src/Protocol/Feed.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2023, the Friendica project
   4  *
   5  * @license GNU AGPL version 3 or any later version
   6  *
   7  * This program is free software: you can redistribute it and/or modify
   8  * it under the terms of the GNU Affero General Public License as
   9  * published by the Free Software Foundation, either version 3 of the
  10  * License, or (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU Affero General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Affero General Public License
  18  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  19  *
  20  */
  21
  22 namespace Friendica\Protocol;
  23
  24 use DOMDocument;
  25 use DOMElement;
  26 use DOMXPath;
  27 use Friendica\App;
  28 use Friendica\Content\PageInfo;
  29 use Friendica\Content\Text\BBCode;
  30 use Friendica\Content\Text\HTML;
  31 use Friendica\Core\Cache\Enum\Duration;
  32 use Friendica\Core\Logger;
  33 use Friendica\Core\Protocol;
  34 use Friendica\Core\Worker;
  35 use Friendica\Database\DBA;
  36 use Friendica\DI;
  37 use Friendica\Model\Contact;
  38 use Friendica\Model\Conversation;
  39 use Friendica\Model\Item;
  40 use Friendica\Model\Post;
  41 use Friendica\Model\Tag;
  42 use Friendica\Model\User;
  43 use Friendica\Network\HTTPException;
  44 use Friendica\Util\DateTimeFormat;
  45 use Friendica\Util\Network;
  46 use Friendica\Util\ParseUrl;
  47 use Friendica\Util\Proxy;
  48 use Friendica\Util\Strings;
  49 use Friendica\Util\XML;
  50 use GuzzleHttp\Exception\TransferException;
  51
  52 /**
  53  * This class contain functions to import feeds (RSS/RDF/Atom)
  54  */
  55 class Feed
  56 {
  57         /**
  58          * Read a RSS/RDF/Atom feed and create an item entry for it
  59          *
  60          * @param string $xml      The feed data
  61          * @param array  $importer The user record of the importer
  62          * @param array  $contact  The contact record of the feed
  63          *
  64          * @return array Returns the header and the first item in dry run mode
  65          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  66          */
  67         public static function import(string $xml, array $importer = [], array $contact = []): array
  68         {
  69                 $dryRun = empty($importer) && empty($contact);
  70
  71                 if ($dryRun) {
  72                         Logger::info("Test Atom/RSS feed");
  73                 } else {
  74                         Logger::info('Import Atom/RSS feed "' . $contact['name'] . '" (Contact ' . $contact['id'] . ') for user ' . $importer['uid']);
  75                 }
  76
  77                 $xml = trim($xml);
  78
  79                 if (empty($xml)) {
  80                         Logger::info('XML is empty.');
  81                         return [];
  82                 }
  83
  84                 if (!empty($contact['poll'])) {
  85                         $basepath = $contact['poll'];
  86                 } elseif (!empty($contact['url'])) {
  87                         $basepath = $contact['url'];
  88                 } else {
  89                         $basepath = '';
  90                 }
  91
  92                 $doc = new DOMDocument();
  93                 @$doc->loadXML($xml);
  94                 $xpath = new DOMXPath($doc);
  95
  96                 $xpath->registerNamespace('atom', ActivityNamespace::ATOM1);
  97                 $xpath->registerNamespace('atom03', ActivityNamespace::ATOM03);
  98                 $xpath->registerNamespace('dc', 'http://purl.org/dc/elements/1.1/');
  99                 $xpath->registerNamespace('content', 'http://purl.org/rss/1.0/modules/content/');
 100                 $xpath->registerNamespace('rdf', 'http://www.w3.org/1999/02/22-rdf-syntax-ns#');
 101                 $xpath->registerNamespace('rss', 'http://purl.org/rss/1.0/');
 102                 $xpath->registerNamespace('media', 'http://search.yahoo.com/mrss/');
 103                 $xpath->registerNamespace('poco', ActivityNamespace::POCO);
 104
 105                 $author = [];
 106                 $atomns = 'atom';
 107                 $entries = null;
 108                 $protocol = Conversation::PARCEL_UNKNOWN;
 109
 110                 // Is it RDF?
 111                 if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
 112                         $protocol = Conversation::PARCEL_RDF;
 113                         $author['author-link'] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:link/text()');
 114                         $author['author-name'] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:title/text()');
 115
 116                         if (empty($author['author-name'])) {
 117                                 $author['author-name'] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:description/text()');
 118                         }
 119                         $entries = $xpath->query('/rdf:RDF/rss:item');
 120                 }
 121
 122                 if ($xpath->query('/opml')->length > 0) {
 123                         $protocol = Conversation::PARCEL_OPML;
 124                         $author['author-name'] = XML::getFirstNodeValue($xpath, '/opml/head/title/text()');
 125                         $entries = $xpath->query('/opml/body/outline');
 126                 }
 127
 128                 // Is it Atom?
 129                 if ($xpath->query('/atom:feed')->length > 0) {
 130                         $protocol = Conversation::PARCEL_ATOM;
 131                 } elseif ($xpath->query('/atom03:feed')->length > 0) {
 132                         $protocol = Conversation::PARCEL_ATOM03;
 133                         $atomns = 'atom03';
 134                 }
 135
 136                 if (in_array($protocol, [Conversation::PARCEL_ATOM, Conversation::PARCEL_ATOM03])) {
 137                         $alternate = XML::getFirstAttributes($xpath, $atomns . ":link[@rel='alternate']");
 138                         if (is_object($alternate)) {
 139                                 foreach ($alternate as $attribute) {
 140                                         if ($attribute->name == 'href') {
 141                                                 $author['author-link'] = $attribute->textContent;
 142                                         }
 143                                 }
 144                         }
 145
 146                         if (empty($author['author-link'])) {
 147                                 $self = XML::getFirstAttributes($xpath, $atomns . ":link[@rel='self']");
 148                                 if (is_object($self)) {
 149                                         foreach ($self as $attribute) {
 150                                                 if ($attribute->name == 'href') {
 151                                                         $author['author-link'] = $attribute->textContent;
 152                                                 }
 153                                         }
 154                                 }
 155                         }
 156
 157                         if (empty($author['author-link'])) {
 158                                 $author['author-link'] = XML::getFirstNodeValue($xpath, '/' . $atomns . ':feed/' . $atomns . ':id/text()');
 159                         }
 160                         $author['author-avatar'] = XML::getFirstNodeValue($xpath, '/' . $atomns . ':feed/' . $atomns . ':logo/text()');
 161
 162                         $author['author-name'] = XML::getFirstNodeValue($xpath, '/' . $atomns . ':feed/' . $atomns . ':title/text()');
 163
 164                         if (empty($author['author-name'])) {
 165                                 $author['author-name'] = XML::getFirstNodeValue($xpath, '/' . $atomns . ':feed/' . $atomns . ':subtitle/text()');
 166                         }
 167
 168                         if (empty($author['author-name'])) {
 169                                 $author['author-name'] = XML::getFirstNodeValue($xpath, '/' . $atomns . ':feed/' . $atomns . ':author/' . $atomns . ':name/text()');
 170                         }
 171
 172                         $value = XML::getFirstNodeValue($xpath, '' . $atomns . ':author/poco:displayName/text()');
 173                         if ($value != '') {
 174                                 $author['author-name'] = $value;
 175                         }
 176
 177                         if ($dryRun) {
 178                                 $author['author-id'] = XML::getFirstNodeValue($xpath, '/' . $atomns . ':feed/' . $atomns . ':author/' . $atomns . ':id/text()');
 179
 180                                 // See https://tools.ietf.org/html/rfc4287#section-3.2.2
 181                                 $value = XML::getFirstNodeValue($xpath, $atomns . ':author/' . $atomns . ':uri/text()');
 182                                 if ($value != '') {
 183                                         $author['author-link'] = $value;
 184                                 }
 185
 186                                 $value = XML::getFirstNodeValue($xpath, $atomns . ':author/poco:preferredUsername/text()');
 187                                 if ($value != '') {
 188                                         $author['author-nick'] = $value;
 189                                 }
 190
 191                                 $value = XML::getFirstNodeValue($xpath, $atomns . ':author/poco:address/poco:formatted/text()');
 192                                 if ($value != '') {
 193                                         $author['author-location'] = $value;
 194                                 }
 195
 196                                 $value = XML::getFirstNodeValue($xpath, $atomns . ':author/poco:note/text()');
 197                                 if ($value != '') {
 198                                         $author['author-about'] = $value;
 199                                 }
 200
 201                                 $avatar = XML::getFirstAttributes($xpath, $atomns . ":author/' . $atomns . ':link[@rel='avatar']");
 202                                 if (is_object($avatar)) {
 203                                         foreach ($avatar as $attribute) {
 204                                                 if ($attribute->name == 'href') {
 205                                                         $author['author-avatar'] = $attribute->textContent;
 206                                                 }
 207                                         }
 208                                 }
 209                         }
 210
 211                         $author['edited'] = $author['created'] = XML::getFirstNodeValue($xpath, '/' . $atomns . ':feed/' . $atomns . ':updated/text()');
 212
 213                         $author['app'] = XML::getFirstNodeValue($xpath, '/' . $atomns . ':feed/' . $atomns . ':generator/text()');
 214
 215                         $entries = $xpath->query('/' . $atomns . ':feed/' . $atomns . ':entry');
 216                 }
 217
 218                 // Is it RSS?
 219                 if ($xpath->query('/rss/channel')->length > 0) {
 220                         $protocol = Conversation::PARCEL_RSS;
 221                         $author['author-link'] = XML::getFirstNodeValue($xpath, '/rss/channel/link/text()');
 222
 223                         $author['author-name'] = XML::getFirstNodeValue($xpath, '/rss/channel/title/text()');
 224
 225                         if (empty($author['author-name'])) {
 226                                 $author['author-name'] = XML::getFirstNodeValue($xpath, '/rss/channel/copyright/text()');
 227                         }
 228
 229                         if (empty($author['author-name'])) {
 230                                 $author['author-name'] = XML::getFirstNodeValue($xpath, '/rss/channel/description/text()');
 231                         }
 232
 233                         $author['author-avatar'] = XML::getFirstNodeValue($xpath, '/rss/channel/image/url/text()');
 234
 235                         if (empty($author['author-avatar'])) {
 236                                 $avatar = XML::getFirstAttributes($xpath, '/rss/channel/itunes:image');
 237                                 if (is_object($avatar)) {
 238                                         foreach ($avatar as $attribute) {
 239                                                 if ($attribute->name == 'href') {
 240                                                         $author['author-avatar'] = $attribute->textContent;
 241                                                 }
 242                                         }
 243                                 }
 244                         }
 245
 246                         $author['author-about'] = HTML::toBBCode(XML::getFirstNodeValue($xpath, '/rss/channel/description/text()'), $basepath);
 247
 248                         if (empty($author['author-about'])) {
 249                                 $author['author-about'] = XML::getFirstNodeValue($xpath, '/rss/channel/itunes:summary/text()');
 250                         }
 251
 252                         $author['edited'] = $author['created'] = XML::getFirstNodeValue($xpath, '/rss/channel/pubDate/text()');
 253
 254                         $author['app'] = XML::getFirstNodeValue($xpath, '/rss/channel/generator/text()');
 255
 256                         $entries = $xpath->query('/rss/channel/item');
 257                 }
 258
 259                 if (!$dryRun) {
 260                         $author['author-link'] = $contact['url'];
 261
 262                         if (empty($author['author-name'])) {
 263                                 $author['author-name'] = $contact['name'];
 264                         }
 265
 266                         $author['author-avatar'] = $contact['thumb'];
 267
 268                         $author['owner-link'] = $contact['url'];
 269                         $author['owner-name'] = $contact['name'];
 270                         $author['owner-avatar'] = $contact['thumb'];
 271                 }
 272
 273                 $header = [
 274                         'uid'         => $importer['uid'] ?? 0,
 275                         'network'     => Protocol::FEED,
 276                         'wall'        => 0,
 277                         'origin'      => 0,
 278                         'gravity'     => Item::GRAVITY_PARENT,
 279                         'private'     => Item::PUBLIC,
 280                         'verb'        => Activity::POST,
 281                         'object-type' => Activity\ObjectType::NOTE,
 282                         'post-type'   => Item::PT_ARTICLE,
 283                         'contact-id'  => $contact['id'] ?? 0,
 284                 ];
 285
 286                 $datarray['protocol'] = $protocol;
 287                 $datarray['direction'] = Conversation::PULL;
 288
 289                 if (!is_object($entries)) {
 290                         Logger::info("There are no entries in this feed.");
 291                         return [];
 292                 }
 293
 294                 $items = [];
 295                 $creation_dates = [];
 296
 297                 // Limit the number of items that are about to be fetched
 298                 $total_items = ($entries->length - 1);
 299                 $max_items = DI::config()->get('system', 'max_feed_items');
 300                 if (($max_items > 0) && ($total_items > $max_items)) {
 301                         $total_items = $max_items;
 302                 }
 303
 304                 $postings = [];
 305
 306                 // Importing older entries first
 307                 for ($i = $total_items; $i >= 0; --$i) {
 308                         $entry = $entries->item($i);
 309
 310                         $item = array_merge($header, $author);
 311
 312                         $alternate = XML::getFirstAttributes($xpath, $atomns . ":link[@rel='alternate']", $entry);
 313                         if (!is_object($alternate)) {
 314                                 $alternate = XML::getFirstAttributes($xpath, $atomns . ':link', $entry);
 315                         }
 316                         if (is_object($alternate)) {
 317                                 foreach ($alternate as $attribute) {
 318                                         if ($attribute->name == 'href') {
 319                                                 $item['plink'] = $attribute->textContent;
 320                                         }
 321                                 }
 322                         }
 323
 324                         if ($entry->nodeName == 'outline') {
 325                                 $isrss = false;
 326                                 $plink = '';
 327                                 $uri = '';
 328                                 foreach ($entry->attributes as $attribute) {
 329                                         switch ($attribute->nodeName) {
 330                                                 case 'title':
 331                                                         $item['title'] = $attribute->nodeValue;
 332                                                         break;
 333
 334                                                 case 'text':
 335                                                         $body = $attribute->nodeValue;
 336                                                         break;
 337
 338                                                 case 'htmlUrl':
 339                                                         $plink = $attribute->nodeValue;
 340                                                         break;
 341
 342                                                 case 'xmlUrl':
 343                                                         $uri = $attribute->nodeValue;
 344                                                         break;
 345
 346                                                 case 'type':
 347                                                         $isrss = $attribute->nodeValue == 'rss';
 348                                                         break;
 349                                         }
 350                                 }
 351                                 $item['plink'] = $plink ?: $uri;
 352                                 $item['uri'] = $uri ?: $plink;
 353                                 if (!$isrss || empty($item['uri'])) {
 354                                         continue;
 355                                 }
 356                         }
 357
 358                         if (empty($item['plink'])) {
 359                                 $item['plink'] = XML::getFirstNodeValue($xpath, 'link/text()', $entry);
 360                         }
 361
 362                         if (empty($item['plink'])) {
 363                                 $item['plink'] = XML::getFirstNodeValue($xpath, 'rss:link/text()', $entry);
 364                         }
 365
 366                         // Add the base path if missing
 367                         $item['plink'] = Network::addBasePath($item['plink'], $basepath);
 368
 369                         if (empty($item['uri'])) {
 370                                 $item['uri'] = XML::getFirstNodeValue($xpath, $atomns . ':id/text()', $entry);
 371                         }
 372
 373                         $guid = XML::getFirstNodeValue($xpath, 'guid/text()', $entry);
 374                         if (!empty($guid)) {
 375                                 $item['uri'] = $guid;
 376
 377                                 // Don't use the GUID value directly but instead use it as a basis for the GUID
 378                                 $item['guid'] = Item::guidFromUri($guid, parse_url($guid, PHP_URL_HOST) ?? parse_url($item['plink'], PHP_URL_HOST));
 379                         }
 380
 381                         if (empty($item['uri'])) {
 382                                 $item['uri'] = $item['plink'];
 383                         }
 384
 385                         $orig_plink = $item['plink'];
 386
 387                         try {
 388                                 $item['plink'] = DI::httpClient()->finalUrl($item['plink']);
 389                         } catch (TransferException $exception) {
 390                                 Logger::notice('Item URL couldn\'t get expanded', ['url' => $item['plink'], 'exception' => $exception]);
 391                         }
 392
 393                         if (empty($item['title'])) {
 394                                 $item['title'] = XML::getFirstNodeValue($xpath, $atomns . ':title/text()', $entry);
 395                         }
 396
 397                         if (empty($item['title'])) {
 398                                 $item['title'] = XML::getFirstNodeValue($xpath, 'title/text()', $entry);
 399                         }
 400
 401                         if (empty($item['title'])) {
 402                                 $item['title'] = XML::getFirstNodeValue($xpath, 'rss:title/text()', $entry);
 403                         }
 404
 405                         if (empty($item['title'])) {
 406                                 $item['title'] = XML::getFirstNodeValue($xpath, 'itunes:title/text()', $entry);
 407                         }
 408
 409                         $item['title'] = html_entity_decode($item['title'], ENT_QUOTES, 'UTF-8');
 410
 411                         $published = XML::getFirstNodeValue($xpath, $atomns . ':published/text()', $entry);
 412
 413                         if (empty($published)) {
 414                                 $published = XML::getFirstNodeValue($xpath, 'pubDate/text()', $entry);
 415                         }
 416
 417                         if (empty($published)) {
 418                                 $published = XML::getFirstNodeValue($xpath, 'dc:date/text()', $entry);
 419                         }
 420
 421                         $updated = XML::getFirstNodeValue($xpath, $atomns . ':updated/text()', $entry);
 422
 423                         if (empty($updated) && !empty($published)) {
 424                                 $updated = $published;
 425                         }
 426
 427                         if (empty($published) && !empty($updated)) {
 428                                 $published = $updated;
 429                         }
 430
 431                         if ($published != '') {
 432                                 $item['created'] = trim($published);
 433                         }
 434
 435                         if ($updated != '') {
 436                                 $item['edited'] = trim($updated);
 437                         }
 438
 439                         if (!$dryRun) {
 440                                 $condition = [
 441                                         "`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
 442                                         $importer['uid'], $item['uri'], Protocol::FEED, Protocol::DFRN
 443                                 ];
 444                                 $previous = Post::selectFirst(['id', 'created'], $condition);
 445                                 if (DBA::isResult($previous)) {
 446                                         // Use the creation date when the post had been stored. It can happen this date changes in the feed.
 447                                         $creation_dates[] = $previous['created'];
 448                                         Logger::info('Item with URI ' . $item['uri'] . ' for user ' . $importer['uid'] . ' already existed under id ' . $previous['id']);
 449                                         continue;
 450                                 }
 451                                 $creation_dates[] = DateTimeFormat::utc($item['created']);
 452                         }
 453
 454                         $creator = XML::getFirstNodeValue($xpath, 'author/text()', $entry);
 455
 456                         if (empty($creator)) {
 457                                 $creator = XML::getFirstNodeValue($xpath, $atomns . ':author/' . $atomns . ':name/text()', $entry);
 458                         }
 459
 460                         if (empty($creator)) {
 461                                 $creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
 462                         }
 463
 464                         if ($creator != '') {
 465                                 $item['author-name'] = $creator;
 466                         }
 467
 468                         $creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
 469
 470                         if ($creator != '') {
 471                                 $item['author-name'] = $creator;
 472                         }
 473
 474                         /// @TODO ?
 475                         // <category>Ausland</category>
 476                         // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 477
 478                         $attachments = [];
 479
 480                         $enclosures = $xpath->query("enclosure|' . $atomns . ':link[@rel='enclosure']", $entry);
 481                         if (!empty($enclosures)) {
 482                                 foreach ($enclosures as $enclosure) {
 483                                         $href = '';
 484                                         $length = null;
 485                                         $type = null;
 486
 487                                         foreach ($enclosure->attributes as $attribute) {
 488                                                 if (in_array($attribute->name, ['url', 'href'])) {
 489                                                         $href = $attribute->textContent;
 490                                                 } elseif ($attribute->name == 'length') {
 491                                                         $length = (int)$attribute->textContent;
 492                                                 } elseif ($attribute->name == 'type') {
 493                                                         $type = $attribute->textContent;
 494                                                 }
 495                                         }
 496
 497                                         if (!empty($href)) {
 498                                                 $attachment = ['uri-id' => -1, 'type' => Post\Media::UNKNOWN, 'url' => $href, 'mimetype' => $type, 'size' => $length];
 499
 500                                                 $attachment = Post\Media::fetchAdditionalData($attachment);
 501
 502                                                 // By now we separate the visible media types (audio, video, image) from the rest
 503                                                 // In the future we should try to avoid the DOCUMENT type and only use the real one - but not in the RC phase.
 504                                                 if (!in_array($attachment['type'], [Post\Media::AUDIO, Post\Media::IMAGE, Post\Media::VIDEO])) {
 505                                                         $attachment['type'] = Post\Media::DOCUMENT;
 506                                                 }
 507                                                 $attachments[] = $attachment;
 508                                         }
 509                                 }
 510                         }
 511
 512                         $taglist = [];
 513                         $categories = $xpath->query('category', $entry);
 514                         foreach ($categories as $category) {
 515                                 $taglist[] = $category->nodeValue;
 516                         }
 517
 518                         if (empty($body)) {
 519                                 $body = trim(XML::getFirstNodeValue($xpath, $atomns . ':content/text()', $entry));
 520                         }
 521
 522                         if (empty($body)) {
 523                                 $body = trim(XML::getFirstNodeValue($xpath, 'content:encoded/text()', $entry));
 524                         }
 525
 526                         $summary = trim(XML::getFirstNodeValue($xpath, $atomns . ':summary/text()', $entry));
 527
 528                         if (empty($summary)) {
 529                                 $summary = trim(XML::getFirstNodeValue($xpath, 'description/text()', $entry));
 530                         }
 531
 532                         if (empty($body)) {
 533                                 $body = $summary;
 534                                 $summary = '';
 535                         }
 536
 537                         if ($body == $summary) {
 538                                 $summary = '';
 539                         }
 540
 541                         // remove the content of the title if it is identically to the body
 542                         // This helps with auto generated titles e.g. from tumblr
 543                         if (self::titleIsBody($item['title'], $body)) {
 544                                 $item['title'] = '';
 545                         }
 546                         $item['body'] = HTML::toBBCode($body, $basepath);
 547
 548                         // Remove tracking pixels
 549                         $item['body'] = preg_replace("/\[img=1x1\]([^\[\]]*)\[\/img\]/Usi", '', $item['body']);
 550
 551                         if (($item['body'] == '') && ($item['title'] != '')) {
 552                                 $item['body'] = $item['title'];
 553                                 $item['title'] = '';
 554                         }
 555
 556                         if ($dryRun) {
 557                                 $item['attachments'] = $attachments;
 558                                 $items[] = $item;
 559                                 break;
 560                         } elseif (!Item::isValid($item)) {
 561                                 Logger::info('Feed item is invalid', ['created' => $item['created'], 'uid' => $item['uid'], 'uri' => $item['uri']]);
 562                                 continue;
 563                         } elseif (Item::isTooOld($item)) {
 564                                 Logger::info('Feed is too old', ['created' => $item['created'], 'uid' => $item['uid'], 'uri' => $item['uri']]);
 565                                 continue;
 566                         }
 567
 568                         $preview = '';
 569                         if (!empty($contact['fetch_further_information']) && ($contact['fetch_further_information'] < 3)) {
 570                                 // Handle enclosures and treat them as preview picture
 571                                 foreach ($attachments as $attachment) {
 572                                         if ($attachment['mimetype'] == 'image/jpeg') {
 573                                                 $preview = $attachment['url'];
 574                                         }
 575                                 }
 576
 577                                 // Remove a possible link to the item itself
 578                                 $item['body'] = str_replace($item['plink'], '', $item['body']);
 579                                 $item['body'] = trim(preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item['body']));
 580
 581                                 // Replace the content when the title is longer than the body
 582                                 $replace = (strlen($item['title']) > strlen($item['body']));
 583
 584                                 // Replace it, when there is an image in the body
 585                                 if (strstr($item['body'], '[/img]')) {
 586                                         $replace = true;
 587                                 }
 588
 589                                 // Replace it, when there is a link in the body
 590                                 if (strstr($item['body'], '[/url]')) {
 591                                         $replace = true;
 592                                 }
 593
 594                                 $saved_body = $item['body'];
 595                                 $saved_title = $item['title'];
 596
 597                                 if ($replace) {
 598                                         $item['body'] = trim($item['title']);
 599                                 }
 600
 601                                 $data = ParseUrl::getSiteinfoCached($item['plink']);
 602                                 if (!empty($data['text']) && !empty($data['title']) && (mb_strlen($item['body']) < mb_strlen($data['text']))) {
 603                                         // When the fetched page info text is longer than the body, we do try to enhance the body
 604                                         if (!empty($item['body']) && (strpos($data['title'], $item['body']) === false) && (strpos($data['text'], $item['body']) === false)) {
 605                                                 // The body is not part of the fetched page info title or page info text. So we add the text to the body
 606                                                 $item['body'] .= "\n\n" . $data['text'];
 607                                         } else {
 608                                                 // Else we replace the body with the page info text
 609                                                 $item['body'] = $data['text'];
 610                                         }
 611                                 }
 612
 613                                 $data = PageInfo::queryUrl($item['plink'], false, $preview, ($contact['fetch_further_information'] == 2), $contact['ffi_keyword_denylist'] ?? '');
 614
 615                                 if (!empty($data)) {
 616                                         // Take the data that was provided by the feed if the query is empty
 617                                         if (($data['type'] == 'link') && empty($data['title']) && empty($data['text'])) {
 618                                                 $data['title'] = $saved_title;
 619                                                 $item['body'] = $saved_body;
 620                                         }
 621
 622                                         $data_text = strip_tags(trim($data['text'] ?? ''));
 623                                         $item_body = strip_tags(trim($item['body'] ?? ''));
 624
 625                                         if (!empty($data_text) && (($data_text == $item_body) || strstr($item_body, $data_text))) {
 626                                                 $data['text'] = '';
 627                                         }
 628
 629                                         // We always strip the title since it will be added in the page information
 630                                         $item['title'] = '';
 631                                         $item['body'] = $item['body'] . "\n" . PageInfo::getFooterFromData($data, false);
 632                                         $taglist = $contact['fetch_further_information'] == 2 ? PageInfo::getTagsFromUrl($item['plink'], $preview, $contact['ffi_keyword_denylist'] ?? '') : [];
 633                                         $item['object-type'] = Activity\ObjectType::BOOKMARK;
 634                                         $attachments = [];
 635
 636                                         foreach (['audio', 'video'] as $elementname) {
 637                                                 if (!empty($data[$elementname])) {
 638                                                         foreach ($data[$elementname] as $element) {
 639                                                                 if (!empty($element['src'])) {
 640                                                                         $src = $element['src'];
 641                                                                 } elseif (!empty($element['content'])) {
 642                                                                         $src = $element['content'];
 643                                                                 } else {
 644                                                                         continue;
 645                                                                 }
 646
 647                                                                 $attachments[] = [
 648                                                                         'type'        => ($elementname == 'audio') ? Post\Media::AUDIO : Post\Media::VIDEO,
 649                                                                         'url'         => $src,
 650                                                                         'preview'     => $element['image']       ?? null,
 651                                                                         'mimetype'    => $element['contenttype'] ?? null,
 652                                                                         'name'        => $element['name']        ?? null,
 653                                                                         'description' => $element['description'] ?? null,
 654                                                                 ];
 655                                                         }
 656                                                 }
 657                                         }
 658                                 }
 659                         } else {
 660                                 if (!empty($summary)) {
 661                                         $item['body'] = '[abstract]' . HTML::toBBCode($summary, $basepath) . "[/abstract]\n" . $item['body'];
 662                                 }
 663
 664                                 if (!empty($contact['fetch_further_information']) && ($contact['fetch_further_information'] == 3)) {
 665                                         if (empty($taglist)) {
 666                                                 $taglist = PageInfo::getTagsFromUrl($item['plink'], $preview, $contact['ffi_keyword_denylist'] ?? '');
 667                                         }
 668                                         $item['body'] .= "\n" . self::tagToString($taglist);
 669                                 } else {
 670                                         $taglist = [];
 671                                 }
 672
 673                                 // Add the link to the original feed entry if not present in feed
 674                                 if (($item['plink'] != '') && !strstr($item['body'], $item['plink']) && !in_array($item['plink'], array_column($attachments, 'url'))) {
 675                                         $item['body'] .= '[hr][url]' . $item['plink'] . '[/url]';
 676                                 }
 677                         }
 678
 679                         if (empty($item['title'])) {
 680                                 $item['post-type'] = Item::PT_NOTE;
 681                         }
 682
 683                         Logger::info('Stored feed', ['item' => $item]);
 684
 685                         $notify = Item::isRemoteSelf($contact, $item);
 686                         $item['wall'] = (bool)$notify;
 687
 688                         // Distributed items should have a well-formatted URI.
 689                         // Additionally, we have to avoid conflicts with identical URI between imported feeds and these items.
 690                         if ($notify) {
 691                                 $item['guid'] = Item::guidFromUri($orig_plink, DI::baseUrl()->getHost());
 692                                 $item['uri']  = Item::newURI($item['guid']);
 693                                 unset($item['plink']);
 694                                 unset($item['thr-parent']);
 695                                 unset($item['parent-uri']);
 696
 697                                 // Set the delivery priority for "remote self" to "medium"
 698                                 $notify = Worker::PRIORITY_MEDIUM;
 699                         }
 700
 701                         $condition = ['uid' => $item['uid'], 'uri' => $item['uri']];
 702                         if (!Post::exists($condition) && !Post\Delayed::exists($item['uri'], $item['uid'])) {
 703                                 if (!$notify) {
 704                                         Post\Delayed::publish($item, $notify, $taglist, $attachments);
 705                                 } else {
 706                                         $postings[] = [
 707                                                 'item' => $item, 'notify' => $notify,
 708                                                 'taglist' => $taglist, 'attachments' => $attachments
 709                                         ];
 710                                 }
 711                         } else {
 712                                 Logger::info('Post already created or exists in the delayed posts queue', ['uid' => $item['uid'], 'uri' => $item['uri']]);
 713                         }
 714                 }
 715
 716                 if (!empty($postings)) {
 717                         $min_posting = DI::config()->get('system', 'minimum_posting_interval', 0);
 718                         $total = count($postings);
 719                         if ($total > 1) {
 720                                 // Posts shouldn't be delayed more than a day
 721                                 $interval = min(1440, self::getPollInterval($contact));
 722                                 $delay = max(round(($interval * 60) / $total), 60 * $min_posting);
 723                                 Logger::info('Got posting delay', ['delay' => $delay, 'interval' => $interval, 'items' => $total, 'cid' => $contact['id'], 'url' => $contact['url']]);
 724                         } else {
 725                                 $delay = 0;
 726                         }
 727
 728                         $post_delay = 0;
 729
 730                         foreach ($postings as $posting) {
 731                                 if ($delay > 0) {
 732                                         $publish_time = time() + $post_delay;
 733                                         $post_delay += $delay;
 734                                 } else {
 735                                         $publish_time = time();
 736                                 }
 737
 738                                 $last_publish = DI::pConfig()->get($posting['item']['uid'], 'system', 'last_publish', 0, true);
 739                                 $next_publish = max($last_publish + (60 * $min_posting), time());
 740                                 if ($publish_time < $next_publish) {
 741                                         $publish_time = $next_publish;
 742                                 }
 743                                 $publish_at = date(DateTimeFormat::MYSQL, $publish_time);
 744
 745                                 if (Post\Delayed::add($posting['item']['uri'], $posting['item'], $posting['notify'], Post\Delayed::PREPARED, $publish_at, $posting['taglist'], $posting['attachments'])) {
 746                                         DI::pConfig()->set($item['uid'], 'system', 'last_publish', $publish_time);
 747                                 }
 748                         }
 749                 }
 750
 751                 if (!$dryRun && DI::config()->get('system', 'adjust_poll_frequency')) {
 752                         self::adjustPollFrequency($contact, $creation_dates);
 753                 }
 754
 755                 return ['header' => $author, 'items' => $items];
 756         }
 757
 758         /**
 759          * Automatically adjust the poll frequency according to the post frequency
 760          *
 761          * @param array $contact Contact array
 762          * @param array $creation_dates
 763          * @return void
 764          */
 765         private static function adjustPollFrequency(array $contact, array $creation_dates)
 766         {
 767                 if ($contact['network'] != Protocol::FEED) {
 768                         Logger::info('Contact is no feed, skip.', ['id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url'], 'network' => $contact['network']]);
 769                         return;
 770                 }
 771
 772                 if (!empty($creation_dates)) {
 773                         // Count the post frequency and the earliest and latest post date
 774                         $frequency = [];
 775                         $oldest = time();
 776                         $newest = 0;
 777                         $oldest_date = $newest_date = '';
 778
 779                         foreach ($creation_dates as $date) {
 780                                 $timestamp = strtotime($date);
 781                                 $day = intdiv($timestamp, 86400);
 782                                 $hour = $timestamp % 86400;
 783
 784                                 // Only have a look at values from the last seven days
 785                                 if (((time() / 86400) - $day) < 7) {
 786                                         if (empty($frequency[$day])) {
 787                                                 $frequency[$day] = ['count' => 1, 'low' => $hour, 'high' => $hour];
 788                                         } else {
 789                                                 ++$frequency[$day]['count'];
 790                                                 if ($frequency[$day]['low'] > $hour) {
 791                                                         $frequency[$day]['low'] = $hour;
 792                                                 }
 793                                                 if ($frequency[$day]['high'] < $hour) {
 794                                                         $frequency[$day]['high'] = $hour;
 795                                                 }
 796                                         }
 797                                 }
 798                                 if ($oldest > $day) {
 799                                         $oldest = $day;
 800                                         $oldest_date = $date;
 801                                 }
 802
 803                                 if ($newest < $day) {
 804                                         $newest = $day;
 805                                         $newest_date = $date;
 806                                 }
 807                         }
 808
 809                         if (count($creation_dates) == 1) {
 810                                 Logger::info('Feed had posted a single time, switching to daily polling', ['newest' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 811                                 $priority = 8; // Poll once a day
 812                         }
 813
 814                         if (empty($priority) && (((time() / 86400) - $newest) > 730)) {
 815                                 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']]);
 816                                 $priority = 10; // Poll every month
 817                         }
 818
 819                         if (empty($priority) && (((time() / 86400) - $newest) > 365)) {
 820                                 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']]);
 821                                 $priority = 9; // Poll every week
 822                         }
 823
 824                         if (empty($priority) && empty($frequency)) {
 825                                 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']]);
 826                                 $priority = 8; // Poll once a day
 827                         }
 828
 829                         if (empty($priority)) {
 830                                 // Calculate the highest "posts per day" value
 831                                 $max = 0;
 832                                 foreach ($frequency as $entry) {
 833                                         if (($entry['count'] == 1) || ($entry['high'] == $entry['low'])) {
 834                                                 continue;
 835                                         }
 836
 837                                         // We take the earliest and latest post day and interpolate the number of post per day
 838                                         // that would had been created with this post frequency
 839
 840                                         // Assume at least four hours between oldest and newest post per day - should be okay for news outlets
 841                                         $duration = max($entry['high'] - $entry['low'], 14400);
 842                                         $ppd = (86400 / $duration) * $entry['count'];
 843                                         if ($ppd > $max) {
 844                                                 $max = $ppd;
 845                                         }
 846                                 }
 847                                 if ($max > 48) {
 848                                         $priority = 1; // Poll every quarter hour
 849                                 } elseif ($max > 24) {
 850                                         $priority = 2; // Poll half an hour
 851                                 } elseif ($max > 12) {
 852                                         $priority = 3; // Poll hourly
 853                                 } elseif ($max > 8) {
 854                                         $priority = 4; // Poll every two hours
 855                                 } elseif ($max > 4) {
 856                                         $priority = 5; // Poll every three hours
 857                                 } elseif ($max > 2) {
 858                                         $priority = 6; // Poll every six hours
 859                                 } else {
 860                                         $priority = 7; // Poll twice a day
 861                                 }
 862                                 Logger::info('Calculated priority by the posts per day', ['priority' => $priority, 'max' => round($max, 2), 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 863                         }
 864                 } else {
 865                         Logger::info('No posts, switching to daily polling', ['id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 866                         $priority = 8; // Poll once a day
 867                 }
 868
 869                 if ($contact['rating'] != $priority) {
 870                         Logger::notice('Adjusting priority', ['old' => $contact['rating'], 'new' => $priority, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 871                         Contact::update(['rating' => $priority], ['id' => $contact['id']]);
 872                 }
 873         }
 874
 875         /**
 876          * Get the poll interval for the given contact array
 877          *
 878          * @param array $contact
 879          * @return int Poll interval in minutes
 880          */
 881         public static function getPollInterval(array $contact): int
 882         {
 883                 if (in_array($contact['network'], [Protocol::MAIL, Protocol::FEED])) {
 884                         $ratings = [0, 3, 7, 8, 9, 10];
 885                         if (DI::config()->get('system', 'adjust_poll_frequency') && ($contact['network'] == Protocol::FEED)) {
 886                                 $rating = $contact['rating'];
 887                         } elseif (array_key_exists($contact['priority'], $ratings)) {
 888                                 $rating = $ratings[$contact['priority']];
 889                         } else {
 890                                 $rating = -1;
 891                         }
 892                 } else {
 893                         // Check once a week per default for all other networks
 894                         $rating = 9;
 895                 }
 896
 897                 // Friendica and OStatus are checked once a day
 898                 if (in_array($contact['network'], [Protocol::DFRN, Protocol::OSTATUS])) {
 899                         $rating = 8;
 900                 }
 901
 902                 // Check archived contacts or contacts with unsupported protocols once a month
 903                 if ($contact['archive'] || in_array($contact['network'], [Protocol::ZOT, Protocol::PHANTOM])) {
 904                         $rating = 10;
 905                 }
 906
 907                 if ($rating < 0) {
 908                         return 0;
 909                 }
 910                 /*
 911                  * Based on $contact['priority'], should we poll this site now? Or later?
 912                  */
 913
 914                 $min_poll_interval = max(1, DI::config()->get('system', 'min_poll_interval'));
 915
 916                 $poll_intervals = [$min_poll_interval, 15, 30, 60, 120, 180, 360, 720, 1440, 10080, 43200];
 917
 918                 //$poll_intervals = [$min_poll_interval . ' minute', '15 minute', '30 minute',
 919                 //      '1 hour', '2 hour', '3 hour', '6 hour', '12 hour' ,'1 day', '1 week', '1 month'];
 920
 921                 return $poll_intervals[$rating];
 922         }
 923
 924         /**
 925          * Convert a tag array to a tag string
 926          *
 927          * @param array $tags
 928          * @return string tag string
 929          */
 930         private static function tagToString(array $tags): string
 931         {
 932                 $tagstr = '';
 933
 934                 foreach ($tags as $tag) {
 935                         if ($tagstr != '') {
 936                                 $tagstr .= ', ';
 937                         }
 938
 939                         $tagstr .= '#[url=' . DI::baseUrl() . '/search?tag=' . urlencode($tag) . ']' . $tag . '[/url]';
 940                 }
 941
 942                 return $tagstr;
 943         }
 944
 945         private static function titleIsBody(string $title, string $body): bool
 946         {
 947                 $title = strip_tags($title);
 948                 $title = trim($title);
 949                 $title = html_entity_decode($title, ENT_QUOTES, 'UTF-8');
 950                 $title = str_replace(["\n", "\r", "\t", " "], ['', '', '', ''], $title);
 951
 952                 $body = strip_tags($body);
 953                 $body = trim($body);
 954                 $body = html_entity_decode($body, ENT_QUOTES, 'UTF-8');
 955                 $body = str_replace(["\n", "\r", "\t", " "], ['', '', '', ''], $body);
 956
 957                 if (strlen($title) < strlen($body)) {
 958                         $body = substr($body, 0, strlen($title));
 959                 }
 960
 961                 if (($title != $body) && (substr($title, -3) == '...')) {
 962                         $pos = strrpos($title, '...');
 963                         if ($pos > 0) {
 964                                 $title = substr($title, 0, $pos);
 965                                 $body = substr($body, 0, $pos);
 966                         }
 967                 }
 968                 return ($title == $body);
 969         }
 970
 971         /**
 972          * Creates the Atom feed for a given nickname
 973          *
 974          * Supported filters:
 975          * - activity (default): all the public posts
 976          * - posts: all the public top-level posts
 977          * - comments: all the public replies
 978          *
 979          * Updates the provided last_update parameter if the result comes from the
 980          * cache or it is empty
 981          *
 982          * @param array   $owner       owner-view record of the feed owner
 983          * @param string  $last_update Date of the last update
 984          * @param integer $max_items   Number of maximum items to fetch
 985          * @param string  $filter      Feed items filter (activity, posts or comments)
 986          * @param boolean $nocache     Wether to bypass caching
 987          *
 988          * @return string Atom feed
 989          * @throws HTTPException\InternalServerErrorException
 990          * @throws \ImagickException
 991          */
 992         public static function atom(array $owner, string $last_update, int $max_items = 300, string $filter = 'activity', bool $nocache = false)
 993         {
 994                 $stamp = microtime(true);
 995
 996                 $cachekey = 'feed:feed:' . $owner['nickname'] . ':' . $filter . ':' . $last_update;
 997
 998                 // Display events in the user's timezone
 999                 if (strlen($owner['timezone'])) {
1000                         DI::app()->setTimeZone($owner['timezone']);
1001                 }
1002
1003                 $previous_created = $last_update;
1004
1005                 // Don't cache when the last item was posted less than 15 minutes ago (Cache duration)
1006                 if ((time() - strtotime($owner['last-item'])) < 15 * 60) {
1007                         $result = DI::cache()->get($cachekey);
1008                         if (!$nocache && !is_null($result)) {
1009                                 Logger::info('Cached feed duration', ['seconds' => number_format(microtime(true) - $stamp, 3), 'nick' => $owner['nickname'], 'filter' => $filter, 'created' => $previous_created]);
1010                                 return $result['feed'];
1011                         }
1012                 }
1013
1014                 $check_date = empty($last_update) ? '' : DateTimeFormat::utc($last_update);
1015                 $authorid = Contact::getIdForURL($owner['url']);
1016
1017                 $condition = [
1018                         "`uid` = ? AND `received` > ? AND NOT `deleted` AND `gravity` IN (?, ?)
1019                         AND `private` != ? AND `visible` AND `wall` AND `parent-network` IN (?, ?, ?, ?)",
1020                         $owner['uid'], $check_date, Item::GRAVITY_PARENT, Item::GRAVITY_COMMENT,
1021                         Item::PRIVATE, Protocol::ACTIVITYPUB,
1022                         Protocol::OSTATUS, Protocol::DFRN, Protocol::DIASPORA
1023                 ];
1024
1025                 if ($filter === 'comments') {
1026                         $condition[0] .= " AND `gravity` = ? ";
1027                         $condition[] = Item::GRAVITY_COMMENT;
1028                 }
1029
1030                 if ($owner['account-type'] != User::ACCOUNT_TYPE_COMMUNITY) {
1031                         $condition[0] .= " AND `contact-id` = ? AND `author-id` = ?";
1032                         $condition[] = $owner['id'];
1033                         $condition[] = $authorid;
1034                 }
1035
1036                 $params = ['order' => ['received' => true], 'limit' => $max_items];
1037
1038                 if ($filter === 'posts') {
1039                         $ret = Post::selectThread(Item::DELIVER_FIELDLIST, $condition, $params);
1040                 } else {
1041                         $ret = Post::select(Item::DELIVER_FIELDLIST, $condition, $params);
1042                 }
1043
1044                 $items = Post::toArray($ret);
1045
1046                 $doc = new DOMDocument('1.0', 'utf-8');
1047                 $doc->formatOutput = true;
1048
1049                 $root = self::addHeader($doc, $owner, $filter);
1050
1051                 foreach ($items as $item) {
1052                         $entry = self::noteEntry($doc, $item, $owner);
1053                         $root->appendChild($entry);
1054
1055                         if ($last_update < $item['created']) {
1056                                 $last_update = $item['created'];
1057                         }
1058                 }
1059
1060                 $feeddata = trim($doc->saveXML());
1061
1062                 $msg = ['feed' => $feeddata, 'last_update' => $last_update];
1063                 DI::cache()->set($cachekey, $msg, Duration::QUARTER_HOUR);
1064
1065                 Logger::info('Feed duration', ['seconds' => number_format(microtime(true) - $stamp, 3), 'nick' => $owner['nickname'], 'filter' => $filter, 'created' => $previous_created]);
1066
1067                 return $feeddata;
1068         }
1069
1070         /**
1071          * Adds the header elements to the XML document
1072          *
1073          * @param DOMDocument $doc       XML document
1074          * @param array       $owner     Contact data of the poster
1075          * @param string      $filter    The related feed filter (activity, posts or comments)
1076          *
1077          * @return DOMElement Header root element
1078          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1079          */
1080         private static function addHeader(DOMDocument $doc, array $owner, string $filter): DOMElement
1081         {
1082                 $root = $doc->createElementNS(ActivityNamespace::ATOM1, 'feed');
1083                 $doc->appendChild($root);
1084
1085                 $title = '';
1086                 $selfUri = '/feed/' . $owner['nick'] . '/';
1087                 switch ($filter) {
1088                         case 'activity':
1089                                 $title = DI::l10n()->t('%s\'s timeline', $owner['name']);
1090                                 $selfUri .= $filter;
1091                                 break;
1092                         case 'posts':
1093                                 $title = DI::l10n()->t('%s\'s posts', $owner['name']);
1094                                 break;
1095                         case 'comments':
1096                                 $title = DI::l10n()->t('%s\'s comments', $owner['name']);
1097                                 $selfUri .= $filter;
1098                                 break;
1099                 }
1100
1101                 $attributes = ['uri' => 'https://friendi.ca', 'version' => App::VERSION . '-' . DB_UPDATE_VERSION];
1102                 XML::addElement($doc, $root, 'generator', App::PLATFORM, $attributes);
1103                 XML::addElement($doc, $root, 'id', DI::baseUrl() . '/profile/' . $owner['nick']);
1104                 XML::addElement($doc, $root, 'title', $title);
1105                 XML::addElement($doc, $root, 'subtitle', sprintf("Updates from %s on %s", $owner['name'], DI::config()->get('config', 'sitename')));
1106                 XML::addElement($doc, $root, 'logo', User::getAvatarUrl($owner, Proxy::SIZE_SMALL));
1107                 XML::addElement($doc, $root, 'updated', DateTimeFormat::utcNow(DateTimeFormat::ATOM));
1108
1109                 $author = self::addAuthor($doc, $owner);
1110                 $root->appendChild($author);
1111
1112                 $attributes = ['href' => $owner['url'], 'rel' => 'alternate', 'type' => 'text/html'];
1113                 XML::addElement($doc, $root, 'link', '', $attributes);
1114
1115                 OStatus::addHubLink($doc, $root, $owner['nick']);
1116
1117                 $attributes = ['href' => DI::baseUrl() . $selfUri, 'rel' => 'self', 'type' => 'application/atom+xml'];
1118                 XML::addElement($doc, $root, 'link', '', $attributes);
1119
1120                 return $root;
1121         }
1122
1123         /**
1124          * Adds the author element to the XML document
1125          *
1126          * @param DOMDocument $doc          XML document
1127          * @param array       $owner        Contact data of the poster
1128          * @return DOMElement author element
1129          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1130          */
1131         private static function addAuthor(DOMDocument $doc, array $owner): DOMElement
1132         {
1133                 $author = $doc->createElement('author');
1134                 XML::addElement($doc, $author, 'uri', $owner['url']);
1135                 XML::addElement($doc, $author, 'name', $owner['nick']);
1136                 XML::addElement($doc, $author, 'email', $owner['addr']);
1137
1138                 return $author;
1139         }
1140
1141         /**
1142          * Adds a regular entry element
1143          *
1144          * @param DOMDocument $doc       XML document
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        $toplevel  Is it for en entry element (false) or a feed entry (true)?
1148          * @return DOMElement Entry element
1149          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1150          * @throws \ImagickException
1151          */
1152         private static function noteEntry(DOMDocument $doc, array $item, array $owner): DOMElement
1153         {
1154                 if (($item['gravity'] != Item::GRAVITY_PARENT) && (Strings::normaliseLink($item['author-link']) != Strings::normaliseLink($owner['url']))) {
1155                         Logger::info('Feed entry author does not match feed owner', ['owner' => $owner['url'], 'author' => $item['author-link']]);
1156                 }
1157
1158                 $entry = OStatus::entryHeader($doc, $owner, $item, false);
1159
1160                 self::entryContent($doc, $entry, $item, self::getTitle($item), '', true);
1161
1162                 self::entryFooter($doc, $entry, $item, $owner);
1163
1164                 return $entry;
1165         }
1166
1167         /**
1168          * Adds elements to the XML document
1169          *
1170          * @param DOMDocument $doc       XML document
1171          * @param \DOMElement $entry     Entry element where the content is added
1172          * @param array       $item      Data of the item that is to be posted
1173          * @param array       $owner     Contact data of the poster
1174          * @param string      $title     Title for the post
1175          * @param string      $verb      The activity verb
1176          * @param bool        $complete  Add the "status_net" element?
1177          * @return void
1178          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1179          */
1180         private static function entryContent(DOMDocument $doc, DOMElement $entry, array $item, $title, string $verb = '', bool $complete = true)
1181         {
1182                 if ($verb == '') {
1183                         $verb = OStatus::constructVerb($item);
1184                 }
1185
1186                 XML::addElement($doc, $entry, 'id', $item['uri']);
1187                 XML::addElement($doc, $entry, 'title', html_entity_decode($title, ENT_QUOTES, 'UTF-8'));
1188
1189                 $body = Post\Media::addAttachmentsToBody($item['uri-id'], DI::contentItem()->addSharedPost($item));
1190                 $body = Post\Media::addHTMLLinkToBody($item['uri-id'], $body);
1191
1192                 $body = BBCode::convertForUriId($item['uri-id'], $body, BBCode::ACTIVITYPUB);
1193
1194                 XML::addElement($doc, $entry, 'content', $body, ['type' => 'html']);
1195
1196                 XML::addElement(
1197                         $doc,
1198                         $entry,
1199                         'link',
1200                         '',
1201                         [
1202                                 'rel' => 'alternate', 'type' => 'text/html',
1203                                 'href' => DI::baseUrl() . '/display/' . $item['guid']
1204                         ]
1205                 );
1206
1207                 XML::addElement($doc, $entry, 'published', DateTimeFormat::utc($item['created'] . '+00:00', DateTimeFormat::ATOM));
1208                 XML::addElement($doc, $entry, 'updated', DateTimeFormat::utc($item['edited'] . '+00:00', DateTimeFormat::ATOM));
1209         }
1210
1211         /**
1212          * Adds the elements at the foot of an entry to the XML document
1213          *
1214          * @param DOMDocument $doc       XML document
1215          * @param object      $entry     The entry element where the elements are added
1216          * @param array       $item      Data of the item that is to be posted
1217          * @param array       $owner     Contact data of the poster
1218          * @param bool        $complete  default true
1219          * @return void
1220          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1221          */
1222         private static function entryFooter(DOMDocument $doc, $entry, array $item, array $owner)
1223         {
1224                 $mentioned = [];
1225
1226                 if ($item['gravity'] != Item::GRAVITY_PARENT) {
1227                         $parent = Post::selectFirst(['guid', 'author-link', 'owner-link'], ['id' => $item['parent']]);
1228
1229                         $thrparent = Post::selectFirst(['guid', 'author-link', 'owner-link', 'plink'], ['uid' => $owner['uid'], 'uri' => $item['thr-parent']]);
1230
1231                         if (DBA::isResult($thrparent)) {
1232                                 $mentioned[$thrparent['author-link']] = $thrparent['author-link'];
1233                                 $mentioned[$thrparent['owner-link']]  = $thrparent['owner-link'];
1234                                 $parent_plink                         = $thrparent['plink'];
1235                         } elseif (DBA::isResult($parent)) {
1236                                 $mentioned[$parent['author-link']] = $parent['author-link'];
1237                                 $mentioned[$parent['owner-link']]  = $parent['owner-link'];
1238                                 $parent_plink                      = DI::baseUrl() . '/display/' . $parent['guid'];
1239                         } else {
1240                                 DI::logger()->notice('Missing parent and thr-parent for child item', ['item' => $item]);
1241                         }
1242
1243                         if (isset($parent_plink)) {
1244                                 $attributes = [
1245                                         'ref'  => $item['thr-parent'],
1246                                         'href' => $parent_plink
1247                                 ];
1248                                 XML::addElement($doc, $entry, 'thr:in-reply-to', '', $attributes);
1249
1250                                 $attributes = [
1251                                         'rel'  => 'related',
1252                                         'href' => $parent_plink
1253                                 ];
1254                                 XML::addElement($doc, $entry, 'link', '', $attributes);
1255                         }
1256                 }
1257
1258                 // uri-id isn't present for follow entry pseudo-items
1259                 $tags = Tag::getByURIId($item['uri-id'] ?? 0);
1260                 foreach ($tags as $tag) {
1261                         $mentioned[$tag['url']] = $tag['url'];
1262                 }
1263
1264                 foreach ($tags as $tag) {
1265                         if ($tag['type'] == Tag::HASHTAG) {
1266                                 XML::addElement($doc, $entry, 'category', '', ['term' => $tag['name']]);
1267                         }
1268                 }
1269
1270                 OStatus::getAttachment($doc, $entry, $item);
1271         }
1272
1273         /**
1274          * Fetch or create title for feed entry
1275          *
1276          * @param array $item
1277          * @return string title
1278          */
1279         private static function getTitle(array $item): string
1280         {
1281                 if ($item['title'] != '') {
1282                         return BBCode::convertForUriId($item['uri-id'], $item['title'], BBCode::ACTIVITYPUB);
1283                 }
1284
1285                 // Fetch information about the post
1286                 $media = Post\Media::getByURIId($item['uri-id'], [Post\Media::HTML]);
1287                 if (!empty($media) && !empty($media[0]['name']) && ($media[0]['name'] != $media[0]['url'])) {
1288                         return $media[0]['name'];
1289                 }
1290
1291                 // If no bookmark is found then take the first line
1292                 // Remove the share element before fetching the first line
1293                 $title = trim(preg_replace("/\[share.*?\](.*?)\[\/share\]/ism", "\n$1\n", $item['body']));
1294
1295                 $title = BBCode::toPlaintext($title) . "\n";
1296                 $pos = strpos($title, "\n");
1297                 $trailer = '';
1298                 if (($pos == 0) || ($pos > 100)) {
1299                         $pos = 100;
1300                         $trailer = '...';
1301                 }
1302
1303                 return substr($title, 0, $pos) . $trailer;
1304         }
1305 }