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