src/Protocol/Feed.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2024, 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\Contact\LocalRelationship\Entity\LocalRelationship;
  29 use Friendica\Content\PageInfo;
  30 use Friendica\Content\Text\BBCode;
  31 use Friendica\Content\Text\HTML;
  32 use Friendica\Core\Cache\Enum\Duration;
  33 use Friendica\Core\Logger;
  34 use Friendica\Core\Protocol;
  35 use Friendica\Core\Worker;
  36 use Friendica\Database\DBA;
  37 use Friendica\DI;
  38 use Friendica\Model\Contact;
  39 use Friendica\Model\Conversation;
  40 use Friendica\Model\Item;
  41 use Friendica\Model\Post;
  42 use Friendica\Model\Tag;
  43 use Friendica\Model\User;
  44 use Friendica\Network\HTTPException;
  45 use Friendica\Util\DateTimeFormat;
  46 use Friendica\Util\Network;
  47 use Friendica\Util\ParseUrl;
  48 use Friendica\Util\Proxy;
  49 use Friendica\Util\Strings;
  50 use Friendica\Util\XML;
  51 use GuzzleHttp\Exception\TransferException;
  52
  53 /**
  54  * This class contain functions to import feeds (RSS/RDF/Atom)
  55  */
  56 class Feed
  57 {
  58         /**
  59          * Read a RSS/RDF/Atom feed and create an item entry for it
  60          *
  61          * @param string $xml      The feed data
  62          * @param array  $importer The user record of the importer
  63          * @param array  $contact  The contact record of the feed
  64          *
  65          * @return array Returns the header and the first item in dry run mode
  66          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  67          */
  68         public static function import(string $xml, array $importer = [], array $contact = []): array
  69         {
  70                 $dryRun = empty($importer) && empty($contact);
  71
  72                 if ($dryRun) {
  73                         Logger::info("Test Atom/RSS feed");
  74                 } else {
  75                         Logger::info('Import Atom/RSS feed "' . $contact['name'] . '" (Contact ' . $contact['id'] . ') for user ' . $importer['uid']);
  76                 }
  77
  78                 $xml = trim($xml);
  79
  80                 if (empty($xml)) {
  81                         Logger::info('XML is empty.');
  82                         return [];
  83                 }
  84
  85                 if (!empty($contact['poll'])) {
  86                         $basepath = $contact['poll'];
  87                 } elseif (!empty($contact['url'])) {
  88                         $basepath = $contact['url'];
  89                 } else {
  90                         $basepath = '';
  91                 }
  92
  93                 $doc = new DOMDocument();
  94                 @$doc->loadXML($xml);
  95                 $xpath = new DOMXPath($doc);
  96
  97                 $xpath->registerNamespace('atom', ActivityNamespace::ATOM1);
  98                 $xpath->registerNamespace('atom03', ActivityNamespace::ATOM03);
  99                 $xpath->registerNamespace('dc', 'http://purl.org/dc/elements/1.1/');
 100                 $xpath->registerNamespace('content', 'http://purl.org/rss/1.0/modules/content/');
 101                 $xpath->registerNamespace('rdf', 'http://www.w3.org/1999/02/22-rdf-syntax-ns#');
 102                 $xpath->registerNamespace('rss', 'http://purl.org/rss/1.0/');
 103                 $xpath->registerNamespace('media', 'http://search.yahoo.com/mrss/');
 104                 $xpath->registerNamespace('poco', ActivityNamespace::POCO);
 105
 106                 $author = [];
 107                 $atomns = 'atom';
 108                 $entries = null;
 109                 $protocol = Conversation::PARCEL_UNKNOWN;
 110
 111                 // Is it RDF?
 112                 if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
 113                         $protocol = Conversation::PARCEL_RDF;
 114                         $author['author-link'] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:link/text()');
 115                         $author['author-name'] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:title/text()');
 116
 117                         if (empty($author['author-name'])) {
 118                                 $author['author-name'] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:description/text()');
 119                         }
 120                         $entries = $xpath->query('/rdf:RDF/rss:item');
 121                 }
 122
 123                 if ($xpath->query('/opml')->length > 0) {
 124                         $protocol = Conversation::PARCEL_OPML;
 125                         $author['author-name'] = XML::getFirstNodeValue($xpath, '/opml/head/title/text()');
 126                         $entries = $xpath->query('/opml/body/outline');
 127                 }
 128
 129                 // Is it Atom?
 130                 if ($xpath->query('/atom:feed')->length > 0) {
 131                         $protocol = Conversation::PARCEL_ATOM;
 132                 } elseif ($xpath->query('/atom03:feed')->length > 0) {
 133                         $protocol = Conversation::PARCEL_ATOM03;
 134                         $atomns = 'atom03';
 135                 }
 136
 137                 if (in_array($protocol, [Conversation::PARCEL_ATOM, Conversation::PARCEL_ATOM03])) {
 138                         $alternate = XML::getFirstAttributes($xpath, $atomns . ":link[@rel='alternate']");
 139                         if (is_object($alternate)) {
 140                                 foreach ($alternate as $attribute) {
 141                                         if ($attribute->name == 'href') {
 142                                                 $author['author-link'] = $attribute->textContent;
 143                                         }
 144                                 }
 145                         }
 146
 147                         if (empty($author['author-link'])) {
 148                                 $self = XML::getFirstAttributes($xpath, $atomns . ":link[@rel='self']");
 149                                 if (is_object($self)) {
 150                                         foreach ($self as $attribute) {
 151                                                 if ($attribute->name == 'href') {
 152                                                         $author['author-link'] = $attribute->textContent;
 153                                                 }
 154                                         }
 155                                 }
 156                         }
 157
 158                         if (empty($author['author-link'])) {
 159                                 $author['author-link'] = XML::getFirstNodeValue($xpath, '/' . $atomns . ':feed/' . $atomns . ':id/text()');
 160                         }
 161                         $author['author-avatar'] = XML::getFirstNodeValue($xpath, '/' . $atomns . ':feed/' . $atomns . ':logo/text()');
 162
 163                         $author['author-name'] = XML::getFirstNodeValue($xpath, '/' . $atomns . ':feed/' . $atomns . ':title/text()');
 164
 165                         if (empty($author['author-name'])) {
 166                                 $author['author-name'] = XML::getFirstNodeValue($xpath, '/' . $atomns . ':feed/' . $atomns . ':subtitle/text()');
 167                         }
 168
 169                         if (empty($author['author-name'])) {
 170                                 $author['author-name'] = XML::getFirstNodeValue($xpath, '/' . $atomns . ':feed/' . $atomns . ':author/' . $atomns . ':name/text()');
 171                         }
 172
 173                         $value = XML::getFirstNodeValue($xpath, '' . $atomns . ':author/poco:displayName/text()');
 174                         if ($value != '') {
 175                                 $author['author-name'] = $value;
 176                         }
 177
 178                         if ($dryRun) {
 179                                 $author['author-id'] = XML::getFirstNodeValue($xpath, '/' . $atomns . ':feed/' . $atomns . ':author/' . $atomns . ':id/text()');
 180
 181                                 // See https://tools.ietf.org/html/rfc4287#section-3.2.2
 182                                 $value = XML::getFirstNodeValue($xpath, $atomns . ':author/' . $atomns . ':uri/text()');
 183                                 if ($value != '') {
 184                                         $author['author-link'] = $value;
 185                                 }
 186
 187                                 $value = XML::getFirstNodeValue($xpath, $atomns . ':author/poco:preferredUsername/text()');
 188                                 if ($value != '') {
 189                                         $author['author-nick'] = $value;
 190                                 }
 191
 192                                 $value = XML::getFirstNodeValue($xpath, $atomns . ':author/poco:address/poco:formatted/text()');
 193                                 if ($value != '') {
 194                                         $author['author-location'] = $value;
 195                                 }
 196
 197                                 $value = XML::getFirstNodeValue($xpath, $atomns . ':author/poco:note/text()');
 198                                 if ($value != '') {
 199                                         $author['author-about'] = $value;
 200                                 }
 201
 202                                 $avatar = XML::getFirstAttributes($xpath, $atomns . ":author/' . $atomns . ':link[@rel='avatar']");
 203                                 if (is_object($avatar)) {
 204                                         foreach ($avatar as $attribute) {
 205                                                 if ($attribute->name == 'href') {
 206                                                         $author['author-avatar'] = $attribute->textContent;
 207                                                 }
 208                                         }
 209                                 }
 210                         }
 211
 212                         $author['edited'] = $author['created'] = XML::getFirstNodeValue($xpath, '/' . $atomns . ':feed/' . $atomns . ':updated/text()');
 213
 214                         $author['app'] = XML::getFirstNodeValue($xpath, '/' . $atomns . ':feed/' . $atomns . ':generator/text()');
 215
 216                         $entries = $xpath->query('/' . $atomns . ':feed/' . $atomns . ':entry');
 217                 }
 218
 219                 // Is it RSS?
 220                 if ($xpath->query('/rss/channel')->length > 0) {
 221                         $protocol = Conversation::PARCEL_RSS;
 222                         $author['author-link'] = XML::getFirstNodeValue($xpath, '/rss/channel/link/text()');
 223
 224                         $author['author-name'] = XML::getFirstNodeValue($xpath, '/rss/channel/title/text()');
 225
 226                         if (empty($author['author-name'])) {
 227                                 $author['author-name'] = XML::getFirstNodeValue($xpath, '/rss/channel/copyright/text()');
 228                         }
 229
 230                         if (empty($author['author-name'])) {
 231                                 $author['author-name'] = XML::getFirstNodeValue($xpath, '/rss/channel/description/text()');
 232                         }
 233
 234                         $author['author-avatar'] = XML::getFirstNodeValue($xpath, '/rss/channel/image/url/text()');
 235
 236                         if (empty($author['author-avatar'])) {
 237                                 $avatar = XML::getFirstAttributes($xpath, '/rss/channel/itunes:image');
 238                                 if (is_object($avatar)) {
 239                                         foreach ($avatar as $attribute) {
 240                                                 if ($attribute->name == 'href') {
 241                                                         $author['author-avatar'] = $attribute->textContent;
 242                                                 }
 243                                         }
 244                                 }
 245                         }
 246
 247                         $author['author-about'] = HTML::toBBCode(XML::getFirstNodeValue($xpath, '/rss/channel/description/text()'), $basepath);
 248
 249                         if (empty($author['author-about'])) {
 250                                 $author['author-about'] = XML::getFirstNodeValue($xpath, '/rss/channel/itunes:summary/text()');
 251                         }
 252
 253                         $author['edited'] = $author['created'] = XML::getFirstNodeValue($xpath, '/rss/channel/pubDate/text()');
 254
 255                         $author['app'] = XML::getFirstNodeValue($xpath, '/rss/channel/generator/text()');
 256
 257                         $entries = $xpath->query('/rss/channel/item');
 258                 }
 259
 260                 if (!$dryRun) {
 261                         $author['author-link'] = $contact['url'];
 262
 263                         if (empty($author['author-name'])) {
 264                                 $author['author-name'] = $contact['name'];
 265                         }
 266
 267                         $author['author-avatar'] = $contact['thumb'];
 268
 269                         $author['owner-link'] = $contact['url'];
 270                         $author['owner-name'] = $contact['name'];
 271                         $author['owner-avatar'] = $contact['thumb'];
 272                 }
 273
 274                 $header = [
 275                         'uid'         => $importer['uid'] ?? 0,
 276                         'network'     => Protocol::FEED,
 277                         'wall'        => 0,
 278                         'origin'      => 0,
 279                         'gravity'     => Item::GRAVITY_PARENT,
 280                         'private'     => Item::PUBLIC,
 281                         'verb'        => Activity::POST,
 282                         'object-type' => Activity\ObjectType::NOTE,
 283                         'post-type'   => Item::PT_ARTICLE,
 284                         'contact-id'  => $contact['id'] ?? 0,
 285                 ];
 286
 287                 $datarray['protocol'] = $protocol;
 288                 $datarray['direction'] = Conversation::PULL;
 289
 290                 if (!is_object($entries)) {
 291                         Logger::info("There are no entries in this feed.");
 292                         return [];
 293                 }
 294
 295                 $items = [];
 296                 $creation_dates = [];
 297
 298                 // Limit the number of items that are about to be fetched
 299                 $total_items = ($entries->length - 1);
 300                 $max_items = DI::config()->get('system', 'max_feed_items');
 301                 if (($max_items > 0) && ($total_items > $max_items)) {
 302                         $total_items = $max_items;
 303                 }
 304
 305                 $postings = [];
 306
 307                 // Importing older entries first
 308                 for ($i = $total_items; $i >= 0; --$i) {
 309                         $entry = $entries->item($i);
 310
 311                         $item = array_merge($header, $author);
 312                         $body = '';
 313
 314                         $alternate = XML::getFirstAttributes($xpath, $atomns . ":link[@rel='alternate']", $entry);
 315                         if (!is_object($alternate)) {
 316                                 $alternate = XML::getFirstAttributes($xpath, $atomns . ':link', $entry);
 317                         }
 318                         if (is_object($alternate)) {
 319                                 foreach ($alternate as $attribute) {
 320                                         if ($attribute->name == 'href') {
 321                                                 $item['plink'] = $attribute->textContent;
 322                                         }
 323                                 }
 324                         }
 325
 326                         if ($entry->nodeName == 'outline') {
 327                                 $isrss = false;
 328                                 $plink = '';
 329                                 $uri = '';
 330                                 foreach ($entry->attributes as $attribute) {
 331                                         switch ($attribute->nodeName) {
 332                                                 case 'title':
 333                                                         $item['title'] = $attribute->nodeValue;
 334                                                         break;
 335
 336                                                 case 'text':
 337                                                         $body = $attribute->nodeValue;
 338                                                         break;
 339
 340                                                 case 'htmlUrl':
 341                                                         $plink = $attribute->nodeValue;
 342                                                         break;
 343
 344                                                 case 'xmlUrl':
 345                                                         $uri = $attribute->nodeValue;
 346                                                         break;
 347
 348                                                 case 'type':
 349                                                         $isrss = $attribute->nodeValue == 'rss';
 350                                                         break;
 351                                         }
 352                                 }
 353                                 $item['plink'] = $plink ?: $uri;
 354                                 $item['uri'] = $uri ?: $plink;
 355                                 if (!$isrss || empty($item['uri'])) {
 356                                         continue;
 357                                 }
 358                         }
 359
 360                         if (empty($item['plink'])) {
 361                                 $item['plink'] = XML::getFirstNodeValue($xpath, 'link/text()', $entry);
 362                         }
 363
 364                         if (empty($item['plink'])) {
 365                                 $item['plink'] = XML::getFirstNodeValue($xpath, 'rss:link/text()', $entry);
 366                         }
 367
 368                         // Add the base path if missing
 369                         $item['plink'] = Network::addBasePath($item['plink'], $basepath);
 370
 371                         if (empty($item['uri'])) {
 372                                 $item['uri'] = XML::getFirstNodeValue($xpath, $atomns . ':id/text()', $entry);
 373                         }
 374
 375                         $guid = XML::getFirstNodeValue($xpath, 'guid/text()', $entry);
 376                         if (!empty($guid)) {
 377                                 $item['uri'] = $guid;
 378
 379                                 // Don't use the GUID value directly but instead use it as a basis for the GUID
 380                                 $item['guid'] = Item::guidFromUri($guid, parse_url($guid, PHP_URL_HOST) ?? parse_url($item['plink'], PHP_URL_HOST));
 381                         }
 382
 383                         if (empty($item['uri'])) {
 384                                 $item['uri'] = $item['plink'];
 385                         }
 386
 387                         $orig_plink = $item['plink'];
 388
 389                         try {
 390                                 $item['plink'] = DI::httpClient()->finalUrl($item['plink']);
 391                         } catch (TransferException $exception) {
 392                                 Logger::notice('Item URL couldn\'t get expanded', ['url' => $item['plink'], 'exception' => $exception]);
 393                         }
 394
 395                         if (empty($item['title'])) {
 396                                 $item['title'] = XML::getFirstNodeValue($xpath, $atomns . ':title/text()', $entry);
 397                         }
 398
 399                         if (empty($item['title'])) {
 400                                 $item['title'] = XML::getFirstNodeValue($xpath, 'title/text()', $entry);
 401                         }
 402
 403                         if (empty($item['title'])) {
 404                                 $item['title'] = XML::getFirstNodeValue($xpath, 'rss:title/text()', $entry);
 405                         }
 406
 407                         if (empty($item['title'])) {
 408                                 $item['title'] = XML::getFirstNodeValue($xpath, 'itunes:title/text()', $entry);
 409                         }
 410
 411                         $item['title'] = html_entity_decode($item['title'], ENT_QUOTES, 'UTF-8');
 412
 413                         $published = XML::getFirstNodeValue($xpath, $atomns . ':published/text()', $entry);
 414
 415                         if (empty($published)) {
 416                                 $published = XML::getFirstNodeValue($xpath, 'pubDate/text()', $entry);
 417                         }
 418
 419                         if (empty($published)) {
 420                                 $published = XML::getFirstNodeValue($xpath, 'dc:date/text()', $entry);
 421                         }
 422
 423                         $updated = XML::getFirstNodeValue($xpath, $atomns . ':updated/text()', $entry);
 424
 425                         if (empty($updated) && !empty($published)) {
 426                                 $updated = $published;
 427                         }
 428
 429                         if (empty($published) && !empty($updated)) {
 430                                 $published = $updated;
 431                         }
 432
 433                         if ($published != '') {
 434                                 $item['created'] = trim($published);
 435                         }
 436
 437                         if ($updated != '') {
 438                                 $item['edited'] = trim($updated);
 439                         }
 440
 441                         if (!$dryRun) {
 442                                 $condition = [
 443                                         "`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
 444                                         $importer['uid'], $item['uri'], Protocol::FEED, Protocol::DFRN
 445                                 ];
 446                                 $previous = Post::selectFirst(['id', 'created'], $condition);
 447                                 if (DBA::isResult($previous)) {
 448                                         // Use the creation date when the post had been stored. It can happen this date changes in the feed.
 449                                         $creation_dates[] = $previous['created'];
 450                                         Logger::info('Item with URI ' . $item['uri'] . ' for user ' . $importer['uid'] . ' already existed under id ' . $previous['id']);
 451                                         continue;
 452                                 }
 453                                 $creation_dates[] = DateTimeFormat::utc($item['created']);
 454                         }
 455
 456                         $creator = XML::getFirstNodeValue($xpath, 'author/text()', $entry);
 457
 458                         if (empty($creator)) {
 459                                 $creator = XML::getFirstNodeValue($xpath, $atomns . ':author/' . $atomns . ':name/text()', $entry);
 460                         }
 461
 462                         if (empty($creator)) {
 463                                 $creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
 464                         }
 465
 466                         if ($creator != '') {
 467                                 $item['author-name'] = $creator;
 468                         }
 469
 470                         $creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
 471
 472                         if ($creator != '') {
 473                                 $item['author-name'] = $creator;
 474                         }
 475
 476                         /// @TODO ?
 477                         // <category>Ausland</category>
 478                         // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 479
 480                         $attachments = [];
 481
 482                         $enclosures = $xpath->query("enclosure|$atomns:link[@rel='enclosure']", $entry);
 483                         if (!empty($enclosures)) {
 484                                 foreach ($enclosures as $enclosure) {
 485                                         $href = '';
 486                                         $length = null;
 487                                         $type = null;
 488
 489                                         foreach ($enclosure->attributes as $attribute) {
 490                                                 if (in_array($attribute->name, ['url', 'href'])) {
 491                                                         $href = $attribute->textContent;
 492                                                 } elseif ($attribute->name == 'length') {
 493                                                         $length = (int)$attribute->textContent;
 494                                                 } elseif ($attribute->name == 'type') {
 495                                                         $type = $attribute->textContent;
 496                                                 }
 497                                         }
 498
 499                                         if (!empty($href)) {
 500                                                 $attachment = ['uri-id' => -1, 'type' => Post\Media::UNKNOWN, 'url' => $href, 'mimetype' => $type, 'size' => $length];
 501
 502                                                 $attachment = Post\Media::fetchAdditionalData($attachment);
 503
 504                                                 // By now we separate the visible media types (audio, video, image) from the rest
 505                                                 // In the future we should try to avoid the DOCUMENT type and only use the real one - but not in the RC phase.
 506                                                 if (!in_array($attachment['type'], [Post\Media::AUDIO, Post\Media::IMAGE, Post\Media::VIDEO])) {
 507                                                         $attachment['type'] = Post\Media::DOCUMENT;
 508                                                 }
 509                                                 $attachments[] = $attachment;
 510                                         }
 511                                 }
 512                         }
 513
 514                         $taglist = [];
 515                         $categories = $xpath->query('category', $entry);
 516                         foreach ($categories as $category) {
 517                                 $taglist[] = $category->nodeValue;
 518                         }
 519
 520                         if (empty($body)) {
 521                                 $body = trim(XML::getFirstNodeValue($xpath, $atomns . ':content/text()', $entry));
 522                         }
 523
 524                         if (empty($body)) {
 525                                 $body = trim(XML::getFirstNodeValue($xpath, 'content:encoded/text()', $entry));
 526                         }
 527
 528                         $summary = trim(XML::getFirstNodeValue($xpath, $atomns . ':summary/text()', $entry));
 529
 530                         if (empty($summary)) {
 531                                 $summary = trim(XML::getFirstNodeValue($xpath, 'description/text()', $entry));
 532                         }
 533
 534                         if (empty($body)) {
 535                                 $body = $summary;
 536                                 $summary = '';
 537                         }
 538
 539                         if ($body == $summary) {
 540                                 $summary = '';
 541                         }
 542
 543                         // remove the content of the title if it is identically to the body
 544                         // This helps with auto generated titles e.g. from tumblr
 545                         if (self::titleIsBody($item['title'], $body)) {
 546                                 $item['title'] = '';
 547                         }
 548                         $item['body'] = HTML::toBBCode($body, $basepath);
 549
 550                         // Remove tracking pixels
 551                         $item['body'] = preg_replace("/\[img=1x1\]([^\[\]]*)\[\/img\]/Usi", '', $item['body']);
 552
 553                         if (($item['body'] == '') && ($item['title'] != '')) {
 554                                 $item['body'] = $item['title'];
 555                                 $item['title'] = '';
 556                         }
 557
 558                         if ($dryRun) {
 559                                 $item['attachments'] = $attachments;
 560                                 $items[] = $item;
 561                                 break;
 562                         } elseif (!Item::isValid($item)) {
 563                                 Logger::info('Feed item is invalid', ['created' => $item['created'], 'uid' => $item['uid'], 'uri' => $item['uri']]);
 564                                 continue;
 565                         } elseif (Item::isTooOld($item)) {
 566                                 Logger::info('Feed is too old', ['created' => $item['created'], 'uid' => $item['uid'], 'uri' => $item['uri']]);
 567                                 continue;
 568                         }
 569
 570                         $fetch_further_information = $contact['fetch_further_information'] ?? LocalRelationship::FFI_NONE;
 571
 572                         $preview = '';
 573                         if (in_array($fetch_further_information, [LocalRelationship::FFI_INFORMATION, LocalRelationship::FFI_BOTH])) {
 574                                 // Handle enclosures and treat them as preview picture
 575                                 foreach ($attachments as $attachment) {
 576                                         if ($attachment['mimetype'] == 'image/jpeg') {
 577                                                 $preview = $attachment['url'];
 578                                         }
 579                                 }
 580
 581                                 // Remove a possible link to the item itself
 582                                 $item['body'] = str_replace($item['plink'], '', $item['body']);
 583                                 $item['body'] = trim(preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item['body']));
 584
 585                                 // Replace the content when the title is longer than the body
 586                                 $replace = (strlen($item['title']) > strlen($item['body']));
 587
 588                                 // Replace it, when there is an image in the body
 589                                 if (strstr($item['body'], '[/img]')) {
 590                                         $replace = true;
 591                                 }
 592
 593                                 // Replace it, when there is a link in the body
 594                                 if (strstr($item['body'], '[/url]')) {
 595                                         $replace = true;
 596                                 }
 597
 598                                 $saved_body = $item['body'];
 599                                 $saved_title = $item['title'];
 600
 601                                 if ($replace) {
 602                                         $item['body'] = trim($item['title']);
 603                                 }
 604
 605                                 $data = ParseUrl::getSiteinfoCached($item['plink']);
 606                                 if (!empty($data['text']) && !empty($data['title']) && (mb_strlen($item['body']) < mb_strlen($data['text']))) {
 607                                         // When the fetched page info text is longer than the body, we do try to enhance the body
 608                                         if (!empty($item['body']) && (strpos($data['title'], $item['body']) === false) && (strpos($data['text'], $item['body']) === false)) {
 609                                                 // The body is not part of the fetched page info title or page info text. So we add the text to the body
 610                                                 $item['body'] .= "\n\n" . $data['text'];
 611                                         } else {
 612                                                 // Else we replace the body with the page info text
 613                                                 $item['body'] = $data['text'];
 614                                         }
 615                                 }
 616
 617                                 $data = PageInfo::queryUrl(
 618                                         $item['plink'],
 619                                         false,
 620                                         $preview,
 621                                         ($contact['fetch_further_information'] == 2),
 622                                         $contact['ffi_keyword_blacklist'] ?? ''
 623                                 );
 624
 625                                 if (!empty($data)) {
 626                                         // Take the data that was provided by the feed if the query is empty
 627                                         if (($data['type'] == 'link') && empty($data['title']) && empty($data['text'])) {
 628                                                 $data['title'] = $saved_title;
 629                                                 $item['body'] = $saved_body;
 630                                         }
 631
 632                                         $data_text = strip_tags(trim($data['text'] ?? ''));
 633                                         $item_body = strip_tags(trim($item['body'] ?? ''));
 634
 635                                         if (!empty($data_text) && (($data_text == $item_body) || strstr($item_body, $data_text))) {
 636                                                 $data['text'] = '';
 637                                         }
 638
 639                                         // We always strip the title since it will be added in the page information
 640                                         $item['title'] = '';
 641                                         $item['body'] = $item['body'] . "\n" . PageInfo::getFooterFromData($data, false);
 642                                         $taglist = $contact['fetch_further_information'] == 2 ? PageInfo::getTagsFromUrl($item['plink'], $preview, $contact['ffi_keyword_blacklist'] ?? '') : [];
 643                                         $item['object-type'] = Activity\ObjectType::BOOKMARK;
 644                                         $attachments = [];
 645
 646                                         foreach (['audio', 'video'] as $elementname) {
 647                                                 if (!empty($data[$elementname])) {
 648                                                         foreach ($data[$elementname] as $element) {
 649                                                                 if (!empty($element['src'])) {
 650                                                                         $src = $element['src'];
 651                                                                 } elseif (!empty($element['content'])) {
 652                                                                         $src = $element['content'];
 653                                                                 } else {
 654                                                                         continue;
 655                                                                 }
 656
 657                                                                 $attachments[] = [
 658                                                                         'type'        => ($elementname == 'audio') ? Post\Media::AUDIO : Post\Media::VIDEO,
 659                                                                         'url'         => $src,
 660                                                                         'preview'     => $element['image']       ?? null,
 661                                                                         'mimetype'    => $element['contenttype'] ?? null,
 662                                                                         'name'        => $element['name']        ?? null,
 663                                                                         'description' => $element['description'] ?? null,
 664                                                                 ];
 665                                                         }
 666                                                 }
 667                                         }
 668                                 }
 669                         } else {
 670                                 if (!empty($summary)) {
 671                                         $item['body'] = '[abstract]' . HTML::toBBCode($summary, $basepath) . "[/abstract]\n" . $item['body'];
 672                                 }
 673
 674                                 if ($fetch_further_information == LocalRelationship::FFI_KEYWORD) {
 675                                         if (empty($taglist)) {
 676                                                 $taglist = PageInfo::getTagsFromUrl($item['plink'], $preview, $contact['ffi_keyword_blacklist'] ?? '');
 677                                         }
 678                                         $item['body'] .= "\n" . self::tagToString($taglist);
 679                                 } else {
 680                                         $taglist = [];
 681                                 }
 682
 683                                 // Add the link to the original feed entry if not present in feed
 684                                 if (($item['plink'] != '') && !strstr($item['body'], $item['plink']) && !in_array($item['plink'], array_column($attachments, 'url'))) {
 685                                         $item['body'] .= '[hr][url]' . $item['plink'] . '[/url]';
 686                                 }
 687                         }
 688
 689                         if (empty($item['title'])) {
 690                                 $item['post-type'] = Item::PT_NOTE;
 691                         }
 692
 693                         Logger::info('Stored feed', ['item' => $item]);
 694
 695                         $notify = Item::isRemoteSelf($contact, $item);
 696                         $item['wall'] = (bool)$notify;
 697
 698                         // Distributed items should have a well-formatted URI.
 699                         // Additionally, we have to avoid conflicts with identical URI between imported feeds and these items.
 700                         if ($notify) {
 701                                 $item['guid'] = Item::guidFromUri($orig_plink, DI::baseUrl()->getHost());
 702                                 $item['uri']  = Item::newURI($item['guid']);
 703                                 unset($item['plink']);
 704                                 unset($item['thr-parent']);
 705                                 unset($item['parent-uri']);
 706
 707                                 // Set the delivery priority for "remote self" to "medium"
 708                                 $notify = Worker::PRIORITY_MEDIUM;
 709                         }
 710
 711                         $condition = ['uid' => $item['uid'], 'uri' => $item['uri']];
 712                         if (!Post::exists($condition) && !Post\Delayed::exists($item['uri'], $item['uid'])) {
 713                                 if (!$notify) {
 714                                         Post\Delayed::publish($item, $notify, $taglist, $attachments);
 715                                 } else {
 716                                         $postings[] = [
 717                                                 'item' => $item, 'notify' => $notify,
 718                                                 'taglist' => $taglist, 'attachments' => $attachments
 719                                         ];
 720                                 }
 721                         } else {
 722                                 Logger::info('Post already created or exists in the delayed posts queue', ['uid' => $item['uid'], 'uri' => $item['uri']]);
 723                         }
 724                 }
 725
 726                 if (!empty($postings)) {
 727                         $min_posting = DI::config()->get('system', 'minimum_posting_interval', 0);
 728                         $total = count($postings);
 729                         if ($total > 1) {
 730                                 // Posts shouldn't be delayed more than a day
 731                                 $interval = min(1440, self::getPollInterval($contact));
 732                                 $delay = max(round(($interval * 60) / $total), 60 * $min_posting);
 733                                 Logger::info('Got posting delay', ['delay' => $delay, 'interval' => $interval, 'items' => $total, 'cid' => $contact['id'], 'url' => $contact['url']]);
 734                         } else {
 735                                 $delay = 0;
 736                         }
 737
 738                         $post_delay = 0;
 739
 740                         foreach ($postings as $posting) {
 741                                 if ($delay > 0) {
 742                                         $publish_time = time() + $post_delay;
 743                                         $post_delay += $delay;
 744                                 } else {
 745                                         $publish_time = time();
 746                                 }
 747
 748                                 $last_publish = DI::pConfig()->get($posting['item']['uid'], 'system', 'last_publish', 0, true);
 749                                 $next_publish = max($last_publish + (60 * $min_posting), time());
 750                                 if ($publish_time < $next_publish) {
 751                                         $publish_time = $next_publish;
 752                                 }
 753                                 $publish_at = date(DateTimeFormat::MYSQL, $publish_time);
 754
 755                                 if (Post\Delayed::add($posting['item']['uri'], $posting['item'], $posting['notify'], Post\Delayed::PREPARED, $publish_at, $posting['taglist'], $posting['attachments'])) {
 756                                         DI::pConfig()->set($item['uid'], 'system', 'last_publish', $publish_time);
 757                                 }
 758                         }
 759                 }
 760
 761                 if (!$dryRun && DI::config()->get('system', 'adjust_poll_frequency')) {
 762                         self::adjustPollFrequency($contact, $creation_dates);
 763                 }
 764
 765                 return ['header' => $author, 'items' => $items];
 766         }
 767
 768         /**
 769          * Automatically adjust the poll frequency according to the post frequency
 770          *
 771          * @param array $contact Contact array
 772          * @param array $creation_dates
 773          * @return void
 774          */
 775         private static function adjustPollFrequency(array $contact, array $creation_dates)
 776         {
 777                 if ($contact['network'] != Protocol::FEED) {
 778                         Logger::info('Contact is no feed, skip.', ['id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url'], 'network' => $contact['network']]);
 779                         return;
 780                 }
 781
 782                 if (!empty($creation_dates)) {
 783                         // Count the post frequency and the earliest and latest post date
 784                         $frequency = [];
 785                         $oldest = time();
 786                         $newest = 0;
 787                         $oldest_date = $newest_date = '';
 788
 789                         foreach ($creation_dates as $date) {
 790                                 $timestamp = strtotime($date);
 791                                 $day = intdiv($timestamp, 86400);
 792                                 $hour = $timestamp % 86400;
 793
 794                                 // Only have a look at values from the last seven days
 795                                 if (((time() / 86400) - $day) < 7) {
 796                                         if (empty($frequency[$day])) {
 797                                                 $frequency[$day] = ['count' => 1, 'low' => $hour, 'high' => $hour];
 798                                         } else {
 799                                                 ++$frequency[$day]['count'];
 800                                                 if ($frequency[$day]['low'] > $hour) {
 801                                                         $frequency[$day]['low'] = $hour;
 802                                                 }
 803                                                 if ($frequency[$day]['high'] < $hour) {
 804                                                         $frequency[$day]['high'] = $hour;
 805                                                 }
 806                                         }
 807                                 }
 808                                 if ($oldest > $day) {
 809                                         $oldest = $day;
 810                                         $oldest_date = $date;
 811                                 }
 812
 813                                 if ($newest < $day) {
 814                                         $newest = $day;
 815                                         $newest_date = $date;
 816                                 }
 817                         }
 818
 819                         if (count($creation_dates) == 1) {
 820                                 Logger::info('Feed had posted a single time, switching to daily polling', ['newest' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 821                                 $priority = 8; // Poll once a day
 822                         }
 823
 824                         if (empty($priority) && (((time() / 86400) - $newest) > 730)) {
 825                                 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']]);
 826                                 $priority = 10; // Poll every month
 827                         }
 828
 829                         if (empty($priority) && (((time() / 86400) - $newest) > 365)) {
 830                                 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']]);
 831                                 $priority = 9; // Poll every week
 832                         }
 833
 834                         if (empty($priority) && empty($frequency)) {
 835                                 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']]);
 836                                 $priority = 8; // Poll once a day
 837                         }
 838
 839                         if (empty($priority)) {
 840                                 // Calculate the highest "posts per day" value
 841                                 $max = 0;
 842                                 foreach ($frequency as $entry) {
 843                                         if (($entry['count'] == 1) || ($entry['high'] == $entry['low'])) {
 844                                                 continue;
 845                                         }
 846
 847                                         // We take the earliest and latest post day and interpolate the number of post per day
 848                                         // that would had been created with this post frequency
 849
 850                                         // Assume at least four hours between oldest and newest post per day - should be okay for news outlets
 851                                         $duration = max($entry['high'] - $entry['low'], 14400);
 852                                         $ppd = (86400 / $duration) * $entry['count'];
 853                                         if ($ppd > $max) {
 854                                                 $max = $ppd;
 855                                         }
 856                                 }
 857                                 if ($max > 48) {
 858                                         $priority = 1; // Poll every quarter hour
 859                                 } elseif ($max > 24) {
 860                                         $priority = 2; // Poll half an hour
 861                                 } elseif ($max > 12) {
 862                                         $priority = 3; // Poll hourly
 863                                 } elseif ($max > 8) {
 864                                         $priority = 4; // Poll every two hours
 865                                 } elseif ($max > 4) {
 866                                         $priority = 5; // Poll every three hours
 867                                 } elseif ($max > 2) {
 868                                         $priority = 6; // Poll every six hours
 869                                 } else {
 870                                         $priority = 7; // Poll twice a day
 871                                 }
 872                                 Logger::info('Calculated priority by the posts per day', ['priority' => $priority, 'max' => round($max, 2), 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 873                         }
 874                 } else {
 875                         Logger::info('No posts, switching to daily polling', ['id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 876                         $priority = 8; // Poll once a day
 877                 }
 878
 879                 if ($contact['rating'] != $priority) {
 880                         Logger::notice('Adjusting priority', ['old' => $contact['rating'], 'new' => $priority, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 881                         Contact::update(['rating' => $priority], ['id' => $contact['id']]);
 882                 }
 883         }
 884
 885         /**
 886          * Get the poll interval for the given contact array
 887          *
 888          * @param array $contact
 889          * @return int Poll interval in minutes
 890          */
 891         public static function getPollInterval(array $contact): int
 892         {
 893                 if (in_array($contact['network'], [Protocol::MAIL, Protocol::FEED])) {
 894                         $ratings = [0, 3, 7, 8, 9, 10];
 895                         if (DI::config()->get('system', 'adjust_poll_frequency') && ($contact['network'] == Protocol::FEED)) {
 896                                 $rating = $contact['rating'];
 897                         } elseif (array_key_exists($contact['priority'], $ratings)) {
 898                                 $rating = $ratings[$contact['priority']];
 899                         } else {
 900                                 $rating = -1;
 901                         }
 902                 } else {
 903                         // Check once a week per default for all other networks
 904                         $rating = 9;
 905                 }
 906
 907                 // Friendica and OStatus are checked once a day
 908                 if (in_array($contact['network'], [Protocol::DFRN, Protocol::OSTATUS])) {
 909                         $rating = 8;
 910                 }
 911
 912                 // Check archived contacts or contacts with unsupported protocols once a month
 913                 if ($contact['archive'] || in_array($contact['network'], [Protocol::ZOT, Protocol::PHANTOM])) {
 914                         $rating = 10;
 915                 }
 916
 917                 if ($rating < 0) {
 918                         return 0;
 919                 }
 920                 /*
 921                  * Based on $contact['priority'], should we poll this site now? Or later?
 922                  */
 923
 924                 $min_poll_interval = max(1, DI::config()->get('system', 'min_poll_interval'));
 925
 926                 $poll_intervals = [$min_poll_interval, 15, 30, 60, 120, 180, 360, 720, 1440, 10080, 43200];
 927
 928                 //$poll_intervals = [$min_poll_interval . ' minute', '15 minute', '30 minute',
 929                 //      '1 hour', '2 hour', '3 hour', '6 hour', '12 hour' ,'1 day', '1 week', '1 month'];
 930
 931                 return $poll_intervals[$rating];
 932         }
 933
 934         /**
 935          * Convert a tag array to a tag string
 936          *
 937          * @param array $tags
 938          * @return string tag string
 939          */
 940         private static function tagToString(array $tags): string
 941         {
 942                 $tagstr = '';
 943
 944                 foreach ($tags as $tag) {
 945                         if ($tagstr != '') {
 946                                 $tagstr .= ', ';
 947                         }
 948
 949                         $tagstr .= '#[url=' . DI::baseUrl() . '/search?tag=' . urlencode($tag) . ']' . $tag . '[/url]';
 950                 }
 951
 952                 return $tagstr;
 953         }
 954
 955         private static function titleIsBody(string $title, string $body): bool
 956         {
 957                 $title = strip_tags($title);
 958                 $title = trim($title);
 959                 $title = html_entity_decode($title, ENT_QUOTES, 'UTF-8');
 960                 $title = str_replace(["\n", "\r", "\t", " "], ['', '', '', ''], $title);
 961
 962                 $body = strip_tags($body);
 963                 $body = trim($body);
 964                 $body = html_entity_decode($body, ENT_QUOTES, 'UTF-8');
 965                 $body = str_replace(["\n", "\r", "\t", " "], ['', '', '', ''], $body);
 966
 967                 if (strlen($title) < strlen($body)) {
 968                         $body = substr($body, 0, strlen($title));
 969                 }
 970
 971                 if (($title != $body) && (substr($title, -3) == '...')) {
 972                         $pos = strrpos($title, '...');
 973                         if ($pos > 0) {
 974                                 $title = substr($title, 0, $pos);
 975                                 $body = substr($body, 0, $pos);
 976                         }
 977                 }
 978                 return ($title == $body);
 979         }
 980
 981         /**
 982          * Creates the Atom feed for a given nickname
 983          *
 984          * Supported filters:
 985          * - activity (default): all the public posts
 986          * - posts: all the public top-level posts
 987          * - comments: all the public replies
 988          *
 989          * Updates the provided last_update parameter if the result comes from the
 990          * cache or it is empty
 991          *
 992          * @param array   $owner       owner-view record of the feed owner
 993          * @param string  $last_update Date of the last update
 994          * @param integer $max_items   Number of maximum items to fetch
 995          * @param string  $filter      Feed items filter (activity, posts or comments)
 996          * @param boolean $nocache     Wether to bypass caching
 997          *
 998          * @return string Atom feed
 999          * @throws HTTPException\InternalServerErrorException
1000          * @throws \ImagickException
1001          */
1002         public static function atom(array $owner, string $last_update, int $max_items = 300, string $filter = 'activity', bool $nocache = false)
1003         {
1004                 $stamp = microtime(true);
1005
1006                 $cachekey = 'feed:feed:' . $owner['nickname'] . ':' . $filter . ':' . $last_update;
1007
1008                 // Display events in the user's timezone
1009                 if (strlen($owner['timezone'])) {
1010                         DI::app()->setTimeZone($owner['timezone']);
1011                 }
1012
1013                 $previous_created = $last_update;
1014
1015                 // Don't cache when the last item was posted less than 15 minutes ago (Cache duration)
1016                 if ((time() - strtotime($owner['last-item'])) < 15 * 60) {
1017                         $result = DI::cache()->get($cachekey);
1018                         if (!$nocache && !is_null($result)) {
1019                                 Logger::info('Cached feed duration', ['seconds' => number_format(microtime(true) - $stamp, 3), 'nick' => $owner['nickname'], 'filter' => $filter, 'created' => $previous_created]);
1020                                 return $result['feed'];
1021                         }
1022                 }
1023
1024                 $check_date = empty($last_update) ? '' : DateTimeFormat::utc($last_update);
1025                 $authorid = Contact::getIdForURL($owner['url']);
1026
1027                 $condition = [
1028                         "`uid` = ? AND `received` > ? AND NOT `deleted` AND `gravity` IN (?, ?)
1029                         AND `private` != ? AND `visible` AND `wall` AND `parent-network` IN (?, ?, ?, ?)",
1030                         $owner['uid'], $check_date, Item::GRAVITY_PARENT, Item::GRAVITY_COMMENT,
1031                         Item::PRIVATE, Protocol::ACTIVITYPUB,
1032                         Protocol::OSTATUS, Protocol::DFRN, Protocol::DIASPORA
1033                 ];
1034
1035                 if ($filter === 'comments') {
1036                         $condition[0] .= " AND `gravity` = ? ";
1037                         $condition[] = Item::GRAVITY_COMMENT;
1038                 }
1039
1040                 if ($owner['account-type'] != User::ACCOUNT_TYPE_COMMUNITY) {
1041                         $condition[0] .= " AND `contact-id` = ? AND `author-id` = ?";
1042                         $condition[] = $owner['id'];
1043                         $condition[] = $authorid;
1044                 }
1045
1046                 $params = ['order' => ['received' => true], 'limit' => $max_items];
1047
1048                 if ($filter === 'posts') {
1049                         $ret = Post::selectThread(Item::DELIVER_FIELDLIST, $condition, $params);
1050                 } else {
1051                         $ret = Post::select(Item::DELIVER_FIELDLIST, $condition, $params);
1052                 }
1053
1054                 $items = Post::toArray($ret);
1055
1056                 $doc = new DOMDocument('1.0', 'utf-8');
1057                 $doc->formatOutput = true;
1058
1059                 $root = self::addHeader($doc, $owner, $filter);
1060
1061                 foreach ($items as $item) {
1062                         $entry = self::noteEntry($doc, $item, $owner);
1063                         $root->appendChild($entry);
1064
1065                         if ($last_update < $item['created']) {
1066                                 $last_update = $item['created'];
1067                         }
1068                 }
1069
1070                 $feeddata = trim($doc->saveXML());
1071
1072                 $msg = ['feed' => $feeddata, 'last_update' => $last_update];
1073                 DI::cache()->set($cachekey, $msg, Duration::QUARTER_HOUR);
1074
1075                 Logger::info('Feed duration', ['seconds' => number_format(microtime(true) - $stamp, 3), 'nick' => $owner['nickname'], 'filter' => $filter, 'created' => $previous_created]);
1076
1077                 return $feeddata;
1078         }
1079
1080         /**
1081          * Adds the header elements to the XML document
1082          *
1083          * @param DOMDocument $doc       XML document
1084          * @param array       $owner     Contact data of the poster
1085          * @param string      $filter    The related feed filter (activity, posts or comments)
1086          *
1087          * @return DOMElement Header root element
1088          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1089          */
1090         private static function addHeader(DOMDocument $doc, array $owner, string $filter): DOMElement
1091         {
1092                 $root = $doc->createElementNS(ActivityNamespace::ATOM1, 'feed');
1093                 $doc->appendChild($root);
1094
1095                 $title = '';
1096                 $selfUri = '/feed/' . $owner['nick'] . '/';
1097                 switch ($filter) {
1098                         case 'activity':
1099                                 $title = DI::l10n()->t('%s\'s timeline', $owner['name']);
1100                                 $selfUri .= $filter;
1101                                 break;
1102                         case 'posts':
1103                                 $title = DI::l10n()->t('%s\'s posts', $owner['name']);
1104                                 break;
1105                         case 'comments':
1106                                 $title = DI::l10n()->t('%s\'s comments', $owner['name']);
1107                                 $selfUri .= $filter;
1108                                 break;
1109                 }
1110
1111                 $attributes = ['uri' => 'https://friendi.ca', 'version' => App::VERSION . '-' . DB_UPDATE_VERSION];
1112                 XML::addElement($doc, $root, 'generator', App::PLATFORM, $attributes);
1113                 XML::addElement($doc, $root, 'id', DI::baseUrl() . '/profile/' . $owner['nick']);
1114                 XML::addElement($doc, $root, 'title', $title);
1115                 XML::addElement($doc, $root, 'subtitle', sprintf("Updates from %s on %s", $owner['name'], DI::config()->get('config', 'sitename')));
1116                 XML::addElement($doc, $root, 'logo', User::getAvatarUrl($owner, Proxy::SIZE_SMALL));
1117                 XML::addElement($doc, $root, 'updated', DateTimeFormat::utcNow(DateTimeFormat::ATOM));
1118
1119                 $author = self::addAuthor($doc, $owner);
1120                 $root->appendChild($author);
1121
1122                 $attributes = ['href' => $owner['url'], 'rel' => 'alternate', 'type' => 'text/html'];
1123                 XML::addElement($doc, $root, 'link', '', $attributes);
1124
1125                 OStatus::addHubLink($doc, $root, $owner['nick']);
1126
1127                 $attributes = ['href' => DI::baseUrl() . $selfUri, 'rel' => 'self', 'type' => 'application/atom+xml'];
1128                 XML::addElement($doc, $root, 'link', '', $attributes);
1129
1130                 return $root;
1131         }
1132
1133         /**
1134          * Adds the author element to the XML document
1135          *
1136          * @param DOMDocument $doc          XML document
1137          * @param array       $owner        Contact data of the poster
1138          * @return DOMElement author element
1139          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1140          */
1141         private static function addAuthor(DOMDocument $doc, array $owner): DOMElement
1142         {
1143                 $author = $doc->createElement('author');
1144                 XML::addElement($doc, $author, 'uri', $owner['url']);
1145                 XML::addElement($doc, $author, 'name', $owner['nick']);
1146                 XML::addElement($doc, $author, 'email', $owner['addr']);
1147
1148                 return $author;
1149         }
1150
1151         /**
1152          * Adds a regular entry element
1153          *
1154          * @param DOMDocument $doc       XML document
1155          * @param array       $item      Data of the item that is to be posted
1156          * @param array       $owner     Contact data of the poster
1157          * @param bool        $toplevel  Is it for en entry element (false) or a feed entry (true)?
1158          * @return DOMElement Entry element
1159          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1160          * @throws \ImagickException
1161          */
1162         private static function noteEntry(DOMDocument $doc, array $item, array $owner): DOMElement
1163         {
1164                 if (($item['gravity'] != Item::GRAVITY_PARENT) && (Strings::normaliseLink($item['author-link']) != Strings::normaliseLink($owner['url']))) {
1165                         Logger::info('Feed entry author does not match feed owner', ['owner' => $owner['url'], 'author' => $item['author-link']]);
1166                 }
1167
1168                 $entry = OStatus::entryHeader($doc, $owner, $item, false);
1169
1170                 self::entryContent($doc, $entry, $item, self::getTitle($item), '', true);
1171
1172                 self::entryFooter($doc, $entry, $item, $owner);
1173
1174                 return $entry;
1175         }
1176
1177         /**
1178          * Adds elements to the XML document
1179          *
1180          * @param DOMDocument $doc       XML document
1181          * @param \DOMElement $entry     Entry element where the content is added
1182          * @param array       $item      Data of the item that is to be posted
1183          * @param array       $owner     Contact data of the poster
1184          * @param string      $title     Title for the post
1185          * @param string      $verb      The activity verb
1186          * @param bool        $complete  Add the "status_net" element?
1187          * @return void
1188          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1189          */
1190         private static function entryContent(DOMDocument $doc, DOMElement $entry, array $item, $title, string $verb = '', bool $complete = true)
1191         {
1192                 if ($verb == '') {
1193                         $verb = OStatus::constructVerb($item);
1194                 }
1195
1196                 XML::addElement($doc, $entry, 'id', $item['uri']);
1197                 XML::addElement($doc, $entry, 'title', html_entity_decode($title, ENT_QUOTES, 'UTF-8'));
1198
1199                 $body = Post\Media::addAttachmentsToBody($item['uri-id'], DI::contentItem()->addSharedPost($item));
1200                 $body = Post\Media::addHTMLLinkToBody($item['uri-id'], $body);
1201
1202                 $body = BBCode::convertForUriId($item['uri-id'], $body, BBCode::ACTIVITYPUB);
1203
1204                 XML::addElement($doc, $entry, 'content', $body, ['type' => 'html']);
1205
1206                 XML::addElement(
1207                         $doc,
1208                         $entry,
1209                         'link',
1210                         '',
1211                         [
1212                                 'rel' => 'alternate', 'type' => 'text/html',
1213                                 'href' => DI::baseUrl() . '/display/' . $item['guid']
1214                         ]
1215                 );
1216
1217                 XML::addElement($doc, $entry, 'published', DateTimeFormat::utc($item['created'] . '+00:00', DateTimeFormat::ATOM));
1218                 XML::addElement($doc, $entry, 'updated', DateTimeFormat::utc($item['edited'] . '+00:00', DateTimeFormat::ATOM));
1219         }
1220
1221         /**
1222          * Adds the elements at the foot of an entry to the XML document
1223          *
1224          * @param DOMDocument $doc       XML document
1225          * @param object      $entry     The entry element where the elements are added
1226          * @param array       $item      Data of the item that is to be posted
1227          * @param array       $owner     Contact data of the poster
1228          * @param bool        $complete  default true
1229          * @return void
1230          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1231          */
1232         private static function entryFooter(DOMDocument $doc, $entry, array $item, array $owner)
1233         {
1234                 $mentioned = [];
1235
1236                 if ($item['gravity'] != Item::GRAVITY_PARENT) {
1237                         $parent = Post::selectFirst(['guid', 'author-link', 'owner-link'], ['id' => $item['parent']]);
1238
1239                         $thrparent = Post::selectFirst(['guid', 'author-link', 'owner-link', 'plink'], ['uid' => $owner['uid'], 'uri' => $item['thr-parent']]);
1240
1241                         if (DBA::isResult($thrparent)) {
1242                                 $mentioned[$thrparent['author-link']] = $thrparent['author-link'];
1243                                 $mentioned[$thrparent['owner-link']]  = $thrparent['owner-link'];
1244                                 $parent_plink                         = $thrparent['plink'];
1245                         } elseif (DBA::isResult($parent)) {
1246                                 $mentioned[$parent['author-link']] = $parent['author-link'];
1247                                 $mentioned[$parent['owner-link']]  = $parent['owner-link'];
1248                                 $parent_plink                      = DI::baseUrl() . '/display/' . $parent['guid'];
1249                         } else {
1250                                 DI::logger()->notice('Missing parent and thr-parent for child item', ['item' => $item]);
1251                         }
1252
1253                         if (isset($parent_plink)) {
1254                                 $attributes = [
1255                                         'ref'  => $item['thr-parent'],
1256                                         'href' => $parent_plink
1257                                 ];
1258                                 XML::addElement($doc, $entry, 'thr:in-reply-to', '', $attributes);
1259
1260                                 $attributes = [
1261                                         'rel'  => 'related',
1262                                         'href' => $parent_plink
1263                                 ];
1264                                 XML::addElement($doc, $entry, 'link', '', $attributes);
1265                         }
1266                 }
1267
1268                 // uri-id isn't present for follow entry pseudo-items
1269                 $tags = Tag::getByURIId($item['uri-id'] ?? 0);
1270                 foreach ($tags as $tag) {
1271                         $mentioned[$tag['url']] = $tag['url'];
1272                 }
1273
1274                 foreach ($tags as $tag) {
1275                         if ($tag['type'] == Tag::HASHTAG) {
1276                                 XML::addElement($doc, $entry, 'category', '', ['term' => $tag['name']]);
1277                         }
1278                 }
1279
1280                 OStatus::getAttachment($doc, $entry, $item);
1281         }
1282
1283         /**
1284          * Fetch or create title for feed entry
1285          *
1286          * @param array $item
1287          * @return string title
1288          */
1289         private static function getTitle(array $item): string
1290         {
1291                 if ($item['title'] != '') {
1292                         return BBCode::convertForUriId($item['uri-id'], $item['title'], BBCode::ACTIVITYPUB);
1293                 }
1294
1295                 // Fetch information about the post
1296                 $media = Post\Media::getByURIId($item['uri-id'], [Post\Media::HTML]);
1297                 if (!empty($media) && !empty($media[0]['name']) && ($media[0]['name'] != $media[0]['url'])) {
1298                         return $media[0]['name'];
1299                 }
1300
1301                 // If no bookmark is found then take the first line
1302                 // Remove the share element before fetching the first line
1303                 $title = trim(preg_replace("/\[share.*?\](.*?)\[\/share\]/ism", "\n$1\n", $item['body']));
1304
1305                 $title = BBCode::toPlaintext($title) . "\n";
1306                 $pos = strpos($title, "\n");
1307                 $trailer = '';
1308                 if (($pos == 0) || ($pos > 100)) {
1309                         $pos = 100;
1310                         $trailer = '...';
1311                 }
1312
1313                 return substr($title, 0, $pos) . $trailer;
1314         }
1315 }