src/Protocol/Feed.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2021, 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 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\Database\DBA;
  33 use Friendica\DI;
  34 use Friendica\Model\Contact;
  35 use Friendica\Model\Item;
  36 use Friendica\Model\Post;
  37 use Friendica\Model\Tag;
  38 use Friendica\Model\User;
  39 use Friendica\Util\DateTimeFormat;
  40 use Friendica\Util\Network;
  41 use Friendica\Util\ParseUrl;
  42 use Friendica\Util\Proxy;
  43 use Friendica\Util\Strings;
  44 use Friendica\Util\XML;
  45 use GuzzleHttp\Exception\TransferException;
  46
  47 /**
  48  * This class contain functions to import feeds (RSS/RDF/Atom)
  49  */
  50 class Feed
  51 {
  52         /**
  53          * Read a RSS/RDF/Atom feed and create an item entry for it
  54          *
  55          * @param string $xml      The feed data
  56          * @param array  $importer The user record of the importer
  57          * @param array  $contact  The contact record of the feed
  58          *
  59          * @return array Returns the header and the first item in dry run mode
  60          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  61          */
  62         public static function import($xml, array $importer = [], array $contact = [])
  63         {
  64                 $dryRun = empty($importer) && empty($contact);
  65
  66                 if ($dryRun) {
  67                         Logger::info("Test Atom/RSS feed");
  68                 } else {
  69                         Logger::info("Import Atom/RSS feed '" . $contact["name"] . "' (Contact " . $contact["id"] . ") for user " . $importer["uid"]);
  70                 }
  71
  72                 $xml = trim($xml);
  73
  74                 if (empty($xml)) {
  75                         Logger::info('XML is empty.');
  76                         return [];
  77                 }
  78
  79                 if (!empty($contact['poll'])) {
  80                         $basepath = $contact['poll'];
  81                 } elseif (!empty($contact['url'])) {
  82                         $basepath = $contact['url'];
  83                 } else {
  84                         $basepath = '';
  85                 }
  86
  87                 $doc = new DOMDocument();
  88                 @$doc->loadXML($xml);
  89                 $xpath = new DOMXPath($doc);
  90                 $xpath->registerNamespace('atom', ActivityNamespace::ATOM1);
  91                 $xpath->registerNamespace('dc', "http://purl.org/dc/elements/1.1/");
  92                 $xpath->registerNamespace('content', "http://purl.org/rss/1.0/modules/content/");
  93                 $xpath->registerNamespace('rdf', "http://www.w3.org/1999/02/22-rdf-syntax-ns#");
  94                 $xpath->registerNamespace('rss', "http://purl.org/rss/1.0/");
  95                 $xpath->registerNamespace('media', "http://search.yahoo.com/mrss/");
  96                 $xpath->registerNamespace('poco', ActivityNamespace::POCO);
  97
  98                 $author = [];
  99                 $entries = null;
 100
 101                 // Is it RDF?
 102                 if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
 103                         $author["author-link"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:link/text()');
 104                         $author["author-name"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:title/text()');
 105
 106                         if (empty($author["author-name"])) {
 107                                 $author["author-name"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:description/text()');
 108                         }
 109                         $entries = $xpath->query('/rdf:RDF/rss:item');
 110                 }
 111
 112                 // Is it Atom?
 113                 if ($xpath->query('/atom:feed')->length > 0) {
 114                         $alternate = XML::getFirstAttributes($xpath, "atom:link[@rel='alternate']");
 115                         if (is_object($alternate)) {
 116                                 foreach ($alternate as $attribute) {
 117                                         if ($attribute->name == "href") {
 118                                                 $author["author-link"] = $attribute->textContent;
 119                                         }
 120                                 }
 121                         }
 122
 123                         if (empty($author["author-link"])) {
 124                                 $self = XML::getFirstAttributes($xpath, "atom:link[@rel='self']");
 125                                 if (is_object($self)) {
 126                                         foreach ($self as $attribute) {
 127                                                 if ($attribute->name == "href") {
 128                                                         $author["author-link"] = $attribute->textContent;
 129                                                 }
 130                                         }
 131                                 }
 132                         }
 133
 134                         if (empty($author["author-link"])) {
 135                                 $author["author-link"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:id/text()');
 136                         }
 137                         $author["author-avatar"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:logo/text()');
 138
 139                         $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:title/text()');
 140
 141                         if (empty($author["author-name"])) {
 142                                 $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:subtitle/text()');
 143                         }
 144
 145                         if (empty($author["author-name"])) {
 146                                 $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:name/text()');
 147                         }
 148
 149                         $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:displayName/text()');
 150                         if ($value != "") {
 151                                 $author["author-name"] = $value;
 152                         }
 153
 154                         if ($dryRun) {
 155                                 $author["author-id"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:id/text()');
 156
 157                                 // See https://tools.ietf.org/html/rfc4287#section-3.2.2
 158                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/atom:uri/text()');
 159                                 if ($value != "") {
 160                                         $author["author-link"] = $value;
 161                                 }
 162
 163                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:preferredUsername/text()');
 164                                 if ($value != "") {
 165                                         $author["author-nick"] = $value;
 166                                 }
 167
 168                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:address/poco:formatted/text()');
 169                                 if ($value != "") {
 170                                         $author["author-location"] = $value;
 171                                 }
 172
 173                                 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:note/text()');
 174                                 if ($value != "") {
 175                                         $author["author-about"] = $value;
 176                                 }
 177
 178                                 $avatar = XML::getFirstAttributes($xpath, "atom:author/atom:link[@rel='avatar']");
 179                                 if (is_object($avatar)) {
 180                                         foreach ($avatar as $attribute) {
 181                                                 if ($attribute->name == "href") {
 182                                                         $author["author-avatar"] = $attribute->textContent;
 183                                                 }
 184                                         }
 185                                 }
 186                         }
 187
 188                         $author["edited"] = $author["created"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:updated/text()');
 189
 190                         $author["app"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:generator/text()');
 191
 192                         $entries = $xpath->query('/atom:feed/atom:entry');
 193                 }
 194
 195                 // Is it RSS?
 196                 if ($xpath->query('/rss/channel')->length > 0) {
 197                         $author["author-link"] = XML::getFirstNodeValue($xpath, '/rss/channel/link/text()');
 198
 199                         $author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/title/text()');
 200                         $author["author-avatar"] = XML::getFirstNodeValue($xpath, '/rss/channel/image/url/text()');
 201
 202                         if (empty($author["author-name"])) {
 203                                 $author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/copyright/text()');
 204                         }
 205
 206                         if (empty($author["author-name"])) {
 207                                 $author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/description/text()');
 208                         }
 209
 210                         $author["edited"] = $author["created"] = XML::getFirstNodeValue($xpath, '/rss/channel/pubDate/text()');
 211
 212                         $author["app"] = XML::getFirstNodeValue($xpath, '/rss/channel/generator/text()');
 213
 214                         $entries = $xpath->query('/rss/channel/item');
 215                 }
 216
 217                 if (!$dryRun) {
 218                         $author["author-link"] = $contact["url"];
 219
 220                         if (empty($author["author-name"])) {
 221                                 $author["author-name"] = $contact["name"];
 222                         }
 223
 224                         $author["author-avatar"] = $contact["thumb"];
 225
 226                         $author["owner-link"] = $contact["url"];
 227                         $author["owner-name"] = $contact["name"];
 228                         $author["owner-avatar"] = $contact["thumb"];
 229                 }
 230
 231                 $header = [];
 232                 $header["uid"] = $importer["uid"] ?? 0;
 233                 $header["network"] = Protocol::FEED;
 234                 $header["wall"] = 0;
 235                 $header["origin"] = 0;
 236                 $header["gravity"] = GRAVITY_PARENT;
 237                 $header["private"] = Item::PUBLIC;
 238                 $header["verb"] = Activity::POST;
 239                 $header["object-type"] = Activity\ObjectType::NOTE;
 240                 $header["post-type"] = Item::PT_ARTICLE;
 241
 242                 $header["contact-id"] = $contact["id"] ?? 0;
 243
 244                 if (!is_object($entries)) {
 245                         Logger::info("There are no entries in this feed.");
 246                         return [];
 247                 }
 248
 249                 $items = [];
 250                 $creation_dates = [];
 251
 252                 // Limit the number of items that are about to be fetched
 253                 $total_items = ($entries->length - 1);
 254                 $max_items = DI::config()->get('system', 'max_feed_items');
 255                 if (($max_items > 0) && ($total_items > $max_items)) {
 256                         $total_items = $max_items;
 257                 }
 258
 259                 $postings = [];
 260
 261                 // Importing older entries first
 262                 for ($i = $total_items; $i >= 0; --$i) {
 263                         $entry = $entries->item($i);
 264
 265                         $item = array_merge($header, $author);
 266
 267                         $alternate = XML::getFirstAttributes($xpath, "atom:link[@rel='alternate']", $entry);
 268                         if (!is_object($alternate)) {
 269                                 $alternate = XML::getFirstAttributes($xpath, "atom:link", $entry);
 270                         }
 271                         if (is_object($alternate)) {
 272                                 foreach ($alternate as $attribute) {
 273                                         if ($attribute->name == "href") {
 274                                                 $item["plink"] = $attribute->textContent;
 275                                         }
 276                                 }
 277                         }
 278
 279                         if (empty($item["plink"])) {
 280                                 $item["plink"] = XML::getFirstNodeValue($xpath, 'link/text()', $entry);
 281                         }
 282
 283                         if (empty($item["plink"])) {
 284                                 $item["plink"] = XML::getFirstNodeValue($xpath, 'rss:link/text()', $entry);
 285                         }
 286
 287                         $item["uri"] = XML::getFirstNodeValue($xpath, 'atom:id/text()', $entry);
 288
 289                         if (empty($item["uri"])) {
 290                                 $item["uri"] = XML::getFirstNodeValue($xpath, 'guid/text()', $entry);
 291                         }
 292
 293                         if (empty($item["uri"])) {
 294                                 $item["uri"] = $item["plink"];
 295                         }
 296
 297                         // Add the base path if missing
 298                         $item["uri"] = Network::addBasePath($item["uri"], $basepath);
 299                         $item["plink"] = Network::addBasePath($item["plink"], $basepath);
 300
 301                         $orig_plink = $item["plink"];
 302
 303                         try {
 304                                 $item["plink"] = DI::httpClient()->finalUrl($item["plink"]);
 305                         } catch (TransferException $exception) {
 306                                 Logger::notice('Item URL couldn\'t get expanded', ['url' => $item["plink"], 'exception' => $exception]);
 307                         }
 308
 309                         $item["title"] = XML::getFirstNodeValue($xpath, 'atom:title/text()', $entry);
 310
 311                         if (empty($item["title"])) {
 312                                 $item["title"] = XML::getFirstNodeValue($xpath, 'title/text()', $entry);
 313                         }
 314                         if (empty($item["title"])) {
 315                                 $item["title"] = XML::getFirstNodeValue($xpath, 'rss:title/text()', $entry);
 316                         }
 317
 318                         $item["title"] = html_entity_decode($item["title"], ENT_QUOTES, 'UTF-8');
 319
 320                         $published = XML::getFirstNodeValue($xpath, 'atom:published/text()', $entry);
 321
 322                         if (empty($published)) {
 323                                 $published = XML::getFirstNodeValue($xpath, 'pubDate/text()', $entry);
 324                         }
 325
 326                         if (empty($published)) {
 327                                 $published = XML::getFirstNodeValue($xpath, 'dc:date/text()', $entry);
 328                         }
 329
 330                         $updated = XML::getFirstNodeValue($xpath, 'atom:updated/text()', $entry);
 331
 332                         if (empty($updated) && !empty($published)) {
 333                                 $updated = $published;
 334                         }
 335
 336                         if (empty($published) && !empty($updated)) {
 337                                 $published = $updated;
 338                         }
 339
 340                         if ($published != "") {
 341                                 $item["created"] = $published;
 342                         }
 343
 344                         if ($updated != "") {
 345                                 $item["edited"] = $updated;
 346                         }
 347
 348                         if (!$dryRun) {
 349                                 $condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
 350                                         $importer["uid"], $item["uri"], Protocol::FEED, Protocol::DFRN];
 351                                 $previous = Post::selectFirst(['id', 'created'], $condition);
 352                                 if (DBA::isResult($previous)) {
 353                                         // Use the creation date when the post had been stored. It can happen this date changes in the feed.
 354                                         $creation_dates[] = $previous['created'];
 355                                         Logger::info("Item with uri " . $item["uri"] . " for user " . $importer["uid"] . " already existed under id " . $previous["id"]);
 356                                         continue;
 357                                 }
 358                                 $creation_dates[] = DateTimeFormat::utc($item['created']);
 359                         }
 360
 361                         $creator = XML::getFirstNodeValue($xpath, 'author/text()', $entry);
 362
 363                         if (empty($creator)) {
 364                                 $creator = XML::getFirstNodeValue($xpath, 'atom:author/atom:name/text()', $entry);
 365                         }
 366
 367                         if (empty($creator)) {
 368                                 $creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
 369                         }
 370
 371                         if ($creator != "") {
 372                                 $item["author-name"] = $creator;
 373                         }
 374
 375                         $creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
 376
 377                         if ($creator != "") {
 378                                 $item["author-name"] = $creator;
 379                         }
 380
 381                         /// @TODO ?
 382                         // <category>Ausland</category>
 383                         // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
 384
 385                         $attachments = [];
 386
 387                         $enclosures = $xpath->query("enclosure|atom:link[@rel='enclosure']", $entry);
 388                         foreach ($enclosures as $enclosure) {
 389                                 $href = "";
 390                                 $length = null;
 391                                 $type = null;
 392
 393                                 foreach ($enclosure->attributes as $attribute) {
 394                                         if (in_array($attribute->name, ["url", "href"])) {
 395                                                 $href = $attribute->textContent;
 396                                         } elseif ($attribute->name == "length") {
 397                                                 $length = (int)$attribute->textContent;
 398                                         } elseif ($attribute->name == "type") {
 399                                                 $type = $attribute->textContent;
 400                                         }
 401                                 }
 402
 403                                 if (!empty($href)) {
 404                                         $attachments[] = ['type' => Post\Media::DOCUMENT, 'url' => $href, 'mimetype' => $type, 'size' => $length];
 405                                 }
 406                         }
 407
 408                         $taglist = [];
 409                         $categories = $xpath->query("category", $entry);
 410                         foreach ($categories as $category) {
 411                                 $taglist[] = $category->nodeValue;
 412                         }
 413
 414                         $body = trim(XML::getFirstNodeValue($xpath, 'atom:content/text()', $entry));
 415
 416                         if (empty($body)) {
 417                                 $body = trim(XML::getFirstNodeValue($xpath, 'content:encoded/text()', $entry));
 418                         }
 419
 420                         $summary = trim(XML::getFirstNodeValue($xpath, 'atom:summary/text()', $entry));
 421
 422                         if (empty($summary)) {
 423                                 $summary = trim(XML::getFirstNodeValue($xpath, 'description/text()', $entry));
 424                         }
 425
 426                         if (empty($body)) {
 427                                 $body = $summary;
 428                                 $summary = '';
 429                         }
 430
 431                         if ($body == $summary) {
 432                                 $summary = '';
 433                         }
 434
 435                         // remove the content of the title if it is identically to the body
 436                         // This helps with auto generated titles e.g. from tumblr
 437                         if (self::titleIsBody($item["title"], $body)) {
 438                                 $item["title"] = "";
 439                         }
 440                         $item["body"] = HTML::toBBCode($body, $basepath);
 441
 442                         // Remove tracking pixels
 443                         $item["body"] = preg_replace("/\[img=1x1\]([^\[\]]*)\[\/img\]/Usi", '', $item["body"]);
 444
 445                         if (($item["body"] == '') && ($item["title"] != '')) {
 446                                 $item["body"] = $item["title"];
 447                                 $item["title"] = '';
 448                         }
 449
 450                         if ($dryRun) {
 451                                 $items[] = $item;
 452                                 break;
 453                         } elseif (!Item::isValid($item)) {
 454                                 Logger::info('Feed item is invalid', ['created' => $item['created'], 'uid' => $item['uid'], 'uri' => $item['uri']]);
 455                                 continue;
 456                         } elseif (Item::isTooOld($item)) {
 457                                 Logger::info('Feed is too old', ['created' => $item['created'], 'uid' => $item['uid'], 'uri' => $item['uri']]);
 458                                 continue;
 459                         }
 460
 461                         $preview = '';
 462                         if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) {
 463                                 // Handle enclosures and treat them as preview picture
 464                                 foreach ($attachments as $attachment) {
 465                                         if ($attachment["mimetype"] == "image/jpeg") {
 466                                                 $preview = $attachment["url"];
 467                                         }
 468                                 }
 469
 470                                 // Remove a possible link to the item itself
 471                                 $item["body"] = str_replace($item["plink"], '', $item["body"]);
 472                                 $item["body"] = trim(preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item["body"]));
 473
 474                                 // Replace the content when the title is longer than the body
 475                                 $replace = (strlen($item["title"]) > strlen($item["body"]));
 476
 477                                 // Replace it, when there is an image in the body
 478                                 if (strstr($item["body"], '[/img]')) {
 479                                         $replace = true;
 480                                 }
 481
 482                                 // Replace it, when there is a link in the body
 483                                 if (strstr($item["body"], '[/url]')) {
 484                                         $replace = true;
 485                                 }
 486
 487                                 $saved_body = $item["body"];
 488                                 $saved_title = $item["title"];
 489
 490                                 if ($replace) {
 491                                         $item["body"] = trim($item["title"]);
 492                                 }
 493
 494                                 $data = ParseUrl::getSiteinfoCached($item['plink']);
 495                                 if (!empty($data['text']) && !empty($data['title']) && (mb_strlen($item['body']) < mb_strlen($data['text']))) {
 496                                         // When the fetched page info text is longer than the body, we do try to enhance the body
 497                                         if (!empty($item['body']) && (strpos($data['title'], $item['body']) === false) && (strpos($data['text'], $item['body']) === false)) {
 498                                                 // The body is not part of the fetched page info title or page info text. So we add the text to the body
 499                                                 $item['body'] .= "\n\n" . $data['text'];
 500                                         } else {
 501                                                 // Else we replace the body with the page info text
 502                                                 $item['body'] = $data['text'];
 503                                         }
 504                                 }
 505
 506                                 $data = PageInfo::queryUrl($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_denylist"] ?? '');
 507
 508                                 if (!empty($data)) {
 509                                         // Take the data that was provided by the feed if the query is empty
 510                                         if (($data['type'] == 'link') && empty($data['title']) && empty($data['text'])) {
 511                                                 $data['title'] = $saved_title;
 512                                                 $item["body"] = $saved_body;
 513                                         }
 514
 515                                         $data_text = strip_tags(trim($data['text'] ?? ''));
 516                                         $item_body = strip_tags(trim($item['body'] ?? ''));
 517
 518                                         if (!empty($data_text) && (($data_text == $item_body) || strstr($item_body, $data_text))) {
 519                                                 $data['text'] = '';
 520                                         }
 521
 522                                         // We always strip the title since it will be added in the page information
 523                                         $item["title"] = "";
 524                                         $item["body"] = $item["body"] . "\n" . PageInfo::getFooterFromData($data, false);
 525                                         $taglist = $contact["fetch_further_information"] == 2 ? PageInfo::getTagsFromUrl($item["plink"], $preview, $contact["ffi_keyword_denylist"] ?? '') : [];
 526                                         $item["object-type"] = Activity\ObjectType::BOOKMARK;
 527                                         $attachments = [];
 528                                 }
 529                         } else {
 530                                 if (!empty($summary)) {
 531                                         $item["body"] = '[abstract]' . HTML::toBBCode($summary, $basepath) . "[/abstract]\n" . $item["body"];
 532                                 }
 533
 534                                 if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] == 3)) {
 535                                         if (empty($taglist)) {
 536                                                 $taglist = PageInfo::getTagsFromUrl($item["plink"], $preview, $contact["ffi_keyword_denylist"] ?? '');
 537                                         }
 538                                         $item["body"] .= "\n" . self::tagToString($taglist);
 539                                 } else {
 540                                         $taglist = [];
 541                                 }
 542
 543                                 // Add the link to the original feed entry if not present in feed
 544                                 if (($item['plink'] != '') && !strstr($item["body"], $item['plink'])) {
 545                                         $item["body"] .= "[hr][url]" . $item['plink'] . "[/url]";
 546                                 }
 547                         }
 548
 549                         if (empty($item['title'])) {
 550                                 $item['post-type'] = Item::PT_NOTE;
 551                         }
 552
 553                         Logger::info('Stored feed', ['item' => $item]);
 554
 555                         $notify = Item::isRemoteSelf($contact, $item);
 556
 557                         // Distributed items should have a well formatted URI.
 558                         // Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
 559                         if ($notify) {
 560                                 $item['guid'] = Item::guidFromUri($orig_plink, DI::baseUrl()->getHostname());
 561                                 $item['uri'] = Item::newURI($item['uid'], $item['guid']);
 562                                 unset($item['thr-parent']);
 563                                 unset($item['parent-uri']);
 564
 565                                 // Set the delivery priority for "remote self" to "medium"
 566                                 $notify = PRIORITY_MEDIUM;
 567                         }
 568
 569                         $condition = ['uid' => $item['uid'], 'uri' => $item['uri']];
 570                         if (!Post::exists($condition) && !Post\Delayed::exists($item["uri"], $item['uid'])) {
 571                                 if (!$notify) {
 572                                         Post\Delayed::publish($item, $notify, $taglist, $attachments);
 573                                 } else {
 574                                         $postings[] = ['item' => $item, 'notify' => $notify,
 575                                                 'taglist' => $taglist, 'attachments' => $attachments];
 576                                 }
 577                         } else {
 578                                 Logger::info('Post already created or exists in the delayed posts queue', ['uid' => $item['uid'], 'uri' => $item["uri"]]);
 579                         }
 580                 }
 581
 582                 if (!empty($postings)) {
 583                         $min_posting = DI::config()->get('system', 'minimum_posting_interval', 0);
 584                         $total = count($postings);
 585                         if ($total > 1) {
 586                                 // Posts shouldn't be delayed more than a day
 587                                 $interval = min(1440, self::getPollInterval($contact));
 588                                 $delay = max(round(($interval * 60) / $total), 60 * $min_posting);
 589                                 Logger::info('Got posting delay', ['delay' => $delay, 'interval' => $interval, 'items' => $total, 'cid' => $contact['id'], 'url' => $contact['url']]);
 590                         } else {
 591                                 $delay = 0;
 592                         }
 593
 594                         $post_delay = 0;
 595
 596                         foreach ($postings as $posting) {
 597                                 if ($delay > 0) {
 598                                         $publish_time = time() + $post_delay;
 599                                         $post_delay += $delay;
 600                                 } else {
 601                                         $publish_time = time();
 602                                 }
 603
 604                                 $last_publish = DI::pConfig()->get($posting['item']['uid'], 'system', 'last_publish', 0, true);
 605                                 $next_publish = max($last_publish + (60 * $min_posting), time());
 606                                 if ($publish_time < $next_publish) {
 607                                         $publish_time = $next_publish;
 608                                 }
 609                                 $publish_at = date(DateTimeFormat::MYSQL, $publish_time);
 610
 611                                 if (Post\Delayed::add($posting['item']['uri'], $posting['item'], $posting['notify'], Post\Delayed::PREPARED, $publish_at, $posting['taglist'], $posting['attachments'])) {
 612                                         DI::pConfig()->set($item['uid'], 'system', 'last_publish', $publish_time);
 613                                 }
 614                         }
 615                 }
 616
 617                 if (!$dryRun && DI::config()->get('system', 'adjust_poll_frequency')) {
 618                         self::adjustPollFrequency($contact, $creation_dates);
 619                 }
 620
 621                 return ["header" => $author, "items" => $items];
 622         }
 623
 624         /**
 625          * Automatically adjust the poll frequency according to the post frequency
 626          *
 627          * @param array $contact
 628          * @param array $creation_dates
 629          * @return void
 630          */
 631         private static function adjustPollFrequency(array $contact, array $creation_dates)
 632         {
 633                 if ($contact['network'] != Protocol::FEED) {
 634                         Logger::info('Contact is no feed, skip.', ['id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url'], 'network' => $contact['network']]);
 635                         return;
 636                 }
 637
 638                 if (!empty($creation_dates)) {
 639                         // Count the post frequency and the earliest and latest post date
 640                         $frequency = [];
 641                         $oldest = time();
 642                         $newest = 0;
 643                         $oldest_date = $newest_date = '';
 644
 645                         foreach ($creation_dates as $date) {
 646                                 $timestamp = strtotime($date);
 647                                 $day = intdiv($timestamp, 86400);
 648                                 $hour = $timestamp % 86400;
 649
 650                                 // Only have a look at values from the last seven days
 651                                 if (((time() / 86400) - $day) < 7) {
 652                                         if (empty($frequency[$day])) {
 653                                                 $frequency[$day] = ['count' => 1, 'low' => $hour, 'high' => $hour];
 654                                         } else {
 655                                                 ++$frequency[$day]['count'];
 656                                                 if ($frequency[$day]['low'] > $hour) {
 657                                                         $frequency[$day]['low'] = $hour;
 658                                                 }
 659                                                 if ($frequency[$day]['high'] < $hour) {
 660                                                         $frequency[$day]['high'] = $hour;
 661                                                 }
 662                                         }
 663                                 }
 664                                 if ($oldest > $day) {
 665                                         $oldest = $day;
 666                                         $oldest_date = $date;
 667                                 }
 668
 669                                 if ($newest < $day) {
 670                                         $newest = $day;
 671                                         $newest_date = $date;
 672                                 }
 673                         }
 674
 675                         if (count($creation_dates) == 1) {
 676                                 Logger::info('Feed had posted a single time, switching to daily polling', ['newest' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 677                                 $priority = 8; // Poll once a day
 678                         }
 679
 680                         if (empty($priority) && (((time() / 86400) - $newest) > 730)) {
 681                                 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']]);
 682                                 $priority = 10; // Poll every month
 683                         }
 684
 685                         if (empty($priority) && (((time() / 86400) - $newest) > 365)) {
 686                                 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']]);
 687                                 $priority = 9; // Poll every week
 688                         }
 689
 690                         if (empty($priority) && empty($frequency)) {
 691                                 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']]);
 692                                 $priority = 8; // Poll once a day
 693                         }
 694
 695                         if (empty($priority)) {
 696                                 // Calculate the highest "posts per day" value
 697                                 $max = 0;
 698                                 foreach ($frequency as $entry) {
 699                                         if (($entry['count'] == 1) || ($entry['high'] == $entry['low'])) {
 700                                                 continue;
 701                                         }
 702
 703                                         // We take the earliest and latest post day and interpolate the number of post per day
 704                                         // that would had been created with this post frequency
 705
 706                                         // Assume at least four hours between oldest and newest post per day - should be okay for news outlets
 707                                         $duration = max($entry['high'] - $entry['low'], 14400);
 708                                         $ppd = (86400 / $duration) * $entry['count'];
 709                                         if ($ppd > $max) {
 710                                                 $max = $ppd;
 711                                         }
 712                                 }
 713                                 if ($max > 48) {
 714                                         $priority = 1; // Poll every quarter hour
 715                                 } elseif ($max > 24) {
 716                                         $priority = 2; // Poll half an hour
 717                                 } elseif ($max > 12) {
 718                                         $priority = 3; // Poll hourly
 719                                 } elseif ($max > 8) {
 720                                         $priority = 4; // Poll every two hours
 721                                 } elseif ($max > 4) {
 722                                         $priority = 5; // Poll every three hours
 723                                 } elseif ($max > 2) {
 724                                         $priority = 6; // Poll every six hours
 725                                 } else {
 726                                         $priority = 7; // Poll twice a day
 727                                 }
 728                                 Logger::info('Calculated priority by the posts per day', ['priority' => $priority, 'max' => round($max, 2), 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 729                         }
 730                 } else {
 731                         Logger::info('No posts, switching to daily polling', ['id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 732                         $priority = 8; // Poll once a day
 733                 }
 734
 735                 if ($contact['rating'] != $priority) {
 736                         Logger::notice('Adjusting priority', ['old' => $contact['rating'], 'new' => $priority, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
 737                         Contact::update(['rating' => $priority], ['id' => $contact['id']]);
 738                 }
 739         }
 740
 741         /**
 742          * Get the poll interval for the given contact array
 743          *
 744          * @param array $contact
 745          * @return int Poll interval in minutes
 746          */
 747         public static function getPollInterval(array $contact)
 748         {
 749                 if (in_array($contact['network'], [Protocol::MAIL, Protocol::FEED])) {
 750                         $ratings = [0, 3, 7, 8, 9, 10];
 751                         if (DI::config()->get('system', 'adjust_poll_frequency') && ($contact['network'] == Protocol::FEED)) {
 752                                 $rating = $contact['rating'];
 753                         } elseif (array_key_exists($contact['priority'], $ratings)) {
 754                                 $rating = $ratings[$contact['priority']];
 755                         } else {
 756                                 $rating = -1;
 757                         }
 758                 } else {
 759                         // Check once a week per default for all other networks
 760                         $rating = 9;
 761                 }
 762
 763                 // Friendica and OStatus are checked once a day
 764                 if (in_array($contact['network'], [Protocol::DFRN, Protocol::OSTATUS])) {
 765                         $rating = 8;
 766                 }
 767
 768                 // Check archived contacts or contacts with unsupported protocols once a month
 769                 if ($contact['archive'] || in_array($contact['network'], [Protocol::ZOT, Protocol::PHANTOM])) {
 770                         $rating = 10;
 771                 }
 772
 773                 if ($rating < 0) {
 774                         return 0;
 775                 }
 776                 /*
 777                  * Based on $contact['priority'], should we poll this site now? Or later?
 778                  */
 779
 780                 $min_poll_interval = max(1, DI::config()->get('system', 'min_poll_interval'));
 781
 782                 $poll_intervals = [$min_poll_interval, 15, 30, 60, 120, 180, 360, 720 ,1440, 10080, 43200];
 783
 784                 //$poll_intervals = [$min_poll_interval . ' minute', '15 minute', '30 minute',
 785                 //      '1 hour', '2 hour', '3 hour', '6 hour', '12 hour' ,'1 day', '1 week', '1 month'];
 786
 787                 return $poll_intervals[$rating];
 788         }
 789
 790         /**
 791          * Convert a tag array to a tag string
 792          *
 793          * @param array $tags
 794          * @return string tag string
 795          */
 796         private static function tagToString(array $tags)
 797         {
 798                 $tagstr = '';
 799
 800                 foreach ($tags as $tag) {
 801                         if ($tagstr != "") {
 802                                 $tagstr .= ", ";
 803                         }
 804
 805                         $tagstr .= "#[url=" . DI::baseUrl() . "/search?tag=" . urlencode($tag) . "]" . $tag . "[/url]";
 806                 }
 807
 808                 return $tagstr;
 809         }
 810
 811         private static function titleIsBody($title, $body)
 812         {
 813                 $title = strip_tags($title);
 814                 $title = trim($title);
 815                 $title = html_entity_decode($title, ENT_QUOTES, 'UTF-8');
 816                 $title = str_replace(["\n", "\r", "\t", " "], ["", "", "", ""], $title);
 817
 818                 $body = strip_tags($body);
 819                 $body = trim($body);
 820                 $body = html_entity_decode($body, ENT_QUOTES, 'UTF-8');
 821                 $body = str_replace(["\n", "\r", "\t", " "], ["", "", "", ""], $body);
 822
 823                 if (strlen($title) < strlen($body)) {
 824                         $body = substr($body, 0, strlen($title));
 825                 }
 826
 827                 if (($title != $body) && (substr($title, -3) == "...")) {
 828                         $pos = strrpos($title, "...");
 829                         if ($pos > 0) {
 830                                 $title = substr($title, 0, $pos);
 831                                 $body = substr($body, 0, $pos);
 832                         }
 833                 }
 834                 return ($title == $body);
 835         }
 836
 837         /**
 838          * Creates the Atom feed for a given nickname
 839          *
 840          * Supported filters:
 841          * - activity (default): all the public posts
 842          * - posts: all the public top-level posts
 843          * - comments: all the public replies
 844          *
 845          * Updates the provided last_update parameter if the result comes from the
 846          * cache or it is empty
 847          *
 848          * @param string  $owner_nick  Nickname of the feed owner
 849          * @param string  $last_update Date of the last update
 850          * @param integer $max_items   Number of maximum items to fetch
 851          * @param string  $filter      Feed items filter (activity, posts or comments)
 852          * @param boolean $nocache     Wether to bypass caching
 853          *
 854          * @return string Atom feed
 855          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 856          * @throws \ImagickException
 857          */
 858         public static function atom($owner_nick, $last_update, $max_items = 300, $filter = 'activity', $nocache = false)
 859         {
 860                 $stamp = microtime(true);
 861
 862                 $owner = User::getOwnerDataByNick($owner_nick);
 863                 if (!$owner) {
 864                         return;
 865                 }
 866
 867                 $cachekey = "feed:feed:" . $owner_nick . ":" . $filter . ":" . $last_update;
 868
 869                 $previous_created = $last_update;
 870
 871                 // Don't cache when the last item was posted less then 15 minutes ago (Cache duration)
 872                 if ((time() - strtotime($owner['last-item'])) < 15*60) {
 873                         $result = DI::cache()->get($cachekey);
 874                         if (!$nocache && !is_null($result)) {
 875                                 Logger::info('Cached feed duration', ['seconds' => number_format(microtime(true) - $stamp, 3), 'nick' => $owner_nick, 'filter' => $filter, 'created' => $previous_created]);
 876                                 return $result['feed'];
 877                         }
 878                 }
 879
 880                 $check_date = empty($last_update) ? '' : DateTimeFormat::utc($last_update);
 881                 $authorid = Contact::getIdForURL($owner["url"]);
 882
 883                 $condition = ["`uid` = ? AND `received` > ? AND NOT `deleted` AND `gravity` IN (?, ?)
 884                         AND `private` != ? AND `visible` AND `wall` AND `parent-network` IN (?, ?, ?, ?)",
 885                         $owner["uid"], $check_date, GRAVITY_PARENT, GRAVITY_COMMENT,
 886                         Item::PRIVATE, Protocol::ACTIVITYPUB,
 887                         Protocol::OSTATUS, Protocol::DFRN, Protocol::DIASPORA];
 888
 889                 if ($filter === 'comments') {
 890                         $condition[0] .= " AND `gravity` = ? ";
 891                         $condition[] = GRAVITY_COMMENT;
 892                 }
 893
 894                 if ($owner['account-type'] != User::ACCOUNT_TYPE_COMMUNITY) {
 895                         $condition[0] .= " AND `contact-id` = ? AND `author-id` = ?";
 896                         $condition[] = $owner["id"];
 897                         $condition[] = $authorid;
 898                 }
 899
 900                 $params = ['order' => ['received' => true], 'limit' => $max_items];
 901
 902                 if ($filter === 'posts') {
 903                         $ret = Post::selectThread(Item::DELIVER_FIELDLIST, $condition, $params);
 904                 } else {
 905                         $ret = Post::select(Item::DELIVER_FIELDLIST, $condition, $params);
 906                 }
 907
 908                 $items = Post::toArray($ret);
 909
 910                 $doc = new DOMDocument('1.0', 'utf-8');
 911                 $doc->formatOutput = true;
 912
 913                 $root = self::addHeader($doc, $owner, $filter);
 914
 915                 foreach ($items as $item) {
 916                         $entry = self::noteEntry($doc, $item, $owner);
 917                         $root->appendChild($entry);
 918
 919                         if ($last_update < $item['created']) {
 920                                 $last_update = $item['created'];
 921                         }
 922                 }
 923
 924                 $feeddata = trim($doc->saveXML());
 925
 926                 $msg = ['feed' => $feeddata, 'last_update' => $last_update];
 927                 DI::cache()->set($cachekey, $msg, Duration::QUARTER_HOUR);
 928
 929                 Logger::info('Feed duration', ['seconds' => number_format(microtime(true) - $stamp, 3), 'nick' => $owner_nick, 'filter' => $filter, 'created' => $previous_created]);
 930
 931                 return $feeddata;
 932         }
 933
 934         /**
 935          * Adds the header elements to the XML document
 936          *
 937          * @param DOMDocument $doc       XML document
 938          * @param array       $owner     Contact data of the poster
 939          * @param string      $filter    The related feed filter (activity, posts or comments)
 940          *
 941          * @return object header root element
 942          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 943          */
 944         private static function addHeader(DOMDocument $doc, array $owner, $filter)
 945         {
 946                 $root = $doc->createElementNS(ActivityNamespace::ATOM1, 'feed');
 947                 $doc->appendChild($root);
 948
 949                 $title = '';
 950                 $selfUri = '/feed/' . $owner["nick"] . '/';
 951                 switch ($filter) {
 952                         case 'activity':
 953                                 $title = DI::l10n()->t('%s\'s timeline', $owner['name']);
 954                                 $selfUri .= $filter;
 955                                 break;
 956                         case 'posts':
 957                                 $title = DI::l10n()->t('%s\'s posts', $owner['name']);
 958                                 break;
 959                         case 'comments':
 960                                 $title = DI::l10n()->t('%s\'s comments', $owner['name']);
 961                                 $selfUri .= $filter;
 962                                 break;
 963                 }
 964
 965                 $attributes = ["uri" => "https://friendi.ca", "version" => FRIENDICA_VERSION . "-" . DB_UPDATE_VERSION];
 966                 XML::addElement($doc, $root, "generator", FRIENDICA_PLATFORM, $attributes);
 967                 XML::addElement($doc, $root, "id", DI::baseUrl() . "/profile/" . $owner["nick"]);
 968                 XML::addElement($doc, $root, "title", $title);
 969                 XML::addElement($doc, $root, "subtitle", sprintf("Updates from %s on %s", $owner["name"], DI::config()->get('config', 'sitename')));
 970                 XML::addElement($doc, $root, "logo", User::getAvatarUrl($owner, Proxy::SIZE_SMALL));
 971                 XML::addElement($doc, $root, "updated", DateTimeFormat::utcNow(DateTimeFormat::ATOM));
 972
 973                 $author = self::addAuthor($doc, $owner);
 974                 $root->appendChild($author);
 975
 976                 $attributes = ["href" => $owner["url"], "rel" => "alternate", "type" => "text/html"];
 977                 XML::addElement($doc, $root, "link", "", $attributes);
 978
 979                 OStatus::hublinks($doc, $root, $owner["nick"]);
 980
 981                 $attributes = ["href" => DI::baseUrl() . $selfUri, "rel" => "self", "type" => "application/atom+xml"];
 982                 XML::addElement($doc, $root, "link", "", $attributes);
 983
 984                 return $root;
 985         }
 986
 987         /**
 988          * Adds the author element to the XML document
 989          *
 990          * @param DOMDocument $doc          XML document
 991          * @param array       $owner        Contact data of the poster
 992          *
 993          * @return \DOMElement author element
 994          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 995          */
 996         private static function addAuthor(DOMDocument $doc, array $owner)
 997         {
 998                 $author = $doc->createElement("author");
 999                 XML::addElement($doc, $author, "uri", $owner["url"]);
1000                 XML::addElement($doc, $author, "name", $owner["nick"]);
1001                 XML::addElement($doc, $author, "email", $owner["addr"]);
1002
1003                 return $author;
1004         }
1005
1006         /**
1007          * Adds a regular entry element
1008          *
1009          * @param DOMDocument $doc       XML document
1010          * @param array       $item      Data of the item that is to be posted
1011          * @param array       $owner     Contact data of the poster
1012          * @param bool        $toplevel  Is it for en entry element (false) or a feed entry (true)?
1013          *
1014          * @return \DOMElement Entry element
1015          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1016          * @throws \ImagickException
1017          */
1018         private static function noteEntry(DOMDocument $doc, array $item, array $owner)
1019         {
1020                 if (($item['gravity'] != GRAVITY_PARENT) && (Strings::normaliseLink($item["author-link"]) != Strings::normaliseLink($owner["url"]))) {
1021                         Logger::info('Feed entry author does not match feed owner', ['owner' => $owner["url"], 'author' => $item["author-link"]]);
1022                 }
1023
1024                 $entry = OStatus::entryHeader($doc, $owner, $item, false);
1025
1026                 self::entryContent($doc, $entry, $item, self::getTitle($item), '', true);
1027
1028                 self::entryFooter($doc, $entry, $item, $owner);
1029
1030                 return $entry;
1031         }
1032
1033         /**
1034          * Adds elements to the XML document
1035          *
1036          * @param DOMDocument $doc       XML document
1037          * @param \DOMElement $entry     Entry element where the content is added
1038          * @param array       $item      Data of the item that is to be posted
1039          * @param array       $owner     Contact data of the poster
1040          * @param string      $title     Title for the post
1041          * @param string      $verb      The activity verb
1042          * @param bool        $complete  Add the "status_net" element?
1043          * @param bool        $feed_mode Behave like a regular feed for users if true
1044          * @return void
1045          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1046          */
1047         private static function entryContent(DOMDocument $doc, \DOMElement $entry, array $item, $title, $verb = "", $complete = true)
1048         {
1049                 if ($verb == "") {
1050                         $verb = OStatus::constructVerb($item);
1051                 }
1052
1053                 XML::addElement($doc, $entry, "id", $item["uri"]);
1054                 XML::addElement($doc, $entry, "title", html_entity_decode($title, ENT_QUOTES, 'UTF-8'));
1055
1056                 $body = OStatus::formatPicturePost($item['body'], $item['uri-id']);
1057
1058                 $body = BBCode::convertForUriId($item['uri-id'], $body, BBCode::ACTIVITYPUB);
1059
1060                 XML::addElement($doc, $entry, "content", $body, ["type" => "html"]);
1061
1062                 XML::addElement($doc, $entry, "link", "", ["rel" => "alternate", "type" => "text/html",
1063                                                                 "href" => DI::baseUrl()."/display/".$item["guid"]]
1064                 );
1065
1066                 XML::addElement($doc, $entry, "published", DateTimeFormat::utc($item["created"]."+00:00", DateTimeFormat::ATOM));
1067                 XML::addElement($doc, $entry, "updated", DateTimeFormat::utc($item["edited"]."+00:00", DateTimeFormat::ATOM));
1068         }
1069
1070         /**
1071          * Adds the elements at the foot of an entry to the XML document
1072          *
1073          * @param DOMDocument $doc       XML document
1074          * @param object      $entry     The entry element where the elements are added
1075          * @param array       $item      Data of the item that is to be posted
1076          * @param array       $owner     Contact data of the poster
1077          * @param bool        $complete  default true
1078          * @return void
1079          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1080          */
1081         private static function entryFooter(DOMDocument $doc, $entry, array $item, array $owner)
1082         {
1083                 $mentioned = [];
1084
1085                 if ($item['gravity'] != GRAVITY_PARENT) {
1086                         $parent = Post::selectFirst(['guid', 'author-link', 'owner-link'], ['id' => $item['parent']]);
1087
1088                         $thrparent = Post::selectFirst(['guid', 'author-link', 'owner-link', 'plink'], ['uid' => $owner['uid'], 'uri' => $item['thr-parent']]);
1089
1090                         if (DBA::isResult($thrparent)) {
1091                                 $mentioned[$thrparent['author-link']] = $thrparent['author-link'];
1092                                 $mentioned[$thrparent['owner-link']]  = $thrparent['owner-link'];
1093                                 $parent_plink                         = $thrparent['plink'];
1094                         } elseif (DBA::isResult($parent)) {
1095                                 $mentioned[$parent['author-link']] = $parent['author-link'];
1096                                 $mentioned[$parent['owner-link']]  = $parent['owner-link'];
1097                                 $parent_plink                      = DI::baseUrl() . '/display/' . $parent['guid'];
1098                         } else {
1099                                 DI::logger()->notice('Missing parent and thr-parent for child item', ['item' => $item]);
1100                         }
1101
1102                         if (isset($parent_plink)) {
1103                                 $attributes = [
1104                                         'ref'  => $item['thr-parent'],
1105                                         'href' => $parent_plink];
1106                                 XML::addElement($doc, $entry, 'thr:in-reply-to', '', $attributes);
1107
1108                                 $attributes = [
1109                                         'rel'  => 'related',
1110                                         'href' => $parent_plink];
1111                                 XML::addElement($doc, $entry, 'link', '', $attributes);
1112                         }
1113                 }
1114
1115                 // uri-id isn't present for follow entry pseudo-items
1116                 $tags = Tag::getByURIId($item['uri-id'] ?? 0);
1117                 foreach ($tags as $tag) {
1118                         $mentioned[$tag['url']] = $tag['url'];
1119                 }
1120
1121                 foreach ($tags as $tag) {
1122                         if ($tag['type'] == Tag::HASHTAG) {
1123                                 XML::addElement($doc, $entry, 'category', '', ['term' => $tag['name']]);
1124                         }
1125                 }
1126
1127                 OStatus::getAttachment($doc, $entry, $item);
1128         }
1129
1130         /**
1131          * Fetch or create title for feed entry
1132          *
1133          * @param array $item
1134          * @return string title
1135          */
1136         private static function getTitle(array $item)
1137         {
1138                 if ($item['title'] != '') {
1139                         return BBCode::convertForUriId($item['uri-id'], $item['title'], BBCode::ACTIVITYPUB);
1140                 }
1141
1142                 // Fetch information about the post
1143                 $siteinfo = BBCode::getAttachedData($item["body"]);
1144                 if (isset($siteinfo["title"])) {
1145                         return $siteinfo["title"];
1146                 }
1147
1148                 // If no bookmark is found then take the first line
1149                 // Remove the share element before fetching the first line
1150                 $title = trim(preg_replace("/\[share.*?\](.*?)\[\/share\]/ism","\n$1\n",$item['body']));
1151
1152                 $title = BBCode::toPlaintext($title)."\n";
1153                 $pos = strpos($title, "\n");
1154                 $trailer = "";
1155                 if (($pos == 0) || ($pos > 100)) {
1156                         $pos = 100;
1157                         $trailer = "...";
1158                 }
1159
1160                 return substr($title, 0, $pos) . $trailer;
1161         }
1162 }