src/Protocol/Feed.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2022, 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\Util\DateTimeFormat;
  44 use Friendica\Util\Network;
  45 use Friendica\Util\ParseUrl;
  46 use Friendica\Util\Proxy;
  47 use Friendica\Util\Strings;
  48 use Friendica\Util\XML;
  49 use GuzzleHttp\Exception\TransferException;
  50
  51 /**
  52  * This class contain functions to import feeds (RSS/RDF/Atom)
  53  */
  54 class Feed
  55 {
  56         /**
  57          * Read a RSS/RDF/Atom feed and create an item entry for it
  58          *
  59          * @param string $xml      The feed data
  60          * @param array  $importer The user record of the importer
  61          * @param array  $contact  The contact record of the feed
  62          *
  63          * @return array Returns the header and the first item in dry run mode
  64          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  65          */
  66         public static function import(string $xml, array $importer = [], array $contact = []): array
  67         {
  68                 $dryRun = empty($importer) && empty($contact);
  69
  70                 if ($dryRun) {
  71                         Logger::info("Test Atom/RSS feed");
  72                 } else {
  73                         Logger::info('Import Atom/RSS feed "' . $contact['name'] . '" (Contact ' . $contact['id'] . ') for user ' . $importer['uid']);
  74                 }
  75
  76                 $xml = trim($xml);
  77
  78                 if (empty($xml)) {
  79                         Logger::info('XML is empty.');
  80                         return [];
  81                 }
  82
  83                 if (!empty($contact['poll'])) {
  84                         $basepath = $contact['poll'];
  85                 } elseif (!empty($contact['url'])) {
  86                         $basepath = $contact['url'];
  87                 } else {
  88                         $basepath = '';
  89                 }
  90
  91                 $doc = new DOMDocument();
  92                 @$doc->loadXML($xml);
  93                 $xpath = new DOMXPath($doc);
  94                 $xpath->registerNamespace('atom', ActivityNamespace::ATOM1);
  95                 $xpath->registerNamespace('dc', 'http://purl.org/dc/elements/1.1/');
  96                 $xpath->registerNamespace('content', 'http://purl.org/rss/1.0/modules/content/');
  97                 $xpath->registerNamespace('rdf', 'http://www.w3.org/1999/02/22-rdf-syntax-ns#');
  98                 $xpath->registerNamespace('rss', 'http://purl.org/rss/1.0/');
  99                 $xpath->registerNamespace('media', 'http://search.yahoo.com/mrss/');
 100                 $xpath->registerNamespace('poco', ActivityNamespace::POCO);
 101
 102                 $author = [];
 103                 $entries = null;
 104                 $protocol = Conversation::PARCEL_UNKNOWN;
 105
 106                 // Is it RDF?
 107                 if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
 108                         $protocol = Conversation::PARCEL_RDF;
 109                         $author['author-link'] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:link/text()');
 110                         $author['author-name'] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:title/text()');
 111
 112                         if (empty($author['author-name'])) {
 113                                 $author['author-name'] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:description/text()');
 114                         }
 115                         $entries = $xpath->query('/rdf:RDF/rss:item');
 116                 }
 117
 118                 // Is it Atom?
 119                 if ($xpath->query('/atom:feed')->length > 0) {
 120                         $protocol = Conversation::PARCEL_ATOM;
 121                         $alternate = XML::getFirstAttributes($xpath, "atom:link[@rel='alternate']");
 122                         if (is_object($alternate)) {
 123                                 foreach ($alternate as $attribute) {
 124                                         if ($attribute->name == 'href') {
 125                                                 $author['author-link'] = $attribute->textContent;
 126                                         }
 127                                 }
 128                         }
 129
 130                         if (empty($author['author-link'])) {
 131                                 $self = XML::getFirstAttributes($xpath, "atom:link[@rel='self']");
 132                                 if (is_object($self)) {
 133                                         foreach ($self as $attribute) {
 134                                                 if ($attribute->name == 'href') {
 135                                                         $author['author-link'] = $attribute->textContent;
 136                                                 }
 137                                         }
 138                                 }
 139                         }
 140
 141                         if (empty($author['author-link'])) {
 142                                 $author['author-link'] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:id/text()');
 143                         }
 144                         $author['author-avatar'] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:logo/text()');
 145
 146                         $author['author-name'] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:title/text()');
 147
 148                         if (empty($author['author-name'])) {
 149                                 $author['author-name'] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:subtitle/text()');
 150                         }
 151
 152                         if (empty($author['author-name'])) {
 153                                 $author['author-name'] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:name/text()');
 154                         }
 155
 156                         $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:displayName/text()');
 157                         if ($value != '') {
 158                                 $author['author-name'] = $value;
 159                         }
 160
 161                         if ($dryRun) {
 162                                 $author['author-id'] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:id/text()');
 163
 164                                 // See https://tools.ietf.org/html/rfc4287#section-3.2.2
 165                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/atom:uri/text()');
 166                                 if ($value != '') {
 167                                         $author['author-link'] = $value;
 168                                 }
 169
 170                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:preferredUsername/text()');
 171                                 if ($value != '') {
 172                                         $author['author-nick'] = $value;
 173                                 }
 174
 175                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:address/poco:formatted/text()');
 176                                 if ($value != '') {
 177                                         $author['author-location'] = $value;
 178                                 }
 179
 180                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:note/text()');
 181                                 if ($value != '') {
 182                                         $author['author-about'] = $value;
 183                                 }
 184
 185                                 $avatar = XML::getFirstAttributes($xpath, "atom:author/atom:link[@rel='avatar']");
 186                                 if (is_object($avatar)) {
 187                                         foreach ($avatar as $attribute) {
 188                                                 if ($attribute->name == 'href') {
 189                                                         $author['author-avatar'] = $attribute->textContent;
 190                                                 }
 191                                         }
 192                                 }
 193                         }
 194
 195                         $author['edited'] = $author['created'] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:updated/text()');
 196
 197                         $author['app'] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:generator/text()');
 198
 199                         $entries = $xpath->query('/atom:feed/atom:entry');
 200                 }
 201
 202                 // Is it RSS?
 203                 if ($xpath->query('/rss/channel')->length > 0) {
 204                         $protocol = Conversation::PARCEL_RSS;
 205                         $author['author-link'] = XML::getFirstNodeValue($xpath, '/rss/channel/link/text()');
 206
 207                         $author['author-name'] = XML::getFirstNodeValue($xpath, '/rss/channel/title/text()');
 208
 209                         if (empty($author['author-name'])) {
 210                                 $author['author-name'] = XML::getFirstNodeValue($xpath, '/rss/channel/copyright/text()');
 211                         }
 212
 213                         if (empty($author['author-name'])) {
 214                                 $author['author-name'] = XML::getFirstNodeValue($xpath, '/rss/channel/description/text()');
 215                         }
 216
 217                         $author['author-avatar'] = XML::getFirstNodeValue($xpath, '/rss/channel/image/url/text()');
 218
 219                         if (empty($author['author-avatar'])) {
 220                                 $avatar = XML::getFirstAttributes($xpath, '/rss/channel/itunes:image');
 221                                 if (is_object($avatar)) {
 222                                         foreach ($avatar as $attribute) {
 223                                                 if ($attribute->name == 'href') {
 224                                                         $author['author-avatar'] = $attribute->textContent;
 225                                                 }
 226                                         }
 227                                 }
 228                         }
 229
 230                         $author['author-about'] = HTML::toBBCode(XML::getFirstNodeValue($xpath, '/rss/channel/description/text()'), $basepath);
 231
 232                         if (empty($author['author-about'])) {
 233                                 $author['author-about'] = XML::getFirstNodeValue($xpath, '/rss/channel/itunes:summary/text()');
 234                         }
 235
 236                         $author['edited'] = $author['created'] = XML::getFirstNodeValue($xpath, '/rss/channel/pubDate/text()');
 237
 238                         $author['app'] = XML::getFirstNodeValue($xpath, '/rss/channel/generator/text()');
 239
 240                         $entries = $xpath->query('/rss/channel/item');
 241                 }
 242
 243                 if (!$dryRun) {
 244                         $author['author-link'] = $contact['url'];
 245
 246                         if (empty($author['author-name'])) {
 247                                 $author['author-name'] = $contact['name'];
 248                         }
 249
 250                         $author['author-avatar'] = $contact['thumb'];
 251
 252                         $author['owner-link'] = $contact['url'];
 253                         $author['owner-name'] = $contact['name'];
 254                         $author['owner-avatar'] = $contact['thumb'];
 255                 }
 256
 257                 $header = [
 258                         'uid'         => $importer['uid'] ?? 0,
 259                         'network'     => Protocol::FEED,
 260                         'wall'        => 0,
 261                         'origin'      => 0,
 262                         'gravity'     => Item::GRAVITY_PARENT,
 263                         'private'     => Item::PUBLIC,
 264                         'verb'        => Activity::POST,
 265                         'object-type' => Activity\ObjectType::NOTE,
 266                         'post-type'   => Item::PT_ARTICLE,
 267                         'contact-id'  => $contact['id'] ?? 0,
 268                 ];
 269
 270                 $datarray['protocol'] = $protocol;
 271                 $datarray['direction'] = Conversation::PULL;
 272
 273                 if (!is_object($entries)) {
 274                         Logger::info("There are no entries in this feed.");
 275                         return [];
 276                 }
 277
 278                 $items = [];
 279                 $creation_dates = [];
 280
 281                 // Limit the number of items that are about to be fetched
 282                 $total_items = ($entries->length - 1);
 283                 $max_items = DI::config()->get('system', 'max_feed_items');
 284                 if (($max_items > 0) && ($total_items > $max_items)) {
 285                         $total_items = $max_items;
 286                 }
 287
 288                 $postings = [];
 289
 290                 // Importing older entries first
 291                 for ($i = $total_items; $i >= 0; --$i) {
 292                         $entry = $entries->item($i);
 293
 294                         $item = array_merge($header, $author);
 295
 296                         $alternate = XML::getFirstAttributes($xpath, "atom:link[@rel='alternate']", $entry);
 297                         if (!is_object($alternate)) {
 298                                 $alternate = XML::getFirstAttributes($xpath, 'atom:link', $entry);
 299                         }
 300                         if (is_object($alternate)) {
 301                                 foreach ($alternate as $attribute) {
 302                                         if ($attribute->name == 'href') {
 303                                                 $item['plink'] = $attribute->textContent;
 304                                         }
 305                                 }
 306                         }
 307
 308                         if (empty($item['plink'])) {
 309                                 $item['plink'] = XML::getFirstNodeValue($xpath, 'link/text()', $entry);
 310                         }
 311
 312                         if (empty($item['plink'])) {
 313                                 $item['plink'] = XML::getFirstNodeValue($xpath, 'rss:link/text()', $entry);
 314                         }
 315
 316                         // Add the base path if missing
 317                         $item['plink'] = Network::addBasePath($item['plink'], $basepath);
 318
 319                         $item['uri'] = XML::getFirstNodeValue($xpath, 'atom:id/text()', $entry);
 320
 321                         $guid = XML::getFirstNodeValue($xpath, 'guid/text()', $entry);
 322                         if (!empty($guid)) {
 323                                 $item['uri'] = $guid;
 324
 325                                 // Don't use the GUID value directly but instead use it as a basis for the GUID
 326                                 $item['guid'] = Item::guidFromUri($guid, parse_url($guid, PHP_URL_HOST) ?? parse_url($item['plink'], PHP_URL_HOST));
 327                         }
 328
 329                         if (empty($item['uri'])) {
 330                                 $item['uri'] = $item['plink'];
 331                         }
 332
 333                         $orig_plink = $item['plink'];
 334
 335                         try {
 336                                 $item['plink'] = DI::httpClient()->finalUrl($item['plink']);
 337                         } catch (TransferException $exception) {
 338                                 Logger::notice('Item URL couldn\'t get expanded', ['url' => $item['plink'], 'exception' => $exception]);
 339                         }
 340
 341                         $item['title'] = XML::getFirstNodeValue($xpath, 'atom:title/text()', $entry);
 342
 343                         if (empty($item['title'])) {
 344                                 $item['title'] = XML::getFirstNodeValue($xpath, 'title/text()', $entry);
 345                         }
 346
 347                         if (empty($item['title'])) {
 348                                 $item['title'] = XML::getFirstNodeValue($xpath, 'rss:title/text()', $entry);
 349                         }
 350
 351                         if (empty($item['title'])) {
 352                                 $item['title'] = XML::getFirstNodeValue($xpath, 'itunes:title/text()', $entry);
 353                         }
 354
 355                         $item['title'] = html_entity_decode($item['title'], ENT_QUOTES, 'UTF-8');
 356
 357                         $published = XML::getFirstNodeValue($xpath, 'atom:published/text()', $entry);
 358
 359                         if (empty($published)) {
 360                                 $published = XML::getFirstNodeValue($xpath, 'pubDate/text()', $entry);
 361                         }
 362
 363                         if (empty($published)) {
 364                                 $published = XML::getFirstNodeValue($xpath, 'dc:date/text()', $entry);
 365                         }
 366
 367                         $updated = XML::getFirstNodeValue($xpath, 'atom:updated/text()', $entry);
 368
 369                         if (empty($updated) && !empty($published)) {
 370                                 $updated = $published;
 371                         }
 372
 373                         if (empty($published) && !empty($updated)) {
 374                                 $published = $updated;
 375                         }
 376
 377                         if ($published != '') {
 378                                 $item['created'] = $published;
 379                         }
 380
 381                         if ($updated != '') {
 382                                 $item['edited'] = $updated;
 383                         }
 384
 385                         if (!$dryRun) {
 386                                 $condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
 387                                         $importer['uid'], $item['uri'], Protocol::FEED, Protocol::DFRN];
 388                                 $previous = Post::selectFirst(['id', 'created'], $condition);
 389                                 if (DBA::isResult($previous)) {
 390                                         // Use the creation date when the post had been stored. It can happen this date changes in the feed.
 391                                         $creation_dates[] = $previous['created'];
 392                                         Logger::info('Item with URI ' . $item['uri'] . ' for user ' . $importer['uid'] . ' already existed under id ' . $previous['id']);
 393                                         continue;
 394                                 }
 395                                 $creation_dates[] = DateTimeFormat::utc($item['created']);
 396                         }
 397
 398                         $creator = XML::getFirstNodeValue($xpath, 'author/text()', $entry);
 399
 400                         if (empty($creator)) {
 401                                 $creator = XML::getFirstNodeValue($xpath, 'atom:author/atom:name/text()', $entry);
 402                         }
 403
 404                         if (empty($creator)) {
 405                                 $creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
 406                         }
 407
 408                         if ($creator != '') {
 409                                 $item['author-name'] = $creator;
 410                         }
 411
 412                         $creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
 413
 414                         if ($creator != '') {
 415                                 $item['author-name'] = $creator;
 416                         }
 417
 418                         /// @TODO ?
 419                         // <category>Ausland</category>
 420                         // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 421
 422                         $attachments = [];
 423
 424                         $enclosures = $xpath->query("enclosure|atom:link[@rel='enclosure']", $entry);
 425                         foreach ($enclosures as $enclosure) {
 426                                 $href = '';
 427                                 $length = null;
 428                                 $type = null;
 429
 430                                 foreach ($enclosure->attributes as $attribute) {
 431                                         if (in_array($attribute->name, ['url', 'href'])) {
 432                                                 $href = $attribute->textContent;
 433                                         } elseif ($attribute->name == 'length') {
 434                                                 $length = (int)$attribute->textContent;
 435                                         } elseif ($attribute->name == 'type') {
 436                                                 $type = $attribute->textContent;
 437                                         }
 438                                 }
 439
 440                                 if (!empty($href)) {
 441                                         $attachment = ['uri-id' => -1, 'type' => Post\Media::UNKNOWN, 'url' => $href, 'mimetype' => $type, 'size' => $length];
 442
 443                                         $attachment = Post\Media::fetchAdditionalData($attachment);
 444
 445                                         // By now we separate the visible media types (audio, video, image) from the rest
 446                                         // In the future we should try to avoid the DOCUMENT type and only use the real one - but not in the RC phase.
 447                                         if (!in_array($attachment['type'], [Post\Media::AUDIO, Post\Media::IMAGE, Post\Media::VIDEO])) {
 448                                                 $attachment['type'] = Post\Media::DOCUMENT;
 449                                         }
 450                                         $attachments[] = $attachment;
 451                                 }
 452                         }
 453
 454                         $taglist = [];
 455                         $categories = $xpath->query('category', $entry);
 456                         foreach ($categories as $category) {
 457                                 $taglist[] = $category->nodeValue;
 458                         }
 459
 460                         $body = trim(XML::getFirstNodeValue($xpath, 'atom:content/text()', $entry));
 461
 462                         if (empty($body)) {
 463                                 $body = trim(XML::getFirstNodeValue($xpath, 'content:encoded/text()', $entry));
 464                         }
 465
 466                         $summary = trim(XML::getFirstNodeValue($xpath, 'atom:summary/text()', $entry));
 467
 468                         if (empty($summary)) {
 469                                 $summary = trim(XML::getFirstNodeValue($xpath, 'description/text()', $entry));
 470                         }
 471
 472                         if (empty($body)) {
 473                                 $body = $summary;
 474                                 $summary = '';
 475                         }
 476
 477                         if ($body == $summary) {
 478                                 $summary = '';
 479                         }
 480
 481                         // remove the content of the title if it is identically to the body
 482                         // This helps with auto generated titles e.g. from tumblr
 483                         if (self::titleIsBody($item['title'], $body)) {
 484                                 $item['title'] = '';
 485                         }
 486                         $item['body'] = HTML::toBBCode($body, $basepath);
 487
 488                         // Remove tracking pixels
 489                         $item['body'] = preg_replace("/\[img=1x1\]([^\[\]]*)\[\/img\]/Usi", '', $item['body']);
 490
 491                         if (($item['body'] == '') && ($item['title'] != '')) {
 492                                 $item['body'] = $item['title'];
 493                                 $item['title'] = '';
 494                         }
 495
 496                         if ($dryRun) {
 497                                 $item['attachments'] = $attachments;
 498                                 $items[] = $item;
 499                                 break;
 500                         } elseif (!Item::isValid($item)) {
 501                                 Logger::info('Feed item is invalid', ['created' => $item['created'], 'uid' => $item['uid'], 'uri' => $item['uri']]);
 502                                 continue;
 503                         } elseif (Item::isTooOld($item)) {
 504                                 Logger::info('Feed is too old', ['created' => $item['created'], 'uid' => $item['uid'], 'uri' => $item['uri']]);
 505                                 continue;
 506                         }
 507
 508                         $preview = '';
 509                         if (!empty($contact['fetch_further_information']) && ($contact['fetch_further_information'] < 3)) {
 510                                 // Handle enclosures and treat them as preview picture
 511                                 foreach ($attachments as $attachment) {
 512                                         if ($attachment['mimetype'] == 'image/jpeg') {
 513                                                 $preview = $attachment['url'];
 514                                         }
 515                                 }
 516
 517                                 // Remove a possible link to the item itself
 518                                 $item['body'] = str_replace($item['plink'], '', $item['body']);
 519                                 $item['body'] = trim(preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item['body']));
 520
 521                                 // Replace the content when the title is longer than the body
 522                                 $replace = (strlen($item['title']) > strlen($item['body']));
 523
 524                                 // Replace it, when there is an image in the body
 525                                 if (strstr($item['body'], '[/img]')) {
 526                                         $replace = true;
 527                                 }
 528
 529                                 // Replace it, when there is a link in the body
 530                                 if (strstr($item['body'], '[/url]')) {
 531                                         $replace = true;
 532                                 }
 533
 534                                 $saved_body = $item['body'];
 535                                 $saved_title = $item['title'];
 536
 537                                 if ($replace) {
 538                                         $item['body'] = trim($item['title']);
 539                                 }
 540
 541                                 $data = ParseUrl::getSiteinfoCached($item['plink']);
 542                                 if (!empty($data['text']) && !empty($data['title']) && (mb_strlen($item['body']) < mb_strlen($data['text']))) {
 543                                         // When the fetched page info text is longer than the body, we do try to enhance the body
 544                                         if (!empty($item['body']) && (strpos($data['title'], $item['body']) === false) && (strpos($data['text'], $item['body']) === false)) {
 545                                                 // The body is not part of the fetched page info title or page info text. So we add the text to the body
 546                                                 $item['body'] .= "\n\n" . $data['text'];
 547                                         } else {
 548                                                 // Else we replace the body with the page info text
 549                                                 $item['body'] = $data['text'];
 550                                         }
 551                                 }
 552
 553                                 $data = PageInfo::queryUrl($item['plink'], false, $preview, ($contact['fetch_further_information'] == 2), $contact['ffi_keyword_denylist'] ?? '');
 554
 555                                 if (!empty($data)) {
 556                                         // Take the data that was provided by the feed if the query is empty
 557                                         if (($data['type'] == 'link') && empty($data['title']) && empty($data['text'])) {
 558                                                 $data['title'] = $saved_title;
 559                                                 $item['body'] = $saved_body;
 560                                         }
 561
 562                                         $data_text = strip_tags(trim($data['text'] ?? ''));
 563                                         $item_body = strip_tags(trim($item['body'] ?? ''));
 564
 565                                         if (!empty($data_text) && (($data_text == $item_body) || strstr($item_body, $data_text))) {
 566                                                 $data['text'] = '';
 567                                         }
 568
 569                                         // We always strip the title since it will be added in the page information
 570                                         $item['title'] = '';
 571                                         $item['body'] = $item['body'] . "\n" . PageInfo::getFooterFromData($data, false);
 572                                         $taglist = $contact['fetch_further_information'] == 2 ? PageInfo::getTagsFromUrl($item['plink'], $preview, $contact['ffi_keyword_denylist'] ?? '') : [];
 573                                         $item['object-type'] = Activity\ObjectType::BOOKMARK;
 574                                         $attachments = [];
 575
 576                                         foreach (['audio', 'video'] as $elementname) {
 577                                                 if (!empty($data[$elementname])) {
 578                                                         foreach ($data[$elementname] as $element) {
 579                                                                 if (!empty($element['src'])) {
 580                                                                         $src = $element['src'];
 581                                                                 } elseif (!empty($element['content'])) {
 582                                                                         $src = $element['content'];
 583                                                                 } else {
 584                                                                         continue;
 585                                                                 }
 586
 587                                                                 $attachments[] = [
 588                                                                         'type'        => ($elementname == 'audio') ? Post\Media::AUDIO : Post\Media::VIDEO,
 589                                                                         'url'         => $src,
 590                                                                         'preview'     => $element['image']       ?? null,
 591                                                                         'mimetype'    => $element['contenttype'] ?? null,
 592                                                                         'name'        => $element['name']        ?? null,
 593                                                                         'description' => $element['description'] ?? null,
 594                                                                 ];
 595                                                         }
 596                                                 }
 597                                         }
 598                                 }
 599                         } else {
 600                                 if (!empty($summary)) {
 601                                         $item['body'] = '[abstract]' . HTML::toBBCode($summary, $basepath) . "[/abstract]\n" . $item['body'];
 602                                 }
 603
 604                                 if (!empty($contact['fetch_further_information']) && ($contact['fetch_further_information'] == 3)) {
 605                                         if (empty($taglist)) {
 606                                                 $taglist = PageInfo::getTagsFromUrl($item['plink'], $preview, $contact['ffi_keyword_denylist'] ?? '');
 607                                         }
 608                                         $item['body'] .= "\n" . self::tagToString($taglist);
 609                                 } else {
 610                                         $taglist = [];
 611                                 }
 612
 613                                 // Add the link to the original feed entry if not present in feed
 614                                 if (($item['plink'] != '') && !strstr($item['body'], $item['plink']) && !in_array($item['plink'], array_column($attachments, 'url'))) {
 615                                         $item['body'] .= '[hr][url]' . $item['plink'] . '[/url]';
 616                                 }
 617                         }
 618
 619                         if (empty($item['title'])) {
 620                                 $item['post-type'] = Item::PT_NOTE;
 621                         }
 622
 623                         Logger::info('Stored feed', ['item' => $item]);
 624
 625                         $notify = Item::isRemoteSelf($contact, $item);
 626
 627                         // Distributed items should have a well formatted URI.
 628                         // Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
 629                         if ($notify) {
 630                                 $item['guid'] = Item::guidFromUri($orig_plink, DI::baseUrl()->getHostname());
 631                                 $item['uri'] = Item::newURI($item['guid']);
 632                                 unset($item['thr-parent']);
 633                                 unset($item['parent-uri']);
 634
 635                                 // Set the delivery priority for "remote self" to "medium"
 636                                 $notify = Worker::PRIORITY_MEDIUM;
 637                         }
 638
 639                         $condition = ['uid' => $item['uid'], 'uri' => $item['uri']];
 640                         if (!Post::exists($condition) && !Post\Delayed::exists($item['uri'], $item['uid'])) {
 641                                 if (!$notify) {
 642                                         Post\Delayed::publish($item, $notify, $taglist, $attachments);
 643                                 } else {
 644                                         $postings[] = ['item' => $item, 'notify' => $notify,
 645                                                 'taglist' => $taglist, 'attachments' => $attachments];
 646                                 }
 647                         } else {
 648                                 Logger::info('Post already created or exists in the delayed posts queue', ['uid' => $item['uid'], 'uri' => $item['uri']]);
 649                         }
 650                 }
 651
 652                 if (!empty($postings)) {
 653                         $min_posting = DI::config()->get('system', 'minimum_posting_interval', 0);
 654                         $total = count($postings);
 655                         if ($total > 1) {
 656                                 // Posts shouldn't be delayed more than a day
 657                                 $interval = min(1440, self::getPollInterval($contact));
 658                                 $delay = max(round(($interval * 60) / $total), 60 * $min_posting);
 659                                 Logger::info('Got posting delay', ['delay' => $delay, 'interval' => $interval, 'items' => $total, 'cid' => $contact['id'], 'url' => $contact['url']]);
 660                         } else {
 661                                 $delay = 0;
 662                         }
 663
 664                         $post_delay = 0;
 665
 666                         foreach ($postings as $posting) {
 667                                 if ($delay > 0) {
 668                                         $publish_time = time() + $post_delay;
 669                                         $post_delay += $delay;
 670                                 } else {
 671                                         $publish_time = time();
 672                                 }
 673
 674                                 $last_publish = DI::pConfig()->get($posting['item']['uid'], 'system', 'last_publish', 0, true);
 675                                 $next_publish = max($last_publish + (60 * $min_posting), time());
 676                                 if ($publish_time < $next_publish) {
 677                                         $publish_time = $next_publish;
 678                                 }
 679                                 $publish_at = date(DateTimeFormat::MYSQL, $publish_time);
 680
 681                                 if (Post\Delayed::add($posting['item']['uri'], $posting['item'], $posting['notify'], Post\Delayed::PREPARED, $publish_at, $posting['taglist'], $posting['attachments'])) {
 682                                         DI::pConfig()->set($item['uid'], 'system', 'last_publish', $publish_time);
 683                                 }
 684                         }
 685                 }
 686
 687                 if (!$dryRun && DI::config()->get('system', 'adjust_poll_frequency')) {
 688                         self::adjustPollFrequency($contact, $creation_dates);
 689                 }
 690
 691                 return ['header' => $author, 'items' => $items];
 692         }
 693
 694         /**
 695          * Automatically adjust the poll frequency according to the post frequency
 696          *
 697          * @param array $contact Contact array
 698          * @param array $creation_dates
 699          * @return void
 700          */
 701         private static function adjustPollFrequency(array $contact, array $creation_dates)
 702         {
 703                 if ($contact['network'] != Protocol::FEED) {
 704                         Logger::info('Contact is no feed, skip.', ['id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url'], 'network' => $contact['network']]);
 705                         return;
 706                 }
 707
 708                 if (!empty($creation_dates)) {
 709                         // Count the post frequency and the earliest and latest post date
 710                         $frequency = [];
 711                         $oldest = time();
 712                         $newest = 0;
 713                         $oldest_date = $newest_date = '';
 714
 715                         foreach ($creation_dates as $date) {
 716                                 $timestamp = strtotime($date);
 717                                 $day = intdiv($timestamp, 86400);
 718                                 $hour = $timestamp % 86400;
 719
 720                                 // Only have a look at values from the last seven days
 721                                 if (((time() / 86400) - $day) < 7) {
 722                                         if (empty($frequency[$day])) {
 723                                                 $frequency[$day] = ['count' => 1, 'low' => $hour, 'high' => $hour];
 724                                         } else {
 725                                                 ++$frequency[$day]['count'];
 726                                                 if ($frequency[$day]['low'] > $hour) {
 727                                                         $frequency[$day]['low'] = $hour;
 728                                                 }
 729                                                 if ($frequency[$day]['high'] < $hour) {
 730                                                         $frequency[$day]['high'] = $hour;
 731                                                 }
 732                                         }
 733                                 }
 734                                 if ($oldest > $day) {
 735                                         $oldest = $day;
 736                                         $oldest_date = $date;
 737                                 }
 738
 739                                 if ($newest < $day) {
 740                                         $newest = $day;
 741                                         $newest_date = $date;
 742                                 }
 743                         }
 744
 745                         if (count($creation_dates) == 1) {
 746                                 Logger::info('Feed had posted a single time, switching to daily polling', ['newest' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 747                                 $priority = 8; // Poll once a day
 748                         }
 749
 750                         if (empty($priority) && (((time() / 86400) - $newest) > 730)) {
 751                                 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']]);
 752                                 $priority = 10; // Poll every month
 753                         }
 754
 755                         if (empty($priority) && (((time() / 86400) - $newest) > 365)) {
 756                                 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']]);
 757                                 $priority = 9; // Poll every week
 758                         }
 759
 760                         if (empty($priority) && empty($frequency)) {
 761                                 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']]);
 762                                 $priority = 8; // Poll once a day
 763                         }
 764
 765                         if (empty($priority)) {
 766                                 // Calculate the highest "posts per day" value
 767                                 $max = 0;
 768                                 foreach ($frequency as $entry) {
 769                                         if (($entry['count'] == 1) || ($entry['high'] == $entry['low'])) {
 770                                                 continue;
 771                                         }
 772
 773                                         // We take the earliest and latest post day and interpolate the number of post per day
 774                                         // that would had been created with this post frequency
 775
 776                                         // Assume at least four hours between oldest and newest post per day - should be okay for news outlets
 777                                         $duration = max($entry['high'] - $entry['low'], 14400);
 778                                         $ppd = (86400 / $duration) * $entry['count'];
 779                                         if ($ppd > $max) {
 780                                                 $max = $ppd;
 781                                         }
 782                                 }
 783                                 if ($max > 48) {
 784                                         $priority = 1; // Poll every quarter hour
 785                                 } elseif ($max > 24) {
 786                                         $priority = 2; // Poll half an hour
 787                                 } elseif ($max > 12) {
 788                                         $priority = 3; // Poll hourly
 789                                 } elseif ($max > 8) {
 790                                         $priority = 4; // Poll every two hours
 791                                 } elseif ($max > 4) {
 792                                         $priority = 5; // Poll every three hours
 793                                 } elseif ($max > 2) {
 794                                         $priority = 6; // Poll every six hours
 795                                 } else {
 796                                         $priority = 7; // Poll twice a day
 797                                 }
 798                                 Logger::info('Calculated priority by the posts per day', ['priority' => $priority, 'max' => round($max, 2), 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 799                         }
 800                 } else {
 801                         Logger::info('No posts, switching to daily polling', ['id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 802                         $priority = 8; // Poll once a day
 803                 }
 804
 805                 if ($contact['rating'] != $priority) {
 806                         Logger::notice('Adjusting priority', ['old' => $contact['rating'], 'new' => $priority, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 807                         Contact::update(['rating' => $priority], ['id' => $contact['id']]);
 808                 }
 809         }
 810
 811         /**
 812          * Get the poll interval for the given contact array
 813          *
 814          * @param array $contact
 815          * @return int Poll interval in minutes
 816          */
 817         public static function getPollInterval(array $contact): int
 818         {
 819                 if (in_array($contact['network'], [Protocol::MAIL, Protocol::FEED])) {
 820                         $ratings = [0, 3, 7, 8, 9, 10];
 821                         if (DI::config()->get('system', 'adjust_poll_frequency') && ($contact['network'] == Protocol::FEED)) {
 822                                 $rating = $contact['rating'];
 823                         } elseif (array_key_exists($contact['priority'], $ratings)) {
 824                                 $rating = $ratings[$contact['priority']];
 825                         } else {
 826                                 $rating = -1;
 827                         }
 828                 } else {
 829                         // Check once a week per default for all other networks
 830                         $rating = 9;
 831                 }
 832
 833                 // Friendica and OStatus are checked once a day
 834                 if (in_array($contact['network'], [Protocol::DFRN, Protocol::OSTATUS])) {
 835                         $rating = 8;
 836                 }
 837
 838                 // Check archived contacts or contacts with unsupported protocols once a month
 839                 if ($contact['archive'] || in_array($contact['network'], [Protocol::ZOT, Protocol::PHANTOM])) {
 840                         $rating = 10;
 841                 }
 842
 843                 if ($rating < 0) {
 844                         return 0;
 845                 }
 846                 /*
 847                  * Based on $contact['priority'], should we poll this site now? Or later?
 848                  */
 849
 850                 $min_poll_interval = max(1, DI::config()->get('system', 'min_poll_interval'));
 851
 852                 $poll_intervals = [$min_poll_interval, 15, 30, 60, 120, 180, 360, 720 ,1440, 10080, 43200];
 853
 854                 //$poll_intervals = [$min_poll_interval . ' minute', '15 minute', '30 minute',
 855                 //      '1 hour', '2 hour', '3 hour', '6 hour', '12 hour' ,'1 day', '1 week', '1 month'];
 856
 857                 return $poll_intervals[$rating];
 858         }
 859
 860         /**
 861          * Convert a tag array to a tag string
 862          *
 863          * @param array $tags
 864          * @return string tag string
 865          */
 866         private static function tagToString(array $tags): string
 867         {
 868                 $tagstr = '';
 869
 870                 foreach ($tags as $tag) {
 871                         if ($tagstr != '') {
 872                                 $tagstr .= ', ';
 873                         }
 874
 875                         $tagstr .= '#[url=' . DI::baseUrl() . '/search?tag=' . urlencode($tag) . ']' . $tag . '[/url]';
 876                 }
 877
 878                 return $tagstr;
 879         }
 880
 881         private static function titleIsBody(string $title, string $body): bool
 882         {
 883                 $title = strip_tags($title);
 884                 $title = trim($title);
 885                 $title = html_entity_decode($title, ENT_QUOTES, 'UTF-8');
 886                 $title = str_replace(["\n", "\r", "\t", " "], ['', '', '', ''], $title);
 887
 888                 $body = strip_tags($body);
 889                 $body = trim($body);
 890                 $body = html_entity_decode($body, ENT_QUOTES, 'UTF-8');
 891                 $body = str_replace(["\n", "\r", "\t", " "], ['', '', '', ''], $body);
 892
 893                 if (strlen($title) < strlen($body)) {
 894                         $body = substr($body, 0, strlen($title));
 895                 }
 896
 897                 if (($title != $body) && (substr($title, -3) == '...')) {
 898                         $pos = strrpos($title, '...');
 899                         if ($pos > 0) {
 900                                 $title = substr($title, 0, $pos);
 901                                 $body = substr($body, 0, $pos);
 902                         }
 903                 }
 904                 return ($title == $body);
 905         }
 906
 907         /**
 908          * Creates the Atom feed for a given nickname
 909          *
 910          * Supported filters:
 911          * - activity (default): all the public posts
 912          * - posts: all the public top-level posts
 913          * - comments: all the public replies
 914          *
 915          * Updates the provided last_update parameter if the result comes from the
 916          * cache or it is empty
 917          *
 918          * @param string  $owner_nick  Nickname of the feed owner
 919          * @param string  $last_update Date of the last update
 920          * @param integer $max_items   Number of maximum items to fetch
 921          * @param string  $filter      Feed items filter (activity, posts or comments)
 922          * @param boolean $nocache     Wether to bypass caching
 923          *
 924          * @return string Atom feed
 925          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 926          * @throws \ImagickException
 927          */
 928         public static function atom(string $owner_nick, string $last_update, int $max_items = 300, string $filter = 'activity', bool $nocache = false)
 929         {
 930                 $stamp = microtime(true);
 931
 932                 $owner = User::getOwnerDataByNick($owner_nick);
 933                 if (!$owner) {
 934                         return;
 935                 }
 936
 937                 $cachekey = 'feed:feed:' . $owner_nick . ':' . $filter . ':' . $last_update;
 938
 939                 // Display events in the users's timezone
 940                 if (strlen($owner['timezone'])) {
 941                         DI::app()->setTimeZone($owner['timezone']);
 942                 }
 943
 944                 $previous_created = $last_update;
 945
 946                 // Don't cache when the last item was posted less then 15 minutes ago (Cache duration)
 947                 if ((time() - strtotime($owner['last-item'])) < 15*60) {
 948                         $result = DI::cache()->get($cachekey);
 949                         if (!$nocache && !is_null($result)) {
 950                                 Logger::info('Cached feed duration', ['seconds' => number_format(microtime(true) - $stamp, 3), 'nick' => $owner_nick, 'filter' => $filter, 'created' => $previous_created]);
 951                                 return $result['feed'];
 952                         }
 953                 }
 954
 955                 $check_date = empty($last_update) ? '' : DateTimeFormat::utc($last_update);
 956                 $authorid = Contact::getIdForURL($owner['url']);
 957
 958                 $condition = ["`uid` = ? AND `received` > ? AND NOT `deleted` AND `gravity` IN (?, ?)
 959                         AND `private` != ? AND `visible` AND `wall` AND `parent-network` IN (?, ?, ?, ?)",
 960                         $owner['uid'], $check_date, Item::GRAVITY_PARENT, Item::GRAVITY_COMMENT,
 961                         Item::PRIVATE, Protocol::ACTIVITYPUB,
 962                         Protocol::OSTATUS, Protocol::DFRN, Protocol::DIASPORA];
 963
 964                 if ($filter === 'comments') {
 965                         $condition[0] .= " AND `gravity` = ? ";
 966                         $condition[] = Item::GRAVITY_COMMENT;
 967                 }
 968
 969                 if ($owner['account-type'] != User::ACCOUNT_TYPE_COMMUNITY) {
 970                         $condition[0] .= " AND `contact-id` = ? AND `author-id` = ?";
 971                         $condition[] = $owner['id'];
 972                         $condition[] = $authorid;
 973                 }
 974
 975                 $params = ['order' => ['received' => true], 'limit' => $max_items];
 976
 977                 if ($filter === 'posts') {
 978                         $ret = Post::selectThread(Item::DELIVER_FIELDLIST, $condition, $params);
 979                 } else {
 980                         $ret = Post::select(Item::DELIVER_FIELDLIST, $condition, $params);
 981                 }
 982
 983                 $items = Post::toArray($ret);
 984
 985                 $doc = new DOMDocument('1.0', 'utf-8');
 986                 $doc->formatOutput = true;
 987
 988                 $root = self::addHeader($doc, $owner, $filter);
 989
 990                 foreach ($items as $item) {
 991                         $entry = self::noteEntry($doc, $item, $owner);
 992                         $root->appendChild($entry);
 993
 994                         if ($last_update < $item['created']) {
 995                                 $last_update = $item['created'];
 996                         }
 997                 }
 998
 999                 $feeddata = trim($doc->saveXML());
1000
1001                 $msg = ['feed' => $feeddata, 'last_update' => $last_update];
1002                 DI::cache()->set($cachekey, $msg, Duration::QUARTER_HOUR);
1003
1004                 Logger::info('Feed duration', ['seconds' => number_format(microtime(true) - $stamp, 3), 'nick' => $owner_nick, 'filter' => $filter, 'created' => $previous_created]);
1005
1006                 return $feeddata;
1007         }
1008
1009         /**
1010          * Adds the header elements to the XML document
1011          *
1012          * @param DOMDocument $doc       XML document
1013          * @param array       $owner     Contact data of the poster
1014          * @param string      $filter    The related feed filter (activity, posts or comments)
1015          *
1016          * @return DOMElement Header root element
1017          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1018          */
1019         private static function addHeader(DOMDocument $doc, array $owner, string $filter): DOMElement
1020         {
1021                 $root = $doc->createElementNS(ActivityNamespace::ATOM1, 'feed');
1022                 $doc->appendChild($root);
1023
1024                 $title = '';
1025                 $selfUri = '/feed/' . $owner['nick'] . '/';
1026                 switch ($filter) {
1027                         case 'activity':
1028                                 $title = DI::l10n()->t('%s\'s timeline', $owner['name']);
1029                                 $selfUri .= $filter;
1030                                 break;
1031                         case 'posts':
1032                                 $title = DI::l10n()->t('%s\'s posts', $owner['name']);
1033                                 break;
1034                         case 'comments':
1035                                 $title = DI::l10n()->t('%s\'s comments', $owner['name']);
1036                                 $selfUri .= $filter;
1037                                 break;
1038                 }
1039
1040                 $attributes = ['uri' => 'https://friendi.ca', 'version' => App::VERSION . '-' . DB_UPDATE_VERSION];
1041                 XML::addElement($doc, $root, 'generator', App::PLATFORM, $attributes);
1042                 XML::addElement($doc, $root, 'id', DI::baseUrl() . '/profile/' . $owner['nick']);
1043                 XML::addElement($doc, $root, 'title', $title);
1044                 XML::addElement($doc, $root, 'subtitle', sprintf("Updates from %s on %s", $owner['name'], DI::config()->get('config', 'sitename')));
1045                 XML::addElement($doc, $root, 'logo', User::getAvatarUrl($owner, Proxy::SIZE_SMALL));
1046                 XML::addElement($doc, $root, 'updated', DateTimeFormat::utcNow(DateTimeFormat::ATOM));
1047
1048                 $author = self::addAuthor($doc, $owner);
1049                 $root->appendChild($author);
1050
1051                 $attributes = ['href' => $owner['url'], 'rel' => 'alternate', 'type' => 'text/html'];
1052                 XML::addElement($doc, $root, 'link', '', $attributes);
1053
1054                 OStatus::addHubLink($doc, $root, $owner['nick']);
1055
1056                 $attributes = ['href' => DI::baseUrl() . $selfUri, 'rel' => 'self', 'type' => 'application/atom+xml'];
1057                 XML::addElement($doc, $root, 'link', '', $attributes);
1058
1059                 return $root;
1060         }
1061
1062         /**
1063          * Adds the author element to the XML document
1064          *
1065          * @param DOMDocument $doc          XML document
1066          * @param array       $owner        Contact data of the poster
1067          * @return DOMElement author element
1068          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1069          */
1070         private static function addAuthor(DOMDocument $doc, array $owner): DOMElement
1071         {
1072                 $author = $doc->createElement('author');
1073                 XML::addElement($doc, $author, 'uri', $owner['url']);
1074                 XML::addElement($doc, $author, 'name', $owner['nick']);
1075                 XML::addElement($doc, $author, 'email', $owner['addr']);
1076
1077                 return $author;
1078         }
1079
1080         /**
1081          * Adds a regular entry element
1082          *
1083          * @param DOMDocument $doc       XML document
1084          * @param array       $item      Data of the item that is to be posted
1085          * @param array       $owner     Contact data of the poster
1086          * @param bool        $toplevel  Is it for en entry element (false) or a feed entry (true)?
1087          * @return DOMElement Entry element
1088          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1089          * @throws \ImagickException
1090          */
1091         private static function noteEntry(DOMDocument $doc, array $item, array $owner): DOMElement
1092         {
1093                 if (($item['gravity'] != Item::GRAVITY_PARENT) && (Strings::normaliseLink($item['author-link']) != Strings::normaliseLink($owner['url']))) {
1094                         Logger::info('Feed entry author does not match feed owner', ['owner' => $owner['url'], 'author' => $item['author-link']]);
1095                 }
1096
1097                 $entry = OStatus::entryHeader($doc, $owner, $item, false);
1098
1099                 self::entryContent($doc, $entry, $item, self::getTitle($item), '', true);
1100
1101                 self::entryFooter($doc, $entry, $item, $owner);
1102
1103                 return $entry;
1104         }
1105
1106         /**
1107          * Adds elements to the XML document
1108          *
1109          * @param DOMDocument $doc       XML document
1110          * @param \DOMElement $entry     Entry element where the content is added
1111          * @param array       $item      Data of the item that is to be posted
1112          * @param array       $owner     Contact data of the poster
1113          * @param string      $title     Title for the post
1114          * @param string      $verb      The activity verb
1115          * @param bool        $complete  Add the "status_net" element?
1116          * @return void
1117          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1118          */
1119         private static function entryContent(DOMDocument $doc, DOMElement $entry, array $item, $title, string $verb = '', bool $complete = true)
1120         {
1121                 if ($verb == '') {
1122                         $verb = OStatus::constructVerb($item);
1123                 }
1124
1125                 XML::addElement($doc, $entry, 'id', $item['uri']);
1126                 XML::addElement($doc, $entry, 'title', html_entity_decode($title, ENT_QUOTES, 'UTF-8'));
1127
1128                 $body = OStatus::formatPicturePost($item['body'], $item['uri-id']);
1129
1130                 $body = BBCode::convertForUriId($item['uri-id'], $body, BBCode::ACTIVITYPUB);
1131
1132                 XML::addElement($doc, $entry, 'content', $body, ['type' => 'html']);
1133
1134                 XML::addElement($doc, $entry, 'link', '', ['rel' => 'alternate', 'type' => 'text/html',
1135                                                                 'href' => DI::baseUrl() . '/display/' . $item['guid']]
1136                 );
1137
1138                 XML::addElement($doc, $entry, 'published', DateTimeFormat::utc($item['created'] . '+00:00', DateTimeFormat::ATOM));
1139                 XML::addElement($doc, $entry, 'updated', DateTimeFormat::utc($item['edited'] . '+00:00', DateTimeFormat::ATOM));
1140         }
1141
1142         /**
1143          * Adds the elements at the foot of an entry to the XML document
1144          *
1145          * @param DOMDocument $doc       XML document
1146          * @param object      $entry     The entry element where the elements are added
1147          * @param array       $item      Data of the item that is to be posted
1148          * @param array       $owner     Contact data of the poster
1149          * @param bool        $complete  default true
1150          * @return void
1151          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1152          */
1153         private static function entryFooter(DOMDocument $doc, $entry, array $item, array $owner)
1154         {
1155                 $mentioned = [];
1156
1157                 if ($item['gravity'] != Item::GRAVITY_PARENT) {
1158                         $parent = Post::selectFirst(['guid', 'author-link', 'owner-link'], ['id' => $item['parent']]);
1159
1160                         $thrparent = Post::selectFirst(['guid', 'author-link', 'owner-link', 'plink'], ['uid' => $owner['uid'], 'uri' => $item['thr-parent']]);
1161
1162                         if (DBA::isResult($thrparent)) {
1163                                 $mentioned[$thrparent['author-link']] = $thrparent['author-link'];
1164                                 $mentioned[$thrparent['owner-link']]  = $thrparent['owner-link'];
1165                                 $parent_plink                         = $thrparent['plink'];
1166                         } elseif (DBA::isResult($parent)) {
1167                                 $mentioned[$parent['author-link']] = $parent['author-link'];
1168                                 $mentioned[$parent['owner-link']]  = $parent['owner-link'];
1169                                 $parent_plink                      = DI::baseUrl() . '/display/' . $parent['guid'];
1170                         } else {
1171                                 DI::logger()->notice('Missing parent and thr-parent for child item', ['item' => $item]);
1172                         }
1173
1174                         if (isset($parent_plink)) {
1175                                 $attributes = [
1176                                         'ref'  => $item['thr-parent'],
1177                                         'href' => $parent_plink];
1178                                 XML::addElement($doc, $entry, 'thr:in-reply-to', '', $attributes);
1179
1180                                 $attributes = [
1181                                         'rel'  => 'related',
1182                                         'href' => $parent_plink];
1183                                 XML::addElement($doc, $entry, 'link', '', $attributes);
1184                         }
1185                 }
1186
1187                 // uri-id isn't present for follow entry pseudo-items
1188                 $tags = Tag::getByURIId($item['uri-id'] ?? 0);
1189                 foreach ($tags as $tag) {
1190                         $mentioned[$tag['url']] = $tag['url'];
1191                 }
1192
1193                 foreach ($tags as $tag) {
1194                         if ($tag['type'] == Tag::HASHTAG) {
1195                                 XML::addElement($doc, $entry, 'category', '', ['term' => $tag['name']]);
1196                         }
1197                 }
1198
1199                 OStatus::getAttachment($doc, $entry, $item);
1200         }
1201
1202         /**
1203          * Fetch or create title for feed entry
1204          *
1205          * @param array $item
1206          * @return string title
1207          */
1208         private static function getTitle(array $item): string
1209         {
1210                 if ($item['title'] != '') {
1211                         return BBCode::convertForUriId($item['uri-id'], $item['title'], BBCode::ACTIVITYPUB);
1212                 }
1213
1214                 // Fetch information about the post
1215                 $siteinfo = BBCode::getAttachedData($item['body']);
1216                 if (isset($siteinfo['title'])) {
1217                         return $siteinfo['title'];
1218                 }
1219
1220                 // If no bookmark is found then take the first line
1221                 // Remove the share element before fetching the first line
1222                 $title = trim(preg_replace("/\[share.*?\](.*?)\[\/share\]/ism", "\n$1\n", $item['body']));
1223
1224                 $title = BBCode::toPlaintext($title)."\n";
1225                 $pos = strpos($title, "\n");
1226                 $trailer = '';
1227                 if (($pos == 0) || ($pos > 100)) {
1228                         $pos = 100;
1229                         $trailer = '...';
1230                 }
1231
1232                 return substr($title, 0, $pos) . $trailer;
1233         }
1234 }