src/Protocol/Feed.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2023, the Friendica project
   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 DOMElement;
  26 use DOMXPath;
  27 use Friendica\App;
  28 use Friendica\Content\PageInfo;
  29 use Friendica\Content\Text\BBCode;
  30 use Friendica\Content\Text\HTML;
  31 use Friendica\Core\Cache\Enum\Duration;
  32 use Friendica\Core\Logger;
  33 use Friendica\Core\Protocol;
  34 use Friendica\Core\Worker;
  35 use Friendica\Database\DBA;
  36 use Friendica\DI;
  37 use Friendica\Model\Contact;
  38 use Friendica\Model\Conversation;
  39 use Friendica\Model\Item;
  40 use Friendica\Model\Post;
  41 use Friendica\Model\Tag;
  42 use Friendica\Model\User;
  43 use Friendica\Network\HTTPException;
  44 use Friendica\Util\DateTimeFormat;
  45 use Friendica\Util\Network;
  46 use Friendica\Util\ParseUrl;
  47 use Friendica\Util\Proxy;
  48 use Friendica\Util\Strings;
  49 use Friendica\Util\XML;
  50 use GuzzleHttp\Exception\TransferException;
  51
  52 /**
  53  * This class contain functions to import feeds (RSS/RDF/Atom)
  54  */
  55 class Feed
  56 {
  57         /**
  58          * Read a RSS/RDF/Atom feed and create an item entry for it
  59          *
  60          * @param string $xml      The feed data
  61          * @param array  $importer The user record of the importer
  62          * @param array  $contact  The contact record of the feed
  63          *
  64          * @return array Returns the header and the first item in dry run mode
  65          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  66          */
  67         public static function import(string $xml, array $importer = [], array $contact = []): array
  68         {
  69                 $dryRun = empty($importer) && empty($contact);
  70
  71                 if ($dryRun) {
  72                         Logger::info("Test Atom/RSS feed");
  73                 } else {
  74                         Logger::info('Import Atom/RSS feed "' . $contact['name'] . '" (Contact ' . $contact['id'] . ') for user ' . $importer['uid']);
  75                 }
  76
  77                 $xml = trim($xml);
  78
  79                 if (empty($xml)) {
  80                         Logger::info('XML is empty.');
  81                         return [];
  82                 }
  83
  84                 if (!empty($contact['poll'])) {
  85                         $basepath = $contact['poll'];
  86                 } elseif (!empty($contact['url'])) {
  87                         $basepath = $contact['url'];
  88                 } else {
  89                         $basepath = '';
  90                 }
  91
  92                 $doc = new DOMDocument();
  93                 @$doc->loadXML($xml);
  94                 $xpath = new DOMXPath($doc);
  95                 $xpath->registerNamespace('atom', ActivityNamespace::ATOM1);
  96                 $xpath->registerNamespace('dc', 'http://purl.org/dc/elements/1.1/');
  97                 $xpath->registerNamespace('content', 'http://purl.org/rss/1.0/modules/content/');
  98                 $xpath->registerNamespace('rdf', 'http://www.w3.org/1999/02/22-rdf-syntax-ns#');
  99                 $xpath->registerNamespace('rss', 'http://purl.org/rss/1.0/');
 100                 $xpath->registerNamespace('media', 'http://search.yahoo.com/mrss/');
 101                 $xpath->registerNamespace('poco', ActivityNamespace::POCO);
 102
 103                 $author = [];
 104                 $entries = null;
 105                 $protocol = Conversation::PARCEL_UNKNOWN;
 106
 107                 // Is it RDF?
 108                 if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
 109                         $protocol = Conversation::PARCEL_RDF;
 110                         $author['author-link'] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:link/text()');
 111                         $author['author-name'] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:title/text()');
 112
 113                         if (empty($author['author-name'])) {
 114                                 $author['author-name'] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:description/text()');
 115                         }
 116                         $entries = $xpath->query('/rdf:RDF/rss:item');
 117                 }
 118
 119                 // Is it Atom?
 120                 if ($xpath->query('/atom:feed')->length > 0) {
 121                         $protocol = Conversation::PARCEL_ATOM;
 122                         $alternate = XML::getFirstAttributes($xpath, "atom:link[@rel='alternate']");
 123                         if (is_object($alternate)) {
 124                                 foreach ($alternate as $attribute) {
 125                                         if ($attribute->name == 'href') {
 126                                                 $author['author-link'] = $attribute->textContent;
 127                                         }
 128                                 }
 129                         }
 130
 131                         if (empty($author['author-link'])) {
 132                                 $self = XML::getFirstAttributes($xpath, "atom:link[@rel='self']");
 133                                 if (is_object($self)) {
 134                                         foreach ($self as $attribute) {
 135                                                 if ($attribute->name == 'href') {
 136                                                         $author['author-link'] = $attribute->textContent;
 137                                                 }
 138                                         }
 139                                 }
 140                         }
 141
 142                         if (empty($author['author-link'])) {
 143                                 $author['author-link'] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:id/text()');
 144                         }
 145                         $author['author-avatar'] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:logo/text()');
 146
 147                         $author['author-name'] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:title/text()');
 148
 149                         if (empty($author['author-name'])) {
 150                                 $author['author-name'] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:subtitle/text()');
 151                         }
 152
 153                         if (empty($author['author-name'])) {
 154                                 $author['author-name'] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:name/text()');
 155                         }
 156
 157                         $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:displayName/text()');
 158                         if ($value != '') {
 159                                 $author['author-name'] = $value;
 160                         }
 161
 162                         if ($dryRun) {
 163                                 $author['author-id'] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:id/text()');
 164
 165                                 // See https://tools.ietf.org/html/rfc4287#section-3.2.2
 166                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/atom:uri/text()');
 167                                 if ($value != '') {
 168                                         $author['author-link'] = $value;
 169                                 }
 170
 171                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:preferredUsername/text()');
 172                                 if ($value != '') {
 173                                         $author['author-nick'] = $value;
 174                                 }
 175
 176                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:address/poco:formatted/text()');
 177                                 if ($value != '') {
 178                                         $author['author-location'] = $value;
 179                                 }
 180
 181                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:note/text()');
 182                                 if ($value != '') {
 183                                         $author['author-about'] = $value;
 184                                 }
 185
 186                                 $avatar = XML::getFirstAttributes($xpath, "atom:author/atom:link[@rel='avatar']");
 187                                 if (is_object($avatar)) {
 188                                         foreach ($avatar as $attribute) {
 189                                                 if ($attribute->name == 'href') {
 190                                                         $author['author-avatar'] = $attribute->textContent;
 191                                                 }
 192                                         }
 193                                 }
 194                         }
 195
 196                         $author['edited'] = $author['created'] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:updated/text()');
 197
 198                         $author['app'] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:generator/text()');
 199
 200                         $entries = $xpath->query('/atom:feed/atom:entry');
 201                 }
 202
 203                 // Is it RSS?
 204                 if ($xpath->query('/rss/channel')->length > 0) {
 205                         $protocol = Conversation::PARCEL_RSS;
 206                         $author['author-link'] = XML::getFirstNodeValue($xpath, '/rss/channel/link/text()');
 207
 208                         $author['author-name'] = XML::getFirstNodeValue($xpath, '/rss/channel/title/text()');
 209
 210                         if (empty($author['author-name'])) {
 211                                 $author['author-name'] = XML::getFirstNodeValue($xpath, '/rss/channel/copyright/text()');
 212                         }
 213
 214                         if (empty($author['author-name'])) {
 215                                 $author['author-name'] = XML::getFirstNodeValue($xpath, '/rss/channel/description/text()');
 216                         }
 217
 218                         $author['author-avatar'] = XML::getFirstNodeValue($xpath, '/rss/channel/image/url/text()');
 219
 220                         if (empty($author['author-avatar'])) {
 221                                 $avatar = XML::getFirstAttributes($xpath, '/rss/channel/itunes:image');
 222                                 if (is_object($avatar)) {
 223                                         foreach ($avatar as $attribute) {
 224                                                 if ($attribute->name == 'href') {
 225                                                         $author['author-avatar'] = $attribute->textContent;
 226                                                 }
 227                                         }
 228                                 }
 229                         }
 230
 231                         $author['author-about'] = HTML::toBBCode(XML::getFirstNodeValue($xpath, '/rss/channel/description/text()'), $basepath);
 232
 233                         if (empty($author['author-about'])) {
 234                                 $author['author-about'] = XML::getFirstNodeValue($xpath, '/rss/channel/itunes:summary/text()');
 235                         }
 236
 237                         $author['edited'] = $author['created'] = XML::getFirstNodeValue($xpath, '/rss/channel/pubDate/text()');
 238
 239                         $author['app'] = XML::getFirstNodeValue($xpath, '/rss/channel/generator/text()');
 240
 241                         $entries = $xpath->query('/rss/channel/item');
 242                 }
 243
 244                 if (!$dryRun) {
 245                         $author['author-link'] = $contact['url'];
 246
 247                         if (empty($author['author-name'])) {
 248                                 $author['author-name'] = $contact['name'];
 249                         }
 250
 251                         $author['author-avatar'] = $contact['thumb'];
 252
 253                         $author['owner-link'] = $contact['url'];
 254                         $author['owner-name'] = $contact['name'];
 255                         $author['owner-avatar'] = $contact['thumb'];
 256                 }
 257
 258                 $header = [
 259                         'uid'         => $importer['uid'] ?? 0,
 260                         'network'     => Protocol::FEED,
 261                         'wall'        => 0,
 262                         'origin'      => 0,
 263                         'gravity'     => Item::GRAVITY_PARENT,
 264                         'private'     => Item::PUBLIC,
 265                         'verb'        => Activity::POST,
 266                         'object-type' => Activity\ObjectType::NOTE,
 267                         'post-type'   => Item::PT_ARTICLE,
 268                         'contact-id'  => $contact['id'] ?? 0,
 269                 ];
 270
 271                 $datarray['protocol'] = $protocol;
 272                 $datarray['direction'] = Conversation::PULL;
 273
 274                 if (!is_object($entries)) {
 275                         Logger::info("There are no entries in this feed.");
 276                         return [];
 277                 }
 278
 279                 $items = [];
 280                 $creation_dates = [];
 281
 282                 // Limit the number of items that are about to be fetched
 283                 $total_items = ($entries->length - 1);
 284                 $max_items = DI::config()->get('system', 'max_feed_items');
 285                 if (($max_items > 0) && ($total_items > $max_items)) {
 286                         $total_items = $max_items;
 287                 }
 288
 289                 $postings = [];
 290
 291                 // Importing older entries first
 292                 for ($i = $total_items; $i >= 0; --$i) {
 293                         $entry = $entries->item($i);
 294
 295                         $item = array_merge($header, $author);
 296
 297                         $alternate = XML::getFirstAttributes($xpath, "atom:link[@rel='alternate']", $entry);
 298                         if (!is_object($alternate)) {
 299                                 $alternate = XML::getFirstAttributes($xpath, 'atom:link', $entry);
 300                         }
 301                         if (is_object($alternate)) {
 302                                 foreach ($alternate as $attribute) {
 303                                         if ($attribute->name == 'href') {
 304                                                 $item['plink'] = $attribute->textContent;
 305                                         }
 306                                 }
 307                         }
 308
 309                         if (empty($item['plink'])) {
 310                                 $item['plink'] = XML::getFirstNodeValue($xpath, 'link/text()', $entry);
 311                         }
 312
 313                         if (empty($item['plink'])) {
 314                                 $item['plink'] = XML::getFirstNodeValue($xpath, 'rss:link/text()', $entry);
 315                         }
 316
 317                         // Add the base path if missing
 318                         $item['plink'] = Network::addBasePath($item['plink'], $basepath);
 319
 320                         $item['uri'] = XML::getFirstNodeValue($xpath, 'atom:id/text()', $entry);
 321
 322                         $guid = XML::getFirstNodeValue($xpath, 'guid/text()', $entry);
 323                         if (!empty($guid)) {
 324                                 $item['uri'] = $guid;
 325
 326                                 // Don't use the GUID value directly but instead use it as a basis for the GUID
 327                                 $item['guid'] = Item::guidFromUri($guid, parse_url($guid, PHP_URL_HOST) ?? parse_url($item['plink'], PHP_URL_HOST));
 328                         }
 329
 330                         if (empty($item['uri'])) {
 331                                 $item['uri'] = $item['plink'];
 332                         }
 333
 334                         $orig_plink = $item['plink'];
 335
 336                         try {
 337                                 $item['plink'] = DI::httpClient()->finalUrl($item['plink']);
 338                         } catch (TransferException $exception) {
 339                                 Logger::notice('Item URL couldn\'t get expanded', ['url' => $item['plink'], 'exception' => $exception]);
 340                         }
 341
 342                         $item['title'] = XML::getFirstNodeValue($xpath, 'atom:title/text()', $entry);
 343
 344                         if (empty($item['title'])) {
 345                                 $item['title'] = XML::getFirstNodeValue($xpath, 'title/text()', $entry);
 346                         }
 347
 348                         if (empty($item['title'])) {
 349                                 $item['title'] = XML::getFirstNodeValue($xpath, 'rss:title/text()', $entry);
 350                         }
 351
 352                         if (empty($item['title'])) {
 353                                 $item['title'] = XML::getFirstNodeValue($xpath, 'itunes:title/text()', $entry);
 354                         }
 355
 356                         $item['title'] = html_entity_decode($item['title'], ENT_QUOTES, 'UTF-8');
 357
 358                         $published = XML::getFirstNodeValue($xpath, 'atom:published/text()', $entry);
 359
 360                         if (empty($published)) {
 361                                 $published = XML::getFirstNodeValue($xpath, 'pubDate/text()', $entry);
 362                         }
 363
 364                         if (empty($published)) {
 365                                 $published = XML::getFirstNodeValue($xpath, 'dc:date/text()', $entry);
 366                         }
 367
 368                         $updated = XML::getFirstNodeValue($xpath, 'atom:updated/text()', $entry);
 369
 370                         if (empty($updated) && !empty($published)) {
 371                                 $updated = $published;
 372                         }
 373
 374                         if (empty($published) && !empty($updated)) {
 375                                 $published = $updated;
 376                         }
 377
 378                         if ($published != '') {
 379                                 $item['created'] = trim($published);
 380                         }
 381
 382                         if ($updated != '') {
 383                                 $item['edited'] = trim($updated);
 384                         }
 385
 386                         if (!$dryRun) {
 387                                 $condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
 388                                         $importer['uid'], $item['uri'], Protocol::FEED, Protocol::DFRN];
 389                                 $previous = Post::selectFirst(['id', 'created'], $condition);
 390                                 if (DBA::isResult($previous)) {
 391                                         // Use the creation date when the post had been stored. It can happen this date changes in the feed.
 392                                         $creation_dates[] = $previous['created'];
 393                                         Logger::info('Item with URI ' . $item['uri'] . ' for user ' . $importer['uid'] . ' already existed under id ' . $previous['id']);
 394                                         continue;
 395                                 }
 396                                 $creation_dates[] = DateTimeFormat::utc($item['created']);
 397                         }
 398
 399                         $creator = XML::getFirstNodeValue($xpath, 'author/text()', $entry);
 400
 401                         if (empty($creator)) {
 402                                 $creator = XML::getFirstNodeValue($xpath, 'atom:author/atom:name/text()', $entry);
 403                         }
 404
 405                         if (empty($creator)) {
 406                                 $creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
 407                         }
 408
 409                         if ($creator != '') {
 410                                 $item['author-name'] = $creator;
 411                         }
 412
 413                         $creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
 414
 415                         if ($creator != '') {
 416                                 $item['author-name'] = $creator;
 417                         }
 418
 419                         /// @TODO ?
 420                         // <category>Ausland</category>
 421                         // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 422
 423                         $attachments = [];
 424
 425                         $enclosures = $xpath->query("enclosure|atom:link[@rel='enclosure']", $entry);
 426                         foreach ($enclosures as $enclosure) {
 427                                 $href = '';
 428                                 $length = null;
 429                                 $type = null;
 430
 431                                 foreach ($enclosure->attributes as $attribute) {
 432                                         if (in_array($attribute->name, ['url', 'href'])) {
 433                                                 $href = $attribute->textContent;
 434                                         } elseif ($attribute->name == 'length') {
 435                                                 $length = (int)$attribute->textContent;
 436                                         } elseif ($attribute->name == 'type') {
 437                                                 $type = $attribute->textContent;
 438                                         }
 439                                 }
 440
 441                                 if (!empty($href)) {
 442                                         $attachment = ['uri-id' => -1, 'type' => Post\Media::UNKNOWN, 'url' => $href, 'mimetype' => $type, 'size' => $length];
 443
 444                                         $attachment = Post\Media::fetchAdditionalData($attachment);
 445
 446                                         // By now we separate the visible media types (audio, video, image) from the rest
 447                                         // In the future we should try to avoid the DOCUMENT type and only use the real one - but not in the RC phase.
 448                                         if (!in_array($attachment['type'], [Post\Media::AUDIO, Post\Media::IMAGE, Post\Media::VIDEO])) {
 449                                                 $attachment['type'] = Post\Media::DOCUMENT;
 450                                         }
 451                                         $attachments[] = $attachment;
 452                                 }
 453                         }
 454
 455                         $taglist = [];
 456                         $categories = $xpath->query('category', $entry);
 457                         foreach ($categories as $category) {
 458                                 $taglist[] = $category->nodeValue;
 459                         }
 460
 461                         $body = trim(XML::getFirstNodeValue($xpath, 'atom:content/text()', $entry));
 462
 463                         if (empty($body)) {
 464                                 $body = trim(XML::getFirstNodeValue($xpath, 'content:encoded/text()', $entry));
 465                         }
 466
 467                         $summary = trim(XML::getFirstNodeValue($xpath, 'atom:summary/text()', $entry));
 468
 469                         if (empty($summary)) {
 470                                 $summary = trim(XML::getFirstNodeValue($xpath, 'description/text()', $entry));
 471                         }
 472
 473                         if (empty($body)) {
 474                                 $body = $summary;
 475                                 $summary = '';
 476                         }
 477
 478                         if ($body == $summary) {
 479                                 $summary = '';
 480                         }
 481
 482                         // remove the content of the title if it is identically to the body
 483                         // This helps with auto generated titles e.g. from tumblr
 484                         if (self::titleIsBody($item['title'], $body)) {
 485                                 $item['title'] = '';
 486                         }
 487                         $item['body'] = HTML::toBBCode($body, $basepath);
 488
 489                         // Remove tracking pixels
 490                         $item['body'] = preg_replace("/\[img=1x1\]([^\[\]]*)\[\/img\]/Usi", '', $item['body']);
 491
 492                         if (($item['body'] == '') && ($item['title'] != '')) {
 493                                 $item['body'] = $item['title'];
 494                                 $item['title'] = '';
 495                         }
 496
 497                         if ($dryRun) {
 498                                 $item['attachments'] = $attachments;
 499                                 $items[] = $item;
 500                                 break;
 501                         } elseif (!Item::isValid($item)) {
 502                                 Logger::info('Feed item is invalid', ['created' => $item['created'], 'uid' => $item['uid'], 'uri' => $item['uri']]);
 503                                 continue;
 504                         } elseif (Item::isTooOld($item)) {
 505                                 Logger::info('Feed is too old', ['created' => $item['created'], 'uid' => $item['uid'], 'uri' => $item['uri']]);
 506                                 continue;
 507                         }
 508
 509                         $preview = '';
 510                         if (!empty($contact['fetch_further_information']) && ($contact['fetch_further_information'] < 3)) {
 511                                 // Handle enclosures and treat them as preview picture
 512                                 foreach ($attachments as $attachment) {
 513                                         if ($attachment['mimetype'] == 'image/jpeg') {
 514                                                 $preview = $attachment['url'];
 515                                         }
 516                                 }
 517
 518                                 // Remove a possible link to the item itself
 519                                 $item['body'] = str_replace($item['plink'], '', $item['body']);
 520                                 $item['body'] = trim(preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item['body']));
 521
 522                                 // Replace the content when the title is longer than the body
 523                                 $replace = (strlen($item['title']) > strlen($item['body']));
 524
 525                                 // Replace it, when there is an image in the body
 526                                 if (strstr($item['body'], '[/img]')) {
 527                                         $replace = true;
 528                                 }
 529
 530                                 // Replace it, when there is a link in the body
 531                                 if (strstr($item['body'], '[/url]')) {
 532                                         $replace = true;
 533                                 }
 534
 535                                 $saved_body = $item['body'];
 536                                 $saved_title = $item['title'];
 537
 538                                 if ($replace) {
 539                                         $item['body'] = trim($item['title']);
 540                                 }
 541
 542                                 $data = ParseUrl::getSiteinfoCached($item['plink']);
 543                                 if (!empty($data['text']) && !empty($data['title']) && (mb_strlen($item['body']) < mb_strlen($data['text']))) {
 544                                         // When the fetched page info text is longer than the body, we do try to enhance the body
 545                                         if (!empty($item['body']) && (strpos($data['title'], $item['body']) === false) && (strpos($data['text'], $item['body']) === false)) {
 546                                                 // The body is not part of the fetched page info title or page info text. So we add the text to the body
 547                                                 $item['body'] .= "\n\n" . $data['text'];
 548                                         } else {
 549                                                 // Else we replace the body with the page info text
 550                                                 $item['body'] = $data['text'];
 551                                         }
 552                                 }
 553
 554                                 $data = PageInfo::queryUrl($item['plink'], false, $preview, ($contact['fetch_further_information'] == 2), $contact['ffi_keyword_denylist'] ?? '');
 555
 556                                 if (!empty($data)) {
 557                                         // Take the data that was provided by the feed if the query is empty
 558                                         if (($data['type'] == 'link') && empty($data['title']) && empty($data['text'])) {
 559                                                 $data['title'] = $saved_title;
 560                                                 $item['body'] = $saved_body;
 561                                         }
 562
 563                                         $data_text = strip_tags(trim($data['text'] ?? ''));
 564                                         $item_body = strip_tags(trim($item['body'] ?? ''));
 565
 566                                         if (!empty($data_text) && (($data_text == $item_body) || strstr($item_body, $data_text))) {
 567                                                 $data['text'] = '';
 568                                         }
 569
 570                                         // We always strip the title since it will be added in the page information
 571                                         $item['title'] = '';
 572                                         $item['body'] = $item['body'] . "\n" . PageInfo::getFooterFromData($data, false);
 573                                         $taglist = $contact['fetch_further_information'] == 2 ? PageInfo::getTagsFromUrl($item['plink'], $preview, $contact['ffi_keyword_denylist'] ?? '') : [];
 574                                         $item['object-type'] = Activity\ObjectType::BOOKMARK;
 575                                         $attachments = [];
 576
 577                                         foreach (['audio', 'video'] as $elementname) {
 578                                                 if (!empty($data[$elementname])) {
 579                                                         foreach ($data[$elementname] as $element) {
 580                                                                 if (!empty($element['src'])) {
 581                                                                         $src = $element['src'];
 582                                                                 } elseif (!empty($element['content'])) {
 583                                                                         $src = $element['content'];
 584                                                                 } else {
 585                                                                         continue;
 586                                                                 }
 587
 588                                                                 $attachments[] = [
 589                                                                         'type'        => ($elementname == 'audio') ? Post\Media::AUDIO : Post\Media::VIDEO,
 590                                                                         'url'         => $src,
 591                                                                         'preview'     => $element['image']       ?? null,
 592                                                                         'mimetype'    => $element['contenttype'] ?? null,
 593                                                                         'name'        => $element['name']        ?? null,
 594                                                                         'description' => $element['description'] ?? null,
 595                                                                 ];
 596                                                         }
 597                                                 }
 598                                         }
 599                                 }
 600                         } else {
 601                                 if (!empty($summary)) {
 602                                         $item['body'] = '[abstract]' . HTML::toBBCode($summary, $basepath) . "[/abstract]\n" . $item['body'];
 603                                 }
 604
 605                                 if (!empty($contact['fetch_further_information']) && ($contact['fetch_further_information'] == 3)) {
 606                                         if (empty($taglist)) {
 607                                                 $taglist = PageInfo::getTagsFromUrl($item['plink'], $preview, $contact['ffi_keyword_denylist'] ?? '');
 608                                         }
 609                                         $item['body'] .= "\n" . self::tagToString($taglist);
 610                                 } else {
 611                                         $taglist = [];
 612                                 }
 613
 614                                 // Add the link to the original feed entry if not present in feed
 615                                 if (($item['plink'] != '') && !strstr($item['body'], $item['plink']) && !in_array($item['plink'], array_column($attachments, 'url'))) {
 616                                         $item['body'] .= '[hr][url]' . $item['plink'] . '[/url]';
 617                                 }
 618                         }
 619
 620                         if (empty($item['title'])) {
 621                                 $item['post-type'] = Item::PT_NOTE;
 622                         }
 623
 624                         Logger::info('Stored feed', ['item' => $item]);
 625
 626                         $notify = Item::isRemoteSelf($contact, $item);
 627                         $item['wall'] = (bool)$notify;
 628
 629                         // Distributed items should have a well-formatted URI.
 630                         // Additionally, we have to avoid conflicts with identical URI between imported feeds and these items.
 631                         if ($notify) {
 632                                 $item['guid'] = Item::guidFromUri($orig_plink, DI::baseUrl()->getHostname());
 633                                 $item['uri'] = Item::newURI($item['guid']);
 634                                 unset($item['thr-parent']);
 635                                 unset($item['parent-uri']);
 636
 637                                 // Set the delivery priority for "remote self" to "medium"
 638                                 $notify = Worker::PRIORITY_MEDIUM;
 639                         }
 640
 641                         $condition = ['uid' => $item['uid'], 'uri' => $item['uri']];
 642                         if (!Post::exists($condition) && !Post\Delayed::exists($item['uri'], $item['uid'])) {
 643                                 if (!$notify) {
 644                                         Post\Delayed::publish($item, $notify, $taglist, $attachments);
 645                                 } else {
 646                                         $postings[] = ['item' => $item, 'notify' => $notify,
 647                                                 'taglist' => $taglist, 'attachments' => $attachments];
 648                                 }
 649                         } else {
 650                                 Logger::info('Post already created or exists in the delayed posts queue', ['uid' => $item['uid'], 'uri' => $item['uri']]);
 651                         }
 652                 }
 653
 654                 if (!empty($postings)) {
 655                         $min_posting = DI::config()->get('system', 'minimum_posting_interval', 0);
 656                         $total = count($postings);
 657                         if ($total > 1) {
 658                                 // Posts shouldn't be delayed more than a day
 659                                 $interval = min(1440, self::getPollInterval($contact));
 660                                 $delay = max(round(($interval * 60) / $total), 60 * $min_posting);
 661                                 Logger::info('Got posting delay', ['delay' => $delay, 'interval' => $interval, 'items' => $total, 'cid' => $contact['id'], 'url' => $contact['url']]);
 662                         } else {
 663                                 $delay = 0;
 664                         }
 665
 666                         $post_delay = 0;
 667
 668                         foreach ($postings as $posting) {
 669                                 if ($delay > 0) {
 670                                         $publish_time = time() + $post_delay;
 671                                         $post_delay += $delay;
 672                                 } else {
 673                                         $publish_time = time();
 674                                 }
 675
 676                                 $last_publish = DI::pConfig()->get($posting['item']['uid'], 'system', 'last_publish', 0, true);
 677                                 $next_publish = max($last_publish + (60 * $min_posting), time());
 678                                 if ($publish_time < $next_publish) {
 679                                         $publish_time = $next_publish;
 680                                 }
 681                                 $publish_at = date(DateTimeFormat::MYSQL, $publish_time);
 682
 683                                 if (Post\Delayed::add($posting['item']['uri'], $posting['item'], $posting['notify'], Post\Delayed::PREPARED, $publish_at, $posting['taglist'], $posting['attachments'])) {
 684                                         DI::pConfig()->set($item['uid'], 'system', 'last_publish', $publish_time);
 685                                 }
 686                         }
 687                 }
 688
 689                 if (!$dryRun && DI::config()->get('system', 'adjust_poll_frequency')) {
 690                         self::adjustPollFrequency($contact, $creation_dates);
 691                 }
 692
 693                 return ['header' => $author, 'items' => $items];
 694         }
 695
 696         /**
 697          * Automatically adjust the poll frequency according to the post frequency
 698          *
 699          * @param array $contact Contact array
 700          * @param array $creation_dates
 701          * @return void
 702          */
 703         private static function adjustPollFrequency(array $contact, array $creation_dates)
 704         {
 705                 if ($contact['network'] != Protocol::FEED) {
 706                         Logger::info('Contact is no feed, skip.', ['id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url'], 'network' => $contact['network']]);
 707                         return;
 708                 }
 709
 710                 if (!empty($creation_dates)) {
 711                         // Count the post frequency and the earliest and latest post date
 712                         $frequency = [];
 713                         $oldest = time();
 714                         $newest = 0;
 715                         $oldest_date = $newest_date = '';
 716
 717                         foreach ($creation_dates as $date) {
 718                                 $timestamp = strtotime($date);
 719                                 $day = intdiv($timestamp, 86400);
 720                                 $hour = $timestamp % 86400;
 721
 722                                 // Only have a look at values from the last seven days
 723                                 if (((time() / 86400) - $day) < 7) {
 724                                         if (empty($frequency[$day])) {
 725                                                 $frequency[$day] = ['count' => 1, 'low' => $hour, 'high' => $hour];
 726                                         } else {
 727                                                 ++$frequency[$day]['count'];
 728                                                 if ($frequency[$day]['low'] > $hour) {
 729                                                         $frequency[$day]['low'] = $hour;
 730                                                 }
 731                                                 if ($frequency[$day]['high'] < $hour) {
 732                                                         $frequency[$day]['high'] = $hour;
 733                                                 }
 734                                         }
 735                                 }
 736                                 if ($oldest > $day) {
 737                                         $oldest = $day;
 738                                         $oldest_date = $date;
 739                                 }
 740
 741                                 if ($newest < $day) {
 742                                         $newest = $day;
 743                                         $newest_date = $date;
 744                                 }
 745                         }
 746
 747                         if (count($creation_dates) == 1) {
 748                                 Logger::info('Feed had posted a single time, switching to daily polling', ['newest' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 749                                 $priority = 8; // Poll once a day
 750                         }
 751
 752                         if (empty($priority) && (((time() / 86400) - $newest) > 730)) {
 753                                 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']]);
 754                                 $priority = 10; // Poll every month
 755                         }
 756
 757                         if (empty($priority) && (((time() / 86400) - $newest) > 365)) {
 758                                 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']]);
 759                                 $priority = 9; // Poll every week
 760                         }
 761
 762                         if (empty($priority) && empty($frequency)) {
 763                                 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']]);
 764                                 $priority = 8; // Poll once a day
 765                         }
 766
 767                         if (empty($priority)) {
 768                                 // Calculate the highest "posts per day" value
 769                                 $max = 0;
 770                                 foreach ($frequency as $entry) {
 771                                         if (($entry['count'] == 1) || ($entry['high'] == $entry['low'])) {
 772                                                 continue;
 773                                         }
 774
 775                                         // We take the earliest and latest post day and interpolate the number of post per day
 776                                         // that would had been created with this post frequency
 777
 778                                         // Assume at least four hours between oldest and newest post per day - should be okay for news outlets
 779                                         $duration = max($entry['high'] - $entry['low'], 14400);
 780                                         $ppd = (86400 / $duration) * $entry['count'];
 781                                         if ($ppd > $max) {
 782                                                 $max = $ppd;
 783                                         }
 784                                 }
 785                                 if ($max > 48) {
 786                                         $priority = 1; // Poll every quarter hour
 787                                 } elseif ($max > 24) {
 788                                         $priority = 2; // Poll half an hour
 789                                 } elseif ($max > 12) {
 790                                         $priority = 3; // Poll hourly
 791                                 } elseif ($max > 8) {
 792                                         $priority = 4; // Poll every two hours
 793                                 } elseif ($max > 4) {
 794                                         $priority = 5; // Poll every three hours
 795                                 } elseif ($max > 2) {
 796                                         $priority = 6; // Poll every six hours
 797                                 } else {
 798                                         $priority = 7; // Poll twice a day
 799                                 }
 800                                 Logger::info('Calculated priority by the posts per day', ['priority' => $priority, 'max' => round($max, 2), 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 801                         }
 802                 } else {
 803                         Logger::info('No posts, switching to daily polling', ['id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 804                         $priority = 8; // Poll once a day
 805                 }
 806
 807                 if ($contact['rating'] != $priority) {
 808                         Logger::notice('Adjusting priority', ['old' => $contact['rating'], 'new' => $priority, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 809                         Contact::update(['rating' => $priority], ['id' => $contact['id']]);
 810                 }
 811         }
 812
 813         /**
 814          * Get the poll interval for the given contact array
 815          *
 816          * @param array $contact
 817          * @return int Poll interval in minutes
 818          */
 819         public static function getPollInterval(array $contact): int
 820         {
 821                 if (in_array($contact['network'], [Protocol::MAIL, Protocol::FEED])) {
 822                         $ratings = [0, 3, 7, 8, 9, 10];
 823                         if (DI::config()->get('system', 'adjust_poll_frequency') && ($contact['network'] == Protocol::FEED)) {
 824                                 $rating = $contact['rating'];
 825                         } elseif (array_key_exists($contact['priority'], $ratings)) {
 826                                 $rating = $ratings[$contact['priority']];
 827                         } else {
 828                                 $rating = -1;
 829                         }
 830                 } else {
 831                         // Check once a week per default for all other networks
 832                         $rating = 9;
 833                 }
 834
 835                 // Friendica and OStatus are checked once a day
 836                 if (in_array($contact['network'], [Protocol::DFRN, Protocol::OSTATUS])) {
 837                         $rating = 8;
 838                 }
 839
 840                 // Check archived contacts or contacts with unsupported protocols once a month
 841                 if ($contact['archive'] || in_array($contact['network'], [Protocol::ZOT, Protocol::PHANTOM])) {
 842                         $rating = 10;
 843                 }
 844
 845                 if ($rating < 0) {
 846                         return 0;
 847                 }
 848                 /*
 849                  * Based on $contact['priority'], should we poll this site now? Or later?
 850                  */
 851
 852                 $min_poll_interval = max(1, DI::config()->get('system', 'min_poll_interval'));
 853
 854                 $poll_intervals = [$min_poll_interval, 15, 30, 60, 120, 180, 360, 720 ,1440, 10080, 43200];
 855
 856                 //$poll_intervals = [$min_poll_interval . ' minute', '15 minute', '30 minute',
 857                 //      '1 hour', '2 hour', '3 hour', '6 hour', '12 hour' ,'1 day', '1 week', '1 month'];
 858
 859                 return $poll_intervals[$rating];
 860         }
 861
 862         /**
 863          * Convert a tag array to a tag string
 864          *
 865          * @param array $tags
 866          * @return string tag string
 867          */
 868         private static function tagToString(array $tags): string
 869         {
 870                 $tagstr = '';
 871
 872                 foreach ($tags as $tag) {
 873                         if ($tagstr != '') {
 874                                 $tagstr .= ', ';
 875                         }
 876
 877                         $tagstr .= '#[url=' . DI::baseUrl() . '/search?tag=' . urlencode($tag) . ']' . $tag . '[/url]';
 878                 }
 879
 880                 return $tagstr;
 881         }
 882
 883         private static function titleIsBody(string $title, string $body): bool
 884         {
 885                 $title = strip_tags($title);
 886                 $title = trim($title);
 887                 $title = html_entity_decode($title, ENT_QUOTES, 'UTF-8');
 888                 $title = str_replace(["\n", "\r", "\t", " "], ['', '', '', ''], $title);
 889
 890                 $body = strip_tags($body);
 891                 $body = trim($body);
 892                 $body = html_entity_decode($body, ENT_QUOTES, 'UTF-8');
 893                 $body = str_replace(["\n", "\r", "\t", " "], ['', '', '', ''], $body);
 894
 895                 if (strlen($title) < strlen($body)) {
 896                         $body = substr($body, 0, strlen($title));
 897                 }
 898
 899                 if (($title != $body) && (substr($title, -3) == '...')) {
 900                         $pos = strrpos($title, '...');
 901                         if ($pos > 0) {
 902                                 $title = substr($title, 0, $pos);
 903                                 $body = substr($body, 0, $pos);
 904                         }
 905                 }
 906                 return ($title == $body);
 907         }
 908
 909         /**
 910          * Creates the Atom feed for a given nickname
 911          *
 912          * Supported filters:
 913          * - activity (default): all the public posts
 914          * - posts: all the public top-level posts
 915          * - comments: all the public replies
 916          *
 917          * Updates the provided last_update parameter if the result comes from the
 918          * cache or it is empty
 919          *
 920          * @param array   $owner       owner-view record of the feed owner
 921          * @param string  $last_update Date of the last update
 922          * @param integer $max_items   Number of maximum items to fetch
 923          * @param string  $filter      Feed items filter (activity, posts or comments)
 924          * @param boolean $nocache     Wether to bypass caching
 925          *
 926          * @return string Atom feed
 927          * @throws HTTPException\InternalServerErrorException
 928          * @throws \ImagickException
 929          */
 930         public static function atom(array $owner, string $last_update, int $max_items = 300, string $filter = 'activity', bool $nocache = false)
 931         {
 932                 $stamp = microtime(true);
 933
 934                 $cachekey = 'feed:feed:' . $owner['nickname'] . ':' . $filter . ':' . $last_update;
 935
 936                 // Display events in the user's timezone
 937                 if (strlen($owner['timezone'])) {
 938                         DI::app()->setTimeZone($owner['timezone']);
 939                 }
 940
 941                 $previous_created = $last_update;
 942
 943                 // Don't cache when the last item was posted less then 15 minutes ago (Cache duration)
 944                 if ((time() - strtotime($owner['last-item'])) < 15*60) {
 945                         $result = DI::cache()->get($cachekey);
 946                         if (!$nocache && !is_null($result)) {
 947                                 Logger::info('Cached feed duration', ['seconds' => number_format(microtime(true) - $stamp, 3), 'nick' => $owner['nickname'], 'filter' => $filter, 'created' => $previous_created]);
 948                                 return $result['feed'];
 949                         }
 950                 }
 951
 952                 $check_date = empty($last_update) ? '' : DateTimeFormat::utc($last_update);
 953                 $authorid = Contact::getIdForURL($owner['url']);
 954
 955                 $condition = ["`uid` = ? AND `received` > ? AND NOT `deleted` AND `gravity` IN (?, ?)
 956                         AND `private` != ? AND `visible` AND `wall` AND `parent-network` IN (?, ?, ?, ?)",
 957                         $owner['uid'], $check_date, Item::GRAVITY_PARENT, Item::GRAVITY_COMMENT,
 958                         Item::PRIVATE, Protocol::ACTIVITYPUB,
 959                         Protocol::OSTATUS, Protocol::DFRN, Protocol::DIASPORA];
 960
 961                 if ($filter === 'comments') {
 962                         $condition[0] .= " AND `gravity` = ? ";
 963                         $condition[] = Item::GRAVITY_COMMENT;
 964                 }
 965
 966                 if ($owner['account-type'] != User::ACCOUNT_TYPE_COMMUNITY) {
 967                         $condition[0] .= " AND `contact-id` = ? AND `author-id` = ?";
 968                         $condition[] = $owner['id'];
 969                         $condition[] = $authorid;
 970                 }
 971
 972                 $params = ['order' => ['received' => true], 'limit' => $max_items];
 973
 974                 if ($filter === 'posts') {
 975                         $ret = Post::selectThread(Item::DELIVER_FIELDLIST, $condition, $params);
 976                 } else {
 977                         $ret = Post::select(Item::DELIVER_FIELDLIST, $condition, $params);
 978                 }
 979
 980                 $items = Post::toArray($ret);
 981
 982                 $doc = new DOMDocument('1.0', 'utf-8');
 983                 $doc->formatOutput = true;
 984
 985                 $root = self::addHeader($doc, $owner, $filter);
 986
 987                 foreach ($items as $item) {
 988                         $entry = self::noteEntry($doc, $item, $owner);
 989                         $root->appendChild($entry);
 990
 991                         if ($last_update < $item['created']) {
 992                                 $last_update = $item['created'];
 993                         }
 994                 }
 995
 996                 $feeddata = trim($doc->saveXML());
 997
 998                 $msg = ['feed' => $feeddata, 'last_update' => $last_update];
 999                 DI::cache()->set($cachekey, $msg, Duration::QUARTER_HOUR);
1000
1001                 Logger::info('Feed duration', ['seconds' => number_format(microtime(true) - $stamp, 3), 'nick' => $owner['nickname'], 'filter' => $filter, 'created' => $previous_created]);
1002
1003                 return $feeddata;
1004         }
1005
1006         /**
1007          * Adds the header elements to the XML document
1008          *
1009          * @param DOMDocument $doc       XML document
1010          * @param array       $owner     Contact data of the poster
1011          * @param string      $filter    The related feed filter (activity, posts or comments)
1012          *
1013          * @return DOMElement Header root element
1014          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1015          */
1016         private static function addHeader(DOMDocument $doc, array $owner, string $filter): DOMElement
1017         {
1018                 $root = $doc->createElementNS(ActivityNamespace::ATOM1, 'feed');
1019                 $doc->appendChild($root);
1020
1021                 $title = '';
1022                 $selfUri = '/feed/' . $owner['nick'] . '/';
1023                 switch ($filter) {
1024                         case 'activity':
1025                                 $title = DI::l10n()->t('%s\'s timeline', $owner['name']);
1026                                 $selfUri .= $filter;
1027                                 break;
1028                         case 'posts':
1029                                 $title = DI::l10n()->t('%s\'s posts', $owner['name']);
1030                                 break;
1031                         case 'comments':
1032                                 $title = DI::l10n()->t('%s\'s comments', $owner['name']);
1033                                 $selfUri .= $filter;
1034                                 break;
1035                 }
1036
1037                 $attributes = ['uri' => 'https://friendi.ca', 'version' => App::VERSION . '-' . DB_UPDATE_VERSION];
1038                 XML::addElement($doc, $root, 'generator', App::PLATFORM, $attributes);
1039                 XML::addElement($doc, $root, 'id', DI::baseUrl() . '/profile/' . $owner['nick']);
1040                 XML::addElement($doc, $root, 'title', $title);
1041                 XML::addElement($doc, $root, 'subtitle', sprintf("Updates from %s on %s", $owner['name'], DI::config()->get('config', 'sitename')));
1042                 XML::addElement($doc, $root, 'logo', User::getAvatarUrl($owner, Proxy::SIZE_SMALL));
1043                 XML::addElement($doc, $root, 'updated', DateTimeFormat::utcNow(DateTimeFormat::ATOM));
1044
1045                 $author = self::addAuthor($doc, $owner);
1046                 $root->appendChild($author);
1047
1048                 $attributes = ['href' => $owner['url'], 'rel' => 'alternate', 'type' => 'text/html'];
1049                 XML::addElement($doc, $root, 'link', '', $attributes);
1050
1051                 OStatus::addHubLink($doc, $root, $owner['nick']);
1052
1053                 $attributes = ['href' => DI::baseUrl() . $selfUri, 'rel' => 'self', 'type' => 'application/atom+xml'];
1054                 XML::addElement($doc, $root, 'link', '', $attributes);
1055
1056                 return $root;
1057         }
1058
1059         /**
1060          * Adds the author element to the XML document
1061          *
1062          * @param DOMDocument $doc          XML document
1063          * @param array       $owner        Contact data of the poster
1064          * @return DOMElement author element
1065          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1066          */
1067         private static function addAuthor(DOMDocument $doc, array $owner): DOMElement
1068         {
1069                 $author = $doc->createElement('author');
1070                 XML::addElement($doc, $author, 'uri', $owner['url']);
1071                 XML::addElement($doc, $author, 'name', $owner['nick']);
1072                 XML::addElement($doc, $author, 'email', $owner['addr']);
1073
1074                 return $author;
1075         }
1076
1077         /**
1078          * Adds a regular entry element
1079          *
1080          * @param DOMDocument $doc       XML document
1081          * @param array       $item      Data of the item that is to be posted
1082          * @param array       $owner     Contact data of the poster
1083          * @param bool        $toplevel  Is it for en entry element (false) or a feed entry (true)?
1084          * @return DOMElement Entry element
1085          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1086          * @throws \ImagickException
1087          */
1088         private static function noteEntry(DOMDocument $doc, array $item, array $owner): DOMElement
1089         {
1090                 if (($item['gravity'] != Item::GRAVITY_PARENT) && (Strings::normaliseLink($item['author-link']) != Strings::normaliseLink($owner['url']))) {
1091                         Logger::info('Feed entry author does not match feed owner', ['owner' => $owner['url'], 'author' => $item['author-link']]);
1092                 }
1093
1094                 $entry = OStatus::entryHeader($doc, $owner, $item, false);
1095
1096                 self::entryContent($doc, $entry, $item, self::getTitle($item), '', true);
1097
1098                 self::entryFooter($doc, $entry, $item, $owner);
1099
1100                 return $entry;
1101         }
1102
1103         /**
1104          * Adds elements to the XML document
1105          *
1106          * @param DOMDocument $doc       XML document
1107          * @param \DOMElement $entry     Entry element where the content is added
1108          * @param array       $item      Data of the item that is to be posted
1109          * @param array       $owner     Contact data of the poster
1110          * @param string      $title     Title for the post
1111          * @param string      $verb      The activity verb
1112          * @param bool        $complete  Add the "status_net" element?
1113          * @return void
1114          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1115          */
1116         private static function entryContent(DOMDocument $doc, DOMElement $entry, array $item, $title, string $verb = '', bool $complete = true)
1117         {
1118                 if ($verb == '') {
1119                         $verb = OStatus::constructVerb($item);
1120                 }
1121
1122                 XML::addElement($doc, $entry, 'id', $item['uri']);
1123                 XML::addElement($doc, $entry, 'title', html_entity_decode($title, ENT_QUOTES, 'UTF-8'));
1124
1125                 $body = OStatus::formatPicturePost($item['body'], $item['uri-id']);
1126
1127                 $body = BBCode::convertForUriId($item['uri-id'], $body, BBCode::ACTIVITYPUB);
1128
1129                 XML::addElement($doc, $entry, 'content', $body, ['type' => 'html']);
1130
1131                 XML::addElement($doc, $entry, 'link', '', ['rel' => 'alternate', 'type' => 'text/html',
1132                                                                 'href' => DI::baseUrl() . '/display/' . $item['guid']]
1133                 );
1134
1135                 XML::addElement($doc, $entry, 'published', DateTimeFormat::utc($item['created'] . '+00:00', DateTimeFormat::ATOM));
1136                 XML::addElement($doc, $entry, 'updated', DateTimeFormat::utc($item['edited'] . '+00:00', DateTimeFormat::ATOM));
1137         }
1138
1139         /**
1140          * Adds the elements at the foot of an entry to the XML document
1141          *
1142          * @param DOMDocument $doc       XML document
1143          * @param object      $entry     The entry element where the elements are added
1144          * @param array       $item      Data of the item that is to be posted
1145          * @param array       $owner     Contact data of the poster
1146          * @param bool        $complete  default true
1147          * @return void
1148          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1149          */
1150         private static function entryFooter(DOMDocument $doc, $entry, array $item, array $owner)
1151         {
1152                 $mentioned = [];
1153
1154                 if ($item['gravity'] != Item::GRAVITY_PARENT) {
1155                         $parent = Post::selectFirst(['guid', 'author-link', 'owner-link'], ['id' => $item['parent']]);
1156
1157                         $thrparent = Post::selectFirst(['guid', 'author-link', 'owner-link', 'plink'], ['uid' => $owner['uid'], 'uri' => $item['thr-parent']]);
1158
1159                         if (DBA::isResult($thrparent)) {
1160                                 $mentioned[$thrparent['author-link']] = $thrparent['author-link'];
1161                                 $mentioned[$thrparent['owner-link']]  = $thrparent['owner-link'];
1162                                 $parent_plink                         = $thrparent['plink'];
1163                         } elseif (DBA::isResult($parent)) {
1164                                 $mentioned[$parent['author-link']] = $parent['author-link'];
1165                                 $mentioned[$parent['owner-link']]  = $parent['owner-link'];
1166                                 $parent_plink                      = DI::baseUrl() . '/display/' . $parent['guid'];
1167                         } else {
1168                                 DI::logger()->notice('Missing parent and thr-parent for child item', ['item' => $item]);
1169                         }
1170
1171                         if (isset($parent_plink)) {
1172                                 $attributes = [
1173                                         'ref'  => $item['thr-parent'],
1174                                         'href' => $parent_plink];
1175                                 XML::addElement($doc, $entry, 'thr:in-reply-to', '', $attributes);
1176
1177                                 $attributes = [
1178                                         'rel'  => 'related',
1179                                         'href' => $parent_plink];
1180                                 XML::addElement($doc, $entry, 'link', '', $attributes);
1181                         }
1182                 }
1183
1184                 // uri-id isn't present for follow entry pseudo-items
1185                 $tags = Tag::getByURIId($item['uri-id'] ?? 0);
1186                 foreach ($tags as $tag) {
1187                         $mentioned[$tag['url']] = $tag['url'];
1188                 }
1189
1190                 foreach ($tags as $tag) {
1191                         if ($tag['type'] == Tag::HASHTAG) {
1192                                 XML::addElement($doc, $entry, 'category', '', ['term' => $tag['name']]);
1193                         }
1194                 }
1195
1196                 OStatus::getAttachment($doc, $entry, $item);
1197         }
1198
1199         /**
1200          * Fetch or create title for feed entry
1201          *
1202          * @param array $item
1203          * @return string title
1204          */
1205         private static function getTitle(array $item): string
1206         {
1207                 if ($item['title'] != '') {
1208                         return BBCode::convertForUriId($item['uri-id'], $item['title'], BBCode::ACTIVITYPUB);
1209                 }
1210
1211                 // Fetch information about the post
1212                 $siteinfo = BBCode::getAttachedData($item['body']);
1213                 if (isset($siteinfo['title'])) {
1214                         return $siteinfo['title'];
1215                 }
1216
1217                 // If no bookmark is found then take the first line
1218                 // Remove the share element before fetching the first line
1219                 $title = trim(preg_replace("/\[share.*?\](.*?)\[\/share\]/ism", "\n$1\n", $item['body']));
1220
1221                 $title = BBCode::toPlaintext($title)."\n";
1222                 $pos = strpos($title, "\n");
1223                 $trailer = '';
1224                 if (($pos == 0) || ($pos > 100)) {
1225                         $pos = 100;
1226                         $trailer = '...';
1227                 }
1228
1229                 return substr($title, 0, $pos) . $trailer;
1230         }
1231 }