id"] ?? 0;
 297
 298                 if (!is_object($entries)) {
 299                         Logger::info("There are no entries in this feed.");
 300                         return [];
 301                 }
 302
 303                 $items = [];
 304                 $creation_dates = [];
 305
 306                 // Limit the number of items that are about to be fetched
 307                 $total_items = ($entries->length - 1);
 308                 $max_items = DI::config()->get('system', 'max_feed_items');
 309                 if (($max_items > 0) && ($total_items > $max_items)) {
 310                         $total_items = $max_items;
 311                 }
 312
 313                 // Importing older entries first
 314                 for ($i = $total_items; $i >= 0; --$i) {
 315                         $entry = $entries->item($i);
 316
 317                         $item = array_merge($header, $author);
 318
 319                         $alternate = XML::getFirstAttributes($xpath, "atom:link[@rel='alternate']", $entry);
 320                         if (!is_object($alternate)) {
 321                                 $alternate = XML::getFirstAttributes($xpath, "atom:link", $entry);
 322                         }
 323                         if (is_object($alternate)) {
 324                                 foreach ($alternate AS $attribute) {
 325                                         if ($attribute->name == "href") {
 326                                                 $item["plink"] = $attribute->textContent;
 327                                         }
 328                                 }
 329                         }
 330
 331                         if (empty($item["plink"])) {
 332                                 $item["plink"] = XML::getFirstNodeValue($xpath, 'link/text()', $entry);
 333                         }
 334
 335                         if (empty($item["plink"])) {
 336                                 $item["plink"] = XML::getFirstNodeValue($xpath, 'rss:link/text()', $entry);
 337                         }
 338
 339                         $item["uri"] = XML::getFirstNodeValue($xpath, 'atom:id/text()', $entry);
 340
 341                         if (empty($item["uri"])) {
 342                                 $item["uri"] = XML::getFirstNodeValue($xpath, 'guid/text()', $entry);
 343                         }
 344
 345                         if (empty($item["uri"])) {
 346                                 $item["uri"] = $item["plink"];
 347                         }
 348
 349                         // Add the base path if missing
 350                         $item["uri"] = Network::addBasePath($item["uri"], $basepath);
 351                         $item["plink"] = Network::addBasePath($item["plink"], $basepath);
 352
 353                         $orig_plink = $item["plink"];
 354
 355                         $item["plink"] = DI::httpRequest()->finalUrl($item["plink"]);
 356
 357                         $item["title"] = XML::getFirstNodeValue($xpath, 'atom:title/text()', $entry);
 358
 359                         if (empty($item["title"])) {
 360                                 $item["title"] = XML::getFirstNodeValue($xpath, 'title/text()', $entry);
 361                         }
 362                         if (empty($item["title"])) {
 363                                 $item["title"] = XML::getFirstNodeValue($xpath, 'rss:title/text()', $entry);
 364                         }
 365
 366                         $item["title"] = html_entity_decode($item["title"], ENT_QUOTES, 'UTF-8');
 367
 368                         $published = XML::getFirstNodeValue($xpath, 'atom:published/text()', $entry);
 369
 370                         if (empty($published)) {
 371                                 $published = XML::getFirstNodeValue($xpath, 'pubDate/text()', $entry);
 372                         }
 373
 374                         if (empty($published)) {
 375                                 $published = XML::getFirstNodeValue($xpath, 'dc:date/text()', $entry);
 376                         }
 377
 378                         $updated = XML::getFirstNodeValue($xpath, 'atom:updated/text()', $entry);
 379
 380                         if (empty($updated) && !empty($published)) {
 381                                 $updated = $published;
 382                         }
 383
 384                         if (empty($published) && !empty($updated)) {
 385                                 $published = $updated;
 386                         }
 387
 388                         if ($published != "") {
 389                                 $item["created"] = $published;
 390                         }
 391
 392                         if ($updated != "") {
 393                                 $item["edited"] = $updated;
 394                         }
 395
 396                         if (!$dryRun) {
 397                                 $condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
 398                                         $importer["uid"], $item["uri"], Protocol::FEED, Protocol::DFRN];
 399                                 $previous = Item::selectFirst(['id', 'created'], $condition);
 400                                 if (DBA::isResult($previous)) {
 401                                         // Use the creation date when the post had been stored. It can happen this date changes in the feed.
 402                                         $creation_dates[] = $previous['created'];
 403                                         Logger::info("Item with uri " . $item["uri"] . " for user " . $importer["uid"] . " already existed under id " . $previous["id"]);
 404                                         continue;
 405                                 }
 406                                 $creation_dates[] = DateTimeFormat::utc($item['created']);
 407                         }
 408
 409                         $creator = XML::getFirstNodeValue($xpath, 'author/text()', $entry);
 410
 411                         if (empty($creator)) {
 412                                 $creator = XML::getFirstNodeValue($xpath, 'atom:author/atom:name/text()', $entry);
 413                         }
 414
 415                         if (empty($creator)) {
 416                                 $creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
 417                         }
 418
 419                         if ($creator != "") {
 420                                 $item["author-name"] = $creator;
 421                         }
 422
 423                         $creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
 424
 425                         if ($creator != "") {
 426                                 $item["author-name"] = $creator;
 427                         }
 428
 429                         /// @TODO ?
 430                         // <category>Ausland</category>
 431                         // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 432
 433                         $attachments = [];
 434
 435                         $enclosures = $xpath->query("enclosure|atom:link[@rel='enclosure']", $entry);
 436                         foreach ($enclosures AS $enclosure) {
 437                                 $href = "";
 438                                 $length = null;
 439                                 $type = null;
 440
 441                                 foreach ($enclosure->attributes AS $attribute) {
 442                                         if (in_array($attribute->name, ["url", "href"])) {
 443                                                 $href = $attribute->textContent;
 444                                         } elseif ($attribute->name == "length") {
 445                                                 $length = (int)$attribute->textContent;
 446                                         } elseif ($attribute->name == "type") {
 447                                                 $type = $attribute->textContent;
 448                                         }
 449                                 }
 450
 451                                 if (!empty($href)) {
 452                                         $attachments[] = ['type' => Post\Media::DOCUMENT, 'url' => $href, 'mimetype' => $type, 'size' => $length];
 453                                 }
 454                         }
 455
 456                         $taglist = [];
 457                         $categories = $xpath->query("category", $entry);
 458                         foreach ($categories AS $category) {
 459                                 $taglist[] = $category->nodeValue;
 460                         }
 461
 462                         $body = trim(XML::getFirstNodeValue($xpath, 'atom:content/text()', $entry));
 463
 464                         if (empty($body)) {
 465                                 $body = trim(XML::getFirstNodeValue($xpath, 'content:encoded/text()', $entry));
 466                         }
 467
 468                         $summary = trim(XML::getFirstNodeValue($xpath, 'atom:summary/text()', $entry));
 469
 470                         if (empty($summary)) {
 471                                 $summary = trim(XML::getFirstNodeValue($xpath, 'description/text()', $entry));
 472                         }
 473
 474                         if (empty($body)) {
 475                                 $body = $summary;
 476                                 $summary = '';
 477                         }
 478
 479                         if ($body == $summary) {
 480                                 $summary = '';
 481                         }
 482
 483                         // remove the content of the title if it is identically to the body
 484                         // This helps with auto generated titles e.g. from tumblr
 485                         if (self::titleIsBody($item["title"], $body)) {
 486                                 $item["title"] = "";
 487                         }
 488                         $item["body"] = HTML::toBBCode($body, $basepath);
 489
 490                         // Remove tracking pixels
 491                         $item["body"] = preg_replace("/\[img=1x1\]([^\[\]]*)\[\/img\]/Usi", '', $item["body"]);
 492
 493                         if (($item["body"] == '') && ($item["title"] != '')) {
 494                                 $item["body"] = $item["title"];
 495                                 $item["title"] = '';
 496                         }
 497
 498                         if ($dryRun) {
 499                                 $items[] = $item;
 500                                 break;
 501                         } elseif (!Item::isValid($item)) {
 502                                 Logger::info('Feed is invalid', ['created' => $item['created'], 'uid' => $item['uid'], 'uri' => $item['uri']]);
 503                                 continue;
 504                         }
 505
 506                         $preview = '';
 507                         if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) {
 508                                 // Handle enclosures and treat them as preview picture
 509                                 foreach ($attachments AS $attachment) {
 510                                         if ($attachment["mimetype"] == "image/jpeg") {
 511                                                 $preview = $attachment["url"];
 512                                         }
 513                                 }
 514
 515                                 // Remove a possible link to the item itself
 516                                 $item["body"] = str_replace($item["plink"], '', $item["body"]);
 517                                 $item["body"] = trim(preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item["body"]));
 518
 519                                 // Replace the content when the title is longer than the body
 520                                 $replace = (strlen($item["title"]) > strlen($item["body"]));
 521
 522                                 // Replace it, when there is an image in the body
 523                                 if (strstr($item["body"], '[/img]')) {
 524                                         $replace = true;
 525                                 }
 526
 527                                 // Replace it, when there is a link in the body
 528                                 if (strstr($item["body"], '[/url]')) {
 529                                         $replace = true;
 530                                 }
 531
 532                                 $saved_body = $item["body"];
 533                                 $saved_title = $item["title"];
 534
 535                                 if ($replace) {
 536                                         $item["body"] = trim($item["title"]);
 537                                 }
 538
 539                                 $data = ParseUrl::getSiteinfoCached($item['plink'], true);
 540                                 if (!empty($data['text']) && !empty($data['title']) && (mb_strlen($item['body']) < mb_strlen($data['text']))) {
 541                                         // When the fetched page info text is longer than the body, we do try to enhance the body
 542                                         if (!empty($item['body']) && (strpos($data['title'], $item['body']) === false) && (strpos($data['text'], $item['body']) === false)) {
 543                                                 // The body is not part of the fetched page info title or page info text. So we add the text to the body
 544                                                 $item['body'] .= "\n\n" . $data['text'];
 545                                         } else {
 546                                                 // Else we replace the body with the page info text
 547                                                 $item['body'] = $data['text'];
 548                                         }
 549                                 }
 550
 551                                 $data = PageInfo::queryUrl($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_denylist"] ?? '');
 552
 553                                 if (!empty($data)) {
 554                                         // Take the data that was provided by the feed if the query is empty
 555                                         if (($data['type'] == 'link') && empty($data['title']) && empty($data['text'])) {
 556                                                 $data['title'] = $saved_title;
 557                                                 $item["body"] = $saved_body;
 558                                         }
 559
 560                                         $data_text = strip_tags(trim($data['text'] ?? ''));
 561                                         $item_body = strip_tags(trim($item['body'] ?? ''));
 562
 563                                         if (!empty($data_text) && (($data_text == $item_body) || strstr($item_body, $data_text))) {
 564                                                 $data['text'] = '';
 565                                         }
 566
 567                                         // We always strip the title since it will be added in the page information
 568                                         $item["title"] = "";
 569                                         $item["body"] = $item["body"] . "\n" . PageInfo::getFooterFromData($data, false);
 570                                         $taglist = $contact["fetch_further_information"] == 2 ? PageInfo::getTagsFromUrl($item["plink"], $preview, $contact["ffi_keyword_denylist"] ?? '') : [];
 571                                         $item["object-type"] = Activity\ObjectType::BOOKMARK;
 572                                         $attachments = [];
 573                                 }
 574                         } else {
 575                                 if (!empty($summary)) {
 576                                         $item["body"] = '[abstract]' . HTML::toBBCode($summary, $basepath) . "[/abstract]\n" . $item["body"];
 577                                 }
 578
 579                                 if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] == 3)) {
 580                                         if (empty($taglist)) {
 581                                                 $taglist = PageInfo::getTagsFromUrl($item["plink"], $preview, $contact["ffi_keyword_denylist"] ?? '');
 582                                         }
 583                                         $item["body"] .= "\n" . self::tagToString($taglist);
 584                                 } else {
 585                                         $taglist = [];
 586                                 }
 587
 588                                 // Add the link to the original feed entry if not present in feed
 589                                 if (($item['plink'] != '') && !strstr($item["body"], $item['plink'])) {
 590                                         $item["body"] .= "[hr][url]" . $item['plink'] . "[/url]";
 591                                 }
 592                         }
 593
 594                         Logger::info('Stored feed', ['item' => $item]);
 595
 596                         $notify = Item::isRemoteSelf($contact, $item);
 597
 598                         // Distributed items should have a well formatted URI.
 599                         // Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
 600                         if ($notify) {
 601                                 $item['guid'] = Item::guidFromUri($orig_plink, DI::baseUrl()->getHostname());
 602                                 unset($item['uri']);
 603                                 unset($item['thr-parent']);
 604                                 unset($item['parent-uri']);
 605
 606                                 // Set the delivery priority for "remote self" to "medium"
 607                                 $notify = PRIORITY_MEDIUM;
 608                         }
 609
 610                         $id = Item::insert($item, $notify);
 611
 612                         Logger::info("Feed for contact " . $contact["url"] . " stored under id " . $id);
 613
 614                         if (!empty($id) && (!empty($taglist) || !empty($attachments))) {
 615                                 $feeditem = Item::selectFirst(['uri-id'], ['id' => $id]);
 616                                 foreach ($taglist as $tag) {
 617                                         Tag::store($feeditem['uri-id'], Tag::HASHTAG, $tag);
 618                                 }
 619                                 foreach ($attachments as $attachment) {
 620                                         $attachment['uri-id'] = $feeditem['uri-id'];
 621                                         Post\Media::insert($attachment);
 622                                 }
 623                         }
 624                 }
 625
 626                 if (!$dryRun && DI::config()->get('system', 'adjust_poll_frequency')) {
 627                         self::adjustPollFrequency($contact, $creation_dates);
 628                 }
 629
 630                 return ["header" => $author, "items" => $items];
 631         }
 632
 633         /**
 634          * Automatically adjust the poll frequency according to the post frequency
 635          *
 636          * @param array $contact
 637          * @param array $creation_dates
 638          * @return void
 639          */
 640         private static function adjustPollFrequency(array $contact, array $creation_dates)
 641         {
 642                 if ($contact['network'] != Protocol::FEED) {
 643                         Logger::info('Contact is no feed, skip.', ['id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url'], 'network' => $contact['network']]);
 644                         return;
 645                 }
 646
 647                 if (!empty($creation_dates)) {
 648                         // Count the post frequency and the earliest and latest post date
 649                         $frequency = [];
 650                         $oldest = time();
 651                         $newest = 0;
 652                         $oldest_date = $newest_date = '';
 653
 654                         foreach ($creation_dates as $date) {
 655                                 $timestamp = strtotime($date);
 656                                 $day = intdiv($timestamp, 86400);
 657                                 $hour = $timestamp % 86400;
 658
 659                                 // Only have a look at values from the last seven days
 660                                 if (((time() / 86400) - $day) < 7) {
 661                                         if (empty($frequency[$day])) {
 662                                                 $frequency[$day] = ['count' => 1, 'low' => $hour, 'high' => $hour];
 663                                         } else {
 664                                                 ++$frequency[$day]['count'];
 665                                                 if ($frequency[$day]['low'] > $hour) {
 666                                                         $frequency[$day]['low'] = $hour;
 667                                                 }
 668                                                 if ($frequency[$day]['high'] < $hour) {
 669                                                         $frequency[$day]['high'] = $hour;
 670                                                 }
 671                                         }
 672                                 }
 673                                 if ($oldest > $day) {
 674                                         $oldest = $day;
 675                                         $oldest_date = $date;
 676                                 }
 677
 678                                 if ($newest < $day) {
 679                                         $newest = $day;
 680                                         $newest_date = $date;
 681                                 }
 682                         }
 683
 684                         if (count($creation_dates) == 1) {
 685                                 Logger::info('Feed had posted a single time, switching to daily polling', ['newest' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 686                                 $priority = 8; // Poll once a day
 687                         }
 688
 689                         if (empty($priority) && (((time() / 86400) - $newest) > 730)) {
 690                                 Logger::info('Feed had not posted for two years, switching to monthly polling', ['newest' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 691                                 $priority = 10; // Poll every month
 692                         }
 693
 694                         if (empty($priority) && (((time() / 86400) - $newest) > 365)) {
 695                                 Logger::info('Feed had not posted for a year, switching to weekly polling', ['newest' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 696                                 $priority = 9; // Poll every week
 697                         }
 698
 699                         if (empty($priority) && empty($frequency)) {
 700                                 Logger::info('Feed had not posted for at least a week, switching to daily polling', ['newest' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 701                                 $priority = 8; // Poll once a day
 702                         }
 703
 704                         if (empty($priority)) {
 705                                 // Calculate the highest "posts per day" value
 706                                 $max = 0;
 707                                 foreach ($frequency as $entry) {
 708                                         if (($entry['count'] == 1) || ($entry['high'] == $entry['low'])) {
 709                                                 continue;
 710                                         }
 711
 712                                         // We take the earliest and latest post day and interpolate the number of post per day
 713                                         // that would had been created with this post frequency
 714
 715                                         // Assume at least four hours between oldest and newest post per day - should be okay for news outlets
 716                                         $duration = max($entry['high'] - $entry['low'], 14400);
 717                                         $ppd = (86400 / $duration) * $entry['count'];
 718                                         if ($ppd > $max) {
 719                                                 $max = $ppd;
 720                                         }
 721                                 }
 722                                 if ($max > 48) {
 723                                         $priority = 1; // Poll every quarter hour
 724                                 } elseif ($max > 24) {
 725                                         $priority = 2; // Poll half an hour
 726                                 } elseif ($max > 12) {
 727                                         $priority = 3; // Poll hourly
 728                                 } elseif ($max > 8) {
 729                                         $priority = 4; // Poll every two hours
 730                                 } elseif ($max > 4) {
 731                                         $priority = 5; // Poll every three hours
 732                                 } elseif ($max > 2) {
 733                                         $priority = 6; // Poll every six hours
 734                                 } else {
 735                                         $priority = 7; // Poll twice a day
 736                                 }
 737                                 Logger::info('Calculated priority by the posts per day', ['priority' => $priority, 'max' => round($max, 2), 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 738                         }
 739                 } else {
 740                         Logger::info('No posts, switching to daily polling', ['id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 741                         $priority = 8; // Poll once a day
 742                 }
 743
 744                 if ($contact['rating'] != $priority) {
 745                         Logger::notice('Adjusting priority', ['old' => $contact['rating'], 'new' => $priority, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 746                         DBA::update('contact', ['rating' => $priority], ['id' => $contact['id']]);
 747                 }
 748         }
 749
 750         /**
 751          * Convert a tag array to a tag string
 752          *
 753          * @param array $tags
 754          * @return string tag string
 755          */
 756         private static function tagToString(array $tags)
 757         {
 758                 $tagstr = '';
 759
 760                 foreach ($tags as $tag) {
 761                         if ($tagstr != "") {
 762                                 $tagstr .= ", ";
 763                         }
 764
 765                         $tagstr .= "#[url=" . DI::baseUrl() . "/search?tag=" . urlencode($tag) . "]" . $tag . "[/url]";
 766                 }
 767
 768                 return $tagstr;
 769         }
 770
 771         private static function titleIsBody($title, $body)
 772         {
 773                 $title = strip_tags($title);
 774                 $title = trim($title);
 775                 $title = html_entity_decode($title, ENT_QUOTES, 'UTF-8');
 776                 $title = str_replace(["\n", "\r", "\t", " "], ["", "", "", ""], $title);
 777
 778                 $body = strip_tags($body);
 779                 $body = trim($body);
 780                 $body = html_entity_decode($body, ENT_QUOTES, 'UTF-8');
 781                 $body = str_replace(["\n", "\r", "\t", " "], ["", "", "", ""], $body);
 782
 783                 if (strlen($title) < strlen($body)) {
 784                         $body = substr($body, 0, strlen($title));
 785                 }
 786
 787                 if (($title != $body) && (substr($title, -3) == "...")) {
 788                         $pos = strrpos($title, "...");
 789                         if ($pos > 0) {
 790                                 $title = substr($title, 0, $pos);
 791                                 $body = substr($body, 0, $pos);
 792                         }
 793                 }
 794                 return ($title == $body);
 795         }
 796
 797         /**
 798          * Creates the Atom feed for a given nickname
 799          *
 800          * Supported filters:
 801          * - activity (default): all the public posts
 802          * - posts: all the public top-level posts
 803          * - comments: all the public replies
 804          *
 805          * Updates the provided last_update parameter if the result comes from the
 806          * cache or it is empty
 807          *
 808          * @param string  $owner_nick  Nickname of the feed owner
 809          * @param string  $last_update Date of the last update
 810          * @param integer $max_items   Number of maximum items to fetch
 811          * @param string  $filter      Feed items filter (activity, posts or comments)
 812          * @param boolean $nocache     Wether to bypass caching
 813          *
 814          * @return string Atom feed
 815          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 816          * @throws \ImagickException
 817          */
 818         public static function atom($owner_nick, $last_update, $max_items = 300, $filter = 'activity', $nocache = false)
 819         {
 820                 $stamp = microtime(true);
 821
 822                 $owner = User::getOwnerDataByNick($owner_nick);
 823                 if (!$owner) {
 824                         return;
 825                 }
 826
 827                 $cachekey = "feed:feed:" . $owner_nick . ":" . $filter . ":" . $last_update;
 828
 829                 $previous_created = $last_update;
 830
 831                 // Don't cache when the last item was posted less then 15 minutes ago (Cache duration)
 832                 if ((time() - strtotime($owner['last-item'])) < 15*60) {
 833                         $result = DI::cache()->get($cachekey);
 834                         if (!$nocache && !is_null($result)) {
 835                                 Logger::info('Cached feed duration', ['seconds' => number_format(microtime(true) - $stamp, 3), 'nick' => $owner_nick, 'filter' => $filter, 'created' => $previous_created]);
 836                                 return $result['feed'];
 837                         }
 838                 }
 839
 840                 $check_date = empty($last_update) ? '' : DateTimeFormat::utc($last_update);
 841                 $authorid = Contact::getIdForURL($owner["url"]);
 842
 843                 $condition = ["`uid` = ? AND `received` > ? AND NOT `deleted` AND `gravity` IN (?, ?)
 844                         AND `private` != ? AND `visible` AND `wall` AND `parent-network` IN (?, ?, ?, ?)",
 845                         $owner["uid"], $check_date, GRAVITY_PARENT, GRAVITY_COMMENT,
 846                         Item::PRIVATE, Protocol::ACTIVITYPUB,
 847                         Protocol::OSTATUS, Protocol::DFRN, Protocol::DIASPORA];
 848
 849                 if ($filter === 'comments') {
 850                         $condition[0] .= " AND `object-type` = ? ";
 851                         $condition[] = Activity\ObjectType::COMMENT;
 852                 }
 853
 854                 if ($owner['account-type'] != User::ACCOUNT_TYPE_COMMUNITY) {
 855                         $condition[0] .= " AND `contact-id` = ? AND `author-id` = ?";
 856                         $condition[] = $owner["id"];
 857                         $condition[] = $authorid;
 858                 }
 859
 860                 $params = ['order' => ['received' => true], 'limit' => $max_items];
 861
 862                 if ($filter === 'posts') {
 863                         $ret = Item::selectThread([], $condition, $params);
 864                 } else {
 865                         $ret = Item::select([], $condition, $params);
 866                 }
 867
 868                 $items = Item::inArray($ret);
 869
 870                 $doc = new DOMDocument('1.0', 'utf-8');
 871                 $doc->formatOutput = true;
 872
 873                 $root = self::addHeader($doc, $owner, $filter);
 874
 875                 foreach ($items as $item) {
 876                         $entry = self::entry($doc, $item, $owner);
 877                         $root->appendChild($entry);
 878
 879                         if ($last_update < $item['created']) {
 880                                 $last_update = $item['created'];
 881                         }
 882                 }
 883
 884                 $feeddata = trim($doc->saveXML());
 885
 886                 $msg = ['feed' => $feeddata, 'last_update' => $last_update];
 887                 DI::cache()->set($cachekey, $msg, Duration::QUARTER_HOUR);
 888
 889                 Logger::info('Feed duration', ['seconds' => number_format(microtime(true) - $stamp, 3), 'nick' => $owner_nick, 'filter' => $filter, 'created' => $previous_created]);
 890
 891                 return $feeddata;
 892         }
 893
 894         /**
 895          * Adds the header elements to the XML document
 896          *
 897          * @param DOMDocument $doc       XML document
 898          * @param array       $owner     Contact data of the poster
 899          * @param string      $filter    The related feed filter (activity, posts or comments)
 900          *
 901          * @return object header root element
 902          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 903          */
 904         private static function addHeader(DOMDocument $doc, array $owner, $filter)
 905         {
 906                 $root = $doc->createElementNS(ActivityNamespace::ATOM1, 'feed');
 907                 $doc->appendChild($root);
 908
 909                 $title = '';
 910                 $selfUri = '/feed/' . $owner["nick"] . '/';
 911                 switch ($filter) {
 912                         case 'activity':
 913                                 $title = DI::l10n()->t('%s\'s timeline', $owner['name']);
 914                                 $selfUri .= $filter;
 915                                 break;
 916                         case 'posts':
 917                                 $title = DI::l10n()->t('%s\'s posts', $owner['name']);
 918                                 break;
 919                         case 'comments':
 920                                 $title = DI::l10n()->t('%s\'s comments', $owner['name']);
 921                                 $selfUri .= $filter;
 922                                 break;
 923                 }
 924
 925                 $attributes = ["uri" => "https://friendi.ca", "version" => FRIENDICA_VERSION . "-" . DB_UPDATE_VERSION];
 926                 XML::addElement($doc, $root, "generator", FRIENDICA_PLATFORM, $attributes);
 927                 XML::addElement($doc, $root, "id", DI::baseUrl() . "/profile/" . $owner["nick"]);
 928                 XML::addElement($doc, $root, "title", $title);
 929                 XML::addElement($doc, $root, "subtitle", sprintf("Updates from %s on %s", $owner["name"], DI::config()->get('config', 'sitename')));
 930                 XML::addElement($doc, $root, "logo", $owner["photo"]);
 931                 XML::addElement($doc, $root, "updated", DateTimeFormat::utcNow(DateTimeFormat::ATOM));
 932
 933                 $author = self::addAuthor($doc, $owner);
 934                 $root->appendChild($author);
 935
 936                 $attributes = ["href" => $owner["url"], "rel" => "alternate", "type" => "text/html"];
 937                 XML::addElement($doc, $root, "link", "", $attributes);
 938
 939                 OStatus::hublinks($doc, $root, $owner["nick"]);
 940
 941                 $attributes = ["href" => DI::baseUrl() . $selfUri, "rel" => "self", "type" => "application/atom+xml"];
 942                 XML::addElement($doc, $root, "link", "", $attributes);
 943
 944                 return $root;
 945         }
 946
 947         /**
 948          * Adds the author element to the XML document
 949          *
 950          * @param DOMDocument $doc          XML document
 951          * @param array       $owner        Contact data of the poster
 952          *
 953          * @return \DOMElement author element
 954          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 955          */
 956         private static function addAuthor(DOMDocument $doc, array $owner)
 957         {
 958                 $author = $doc->createElement("author");
 959                 XML::addElement($doc, $author, "uri", $owner["url"]);
 960                 XML::addElement($doc, $author, "name", $owner["nick"]);
 961                 XML::addElement($doc, $author, "email", $owner["addr"]);
 962
 963                 return $author;
 964         }
 965
 966         /**
 967          * Adds an entry element to the XML document
 968          *
 969          * @param DOMDocument $doc       XML document
 970          * @param array       $item      Data of the item that is to be posted
 971          * @param array       $owner     Contact data of the poster
 972          * @param bool        $toplevel  optional default false
 973          *
 974          * @return \DOMElement Entry element
 975          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 976          * @throws \ImagickException
 977          */
 978         private static function entry(DOMDocument $doc, array $item, array $owner)
 979         {
 980                 $xml = null;
 981
 982                 $repeated_guid = OStatus::getResharedGuid($item);
 983                 if ($repeated_guid != "") {
 984                         $xml = self::reshareEntry($doc, $item, $owner, $repeated_guid);
 985                 }
 986
 987                 if ($xml) {
 988                         return $xml;
 989                 }
 990
 991                 return self::noteEntry($doc, $item, $owner);
 992         }
 993
 994                 /**
 995          * Adds an entry element with reshared content
 996          *
 997          * @param DOMDocument $doc           XML document
 998          * @param array       $item          Data of the item that is to be posted
 999          * @param array       $owner         Contact data of the poster
1000          * @param string      $repeated_guid guid
1001          * @param bool        $toplevel      Is it for en entry element (false) or a feed entry (true)?
1002          *
1003          * @return bool Entry element
1004          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1005          * @throws \ImagickException
1006          */
1007         private static function reshareEntry(DOMDocument $doc, array $item, array $owner, $repeated_guid)
1008         {
1009                 if (($item['gravity'] != GRAVITY_PARENT) && (Strings::normaliseLink($item["author-link"]) != Strings::normaliseLink($owner["url"]))) {
1010                         Logger::info('Feed entry author does not match feed owner', ['owner' => $owner["url"], 'author' => $item["author-link"]]);
1011                 }
1012
1013                 $entry = OStatus::entryHeader($doc, $owner, $item, false);
1014
1015                 $condition = ['uid' => $owner["uid"], 'guid' => $repeated_guid, 'private' => [Item::PUBLIC, Item::UNLISTED],
1016                         'network' => Protocol::FEDERATED];
1017                 $repeated_item = Item::selectFirst([], $condition);
1018                 if (!DBA::isResult($repeated_item)) {
1019                         return false;
1020                 }
1021
1022                 self::entryContent($doc, $entry, $item, self::getTitle($repeated_item), Activity::SHARE, false);
1023
1024                 self::entryFooter($doc, $entry, $item, $owner);
1025
1026                 return $entry;
1027         }
1028
1029         /**
1030          * Adds a regular entry element
1031          *
1032          * @param DOMDocument $doc       XML document
1033          * @param array       $item      Data of the item that is to be posted
1034          * @param array       $owner     Contact data of the poster
1035          * @param bool        $toplevel  Is it for en entry element (false) or a feed entry (true)?
1036          *
1037          * @return \DOMElement Entry element
1038          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1039          * @throws \ImagickException
1040          */
1041         private static function noteEntry(DOMDocument $doc, array $item, array $owner)
1042         {
1043                 if (($item['gravity'] != GRAVITY_PARENT) && (Strings::normaliseLink($item["author-link"]) != Strings::normaliseLink($owner["url"]))) {
1044                         Logger::info('Feed entry author does not match feed owner', ['owner' => $owner["url"], 'author' => $item["author-link"]]);
1045                 }
1046
1047                 $entry = OStatus::entryHeader($doc, $owner, $item, false);
1048
1049                 self::entryContent($doc, $entry, $item, self::getTitle($item), '', true);
1050
1051                 self::entryFooter($doc, $entry, $item, $owner);
1052
1053                 return $entry;
1054         }
1055
1056         /**
1057          * Adds elements to the XML document
1058          *
1059          * @param DOMDocument $doc       XML document
1060          * @param \DOMElement $entry     Entry element where the content is added
1061          * @param array       $item      Data of the item that is to be posted
1062          * @param array       $owner     Contact data of the poster
1063          * @param string      $title     Title for the post
1064          * @param string      $verb      The activity verb
1065          * @param bool        $complete  Add the "status_net" element?
1066          * @param bool        $feed_mode Behave like a regular feed for users if true
1067          * @return void
1068          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1069          */
1070         private static function entryContent(DOMDocument $doc, \DOMElement $entry, array $item, $title, $verb = "", $complete = true)
1071         {
1072                 if ($verb == "") {
1073                         $verb = OStatus::constructVerb($item);
1074                 }
1075
1076                 XML::addElement($doc, $entry, "id", $item["uri"]);
1077                 XML::addElement($doc, $entry, "title", html_entity_decode($title, ENT_QUOTES, 'UTF-8'));
1078
1079                 $body = OStatus::formatPicturePost($item['body']);
1080
1081                 $body = BBCode::convert($body, false, BBCode::OSTATUS);
1082
1083                 XML::addElement($doc, $entry, "content", $body, ["type" => "html"]);
1084
1085                 XML::addElement($doc, $entry, "link", "", ["rel" => "alternate", "type" => "text/html",
1086                                                                 "href" => DI::baseUrl()."/display/".$item["guid"]]
1087                 );
1088
1089                 XML::addElement($doc, $entry, "published", DateTimeFormat::utc($item["created"]."+00:00", DateTimeFormat::ATOM));
1090                 XML::addElement($doc, $entry, "updated", DateTimeFormat::utc($item["edited"]."+00:00", DateTimeFormat::ATOM));
1091         }
1092
1093         /**
1094          * Adds the elements at the foot of an entry to the XML document
1095          *
1096          * @param DOMDocument $doc       XML document
1097          * @param object      $entry     The entry element where the elements are added
1098          * @param array       $item      Data of the item that is to be posted
1099          * @param array       $owner     Contact data of the poster
1100          * @param bool        $complete  default true
1101          * @return void
1102          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1103          */
1104         private static function entryFooter(DOMDocument $doc, $entry, array $item, array $owner)
1105         {
1106                 $mentioned = [];
1107
1108                 if ($item['gravity'] != GRAVITY_PARENT) {
1109                         $parent = Item::selectFirst(['guid', 'author-link', 'owner-link'], ['id' => $item['parent']]);
1110
1111                         $thrparent = Item::selectFirst(['guid', 'author-link', 'owner-link', 'plink'], ['uid' => $owner["uid"], 'uri' => $item['thr-parent']]);
1112
1113                         if (DBA::isResult($thrparent)) {
1114                                 $mentioned[$thrparent["author-link"]] = $thrparent["author-link"];
1115                                 $mentioned[$thrparent["owner-link"]] = $thrparent["owner-link"];
1116                                 $parent_plink = $thrparent["plink"];
1117                         } else {
1118                                 $mentioned[$parent["author-link"]] = $parent["author-link"];
1119                                 $mentioned[$parent["owner-link"]] = $parent["owner-link"];
1120                                 $parent_plink = DI::baseUrl()."/display/".$parent["guid"];
1121                         }
1122
1123                         $attributes = [
1124                                         "ref" => $item['thr-parent'],
1125                                         "href" => $parent_plink];
1126                         XML::addElement($doc, $entry, "thr:in-reply-to", "", $attributes);
1127
1128                         $attributes = [
1129                                         "rel" => "related",
1130                                         "href" => $parent_plink];
1131                         XML::addElement($doc, $entry, "link", "", $attributes);
1132                 }
1133
1134                 // uri-id isn't present for follow entry pseudo-items
1135                 $tags = Tag::getByURIId($item['uri-id'] ?? 0);
1136                 foreach ($tags as $tag) {
1137                         $mentioned[$tag['url']] = $tag['url'];
1138                 }
1139
1140                 foreach ($tags as $tag) {
1141                         if ($tag['type'] == Tag::HASHTAG) {
1142                                 XML::addElement($doc, $entry, "category", "", ["term" => $tag['name']]);
1143                         }
1144                 }
1145
1146                 OStatus::getAttachment($doc, $entry, $item);
1147         }
1148
1149         /**
1150          * Fetch or create title for feed entry
1151          *
1152          * @param array $item
1153          * @return string title
1154          */
1155         private static function getTitle(array $item)
1156         {
1157                 if ($item['title'] != '') {
1158                         return BBCode::convert($item['title'], false, BBCode::OSTATUS);
1159                 }
1160
1161                 // Fetch information about the post
1162                 $siteinfo = BBCode::getAttachedData($item["body"]);
1163                 if (isset($siteinfo["title"])) {
1164                         return $siteinfo["title"];
1165                 }
1166
1167                 // If no bookmark is found then take the first line
1168                 // Remove the share element before fetching the first line
1169                 $title = trim(preg_replace("/\[share.*?\](.*?)\[\/share\]/ism","\n$1\n",$item['body']));
1170
1171                 $title = HTML::toPlaintext(BBCode::convert($title, false), 0, true)."\n";
1172                 $pos = strpos($title, "\n");
1173                 $trailer = "";
1174                 if (($pos == 0) || ($pos > 100)) {
1175                         $pos = 100;
1176                         $trailer = "...";
1177                 }
1178
1179                 return substr($title, 0, $pos) . $trailer;
1180         }
1181 }