3 * @copyright Copyright (C) 2010-2022, the Friendica project
5 * @license GNU AGPL version 3 or any later version
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.
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.
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/>.
22 namespace Friendica\Protocol;
26 use Friendica\Content\PageInfo;
27 use Friendica\Content\Text\BBCode;
28 use Friendica\Content\Text\HTML;
29 use Friendica\Core\Cache\Enum\Duration;
30 use Friendica\Core\Logger;
31 use Friendica\Core\Protocol;
32 use Friendica\Database\DBA;
34 use Friendica\Model\Contact;
35 use Friendica\Model\Item;
36 use Friendica\Model\Post;
37 use Friendica\Model\Tag;
38 use Friendica\Model\User;
39 use Friendica\Util\DateTimeFormat;
40 use Friendica\Util\Network;
41 use Friendica\Util\ParseUrl;
42 use Friendica\Util\Proxy;
43 use Friendica\Util\Strings;
44 use Friendica\Util\XML;
45 use GuzzleHttp\Exception\TransferException;
48 * This class contain functions to import feeds (RSS/RDF/Atom)
53 * Read a RSS/RDF/Atom feed and create an item entry for it
55 * @param string $xml The feed data
56 * @param array $importer The user record of the importer
57 * @param array $contact The contact record of the feed
59 * @return array Returns the header and the first item in dry run mode
60 * @throws \Friendica\Network\HTTPException\InternalServerErrorException
62 public static function import($xml, array $importer = [], array $contact = [])
64 $dryRun = empty($importer) && empty($contact);
67 Logger::info("Test Atom/RSS feed");
69 Logger::info("Import Atom/RSS feed '" . $contact["name"] . "' (Contact " . $contact["id"] . ") for user " . $importer["uid"]);
75 Logger::info('XML is empty.');
79 if (!empty($contact['poll'])) {
80 $basepath = $contact['poll'];
81 } elseif (!empty($contact['url'])) {
82 $basepath = $contact['url'];
87 $doc = new DOMDocument();
89 $xpath = new DOMXPath($doc);
90 $xpath->registerNamespace('atom', ActivityNamespace::ATOM1);
91 $xpath->registerNamespace('dc', "http://purl.org/dc/elements/1.1/");
92 $xpath->registerNamespace('content', "http://purl.org/rss/1.0/modules/content/");
93 $xpath->registerNamespace('rdf', "http://www.w3.org/1999/02/22-rdf-syntax-ns#");
94 $xpath->registerNamespace('rss', "http://purl.org/rss/1.0/");
95 $xpath->registerNamespace('media', "http://search.yahoo.com/mrss/");
96 $xpath->registerNamespace('poco', ActivityNamespace::POCO);
102 if ($xpath->query('/rdf:RDF/rss:channel')->length > 0) {
103 $author["author-link"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:link/text()');
104 $author["author-name"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:title/text()');
106 if (empty($author["author-name"])) {
107 $author["author-name"] = XML::getFirstNodeValue($xpath, '/rdf:RDF/rss:channel/rss:description/text()');
109 $entries = $xpath->query('/rdf:RDF/rss:item');
113 if ($xpath->query('/atom:feed')->length > 0) {
114 $alternate = XML::getFirstAttributes($xpath, "atom:link[@rel='alternate']");
115 if (is_object($alternate)) {
116 foreach ($alternate as $attribute) {
117 if ($attribute->name == "href") {
118 $author["author-link"] = $attribute->textContent;
123 if (empty($author["author-link"])) {
124 $self = XML::getFirstAttributes($xpath, "atom:link[@rel='self']");
125 if (is_object($self)) {
126 foreach ($self as $attribute) {
127 if ($attribute->name == "href") {
128 $author["author-link"] = $attribute->textContent;
134 if (empty($author["author-link"])) {
135 $author["author-link"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:id/text()');
137 $author["author-avatar"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:logo/text()');
139 $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:title/text()');
141 if (empty($author["author-name"])) {
142 $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:subtitle/text()');
145 if (empty($author["author-name"])) {
146 $author["author-name"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:name/text()');
149 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:displayName/text()');
151 $author["author-name"] = $value;
155 $author["author-id"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:author/atom:id/text()');
157 // See https://tools.ietf.org/html/rfc4287#section-3.2.2
158 $value = XML::getFirstNodeValue($xpath, 'atom:author/atom:uri/text()');
160 $author["author-link"] = $value;
163 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:preferredUsername/text()');
165 $author["author-nick"] = $value;
168 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:address/poco:formatted/text()');
170 $author["author-location"] = $value;
173 $value = XML::getFirstNodeValue($xpath, 'atom:author/poco:note/text()');
175 $author["author-about"] = $value;
178 $avatar = XML::getFirstAttributes($xpath, "atom:author/atom:link[@rel='avatar']");
179 if (is_object($avatar)) {
180 foreach ($avatar as $attribute) {
181 if ($attribute->name == "href") {
182 $author["author-avatar"] = $attribute->textContent;
188 $author["edited"] = $author["created"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:updated/text()');
190 $author["app"] = XML::getFirstNodeValue($xpath, '/atom:feed/atom:generator/text()');
192 $entries = $xpath->query('/atom:feed/atom:entry');
196 if ($xpath->query('/rss/channel')->length > 0) {
197 $author["author-link"] = XML::getFirstNodeValue($xpath, '/rss/channel/link/text()');
199 $author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/title/text()');
200 $author["author-avatar"] = XML::getFirstNodeValue($xpath, '/rss/channel/image/url/text()');
202 if (empty($author["author-name"])) {
203 $author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/copyright/text()');
206 if (empty($author["author-name"])) {
207 $author["author-name"] = XML::getFirstNodeValue($xpath, '/rss/channel/description/text()');
210 $author["edited"] = $author["created"] = XML::getFirstNodeValue($xpath, '/rss/channel/pubDate/text()');
212 $author["app"] = XML::getFirstNodeValue($xpath, '/rss/channel/generator/text()');
214 $entries = $xpath->query('/rss/channel/item');
218 $author["author-link"] = $contact["url"];
220 if (empty($author["author-name"])) {
221 $author["author-name"] = $contact["name"];
224 $author["author-avatar"] = $contact["thumb"];
226 $author["owner-link"] = $contact["url"];
227 $author["owner-name"] = $contact["name"];
228 $author["owner-avatar"] = $contact["thumb"];
232 $header["uid"] = $importer["uid"] ?? 0;
233 $header["network"] = Protocol::FEED;
235 $header["origin"] = 0;
236 $header["gravity"] = GRAVITY_PARENT;
237 $header["private"] = Item::PUBLIC;
238 $header["verb"] = Activity::POST;
239 $header["object-type"] = Activity\ObjectType::NOTE;
240 $header["post-type"] = Item::PT_ARTICLE;
242 $header["contact-id"] = $contact["id"] ?? 0;
244 if (!is_object($entries)) {
245 Logger::info("There are no entries in this feed.");
250 $creation_dates = [];
252 // Limit the number of items that are about to be fetched
253 $total_items = ($entries->length - 1);
254 $max_items = DI::config()->get('system', 'max_feed_items');
255 if (($max_items > 0) && ($total_items > $max_items)) {
256 $total_items = $max_items;
261 // Importing older entries first
262 for ($i = $total_items; $i >= 0; --$i) {
263 $entry = $entries->item($i);
265 $item = array_merge($header, $author);
267 $alternate = XML::getFirstAttributes($xpath, "atom:link[@rel='alternate']", $entry);
268 if (!is_object($alternate)) {
269 $alternate = XML::getFirstAttributes($xpath, "atom:link", $entry);
271 if (is_object($alternate)) {
272 foreach ($alternate as $attribute) {
273 if ($attribute->name == "href") {
274 $item["plink"] = $attribute->textContent;
279 if (empty($item["plink"])) {
280 $item["plink"] = XML::getFirstNodeValue($xpath, 'link/text()', $entry);
283 if (empty($item["plink"])) {
284 $item["plink"] = XML::getFirstNodeValue($xpath, 'rss:link/text()', $entry);
287 $item["uri"] = XML::getFirstNodeValue($xpath, 'atom:id/text()', $entry);
289 if (empty($item["uri"])) {
290 $item["uri"] = XML::getFirstNodeValue($xpath, 'guid/text()', $entry);
293 if (empty($item["uri"])) {
294 $item["uri"] = $item["plink"];
297 // Add the base path if missing
298 $item["uri"] = Network::addBasePath($item["uri"], $basepath);
299 $item["plink"] = Network::addBasePath($item["plink"], $basepath);
301 $orig_plink = $item["plink"];
304 $item["plink"] = DI::httpClient()->finalUrl($item["plink"]);
305 } catch (TransferException $exception) {
306 Logger::notice('Item URL couldn\'t get expanded', ['url' => $item["plink"], 'exception' => $exception]);
309 $item["title"] = XML::getFirstNodeValue($xpath, 'atom:title/text()', $entry);
311 if (empty($item["title"])) {
312 $item["title"] = XML::getFirstNodeValue($xpath, 'title/text()', $entry);
314 if (empty($item["title"])) {
315 $item["title"] = XML::getFirstNodeValue($xpath, 'rss:title/text()', $entry);
318 $item["title"] = html_entity_decode($item["title"], ENT_QUOTES, 'UTF-8');
320 $published = XML::getFirstNodeValue($xpath, 'atom:published/text()', $entry);
322 if (empty($published)) {
323 $published = XML::getFirstNodeValue($xpath, 'pubDate/text()', $entry);
326 if (empty($published)) {
327 $published = XML::getFirstNodeValue($xpath, 'dc:date/text()', $entry);
330 $updated = XML::getFirstNodeValue($xpath, 'atom:updated/text()', $entry);
332 if (empty($updated) && !empty($published)) {
333 $updated = $published;
336 if (empty($published) && !empty($updated)) {
337 $published = $updated;
340 if ($published != "") {
341 $item["created"] = $published;
344 if ($updated != "") {
345 $item["edited"] = $updated;
349 $condition = ["`uid` = ? AND `uri` = ? AND `network` IN (?, ?)",
350 $importer["uid"], $item["uri"], Protocol::FEED, Protocol::DFRN];
351 $previous = Post::selectFirst(['id', 'created'], $condition);
352 if (DBA::isResult($previous)) {
353 // Use the creation date when the post had been stored. It can happen this date changes in the feed.
354 $creation_dates[] = $previous['created'];
355 Logger::info("Item with uri " . $item["uri"] . " for user " . $importer["uid"] . " already existed under id " . $previous["id"]);
358 $creation_dates[] = DateTimeFormat::utc($item['created']);
361 $creator = XML::getFirstNodeValue($xpath, 'author/text()', $entry);
363 if (empty($creator)) {
364 $creator = XML::getFirstNodeValue($xpath, 'atom:author/atom:name/text()', $entry);
367 if (empty($creator)) {
368 $creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
371 if ($creator != "") {
372 $item["author-name"] = $creator;
375 $creator = XML::getFirstNodeValue($xpath, 'dc:creator/text()', $entry);
377 if ($creator != "") {
378 $item["author-name"] = $creator;
382 // <category>Ausland</category>
383 // <media:thumbnail width="152" height="76" url="http://www.taz.de/picture/667875/192/14388767.jpg"/>
387 $enclosures = $xpath->query("enclosure|atom:link[@rel='enclosure']", $entry);
388 foreach ($enclosures as $enclosure) {
393 foreach ($enclosure->attributes as $attribute) {
394 if (in_array($attribute->name, ["url", "href"])) {
395 $href = $attribute->textContent;
396 } elseif ($attribute->name == "length") {
397 $length = (int)$attribute->textContent;
398 } elseif ($attribute->name == "type") {
399 $type = $attribute->textContent;
404 $attachment = ['type' => Post\Media::UNKNOWN, 'url' => $href, 'mimetype' => $type, 'size' => $length];
406 $attachment = Post\Media::fetchAdditionalData($attachment);
408 // By now we separate the visible media types (audio, video, image) from the rest
409 // In the future we should try to avoid the DOCUMENT type and only use the real one - but not in the RC phase.
410 if (!in_array($attachment['type'], [Post\Media::AUDIO, Post\Media::IMAGE, Post\Media::VIDEO])) {
411 $attachment['type'] = Post\Media::DOCUMENT;
413 $attachments[] = $attachment;
418 $categories = $xpath->query("category", $entry);
419 foreach ($categories as $category) {
420 $taglist[] = $category->nodeValue;
423 $body = trim(XML::getFirstNodeValue($xpath, 'atom:content/text()', $entry));
426 $body = trim(XML::getFirstNodeValue($xpath, 'content:encoded/text()', $entry));
429 $summary = trim(XML::getFirstNodeValue($xpath, 'atom:summary/text()', $entry));
431 if (empty($summary)) {
432 $summary = trim(XML::getFirstNodeValue($xpath, 'description/text()', $entry));
440 if ($body == $summary) {
444 // remove the content of the title if it is identically to the body
445 // This helps with auto generated titles e.g. from tumblr
446 if (self::titleIsBody($item["title"], $body)) {
449 $item["body"] = HTML::toBBCode($body, $basepath);
451 // Remove tracking pixels
452 $item["body"] = preg_replace("/\[img=1x1\]([^\[\]]*)\[\/img\]/Usi", '', $item["body"]);
454 if (($item["body"] == '') && ($item["title"] != '')) {
455 $item["body"] = $item["title"];
462 } elseif (!Item::isValid($item)) {
463 Logger::info('Feed item is invalid', ['created' => $item['created'], 'uid' => $item['uid'], 'uri' => $item['uri']]);
465 } elseif (Item::isTooOld($item)) {
466 Logger::info('Feed is too old', ['created' => $item['created'], 'uid' => $item['uid'], 'uri' => $item['uri']]);
471 if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] < 3)) {
472 // Handle enclosures and treat them as preview picture
473 foreach ($attachments as $attachment) {
474 if ($attachment["mimetype"] == "image/jpeg") {
475 $preview = $attachment["url"];
479 // Remove a possible link to the item itself
480 $item["body"] = str_replace($item["plink"], '', $item["body"]);
481 $item["body"] = trim(preg_replace('/\[url\=\](\w+.*?)\[\/url\]/i', '', $item["body"]));
483 // Replace the content when the title is longer than the body
484 $replace = (strlen($item["title"]) > strlen($item["body"]));
486 // Replace it, when there is an image in the body
487 if (strstr($item["body"], '[/img]')) {
491 // Replace it, when there is a link in the body
492 if (strstr($item["body"], '[/url]')) {
496 $saved_body = $item["body"];
497 $saved_title = $item["title"];
500 $item["body"] = trim($item["title"]);
503 $data = ParseUrl::getSiteinfoCached($item['plink']);
504 if (!empty($data['text']) && !empty($data['title']) && (mb_strlen($item['body']) < mb_strlen($data['text']))) {
505 // When the fetched page info text is longer than the body, we do try to enhance the body
506 if (!empty($item['body']) && (strpos($data['title'], $item['body']) === false) && (strpos($data['text'], $item['body']) === false)) {
507 // The body is not part of the fetched page info title or page info text. So we add the text to the body
508 $item['body'] .= "\n\n" . $data['text'];
510 // Else we replace the body with the page info text
511 $item['body'] = $data['text'];
515 $data = PageInfo::queryUrl($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_denylist"] ?? '');
518 // Take the data that was provided by the feed if the query is empty
519 if (($data['type'] == 'link') && empty($data['title']) && empty($data['text'])) {
520 $data['title'] = $saved_title;
521 $item["body"] = $saved_body;
524 $data_text = strip_tags(trim($data['text'] ?? ''));
525 $item_body = strip_tags(trim($item['body'] ?? ''));
527 if (!empty($data_text) && (($data_text == $item_body) || strstr($item_body, $data_text))) {
531 // We always strip the title since it will be added in the page information
533 $item["body"] = $item["body"] . "\n" . PageInfo::getFooterFromData($data, false);
534 $taglist = $contact["fetch_further_information"] == 2 ? PageInfo::getTagsFromUrl($item["plink"], $preview, $contact["ffi_keyword_denylist"] ?? '') : [];
535 $item["object-type"] = Activity\ObjectType::BOOKMARK;
538 foreach (['audio', 'video'] as $elementname) {
539 if (!empty($data[$elementname])) {
540 foreach ($data[$elementname] as $element) {
541 if (!empty($element['src'])) {
542 $src = $element['src'];
543 } elseif (!empty($element['content'])) {
544 $src = $element['content'];
550 'type' => ($elementname == 'audio') ? Post\Media::AUDIO : Post\Media::VIDEO,
552 'preview' => $element['image'] ?? null,
553 'mimetype' => $element['contenttype'] ?? null,
554 'name' => $element['name'] ?? null,
555 'description' => $element['description'] ?? null,
562 if (!empty($summary)) {
563 $item["body"] = '[abstract]' . HTML::toBBCode($summary, $basepath) . "[/abstract]\n" . $item["body"];
566 if (!empty($contact["fetch_further_information"]) && ($contact["fetch_further_information"] == 3)) {
567 if (empty($taglist)) {
568 $taglist = PageInfo::getTagsFromUrl($item["plink"], $preview, $contact["ffi_keyword_denylist"] ?? '');
570 $item["body"] .= "\n" . self::tagToString($taglist);
575 // Add the link to the original feed entry if not present in feed
576 if (($item['plink'] != '') && !strstr($item["body"], $item['plink']) && !in_array($item['plink'], array_column($attachments, 'url'))) {
577 $item["body"] .= "[hr][url]" . $item['plink'] . "[/url]";
581 if (empty($item['title'])) {
582 $item['post-type'] = Item::PT_NOTE;
585 Logger::info('Stored feed', ['item' => $item]);
587 $notify = Item::isRemoteSelf($contact, $item);
589 // Distributed items should have a well formatted URI.
590 // Additionally we have to avoid conflicts with identical URI between imported feeds and these items.
592 $item['guid'] = Item::guidFromUri($orig_plink, DI::baseUrl()->getHostname());
593 $item['uri'] = Item::newURI($item['uid'], $item['guid']);
594 unset($item['thr-parent']);
595 unset($item['parent-uri']);
597 // Set the delivery priority for "remote self" to "medium"
598 $notify = PRIORITY_MEDIUM;
601 $condition = ['uid' => $item['uid'], 'uri' => $item['uri']];
602 if (!Post::exists($condition) && !Post\Delayed::exists($item["uri"], $item['uid'])) {
604 Post\Delayed::publish($item, $notify, $taglist, $attachments);
606 $postings[] = ['item' => $item, 'notify' => $notify,
607 'taglist' => $taglist, 'attachments' => $attachments];
610 Logger::info('Post already created or exists in the delayed posts queue', ['uid' => $item['uid'], 'uri' => $item["uri"]]);
614 if (!empty($postings)) {
615 $min_posting = DI::config()->get('system', 'minimum_posting_interval', 0);
616 $total = count($postings);
618 // Posts shouldn't be delayed more than a day
619 $interval = min(1440, self::getPollInterval($contact));
620 $delay = max(round(($interval * 60) / $total), 60 * $min_posting);
621 Logger::info('Got posting delay', ['delay' => $delay, 'interval' => $interval, 'items' => $total, 'cid' => $contact['id'], 'url' => $contact['url']]);
628 foreach ($postings as $posting) {
630 $publish_time = time() + $post_delay;
631 $post_delay += $delay;
633 $publish_time = time();
636 $last_publish = DI::pConfig()->get($posting['item']['uid'], 'system', 'last_publish', 0, true);
637 $next_publish = max($last_publish + (60 * $min_posting), time());
638 if ($publish_time < $next_publish) {
639 $publish_time = $next_publish;
641 $publish_at = date(DateTimeFormat::MYSQL, $publish_time);
643 if (Post\Delayed::add($posting['item']['uri'], $posting['item'], $posting['notify'], Post\Delayed::PREPARED, $publish_at, $posting['taglist'], $posting['attachments'])) {
644 DI::pConfig()->set($item['uid'], 'system', 'last_publish', $publish_time);
649 if (!$dryRun && DI::config()->get('system', 'adjust_poll_frequency')) {
650 self::adjustPollFrequency($contact, $creation_dates);
653 return ["header" => $author, "items" => $items];
657 * Automatically adjust the poll frequency according to the post frequency
659 * @param array $contact
660 * @param array $creation_dates
663 private static function adjustPollFrequency(array $contact, array $creation_dates)
665 if ($contact['network'] != Protocol::FEED) {
666 Logger::info('Contact is no feed, skip.', ['id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url'], 'network' => $contact['network']]);
670 if (!empty($creation_dates)) {
671 // Count the post frequency and the earliest and latest post date
675 $oldest_date = $newest_date = '';
677 foreach ($creation_dates as $date) {
678 $timestamp = strtotime($date);
679 $day = intdiv($timestamp, 86400);
680 $hour = $timestamp % 86400;
682 // Only have a look at values from the last seven days
683 if (((time() / 86400) - $day) < 7) {
684 if (empty($frequency[$day])) {
685 $frequency[$day] = ['count' => 1, 'low' => $hour, 'high' => $hour];
687 ++$frequency[$day]['count'];
688 if ($frequency[$day]['low'] > $hour) {
689 $frequency[$day]['low'] = $hour;
691 if ($frequency[$day]['high'] < $hour) {
692 $frequency[$day]['high'] = $hour;
696 if ($oldest > $day) {
698 $oldest_date = $date;
701 if ($newest < $day) {
703 $newest_date = $date;
707 if (count($creation_dates) == 1) {
708 Logger::info('Feed had posted a single time, switching to daily polling', ['newest' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
709 $priority = 8; // Poll once a day
712 if (empty($priority) && (((time() / 86400) - $newest) > 730)) {
713 Logger::info('Feed had not posted for two years, switching to monthly polling', ['newest' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
714 $priority = 10; // Poll every month
717 if (empty($priority) && (((time() / 86400) - $newest) > 365)) {
718 Logger::info('Feed had not posted for a year, switching to weekly polling', ['newest' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
719 $priority = 9; // Poll every week
722 if (empty($priority) && empty($frequency)) {
723 Logger::info('Feed had not posted for at least a week, switching to daily polling', ['newest' => $newest_date, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
724 $priority = 8; // Poll once a day
727 if (empty($priority)) {
728 // Calculate the highest "posts per day" value
730 foreach ($frequency as $entry) {
731 if (($entry['count'] == 1) || ($entry['high'] == $entry['low'])) {
735 // We take the earliest and latest post day and interpolate the number of post per day
736 // that would had been created with this post frequency
738 // Assume at least four hours between oldest and newest post per day - should be okay for news outlets
739 $duration = max($entry['high'] - $entry['low'], 14400);
740 $ppd = (86400 / $duration) * $entry['count'];
746 $priority = 1; // Poll every quarter hour
747 } elseif ($max > 24) {
748 $priority = 2; // Poll half an hour
749 } elseif ($max > 12) {
750 $priority = 3; // Poll hourly
751 } elseif ($max > 8) {
752 $priority = 4; // Poll every two hours
753 } elseif ($max > 4) {
754 $priority = 5; // Poll every three hours
755 } elseif ($max > 2) {
756 $priority = 6; // Poll every six hours
758 $priority = 7; // Poll twice a day
760 Logger::info('Calculated priority by the posts per day', ['priority' => $priority, 'max' => round($max, 2), 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
763 Logger::info('No posts, switching to daily polling', ['id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
764 $priority = 8; // Poll once a day
767 if ($contact['rating'] != $priority) {
768 Logger::notice('Adjusting priority', ['old' => $contact['rating'], 'new' => $priority, 'id' => $contact['id'], 'uid' => $contact['uid'], 'url' => $contact['url']]);
769 Contact::update(['rating' => $priority], ['id' => $contact['id']]);
774 * Get the poll interval for the given contact array
776 * @param array $contact
777 * @return int Poll interval in minutes
779 public static function getPollInterval(array $contact)
781 if (in_array($contact['network'], [Protocol::MAIL, Protocol::FEED])) {
782 $ratings = [0, 3, 7, 8, 9, 10];
783 if (DI::config()->get('system', 'adjust_poll_frequency') && ($contact['network'] == Protocol::FEED)) {
784 $rating = $contact['rating'];
785 } elseif (array_key_exists($contact['priority'], $ratings)) {
786 $rating = $ratings[$contact['priority']];
791 // Check once a week per default for all other networks
795 // Friendica and OStatus are checked once a day
796 if (in_array($contact['network'], [Protocol::DFRN, Protocol::OSTATUS])) {
800 // Check archived contacts or contacts with unsupported protocols once a month
801 if ($contact['archive'] || in_array($contact['network'], [Protocol::ZOT, Protocol::PHANTOM])) {
809 * Based on $contact['priority'], should we poll this site now? Or later?
812 $min_poll_interval = max(1, DI::config()->get('system', 'min_poll_interval'));
814 $poll_intervals = [$min_poll_interval, 15, 30, 60, 120, 180, 360, 720 ,1440, 10080, 43200];
816 //$poll_intervals = [$min_poll_interval . ' minute', '15 minute', '30 minute',
817 // '1 hour', '2 hour', '3 hour', '6 hour', '12 hour' ,'1 day', '1 week', '1 month'];
819 return $poll_intervals[$rating];
823 * Convert a tag array to a tag string
826 * @return string tag string
828 private static function tagToString(array $tags)
832 foreach ($tags as $tag) {
837 $tagstr .= "#[url=" . DI::baseUrl() . "/search?tag=" . urlencode($tag) . "]" . $tag . "[/url]";
843 private static function titleIsBody($title, $body)
845 $title = strip_tags($title);
846 $title = trim($title);
847 $title = html_entity_decode($title, ENT_QUOTES, 'UTF-8');
848 $title = str_replace(["\n", "\r", "\t", " "], ["", "", "", ""], $title);
850 $body = strip_tags($body);
852 $body = html_entity_decode($body, ENT_QUOTES, 'UTF-8');
853 $body = str_replace(["\n", "\r", "\t", " "], ["", "", "", ""], $body);
855 if (strlen($title) < strlen($body)) {
856 $body = substr($body, 0, strlen($title));
859 if (($title != $body) && (substr($title, -3) == "...")) {
860 $pos = strrpos($title, "...");
862 $title = substr($title, 0, $pos);
863 $body = substr($body, 0, $pos);
866 return ($title == $body);
870 * Creates the Atom feed for a given nickname
873 * - activity (default): all the public posts
874 * - posts: all the public top-level posts
875 * - comments: all the public replies
877 * Updates the provided last_update parameter if the result comes from the
878 * cache or it is empty
880 * @param string $owner_nick Nickname of the feed owner
881 * @param string $last_update Date of the last update
882 * @param integer $max_items Number of maximum items to fetch
883 * @param string $filter Feed items filter (activity, posts or comments)
884 * @param boolean $nocache Wether to bypass caching
886 * @return string Atom feed
887 * @throws \Friendica\Network\HTTPException\InternalServerErrorException
888 * @throws \ImagickException
890 public static function atom($owner_nick, $last_update, $max_items = 300, $filter = 'activity', $nocache = false)
892 $stamp = microtime(true);
894 $owner = User::getOwnerDataByNick($owner_nick);
899 $cachekey = "feed:feed:" . $owner_nick . ":" . $filter . ":" . $last_update;
901 $previous_created = $last_update;
903 // Don't cache when the last item was posted less then 15 minutes ago (Cache duration)
904 if ((time() - strtotime($owner['last-item'])) < 15*60) {
905 $result = DI::cache()->get($cachekey);
906 if (!$nocache && !is_null($result)) {
907 Logger::info('Cached feed duration', ['seconds' => number_format(microtime(true) - $stamp, 3), 'nick' => $owner_nick, 'filter' => $filter, 'created' => $previous_created]);
908 return $result['feed'];
912 $check_date = empty($last_update) ? '' : DateTimeFormat::utc($last_update);
913 $authorid = Contact::getIdForURL($owner["url"]);
915 $condition = ["`uid` = ? AND `received` > ? AND NOT `deleted` AND `gravity` IN (?, ?)
916 AND `private` != ? AND `visible` AND `wall` AND `parent-network` IN (?, ?, ?, ?)",
917 $owner["uid"], $check_date, GRAVITY_PARENT, GRAVITY_COMMENT,
918 Item::PRIVATE, Protocol::ACTIVITYPUB,
919 Protocol::OSTATUS, Protocol::DFRN, Protocol::DIASPORA];
921 if ($filter === 'comments') {
922 $condition[0] .= " AND `gravity` = ? ";
923 $condition[] = GRAVITY_COMMENT;
926 if ($owner['account-type'] != User::ACCOUNT_TYPE_COMMUNITY) {
927 $condition[0] .= " AND `contact-id` = ? AND `author-id` = ?";
928 $condition[] = $owner["id"];
929 $condition[] = $authorid;
932 $params = ['order' => ['received' => true], 'limit' => $max_items];
934 if ($filter === 'posts') {
935 $ret = Post::selectThread(Item::DELIVER_FIELDLIST, $condition, $params);
937 $ret = Post::select(Item::DELIVER_FIELDLIST, $condition, $params);
940 $items = Post::toArray($ret);
942 $doc = new DOMDocument('1.0', 'utf-8');
943 $doc->formatOutput = true;
945 $root = self::addHeader($doc, $owner, $filter);
947 foreach ($items as $item) {
948 $entry = self::noteEntry($doc, $item, $owner);
949 $root->appendChild($entry);
951 if ($last_update < $item['created']) {
952 $last_update = $item['created'];
956 $feeddata = trim($doc->saveXML());
958 $msg = ['feed' => $feeddata, 'last_update' => $last_update];
959 DI::cache()->set($cachekey, $msg, Duration::QUARTER_HOUR);
961 Logger::info('Feed duration', ['seconds' => number_format(microtime(true) - $stamp, 3), 'nick' => $owner_nick, 'filter' => $filter, 'created' => $previous_created]);
967 * Adds the header elements to the XML document
969 * @param DOMDocument $doc XML document
970 * @param array $owner Contact data of the poster
971 * @param string $filter The related feed filter (activity, posts or comments)
973 * @return object header root element
974 * @throws \Friendica\Network\HTTPException\InternalServerErrorException
976 private static function addHeader(DOMDocument $doc, array $owner, $filter)
978 $root = $doc->createElementNS(ActivityNamespace::ATOM1, 'feed');
979 $doc->appendChild($root);
982 $selfUri = '/feed/' . $owner["nick"] . '/';
985 $title = DI::l10n()->t('%s\'s timeline', $owner['name']);
989 $title = DI::l10n()->t('%s\'s posts', $owner['name']);
992 $title = DI::l10n()->t('%s\'s comments', $owner['name']);
997 $attributes = ["uri" => "https://friendi.ca", "version" => FRIENDICA_VERSION . "-" . DB_UPDATE_VERSION];
998 XML::addElement($doc, $root, "generator", FRIENDICA_PLATFORM, $attributes);
999 XML::addElement($doc, $root, "id", DI::baseUrl() . "/profile/" . $owner["nick"]);
1000 XML::addElement($doc, $root, "title", $title);
1001 XML::addElement($doc, $root, "subtitle", sprintf("Updates from %s on %s", $owner["name"], DI::config()->get('config', 'sitename')));
1002 XML::addElement($doc, $root, "logo", User::getAvatarUrl($owner, Proxy::SIZE_SMALL));
1003 XML::addElement($doc, $root, "updated", DateTimeFormat::utcNow(DateTimeFormat::ATOM));
1005 $author = self::addAuthor($doc, $owner);
1006 $root->appendChild($author);
1008 $attributes = ["href" => $owner["url"], "rel" => "alternate", "type" => "text/html"];
1009 XML::addElement($doc, $root, "link", "", $attributes);
1011 OStatus::hublinks($doc, $root, $owner["nick"]);
1013 $attributes = ["href" => DI::baseUrl() . $selfUri, "rel" => "self", "type" => "application/atom+xml"];
1014 XML::addElement($doc, $root, "link", "", $attributes);
1020 * Adds the author element to the XML document
1022 * @param DOMDocument $doc XML document
1023 * @param array $owner Contact data of the poster
1025 * @return \DOMElement author element
1026 * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1028 private static function addAuthor(DOMDocument $doc, array $owner)
1030 $author = $doc->createElement("author");
1031 XML::addElement($doc, $author, "uri", $owner["url"]);
1032 XML::addElement($doc, $author, "name", $owner["nick"]);
1033 XML::addElement($doc, $author, "email", $owner["addr"]);
1039 * Adds a regular entry element
1041 * @param DOMDocument $doc XML document
1042 * @param array $item Data of the item that is to be posted
1043 * @param array $owner Contact data of the poster
1044 * @param bool $toplevel Is it for en entry element (false) or a feed entry (true)?
1046 * @return \DOMElement Entry element
1047 * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1048 * @throws \ImagickException
1050 private static function noteEntry(DOMDocument $doc, array $item, array $owner)
1052 if (($item['gravity'] != GRAVITY_PARENT) && (Strings::normaliseLink($item["author-link"]) != Strings::normaliseLink($owner["url"]))) {
1053 Logger::info('Feed entry author does not match feed owner', ['owner' => $owner["url"], 'author' => $item["author-link"]]);
1056 $entry = OStatus::entryHeader($doc, $owner, $item, false);
1058 self::entryContent($doc, $entry, $item, self::getTitle($item), '', true);
1060 self::entryFooter($doc, $entry, $item, $owner);
1066 * Adds elements to the XML document
1068 * @param DOMDocument $doc XML document
1069 * @param \DOMElement $entry Entry element where the content is added
1070 * @param array $item Data of the item that is to be posted
1071 * @param array $owner Contact data of the poster
1072 * @param string $title Title for the post
1073 * @param string $verb The activity verb
1074 * @param bool $complete Add the "status_net" element?
1075 * @param bool $feed_mode Behave like a regular feed for users if true
1077 * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1079 private static function entryContent(DOMDocument $doc, \DOMElement $entry, array $item, $title, $verb = "", $complete = true)
1082 $verb = OStatus::constructVerb($item);
1085 XML::addElement($doc, $entry, "id", $item["uri"]);
1086 XML::addElement($doc, $entry, "title", html_entity_decode($title, ENT_QUOTES, 'UTF-8'));
1088 $body = OStatus::formatPicturePost($item['body'], $item['uri-id']);
1090 $body = BBCode::convertForUriId($item['uri-id'], $body, BBCode::ACTIVITYPUB);
1092 XML::addElement($doc, $entry, "content", $body, ["type" => "html"]);
1094 XML::addElement($doc, $entry, "link", "", ["rel" => "alternate", "type" => "text/html",
1095 "href" => DI::baseUrl()."/display/".$item["guid"]]
1098 XML::addElement($doc, $entry, "published", DateTimeFormat::utc($item["created"]."+00:00", DateTimeFormat::ATOM));
1099 XML::addElement($doc, $entry, "updated", DateTimeFormat::utc($item["edited"]."+00:00", DateTimeFormat::ATOM));
1103 * Adds the elements at the foot of an entry to the XML document
1105 * @param DOMDocument $doc XML document
1106 * @param object $entry The entry element where the elements are added
1107 * @param array $item Data of the item that is to be posted
1108 * @param array $owner Contact data of the poster
1109 * @param bool $complete default true
1111 * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1113 private static function entryFooter(DOMDocument $doc, $entry, array $item, array $owner)
1117 if ($item['gravity'] != GRAVITY_PARENT) {
1118 $parent = Post::selectFirst(['guid', 'author-link', 'owner-link'], ['id' => $item['parent']]);
1120 $thrparent = Post::selectFirst(['guid', 'author-link', 'owner-link', 'plink'], ['uid' => $owner['uid'], 'uri' => $item['thr-parent']]);
1122 if (DBA::isResult($thrparent)) {
1123 $mentioned[$thrparent['author-link']] = $thrparent['author-link'];
1124 $mentioned[$thrparent['owner-link']] = $thrparent['owner-link'];
1125 $parent_plink = $thrparent['plink'];
1126 } elseif (DBA::isResult($parent)) {
1127 $mentioned[$parent['author-link']] = $parent['author-link'];
1128 $mentioned[$parent['owner-link']] = $parent['owner-link'];
1129 $parent_plink = DI::baseUrl() . '/display/' . $parent['guid'];
1131 DI::logger()->notice('Missing parent and thr-parent for child item', ['item' => $item]);
1134 if (isset($parent_plink)) {
1136 'ref' => $item['thr-parent'],
1137 'href' => $parent_plink];
1138 XML::addElement($doc, $entry, 'thr:in-reply-to', '', $attributes);
1142 'href' => $parent_plink];
1143 XML::addElement($doc, $entry, 'link', '', $attributes);
1147 // uri-id isn't present for follow entry pseudo-items
1148 $tags = Tag::getByURIId($item['uri-id'] ?? 0);
1149 foreach ($tags as $tag) {
1150 $mentioned[$tag['url']] = $tag['url'];
1153 foreach ($tags as $tag) {
1154 if ($tag['type'] == Tag::HASHTAG) {
1155 XML::addElement($doc, $entry, 'category', '', ['term' => $tag['name']]);
1159 OStatus::getAttachment($doc, $entry, $item);
1163 * Fetch or create title for feed entry
1165 * @param array $item
1166 * @return string title
1168 private static function getTitle(array $item)
1170 if ($item['title'] != '') {
1171 return BBCode::convertForUriId($item['uri-id'], $item['title'], BBCode::ACTIVITYPUB);
1174 // Fetch information about the post
1175 $siteinfo = BBCode::getAttachedData($item["body"]);
1176 if (isset($siteinfo["title"])) {
1177 return $siteinfo["title"];
1180 // If no bookmark is found then take the first line
1181 // Remove the share element before fetching the first line
1182 $title = trim(preg_replace("/\[share.*?\](.*?)\[\/share\]/ism","\n$1\n",$item['body']));
1184 $title = BBCode::toPlaintext($title)."\n";
1185 $pos = strpos($title, "\n");
1187 if (($pos == 0) || ($pos > 100)) {
1192 return substr($title, 0, $pos) . $trailer;