src/Protocol/Feed.php

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