src/Protocol/Feed.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2020, Friendica
   4  *
   5  * @license GNU AGPL version 3 or any later version
   6  *
   7  * This program is free software: you can redistribute it and/or modify
   8  * it under the terms of the GNU Affero General Public License as
   9  * published by the Free Software Foundation, either version 3 of the
  10  * License, or (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU Affero General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Affero General Public License
  18  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  19  *
  20  */
  21
  22 namespace Friendica\Protocol;
  23
  24 use DOMDocument;
  25 use DOMXPath;
  26 use Friendica\Content\PageInfo;
  27 use Friendica\Content\Text\BBCode;
  28 use Friendica\Content\Text\HTML;
  29 use Friendica\Core\Cache\Duration;
  30 use Friendica\Core\Logger;
  31 use Friendica\Core\Protocol;
  32 use Friendica\Core\Worker;
  33 use Friendica\Database\DBA;
  34 use Friendica\DI;
  35 use Friendica\Model\Contact;
  36 use Friendica\Model\Conversation;
  37 use Friendica\Model\Item;
  38 use Friendica\Model\Post;
  39 use Friendica\Model\Tag;
  40 use Friendica\Model\User;
  41 use Friendica\Util\DateTimeFormat;
  42 use Friendica\Util\Network;
  43 use Friendica\Util\ParseUrl;
  44 use Friendica\Util\Strings;
  45 use Friendica\Util\XML;
  46
  47 /**
  48  * This class contain functions to import feeds (RSS/RDF/Atom)
  49  */
  50 class Feed
  51 {
  52         /**
  53          * consume - process atom feed and update anything/everything we might need to update
  54          *
  55          * $xml = the (atom) feed to consume - RSS isn't as fully supported but may work for simple feeds.
  56          *
  57          * $importer = the contact_record (joined to user_record) of the local user who owns this relationship.
  58          *             It is this person's stuff that is going to be updated.
  59          * $contact =  the person who is sending us stuff. If not set, we MAY be processing a "follow" activity
  60          *             from an external network and MAY create an appropriate contact record. Otherwise, we MUST
  61          *             have a contact record.
  62          * $hub = should we find a hub declation in the feed, pass it back to our calling process, who might (or
  63          *        might not) try and subscribe to it.
  64          * $datedir sorts in reverse order
  65          * $pass - by default ($pass = 0) we cannot guarantee that a parent item has been
  66          *      imported prior to its children being seen in the stream unless we are certain
  67          *      of how the feed is arranged/ordered.
  68          * With $pass = 1, we only pull parent items out of the stream.
  69          * With $pass = 2, we only pull children (comments/likes).
  70          *
  71          * So running this twice, first with pass 1 and then with pass 2 will do the right
  72          * thing regardless of feed ordering. This won't be adequate in a fully-threaded
  73          * model where comments can have sub-threads. That would require some massive sorting
  74          * to get all the feed items into a mostly linear ordering, and might still require
  75          * recursion.
  76          *
  77          * @param       $xml
  78          * @param array $importer
  79          * @param array $contact
  80          * @param       $hub
  81          * @throws ImagickException
  82          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  83          */
  84         public static function consume($xml, array $importer, array $contact, &$hub)
  85         {
  86                 if ($contact['network'] === Protocol::OSTATUS) {
  87                         Logger::info('Consume OStatus messages');
  88                         OStatus::import($xml, $importer, $contact, $hub);
  89
  90                         return;
  91                 }
  92
  93                 if ($contact['network'] === Protocol::FEED) {
  94                         Logger::info('Consume feeds');
  95                         self::import($xml, $importer, $contact);
  96
  97                         return;
  98                 }
  99
 100                 if ($contact['network'] === Protocol::DFRN) {
 101                         Logger::info('Consume DFRN messages');
 102                         $dfrn_importer = DFRN::getImporter($contact['id'], $importer['uid']);
 103                         if (!empty($dfrn_importer)) {
 104                                 Logger::info('Now import the DFRN feed');
 105                                 DFRN::import($xml, $dfrn_importer, true, Conversation::PARCEL_LEGACY_DFRN);
 106                                 return;
 107                         }
 108                 }
 109         }
 110
 111         /**
 112          * Read a RSS/RDF/Atom feed and create an item entry for it
 113          *
 114          * @param string $xml      The feed data
 115          * @param array  $importer The user record of the importer
 116          * @param array  $contact  The contact record of the feed
 117          *
 118          * @return array Returns the header and the first item in dry run mode
 119          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 120          */
 121         public static function import($xml, array $importer = [], array $contact = [])
 122         {
 123                 $dryRun = empty($importer) && empty($contact);
 124
 125                 if ($dryRun) {
 126                         Logger::info("Test Atom/RSS feed");
 127                 } else {
 128                         Logger::info("Import Atom/RSS feed '" . $contact["name"] . "' (Contact " . $contact["id"] . ") for user " . $importer["uid"]);
 129                 }
 130
 131                 if (empty($xml)) {
 132                         Logger::info('XML is empty.');
 133                         return [];
 134                 }
 135
 136                 if (!empty($contact['poll'])) {
 137                         $basepath = $contact['poll'];
 138                 } elseif (!empty($contact['url'])) {
 139                         $basepath = $contact['url'];
 140                 } else {
 141                         $basepath = '';
 142                 }
 143
 144                 $doc = new DOMDocument();
 145                 @$doc->loadXML(trim($xml));
 146                 $xpath = new DOMXPath($doc);
 147                 $xpath->registerNamespace('atom', ActivityNamespace::ATOM1);
 148                 $xpath->registerNamespace('dc', "http://purl.org/dc/elements/1.1/");
 149                 $xpath->registerNamespace('content', "http://purl.org/rss/1.0/modules/content/");
 150                 $xpath->registerNamespace('rdf', "http://www.w3.org/1999/02/22-rdf-syntax-ns#");
 151                 $xpath->registerNamespace('rss', "http://purl.org/rss/1.0/");
 152                 $xpath->registerNamespace('media', "http://search.yahoo.com/mrss/");
 153                 $xpath->registerNamespace('poco', ActivityNamespace::POCO);
 154
 155                 $author = [];
 156                 $entries = null;
 157
 158                 // Is it RDF?
 159                 if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
 160                         $author["author-link"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:link/text()');
 161                         $author["author-name"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:title/text()');
 162
 163                         if (empty($author["author-name"])) {
 164                                 $author["author-name"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:description/text()');
 165                         }
 166                         $entries = $xpath->query('/rdf:RDF/rss:item');
 167                 }
 168
 169                 // Is it Atom?
 170                 if ($xpath->query('/atom:feed')->length > 0) {
 171                         $alternate = XML::getFirstAttributes($xpath, "atom:link[@rel='alternate']");
 172                         if (is_object($alternate)) {
 173                                 foreach ($alternate AS $attribute) {
 174                                         if ($attribute->name == "href") {
 175                                                 $author["author-link"] = $attribute->textContent;
 176                                         }
 177                                 }
 178                         }
 179
 180                         if (empty($author["author-link"])) {
 181                                 $self = XML::getFirstAttributes($xpath, "atom:link[@rel='self']");
 182                                 if (is_object($self)) {
 183                                         foreach ($self AS $attribute) {
 184                                                 if ($attribute->name == "href") {
 185                                                         $author["author-link"] = $attribute->textContent;
 186                                                 }
 187                                         }
 188                                 }
 189                         }
 190
 191                         if (empty($author["author-link"])) {
 192                                 $author["author-link"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:id/text()');
 193                         }
 194                         $author["author-avatar"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:logo/text()');
 195
 196                         $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:title/text()');
 197
 198                         if (empty($author["author-name"])) {
 199                                 $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:subtitle/text()');
 200                         }
 201
 202                         if (empty($author["author-name"])) {
 203                                 $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:name/text()');
 204                         }
 205
 206                         $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:displayName/text()');
 207                         if ($value != "") {
 208                                 $author["author-name"] = $value;
 209                         }
 210
 211                         if ($dryRun) {
 212                                 $author["author-id"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:id/text()');
 213
 214                                 // See https://tools.ietf.org/html/rfc4287#section-3.2.2
 215                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/atom:uri/text()');
 216                                 if ($value != "") {
 217                                         $author["author-link"] = $value;
 218                                 }
 219
 220                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:preferredUsername/text()');
 221                                 if ($value != "") {
 222                                         $author["author-nick"] = $value;
 223                                 }
 224
 225                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:address/poco:formatted/text()');
 226                                 if ($value != "") {
 227                                         $author["author-location"] = $value;
 228                                 }
 229
 230                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:note/text()');
 231                                 if ($value != "") {
 232                                         $author["author-about"] = $value;
 233                                 }
 234
 235                                 $avatar = XML::getFirstAttributes($xpath, "atom:author/atom:link[@rel='avatar']");
 236                                 if (is_object($avatar)) {
 237                                         foreach ($avatar AS $attribute) {
 238                                                 if ($attribute->name == "href") {
 239                                                         $author["author-avatar"] = $attribute->textContent;
 240                                                 }
 241                                         }
 242                                 }
 243                         }
 244
 245                         $author["edited"] = $author["created"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:updated/text()');
 246
 247                         $author["app"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:generator/text()');
 248
 249                         $entries = $xpath->query('/atom:feed/atom:entry');
 250                 }
 251
 252                 // Is it RSS?
 253                 if ($xpath->query('/rss/channel')->length > 0) {
 254                         $author["author-link"] = XML::getFirstNodeValue($xpath, '/rss/channel/link/text()');
 255
 256                         $author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/title/text()');
 257                         $author["author-avatar"] = XML::getFirstNodeValue($xpath, '/rss/channel/image/url/text()');
 258
 259                         if (empty($author["author-name"])) {
 260                                 $author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/copyright/text()');
 261                         }
 262
 263                         if (empty($author["author-name"])) {
 264                                 $author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/description/text()');
 265                         }
 266
 267                         $author["edited"] = $author["created"] = XML::getFirstNodeValue($xpath, '/rss/channel/pubDate/text()');
 268
 269                         $author["app"] = XML::getFirstNodeValue($xpath, '/rss/channel/generator/text()');
 270
 271                         $entries = $xpath->query('/rss/channel/item');
 272                 }
 273
 274                 if (!$dryRun) {
 275                         $author["author-link"] = $contact["url"];
 276
 277                         if (empty($author["author-name"])) {
 278                                 $author["author-name"] = $contact["name"];
 279                         }
 280
 281                         $author["author-avatar"] = $contact["thumb"];
 282
 283                         $author["owner-link"] = $contact["url"];
 284                         $author["owner-name"] = $contact["name"];
 285                         $author["owner-avatar"] = $contact["thumb"];
 286                 }
 287
 288                 $header = [];
 289                 $header["uid"] = $importer["uid"] ?? 0;
 290                 $header["network"] = Protocol::FEED;
 291                 $header["wall"] = 0;
 292                 $header["origin"] = 0;
 293                 $header["gravity"] = GRAVITY_PARENT;
 294                 $header["private"] = Item::PUBLIC;
 295                 $header["verb"] = Activity::POST;
 296                 $header["object-type"] = Activity\ObjectType::NOTE;
 297
 298                 $header["contact-id"] = $contact["id"] ?? 0;
 299
 300                 if (!is_object($entries)) {
 301                         Logger::info("There are no entries in this feed.");
 302                         return [];
 303                 }
 304
 305                 $items = [];
 306                 $creation_dates = [];
 307
 308                 // Limit the number of items that are about to be fetched
 309                 $total_items = ($entries->length - 1);
 310                 $max_items = DI::config()->get('system', 'max_feed_items');
 311                 if (($max_items > 0) && ($total_items > $max_items)) {
 312                         $total_items = $max_items;
 313                 }
 314
 315                 $postings = [];
 316
 317                 // Importing older entries first
 318                 for ($i = $total_items; $i >= 0; --$i) {
 319                         $entry = $entries->item($i);
 320
 321                         $item = array_merge($header, $author);
 322
 323                         $alternate = XML::getFirstAttributes($xpath, "atom:link[@rel='alternate']", $entry);
 324                         if (!is_object($alternate)) {
 325                                 $alternate = XML::getFirstAttributes($xpath, "atom:link", $entry);
 326                         }
 327                         if (is_object($alternate)) {
 328                                 foreach ($alternate AS $attribute) {
 329                                         if ($attribute->name == "href") {
 330                                                 $item["plink"] = $attribute->textContent;
 331                                         }
 332                                 }
 333                         }
 334
 335                         if (empty($item["plink"])) {
 336                                 $item["plink"] = XML::getFirstNodeValue($xpath, 'link/text()', $entry);
 337                         }
 338
 339                         if (empty($item["plink"])) {
 340                                 $item["plink"] = XML::getFirstNodeValue($xpath, 'rss:link/text()', $entry);
 341                         }
 342
 343                         $item["uri"] = XML::getFirstNodeValue($xpath, 'atom:id/text()', $entry);
 344
 345                         if (empty($item["uri"])) {
 346                                 $item["uri"] = XML::getFirstNodeValue($xpath, 'guid/text()', $entry);
 347                         }
 348
 349                         if (empty($item["uri"])) {
 350                                 $item["uri"] = $item["plink"];
 351                         }
 352
 353                         // Add the base path if missing
 354                         $item["uri"] = Network::addBasePath($item["uri"], $basepath);
 355                         $item["plink"] = Network::addBasePath($item["plink"], $basepath);
 356
 357                         $orig_plink = $item["plink"];
 358
 359                         $item["plink"] = DI::httpRequest()->finalUrl($item["plink"]);
 360
 361                         $item["title"] = XML::getFirstNodeValue($xpath, 'atom:title/text()', $entry);
 362
 363                         if (empty($item["title"])) {
 364                                 $item["title"] = XML::getFirstNodeValue($xpath, 'title/text()', $entry);
 365                         }
 366                         if (empty($item["title"])) {
 367                                 $item["title"] = XML::getFirstNodeValue($xpath, 'rss:title/text()', $entry);
 368                         }
 369
 370                         $item["title"] = html_entity_decode($item["title"], ENT_QUOTES, 'UTF-8');
 371
 372                         $published = XML::getFirstNodeValue($xpath, 'atom:published/text()', $entry);
 373
 374                         if (empty($published)) {
 375                                 $published = XML::getFirstNodeValue($xpath, 'pubDate/text()', $entry);
 376                         }
 377
 378                         if (empty($published)) {
 379                                 $published = XML::getFirstNodeValue($xpath, 'dc:date/text()', $entry);
 380                         }
 381
 382                         $updated = XML::getFirstNodeValue($xpath, 'atom:updated/text()', $entry);
 383
 384                         if (empty($updated) && !empty($published)) {
 385                                 $updated = $published;
 386                         }
 387
 388                         if (empty($published) && !empty($updated)) {
 389                                 $published = $updated;
 390                         }
 391
 392                         if ($published != "") {
 393                                 $item["created"] = $published;
 394                         }
 395
 396                         if ($updated != "") {
 397                                 $item["edited"] = $updated;
 398                         }
 399
 400                         if (!$dryRun) {
 401                                 $condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
 402                                         $importer["uid"], $item["uri"], Protocol::FEED, Protocol::DFRN];
 403                                 $previous = Item::selectFirst(['id', 'created'], $condition);
 404                                 if (DBA::isResult($previous)) {
 405                                         // Use the creation date when the post had been stored. It can happen this date changes in the feed.
 406                                         $creation_dates[] = $previous['created'];
 407                                         Logger::info("Item with uri " . $item["uri"] . " for user " . $importer["uid"] . " already existed under id " . $previous["id"]);
 408                                         continue;
 409                                 }
 410                                 $creation_dates[] = DateTimeFormat::utc($item['created']);
 411                         }
 412
 413                         $creator = XML::getFirstNodeValue($xpath, 'author/text()', $entry);
 414
 415                         if (empty($creator)) {
 416                                 $creator = XML::getFirstNodeValue($xpath, 'atom:author/atom:name/text()', $entry);
 417                         }
 418
 419                         if (empty($creator)) {
 420                                 $creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
 421                         }
 422
 423                         if ($creator != "") {
 424                                 $item["author-name"] = $creator;
 425                         }
 426
 427                         $creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
 428
 429                         if ($creator != "") {
 430                                 $item["author-name"] = $creator;
 431                         }
 432
 433                         /// @TODO ?
 434                         // <category>Ausland</category>
 435                         // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 436
 437                         $attachments = [];
 438
 439                         $enclosures = $xpath->query("enclosure|atom:link[@rel='enclosure']", $entry);
 440                         foreach ($enclosures AS $enclosure) {
 441                                 $href = "";
 442                                 $length = null;
 443                                 $type = null;
 444
 445                                 foreach ($enclosure->attributes AS $attribute) {
 446                                         if (in_array($attribute->name, ["url", "href"])) {
 447                                                 $href = $attribute->textContent;
 448                                         } elseif ($attribute->name == "length") {
 449                                                 $length = (int)$attribute->textContent;
 450                                         } elseif ($attribute->name == "type") {
 451                                                 $type = $attribute->textContent;
 452                                         }
 453                                 }
 454
 455                                 if (!empty($href)) {
 456                                         $attachments[] = ['type' => Post\Media::DOCUMENT, 'url' => $href, 'mimetype' => $type, 'size' => $length];
 457                                 }
 458                         }
 459
 460                         $taglist = [];
 461                         $categories = $xpath->query("category", $entry);
 462                         foreach ($categories AS $category) {
 463                                 $taglist[] = $category->nodeValue;
 464                         }
 465
 466                         $body = trim(XML::getFirstNodeValue($xpath, 'atom:content/text()', $entry));
 467
 468                         if (empty($body)) {
 469                                 $body = trim(XML::getFirstNodeValue($xpath, 'content:encoded/text()', $entry));
 470                         }
 471
 472                         $summary = trim(XML::getFirstNodeValue($xpath, 'atom:summary/text()', $entry));
 473
 474                         if (empty($summary)) {
 475                                 $summary = trim(XML::getFirstNodeValue($xpath, 'description/text()', $entry));
 476                         }
 477
 478                         if (empty($body)) {
 479                                 $body = $summary;
 480                                 $summary = '';
 481                         }
 482
 483                         if ($body == $summary) {
 484                                 $summary = '';
 485                         }
 486
 487                         // remove the content of the title if it is identically to the body
 488                         // This helps with auto generated titles e.g. from tumblr
 489                         if (self::titleIsBody($item["title"], $body)) {
 490                                 $item["title"] = "";
 491                         }
 492                         $item["body"] = HTML::toBBCode($body, $basepath);
 493
 494                         // Remove tracking pixels
 495                         $item["body"] = preg_replace("/\[img=1x1\]([^\[\]]*)\[\/img\]/Usi", '', $item["body"]);
 496
 497                         if (($item["body"] == '') && ($item["title"] != '')) {
 498                                 $item["body"] = $item["title"];
 499                                 $item["title"] = '';
 500                         }
 501
 502                         if ($dryRun) {
 503                                 $items[] = $item;
 504                                 break;
 505                         } elseif (!Item::isValid($item)) {
 506                                 Logger::info('Feed is invalid', ['created' => $item['created'], 'uid' => $item['uid'], 'uri' => $item['uri']]);
 507                                 continue;
 508                         }
 509
 510                         $preview = '';
 511                         if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) {
 512                                 // Handle enclosures and treat them as preview picture
 513                                 foreach ($attachments AS $attachment) {
 514                                         if ($attachment["mimetype"] == "image/jpeg") {
 515                                                 $preview = $attachment["url"];
 516                                         }
 517                                 }
 518
 519                                 // Remove a possible link to the item itself
 520                                 $item["body"] = str_replace($item["plink"], '', $item["body"]);
 521                                 $item["body"] = trim(preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item["body"]));
 522
 523                                 // Replace the content when the title is longer than the body
 524                                 $replace = (strlen($item["title"]) > strlen($item["body"]));
 525
 526                                 // Replace it, when there is an image in the body
 527                                 if (strstr($item["body"], '[/img]')) {
 528                                         $replace = true;
 529                                 }
 530
 531                                 // Replace it, when there is a link in the body
 532                                 if (strstr($item["body"], '[/url]')) {
 533                                         $replace = true;
 534                                 }
 535
 536                                 $saved_body = $item["body"];
 537                                 $saved_title = $item["title"];
 538
 539                                 if ($replace) {
 540                                         $item["body"] = trim($item["title"]);
 541                                 }
 542
 543                                 $data = ParseUrl::getSiteinfoCached($item['plink'], true);
 544                                 if (!empty($data['text']) && !empty($data['title']) && (mb_strlen($item['body']) < mb_strlen($data['text']))) {
 545                                         // When the fetched page info text is longer than the body, we do try to enhance the body
 546                                         if (!empty($item['body']) && (strpos($data['title'], $item['body']) === false) && (strpos($data['text'], $item['body']) === false)) {
 547                                                 // The body is not part of the fetched page info title or page info text. So we add the text to the body
 548                                                 $item['body'] .= "\n\n" . $data['text'];
 549                                         } else {
 550                                                 // Else we replace the body with the page info text
 551                                                 $item['body'] = $data['text'];
 552                                         }
 553                                 }
 554
 555                                 $data = PageInfo::queryUrl($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_denylist"] ?? '');
 556
 557                                 if (!empty($data)) {
 558                                         // Take the data that was provided by the feed if the query is empty
 559                                         if (($data['type'] == 'link') && empty($data['title']) && empty($data['text'])) {
 560                                                 $data['title'] = $saved_title;
 561                                                 $item["body"] = $saved_body;
 562                                         }
 563
 564                                         $data_text = strip_tags(trim($data['text'] ?? ''));
 565                                         $item_body = strip_tags(trim($item['body'] ?? ''));
 566
 567                                         if (!empty($data_text) && (($data_text == $item_body) || strstr($item_body, $data_text))) {
 568                                                 $data['text'] = '';
 569                                         }
 570
 571                                         // We always strip the title since it will be added in the page information
 572                                         $item["title"] = "";
 573                                         $item["body"] = $item["body"] . "\n" . PageInfo::getFooterFromData($data, false);
 574                                         $taglist = $contact["fetch_further_information"] == 2 ? PageInfo::getTagsFromUrl($item["plink"], $preview, $contact["ffi_keyword_denylist"] ?? '') : [];
 575                                         $item["object-type"] = Activity\ObjectType::BOOKMARK;
 576                                         $attachments = [];
 577                                 }
 578                         } else {
 579                                 if (!empty($summary)) {
 580                                         $item["body"] = '[abstract]' . HTML::toBBCode($summary, $basepath) . "[/abstract]\n" . $item["body"];
 581                                 }
 582
 583                                 if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] == 3)) {
 584                                         if (empty($taglist)) {
 585                                                 $taglist = PageInfo::getTagsFromUrl($item["plink"], $preview, $contact["ffi_keyword_denylist"] ?? '');
 586                                         }
 587                                         $item["body"] .= "\n" . self::tagToString($taglist);
 588                                 } else {
 589                                         $taglist = [];
 590                                 }
 591
 592                                 // Add the link to the original feed entry if not present in feed
 593                                 if (($item['plink'] != '') && !strstr($item["body"], $item['plink'])) {
 594                                         $item["body"] .= "[hr][url]" . $item['plink'] . "[/url]";
 595                                 }
 596                         }
 597
 598                         Logger::info('Stored feed', ['item' => $item]);
 599
 600                         $notify = Item::isRemoteSelf($contact, $item);
 601
 602                         // Distributed items should have a well formatted URI.
 603                         // Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
 604                         if ($notify) {
 605                                 $item['guid'] = Item::guidFromUri($orig_plink, DI::baseUrl()->getHostname());
 606                                 unset($item['uri']);
 607                                 unset($item['thr-parent']);
 608                                 unset($item['parent-uri']);
 609
 610                                 // Set the delivery priority for "remote self" to "medium"
 611                                 $notify = PRIORITY_MEDIUM;
 612                         }
 613
 614                         $postings[] = ['item' => $item, 'notify' => $notify,
 615                                 'taglist' => $taglist, 'attachments' => $attachments];
 616                 }
 617
 618                 if (!empty($postings)) {
 619                         $total = count($postings);
 620                         if ($total > 1) {
 621                                 // Posts shouldn't be delayed more than a day
 622                                 $interval = min(1440, self::getPollInterval($contact));
 623                                 $delay = round(($interval * 60) / $total);
 624                                 Logger::notice('Got posting delay', ['delay' => $delay, 'interval' => $interval, 'items' => $total, 'cid' => $contact['id'], 'url' => $contact['url']]);
 625                         } else {
 626                                 $delay = 0;
 627                         }
 628
 629                         $post_delay = 0;
 630
 631                         foreach ($postings as $posting) {
 632                                 if ($delay > 0) {
 633                                         $publish_at = DateTimeFormat::utc('now + ' . $post_delay . ' second');
 634                                         Logger::notice('Got publishing date', ['delay' => $delay, 'publish_at' => $publish_at, 'cid' => $contact['id'], 'url' => $contact['url']]);
 635                                         $post_delay += $delay;
 636                                 } else {
 637                                         $publish_at = DBA::NULL_DATETIME;
 638                                 }
 639
 640                                 Worker::add(['priority' => PRIORITY_HIGH, 'delayed' => $publish_at],
 641                                         'DelayedPublish', $posting['item'], $posting['notify'], $posting['taglist'], $posting['attachments']);
 642                         }
 643                 }
 644
 645                 if (!$dryRun && DI::config()->get('system', 'adjust_poll_frequency')) {
 646                         self::adjustPollFrequency($contact, $creation_dates);
 647                 }
 648
 649                 return ["header" => $author, "items" => $items];
 650         }
 651
 652         /**
 653          * Automatically adjust the poll frequency according to the post frequency
 654          *
 655          * @param array $contact
 656          * @param array $creation_dates
 657          * @return void
 658          */
 659         private static function adjustPollFrequency(array $contact, array $creation_dates)
 660         {
 661                 if ($contact['network'] != Protocol::FEED) {
 662                         Logger::info('Contact is no feed, skip.', ['id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url'], 'network' => $contact['network']]);
 663                         return;
 664                 }
 665
 666                 if (!empty($creation_dates)) {
 667                         // Count the post frequency and the earliest and latest post date
 668                         $frequency = [];
 669                         $oldest = time();
 670                         $newest = 0;
 671                         $oldest_date = $newest_date = '';
 672
 673                         foreach ($creation_dates as $date) {
 674                                 $timestamp = strtotime($date);
 675                                 $day = intdiv($timestamp, 86400);
 676                                 $hour = $timestamp % 86400;
 677
 678                                 // Only have a look at values from the last seven days
 679                                 if (((time() / 86400) - $day) < 7) {
 680                                         if (empty($frequency[$day])) {
 681                                                 $frequency[$day] = ['count' => 1, 'low' => $hour, 'high' => $hour];
 682                                         } else {
 683                                                 ++$frequency[$day]['count'];
 684                                                 if ($frequency[$day]['low'] > $hour) {
 685                                                         $frequency[$day]['low'] = $hour;
 686                                                 }
 687                                                 if ($frequency[$day]['high'] < $hour) {
 688                                                         $frequency[$day]['high'] = $hour;
 689                                                 }
 690                                         }
 691                                 }
 692                                 if ($oldest > $day) {
 693                                         $oldest = $day;
 694                                         $oldest_date = $date;
 695                                 }
 696
 697                                 if ($newest < $day) {
 698                                         $newest = $day;
 699                                         $newest_date = $date;
 700                                 }
 701                         }
 702
 703                         if (count($creation_dates) == 1) {
 704                                 Logger::info('Feed had posted a single time, switching to daily polling', ['newest' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 705                                 $priority = 8; // Poll once a day
 706                         }
 707
 708                         if (empty($priority) && (((time() / 86400) - $newest) > 730)) {
 709                                 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']]);
 710                                 $priority = 10; // Poll every month
 711                         }
 712
 713                         if (empty($priority) && (((time() / 86400) - $newest) > 365)) {
 714                                 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']]);
 715                                 $priority = 9; // Poll every week
 716                         }
 717
 718                         if (empty($priority) && empty($frequency)) {
 719                                 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']]);
 720                                 $priority = 8; // Poll once a day
 721                         }
 722
 723                         if (empty($priority)) {
 724                                 // Calculate the highest "posts per day" value
 725                                 $max = 0;
 726                                 foreach ($frequency as $entry) {
 727                                         if (($entry['count'] == 1) || ($entry['high'] == $entry['low'])) {
 728                                                 continue;
 729                                         }
 730
 731                                         // We take the earliest and latest post day and interpolate the number of post per day
 732                                         // that would had been created with this post frequency
 733
 734                                         // Assume at least four hours between oldest and newest post per day - should be okay for news outlets
 735                                         $duration = max($entry['high'] - $entry['low'], 14400);
 736                                         $ppd = (86400 / $duration) * $entry['count'];
 737                                         if ($ppd > $max) {
 738                                                 $max = $ppd;
 739                                         }
 740                                 }
 741                                 if ($max > 48) {
 742                                         $priority = 1; // Poll every quarter hour
 743                                 } elseif ($max > 24) {
 744                                         $priority = 2; // Poll half an hour
 745                                 } elseif ($max > 12) {
 746                                         $priority = 3; // Poll hourly
 747                                 } elseif ($max > 8) {
 748                                         $priority = 4; // Poll every two hours
 749                                 } elseif ($max > 4) {
 750                                         $priority = 5; // Poll every three hours
 751                                 } elseif ($max > 2) {
 752                                         $priority = 6; // Poll every six hours
 753                                 } else {
 754                                         $priority = 7; // Poll twice a day
 755                                 }
 756                                 Logger::info('Calculated priority by the posts per day', ['priority' => $priority, 'max' => round($max, 2), 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 757                         }
 758                 } else {
 759                         Logger::info('No posts, switching to daily polling', ['id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 760                         $priority = 8; // Poll once a day
 761                 }
 762
 763                 if ($contact['rating'] != $priority) {
 764                         Logger::notice('Adjusting priority', ['old' => $contact['rating'], 'new' => $priority, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 765                         DBA::update('contact', ['rating' => $priority], ['id' => $contact['id']]);
 766                 }
 767         }
 768
 769         /**
 770          * Get the poll interval for the given contact array
 771          *
 772          * @param array $contact
 773          * @return int Poll interval in minutes
 774          */
 775         public static function getPollInterval(array $contact)
 776         {
 777                 if (in_array($contact['network'], [Protocol::MAIL, Protocol::FEED])) {
 778                         $ratings = [0, 3, 7, 8, 9, 10];
 779                         if (DI::config()->get('system', 'adjust_poll_frequency') && ($contact['network'] == Protocol::FEED)) {
 780                                 $rating = $contact['rating'];
 781                         } elseif (array_key_exists($contact['priority'], $ratings)) {
 782                                 $rating = $ratings[$contact['priority']];
 783                         } else {
 784                                 $rating = -1;
 785                         }
 786                 } else {
 787                         // Check once a week per default for all other networks
 788                         $rating = 9;
 789                 }
 790
 791                 // Friendica and OStatus are checked once a day
 792                 if (in_array($contact['network'], [Protocol::DFRN, Protocol::OSTATUS])) {
 793                         $rating = 8;
 794                 }
 795
 796                 // Check archived contacts or contacts with unsupported protocols once a month
 797                 if ($contact['archive'] || in_array($contact['network'], [Protocol::ZOT, Protocol::PHANTOM])) {
 798                         $rating = 10;
 799                 }
 800
 801                 if ($rating < 0) {
 802                         return 0;
 803                 }
 804                 /*
 805                  * Based on $contact['priority'], should we poll this site now? Or later?
 806                  */
 807
 808                 $min_poll_interval = max(1, DI::config()->get('system', 'min_poll_interval'));
 809
 810                 $poll_intervals = [$min_poll_interval, 15, 30, 60, 120, 180, 360, 720 ,1440, 10080, 43200];
 811
 812                 //$poll_intervals = [$min_poll_interval . ' minute', '15 minute', '30 minute',
 813                 //      '1 hour', '2 hour', '3 hour', '6 hour', '12 hour' ,'1 day', '1 week', '1 month'];
 814
 815                 return $poll_intervals[$rating];
 816         }
 817
 818         /**
 819          * Convert a tag array to a tag string
 820          *
 821          * @param array $tags
 822          * @return string tag string
 823          */
 824         private static function tagToString(array $tags)
 825         {
 826                 $tagstr = '';
 827
 828                 foreach ($tags as $tag) {
 829                         if ($tagstr != "") {
 830                                 $tagstr .= ", ";
 831                         }
 832
 833                         $tagstr .= "#[url=" . DI::baseUrl() . "/search?tag=" . urlencode($tag) . "]" . $tag . "[/url]";
 834                 }
 835
 836                 return $tagstr;
 837         }
 838
 839         private static function titleIsBody($title, $body)
 840         {
 841                 $title = strip_tags($title);
 842                 $title = trim($title);
 843                 $title = html_entity_decode($title, ENT_QUOTES, 'UTF-8');
 844                 $title = str_replace(["\n", "\r", "\t", " "], ["", "", "", ""], $title);
 845
 846                 $body = strip_tags($body);
 847                 $body = trim($body);
 848                 $body = html_entity_decode($body, ENT_QUOTES, 'UTF-8');
 849                 $body = str_replace(["\n", "\r", "\t", " "], ["", "", "", ""], $body);
 850
 851                 if (strlen($title) < strlen($body)) {
 852                         $body = substr($body, 0, strlen($title));
 853                 }
 854
 855                 if (($title != $body) && (substr($title, -3) == "...")) {
 856                         $pos = strrpos($title, "...");
 857                         if ($pos > 0) {
 858                                 $title = substr($title, 0, $pos);
 859                                 $body = substr($body, 0, $pos);
 860                         }
 861                 }
 862                 return ($title == $body);
 863         }
 864
 865         /**
 866          * Creates the Atom feed for a given nickname
 867          *
 868          * Supported filters:
 869          * - activity (default): all the public posts
 870          * - posts: all the public top-level posts
 871          * - comments: all the public replies
 872          *
 873          * Updates the provided last_update parameter if the result comes from the
 874          * cache or it is empty
 875          *
 876          * @param string  $owner_nick  Nickname of the feed owner
 877          * @param string  $last_update Date of the last update
 878          * @param integer $max_items   Number of maximum items to fetch
 879          * @param string  $filter      Feed items filter (activity, posts or comments)
 880          * @param boolean $nocache     Wether to bypass caching
 881          *
 882          * @return string Atom feed
 883          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 884          * @throws \ImagickException
 885          */
 886         public static function atom($owner_nick, $last_update, $max_items = 300, $filter = 'activity', $nocache = false)
 887         {
 888                 $stamp = microtime(true);
 889
 890                 $owner = User::getOwnerDataByNick($owner_nick);
 891                 if (!$owner) {
 892                         return;
 893                 }
 894
 895                 $cachekey = "feed:feed:" . $owner_nick . ":" . $filter . ":" . $last_update;
 896
 897                 $previous_created = $last_update;
 898
 899                 // Don't cache when the last item was posted less then 15 minutes ago (Cache duration)
 900                 if ((time() - strtotime($owner['last-item'])) < 15*60) {
 901                         $result = DI::cache()->get($cachekey);
 902                         if (!$nocache && !is_null($result)) {
 903                                 Logger::info('Cached feed duration', ['seconds' => number_format(microtime(true) - $stamp, 3), 'nick' => $owner_nick, 'filter' => $filter, 'created' => $previous_created]);
 904                                 return $result['feed'];
 905                         }
 906                 }
 907
 908                 $check_date = empty($last_update) ? '' : DateTimeFormat::utc($last_update);
 909                 $authorid = Contact::getIdForURL($owner["url"]);
 910
 911                 $condition = ["`uid` = ? AND `received` > ? AND NOT `deleted` AND `gravity` IN (?, ?)
 912                         AND `private` != ? AND `visible` AND `wall` AND `parent-network` IN (?, ?, ?, ?)",
 913                         $owner["uid"], $check_date, GRAVITY_PARENT, GRAVITY_COMMENT,
 914                         Item::PRIVATE, Protocol::ACTIVITYPUB,
 915                         Protocol::OSTATUS, Protocol::DFRN, Protocol::DIASPORA];
 916
 917                 if ($filter === 'comments') {
 918                         $condition[0] .= " AND `object-type` = ? ";
 919                         $condition[] = Activity\ObjectType::COMMENT;
 920                 }
 921
 922                 if ($owner['account-type'] != User::ACCOUNT_TYPE_COMMUNITY) {
 923                         $condition[0] .= " AND `contact-id` = ? AND `author-id` = ?";
 924                         $condition[] = $owner["id"];
 925                         $condition[] = $authorid;
 926                 }
 927
 928                 $params = ['order' => ['received' => true], 'limit' => $max_items];
 929
 930                 if ($filter === 'posts') {
 931                         $ret = Item::selectThread([], $condition, $params);
 932                 } else {
 933                         $ret = Item::select([], $condition, $params);
 934                 }
 935
 936                 $items = Item::inArray($ret);
 937
 938                 $doc = new DOMDocument('1.0', 'utf-8');
 939                 $doc->formatOutput = true;
 940
 941                 $root = self::addHeader($doc, $owner, $filter);
 942
 943                 foreach ($items as $item) {
 944                         $entry = self::entry($doc, $item, $owner);
 945                         $root->appendChild($entry);
 946
 947                         if ($last_update < $item['created']) {
 948                                 $last_update = $item['created'];
 949                         }
 950                 }
 951
 952                 $feeddata = trim($doc->saveXML());
 953
 954                 $msg = ['feed' => $feeddata, 'last_update' => $last_update];
 955                 DI::cache()->set($cachekey, $msg, Duration::QUARTER_HOUR);
 956
 957                 Logger::info('Feed duration', ['seconds' => number_format(microtime(true) - $stamp, 3), 'nick' => $owner_nick, 'filter' => $filter, 'created' => $previous_created]);
 958
 959                 return $feeddata;
 960         }
 961
 962         /**
 963          * Adds the header elements to the XML document
 964          *
 965          * @param DOMDocument $doc       XML document
 966          * @param array       $owner     Contact data of the poster
 967          * @param string      $filter    The related feed filter (activity, posts or comments)
 968          *
 969          * @return object header root element
 970          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 971          */
 972         private static function addHeader(DOMDocument $doc, array $owner, $filter)
 973         {
 974                 $root = $doc->createElementNS(ActivityNamespace::ATOM1, 'feed');
 975                 $doc->appendChild($root);
 976
 977                 $title = '';
 978                 $selfUri = '/feed/' . $owner["nick"] . '/';
 979                 switch ($filter) {
 980                         case 'activity':
 981                                 $title = DI::l10n()->t('%s\'s timeline', $owner['name']);
 982                                 $selfUri .= $filter;
 983                                 break;
 984                         case 'posts':
 985                                 $title = DI::l10n()->t('%s\'s posts', $owner['name']);
 986                                 break;
 987                         case 'comments':
 988                                 $title = DI::l10n()->t('%s\'s comments', $owner['name']);
 989                                 $selfUri .= $filter;
 990                                 break;
 991                 }
 992
 993                 $attributes = ["uri" => "https://friendi.ca", "version" => FRIENDICA_VERSION . "-" . DB_UPDATE_VERSION];
 994                 XML::addElement($doc, $root, "generator", FRIENDICA_PLATFORM, $attributes);
 995                 XML::addElement($doc, $root, "id", DI::baseUrl() . "/profile/" . $owner["nick"]);
 996                 XML::addElement($doc, $root, "title", $title);
 997                 XML::addElement($doc, $root, "subtitle", sprintf("Updates from %s on %s", $owner["name"], DI::config()->get('config', 'sitename')));
 998                 XML::addElement($doc, $root, "logo", $owner["photo"]);
 999                 XML::addElement($doc, $root, "updated", DateTimeFormat::utcNow(DateTimeFormat::ATOM));
1000
1001                 $author = self::addAuthor($doc, $owner);
1002                 $root->appendChild($author);
1003
1004                 $attributes = ["href" => $owner["url"], "rel" => "alternate", "type" => "text/html"];
1005                 XML::addElement($doc, $root, "link", "", $attributes);
1006
1007                 OStatus::hublinks($doc, $root, $owner["nick"]);
1008
1009                 $attributes = ["href" => DI::baseUrl() . $selfUri, "rel" => "self", "type" => "application/atom+xml"];
1010                 XML::addElement($doc, $root, "link", "", $attributes);
1011
1012                 return $root;
1013         }
1014
1015         /**
1016          * Adds the author element to the XML document
1017          *
1018          * @param DOMDocument $doc          XML document
1019          * @param array       $owner        Contact data of the poster
1020          *
1021          * @return \DOMElement author element
1022          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1023          */
1024         private static function addAuthor(DOMDocument $doc, array $owner)
1025         {
1026                 $author = $doc->createElement("author");
1027                 XML::addElement($doc, $author, "uri", $owner["url"]);
1028                 XML::addElement($doc, $author, "name", $owner["nick"]);
1029                 XML::addElement($doc, $author, "email", $owner["addr"]);
1030
1031                 return $author;
1032         }
1033
1034         /**
1035          * Adds an entry element to the XML document
1036          *
1037          * @param DOMDocument $doc       XML document
1038          * @param array       $item      Data of the item that is to be posted
1039          * @param array       $owner     Contact data of the poster
1040          * @param bool        $toplevel  optional default false
1041          *
1042          * @return \DOMElement Entry element
1043          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1044          * @throws \ImagickException
1045          */
1046         private static function entry(DOMDocument $doc, array $item, array $owner)
1047         {
1048                 $xml = null;
1049
1050                 $repeated_guid = OStatus::getResharedGuid($item);
1051                 if ($repeated_guid != "") {
1052                         $xml = self::reshareEntry($doc, $item, $owner, $repeated_guid);
1053                 }
1054
1055                 if ($xml) {
1056                         return $xml;
1057                 }
1058
1059                 return self::noteEntry($doc, $item, $owner);
1060         }
1061
1062                 /**
1063          * Adds an entry element with reshared content
1064          *
1065          * @param DOMDocument $doc           XML document
1066          * @param array       $item          Data of the item that is to be posted
1067          * @param array       $owner         Contact data of the poster
1068          * @param string      $repeated_guid guid
1069          * @param bool        $toplevel      Is it for en entry element (false) or a feed entry (true)?
1070          *
1071          * @return bool Entry element
1072          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1073          * @throws \ImagickException
1074          */
1075         private static function reshareEntry(DOMDocument $doc, array $item, array $owner, $repeated_guid)
1076         {
1077                 if (($item['gravity'] != GRAVITY_PARENT) && (Strings::normaliseLink($item["author-link"]) != Strings::normaliseLink($owner["url"]))) {
1078                         Logger::info('Feed entry author does not match feed owner', ['owner' => $owner["url"], 'author' => $item["author-link"]]);
1079                 }
1080
1081                 $entry = OStatus::entryHeader($doc, $owner, $item, false);
1082
1083                 $condition = ['uid' => $owner["uid"], 'guid' => $repeated_guid, 'private' => [Item::PUBLIC, Item::UNLISTED],
1084                         'network' => Protocol::FEDERATED];
1085                 $repeated_item = Item::selectFirst([], $condition);
1086                 if (!DBA::isResult($repeated_item)) {
1087                         return false;
1088                 }
1089
1090                 self::entryContent($doc, $entry, $item, self::getTitle($repeated_item), Activity::SHARE, false);
1091
1092                 self::entryFooter($doc, $entry, $item, $owner);
1093
1094                 return $entry;
1095         }
1096
1097         /**
1098          * Adds a regular entry element
1099          *
1100          * @param DOMDocument $doc       XML document
1101          * @param array       $item      Data of the item that is to be posted
1102          * @param array       $owner     Contact data of the poster
1103          * @param bool        $toplevel  Is it for en entry element (false) or a feed entry (true)?
1104          *
1105          * @return \DOMElement Entry element
1106          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1107          * @throws \ImagickException
1108          */
1109         private static function noteEntry(DOMDocument $doc, array $item, array $owner)
1110         {
1111                 if (($item['gravity'] != GRAVITY_PARENT) && (Strings::normaliseLink($item["author-link"]) != Strings::normaliseLink($owner["url"]))) {
1112                         Logger::info('Feed entry author does not match feed owner', ['owner' => $owner["url"], 'author' => $item["author-link"]]);
1113                 }
1114
1115                 $entry = OStatus::entryHeader($doc, $owner, $item, false);
1116
1117                 self::entryContent($doc, $entry, $item, self::getTitle($item), '', true);
1118
1119                 self::entryFooter($doc, $entry, $item, $owner);
1120
1121                 return $entry;
1122         }
1123
1124         /**
1125          * Adds elements to the XML document
1126          *
1127          * @param DOMDocument $doc       XML document
1128          * @param \DOMElement $entry     Entry element where the content is added
1129          * @param array       $item      Data of the item that is to be posted
1130          * @param array       $owner     Contact data of the poster
1131          * @param string      $title     Title for the post
1132          * @param string      $verb      The activity verb
1133          * @param bool        $complete  Add the "status_net" element?
1134          * @param bool        $feed_mode Behave like a regular feed for users if true
1135          * @return void
1136          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1137          */
1138         private static function entryContent(DOMDocument $doc, \DOMElement $entry, array $item, $title, $verb = "", $complete = true)
1139         {
1140                 if ($verb == "") {
1141                         $verb = OStatus::constructVerb($item);
1142                 }
1143
1144                 XML::addElement($doc, $entry, "id", $item["uri"]);
1145                 XML::addElement($doc, $entry, "title", html_entity_decode($title, ENT_QUOTES, 'UTF-8'));
1146
1147                 $body = OStatus::formatPicturePost($item['body']);
1148
1149                 $body = BBCode::convert($body, false, BBCode::OSTATUS);
1150
1151                 XML::addElement($doc, $entry, "content", $body, ["type" => "html"]);
1152
1153                 XML::addElement($doc, $entry, "link", "", ["rel" => "alternate", "type" => "text/html",
1154                                                                 "href" => DI::baseUrl()."/display/".$item["guid"]]
1155                 );
1156
1157                 XML::addElement($doc, $entry, "published", DateTimeFormat::utc($item["created"]."+00:00", DateTimeFormat::ATOM));
1158                 XML::addElement($doc, $entry, "updated", DateTimeFormat::utc($item["edited"]."+00:00", DateTimeFormat::ATOM));
1159         }
1160
1161         /**
1162          * Adds the elements at the foot of an entry to the XML document
1163          *
1164          * @param DOMDocument $doc       XML document
1165          * @param object      $entry     The entry element where the elements are added
1166          * @param array       $item      Data of the item that is to be posted
1167          * @param array       $owner     Contact data of the poster
1168          * @param bool        $complete  default true
1169          * @return void
1170          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1171          */
1172         private static function entryFooter(DOMDocument $doc, $entry, array $item, array $owner)
1173         {
1174                 $mentioned = [];
1175
1176                 if ($item['gravity'] != GRAVITY_PARENT) {
1177                         $parent = Item::selectFirst(['guid', 'author-link', 'owner-link'], ['id' => $item['parent']]);
1178
1179                         $thrparent = Item::selectFirst(['guid', 'author-link', 'owner-link', 'plink'], ['uid' => $owner["uid"], 'uri' => $item['thr-parent']]);
1180
1181                         if (DBA::isResult($thrparent)) {
1182                                 $mentioned[$thrparent["author-link"]] = $thrparent["author-link"];
1183                                 $mentioned[$thrparent["owner-link"]] = $thrparent["owner-link"];
1184                                 $parent_plink = $thrparent["plink"];
1185                         } else {
1186                                 $mentioned[$parent["author-link"]] = $parent["author-link"];
1187                                 $mentioned[$parent["owner-link"]] = $parent["owner-link"];
1188                                 $parent_plink = DI::baseUrl()."/display/".$parent["guid"];
1189                         }
1190
1191                         $attributes = [
1192                                         "ref" => $item['thr-parent'],
1193                                         "href" => $parent_plink];
1194                         XML::addElement($doc, $entry, "thr:in-reply-to", "", $attributes);
1195
1196                         $attributes = [
1197                                         "rel" => "related",
1198                                         "href" => $parent_plink];
1199                         XML::addElement($doc, $entry, "link", "", $attributes);
1200                 }
1201
1202                 // uri-id isn't present for follow entry pseudo-items
1203                 $tags = Tag::getByURIId($item['uri-id'] ?? 0);
1204                 foreach ($tags as $tag) {
1205                         $mentioned[$tag['url']] = $tag['url'];
1206                 }
1207
1208                 foreach ($tags as $tag) {
1209                         if ($tag['type'] == Tag::HASHTAG) {
1210                                 XML::addElement($doc, $entry, "category", "", ["term" => $tag['name']]);
1211                         }
1212                 }
1213
1214                 OStatus::getAttachment($doc, $entry, $item);
1215         }
1216
1217         /**
1218          * Fetch or create title for feed entry
1219          *
1220          * @param array $item
1221          * @return string title
1222          */
1223         private static function getTitle(array $item)
1224         {
1225                 if ($item['title'] != '') {
1226                         return BBCode::convert($item['title'], false, BBCode::OSTATUS);
1227                 }
1228
1229                 // Fetch information about the post
1230                 $siteinfo = BBCode::getAttachedData($item["body"]);
1231                 if (isset($siteinfo["title"])) {
1232                         return $siteinfo["title"];
1233                 }
1234
1235                 // If no bookmark is found then take the first line
1236                 // Remove the share element before fetching the first line
1237                 $title = trim(preg_replace("/\[share.*?\](.*?)\[\/share\]/ism","\n$1\n",$item['body']));
1238
1239                 $title = HTML::toPlaintext(BBCode::convert($title, false), 0, true)."\n";
1240                 $pos = strpos($title, "\n");
1241                 $trailer = "";
1242                 if (($pos == 0) || ($pos > 100)) {
1243                         $pos = 100;
1244                         $trailer = "...";
1245                 }
1246
1247                 return substr($title, 0, $pos) . $trailer;
1248         }
1249 }