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