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\Model;
24 use Friendica\Content\Text\BBCode;
25 use Friendica\Core\Cache\Enum\Duration;
26 use Friendica\Core\Logger;
27 use Friendica\Core\Protocol;
28 use Friendica\Core\System;
29 use Friendica\Database\Database;
30 use Friendica\Database\DBA;
32 use Friendica\Protocol\ActivityPub;
33 use Friendica\Util\DateTimeFormat;
34 use Friendica\Util\Strings;
39 * This Model class handles tag table interactions.
40 * This tables stores relevant tags related to posts, like hashtags and mentions.
48 * An implicit mention is a mention in a comment body that is redundant with the threading information.
50 const IMPLICIT_MENTION = 8;
52 * An exclusive mention transmits the post only to the target account without transmitting it to the followers, usually a forum.
54 const EXCLUSIVE_MENTION = 9;
62 const GENERAL_COLLECTION = 2;
63 const FOLLOWER_COLLECTION = 3;
64 const PUBLIC_COLLECTION = 4;
66 const TAG_CHARACTER = [
69 self::EXCLUSIVE_MENTION => '!',
70 self::IMPLICIT_MENTION => '%',
74 * Store tag/mention elements
76 * @param integer $uriid
77 * @param integer $type
80 * @param integer $target
82 public static function store(int $uriid, int $type, string $name, string $url = '', int $target = null)
84 if ($type == self::HASHTAG) {
85 // Trim Unicode non-word characters
86 $name = preg_replace('/(^\W+)|(\W+$)/us', '', $name);
88 $tags = explode(self::TAG_CHARACTER[self::HASHTAG], $name);
89 if (count($tags) > 1) {
90 foreach ($tags as $tag) {
91 self::store($uriid, $type, $tag, $url);
104 if (in_array($type, [self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION, self::TO, self::CC, self::BTO, self::BCC])) {
106 // No mention without a contact url
110 if ((substr($url, 0, 7) == 'https//') || (substr($url, 0, 6) == 'http//')) {
111 Logger::notice('Wrong scheme in url', ['url' => $url, 'callstack' => System::callstack(20)]);
114 $cid = Contact::getIdForURL($url, 0, false);
115 Logger::debug('Got id for contact', ['cid' => $cid, 'url' => $url]);
118 $tag = DBA::selectFirst('tag', ['name', 'type'], ['url' => $url]);
120 if ($tag['name'] != substr($name, 0, 96)) {
121 DBA::update('tag', ['name' => substr($name, 0, 96)], ['url' => $url]);
123 if (!empty($target) && ($tag['type'] != $target)) {
124 DBA::update('tag', ['type' => $target], ['url' => $url]);
131 if (!in_array($type, [self::TO, self::CC, self::BTO, self::BCC])) {
132 if (($type != self::HASHTAG) && !empty($url) && ($url != $name)) {
133 $url = strtolower($url);
139 $tagid = self::getID($name, $url, $target);
145 $fields = ['uri-id' => $uriid, 'type' => $type, 'tid' => $tagid, 'cid' => $cid];
147 if (in_array($type, [self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION])) {
148 $condition = $fields;
149 $condition['type'] = [self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION];
150 if (DBA::exists('post-tag', $condition)) {
151 Logger::info('Tag already exists', $fields);
156 DBA::insert('post-tag', $fields, Database::INSERT_IGNORE);
158 Logger::info('Stored tag/mention', ['uri-id' => $uriid, 'tag-id' => $tagid, 'contact-id' => $cid, 'name' => $name, 'type' => $type, 'callstack' => System::callstack(8)]);
162 * Fetch the target type for the given url
165 * @param bool $fetch Fetch information via network operations
168 public static function getTargetType(string $url, bool $fetch = true)
176 $tag = DBA::selectFirst('tag', ['url', 'type'], ['url' => $url]);
177 if (!empty($tag['type'])) {
178 $target = $tag['type'];
179 if ($target != self::GENERAL_COLLECTION) {
180 Logger::debug('Found existing type', ['type' => $tag['type'], 'url' => $url]);
185 if ($url == ActivityPub::PUBLIC_COLLECTION) {
186 $target = self::PUBLIC_COLLECTION;
187 Logger::debug('Public collection', ['url' => $url]);
189 if (DBA::exists('apcontact', ['followers' => $url])) {
190 $target = self::FOLLOWER_COLLECTION;
191 Logger::debug('Found collection via existing apcontact', ['url' => $url]);
192 } elseif (Contact::getIdForURL($url, 0, $fetch ? null : false)) {
193 $target = self::ACCOUNT;
194 Logger::debug('URL is an account', ['url' => $url]);
195 } elseif ($fetch && ($target != self::GENERAL_COLLECTION)) {
196 $content = ActivityPub::fetchContent($url);
197 if (!empty($content['type']) && ($content['type'] == 'OrderedCollection')) {
198 $target = self::GENERAL_COLLECTION;
199 Logger::debug('URL is an ordered collection', ['url' => $url]);
204 if (!empty($target) && !empty($tag['url']) && ($tag['type'] != $target)) {
205 DBA::update('tag', ['type' => $target], ['url' => $url]);
208 if (empty($target)) {
209 Logger::debug('No type could be detected', ['url' => $url]);
216 * Get a tag id for a given tag name and url
218 * @param string $name
223 public static function getID(string $name, string $url = '', int $type = null)
225 $fields = ['name' => substr($name, 0, 96), 'url' => $url];
227 $tag = DBA::selectFirst('tag', ['id', 'type'], $fields);
228 if (DBA::isResult($tag)) {
229 if (empty($tag['type']) && !empty($type)) {
230 DBA::update('tag', ['type' => $type], $fields);
236 $fields['type'] = $type;
239 DBA::insert('tag', $fields, Database::INSERT_IGNORE);
240 $tid = DBA::lastInsertId();
245 Logger::error('No tag id created', $fields);
250 * Store tag/mention elements
252 * @param integer $uriid
253 * @param string $hash
254 * @param string $name
256 * @param boolean $probing
258 public static function storeByHash(int $uriid, string $hash, string $name, string $url = '', $probing = true)
260 $type = self::getTypeForHash($hash);
261 if ($type == self::UNKNOWN) {
265 self::store($uriid, $type, $name, $url, $probing);
269 * Get tags and mentions from the body
271 * @param string $body Body of the post
272 * @param string $tags Accepted tags
274 * @return array Tag list
276 public static function getFromBody(string $body, string $tags = null)
278 if (is_null($tags)) {
279 $tags = self::TAG_CHARACTER[self::HASHTAG] . self::TAG_CHARACTER[self::MENTION] . self::TAG_CHARACTER[self::EXCLUSIVE_MENTION];
282 if (!preg_match_all("/([" . $tags . "])\[url\=([^\[\]]*)\]([^\[\]]*)\[\/url\]/ism", $body, $result, PREG_SET_ORDER)) {
290 * Store tags and mentions from the body
292 * @param integer $uriid URI-Id
293 * @param string $body Body of the post
294 * @param string $tags Accepted tags
295 * @param boolean $probing Perform a probing for contacts, adding them if needed
297 public static function storeFromBody(int $uriid, string $body, string $tags = null, $probing = true)
299 Logger::info('Check for tags', ['uri-id' => $uriid, 'hash' => $tags, 'callstack' => System::callstack()]);
301 if (is_null($tags)) {
302 $tags = self::TAG_CHARACTER[self::HASHTAG] . self::TAG_CHARACTER[self::MENTION] . self::TAG_CHARACTER[self::EXCLUSIVE_MENTION];
305 // Only remove the shared data from "real" reshares
306 $shared = BBCode::fetchShareAttributes($body);
307 if (!empty($shared['guid'])) {
308 if (preg_match("/\s*\[share .*?\](.*?)\[\/share\]\s*/ism", $body, $matches)) {
309 $share_body = $matches[1];
311 $body = preg_replace("/\s*\[share .*?\].*?\[\/share\]\s*/ism", '', $body);
314 foreach (self::getFromBody($body, $tags) as $tag) {
315 self::storeByHash($uriid, $tag[1], $tag[3], $tag[2], $probing);
318 // Search for hashtags in the shared body (but only if hashtags are wanted)
319 if (!empty($share_body) && (strpos($tags, self::TAG_CHARACTER[self::HASHTAG]) !== false)) {
320 foreach (self::getFromBody($share_body, self::TAG_CHARACTER[self::HASHTAG]) as $tag) {
321 self::storeByHash($uriid, $tag[1], $tag[3], $tag[2], $probing);
327 * Store raw tags (not encapsulated in links) from the body
328 * This function is needed in the intermediate phase.
329 * Later we can call item::setHashtags in advance to have all tags converted.
331 * @param integer $uriid URI-Id
332 * @param string $body Body of the post
334 public static function storeRawTagsFromBody(int $uriid, string $body)
336 Logger::info('Check for tags', ['uri-id' => $uriid, 'callstack' => System::callstack()]);
338 $result = BBCode::getTags($body);
339 if (empty($result)) {
343 Logger::info('Found tags', ['uri-id' => $uriid, 'result' => $result]);
345 foreach ($result as $tag) {
346 if (substr($tag, 0, 1) != self::TAG_CHARACTER[self::HASHTAG]) {
349 self::storeByHash($uriid, substr($tag, 0, 1), substr($tag, 1));
354 * Checks for stored hashtags and mentions for the given post
356 * @param integer $uriid
359 public static function existsForPost(int $uriid)
361 return DBA::exists('post-tag', ['uri-id' => $uriid, 'type' => [self::HASHTAG, self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION]]);
367 * @param integer $uriid
368 * @param integer $type
369 * @param string $name
372 public static function remove(int $uriid, int $type, string $name, string $url = '')
374 $condition = ['uri-id' => $uriid, 'type' => $type, 'url' => $url];
375 if ($type == self::HASHTAG) {
376 $condition['name'] = $name;
379 $tag = DBA::selectFirst('tag-view', ['tid', 'cid'], $condition);
380 if (!DBA::isResult($tag)) {
384 Logger::info('Removing tag/mention', ['uri-id' => $uriid, 'tid' => $tag['tid'], 'name' => $name, 'url' => $url, 'callstack' => System::callstack(8)]);
385 DBA::delete('post-tag', ['uri-id' => $uriid, 'type' => $type, 'tid' => $tag['tid'], 'cid' => $tag['cid']]);
391 * @param integer $uriid
392 * @param string $hash
393 * @param string $name
396 public static function removeByHash(int $uriid, string $hash, string $name, string $url = '')
398 $type = self::getTypeForHash($hash);
399 if ($type == self::UNKNOWN) {
403 self::remove($uriid, $type, $name, $url);
407 * Get the type for the given hash
409 * @param string $hash
410 * @return integer type
412 private static function getTypeForHash(string $hash)
414 if ($hash == self::TAG_CHARACTER[self::MENTION]) {
415 return self::MENTION;
416 } elseif ($hash == self::TAG_CHARACTER[self::EXCLUSIVE_MENTION]) {
417 return self::EXCLUSIVE_MENTION;
418 } elseif ($hash == self::TAG_CHARACTER[self::IMPLICIT_MENTION]) {
419 return self::IMPLICIT_MENTION;
420 } elseif ($hash == self::TAG_CHARACTER[self::HASHTAG]) {
421 return self::HASHTAG;
423 return self::UNKNOWN;
428 * Create implicit mentions for a given post
430 * @param integer $uri_id
431 * @param integer $parent_uri_id
433 public static function createImplicitMentions(int $uri_id, int $parent_uri_id)
435 // Always mention the direct parent author
436 $parent = Post::selectFirst(['author-link', 'author-name'], ['uri-id' => $parent_uri_id]);
437 self::store($uri_id, self::IMPLICIT_MENTION, $parent['author-name'], $parent['author-link']);
439 if (DI::config()->get('system', 'disable_implicit_mentions')) {
443 $tags = DBA::select('tag-view', ['name', 'url'], ['uri-id' => $parent_uri_id, 'type' => [self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION]]);
444 while ($tag = DBA::fetch($tags)) {
445 self::store($uri_id, self::IMPLICIT_MENTION, $tag['name'], $tag['url']);
451 * Retrieves the terms from the provided type(s) associated with the provided item ID.
453 * @param int $item_id
454 * @param int|array $type
458 public static function getByURIId(int $uri_id, array $type = [self::HASHTAG, self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION])
460 $condition = ['uri-id' => $uri_id, 'type' => $type];
461 return DBA::selectToArray('tag-view', ['type', 'name', 'url', 'tag-type'], $condition);
465 * Return a string with all tags and mentions
467 * @param integer $uri_id
469 * @return string tags and mentions
472 public static function getCSVByURIId(int $uri_id, array $type = [self::HASHTAG, self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION])
475 $tags = self::getByURIId($uri_id, $type);
476 foreach ($tags as $tag) {
477 $tag_list[] = self::TAG_CHARACTER[$tag['type']] . '[url=' . $tag['url'] . ']' . $tag['name'] . '[/url]';
480 return implode(',', $tag_list);
484 * Sorts an item's tags into mentions, hashtags and other tags. Generate personalized URLs by user and modify the
485 * provided item's body with them.
489 * @throws \Friendica\Network\HTTPException\InternalServerErrorException
490 * @throws \ImagickException
492 public static function populateFromItem(&$item)
498 'implicit_mentions' => [],
501 $searchpath = DI::baseUrl() . "/search?tag=";
503 $taglist = DBA::select('tag-view', ['type', 'name', 'url', 'cid'],
504 ['uri-id' => $item['uri-id'], 'type' => [self::HASHTAG, self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION]]);
505 while ($tag = DBA::fetch($taglist)) {
506 if ($tag['url'] == '') {
507 $tag['url'] = $searchpath . rawurlencode($tag['name']);
510 $orig_tag = $tag['url'];
512 $prefix = self::TAG_CHARACTER[$tag['type']];
513 switch($tag['type']) {
515 if ($orig_tag != $tag['url']) {
516 $item['body'] = str_replace($orig_tag, $tag['url'], $item['body']);
519 $return['hashtags'][] = '<bdi>' . $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['name']) . '</a></bdi>';
520 $return['tags'][] = '<bdi>' . $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['name']) . '</a></bdi>';
523 case self::EXCLUSIVE_MENTION:
524 if (!empty($tag['cid'])) {
525 $tag['url'] = Contact::magicLinkById($tag['cid']);
527 $tag['url'] = Contact::magicLink($tag['url']);
529 $return['mentions'][] = '<bdi>' . $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['name']) . '</a></bdi>';
530 $return['tags'][] = '<bdi>' . $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['name']) . '</a></bdi>';
532 case self::IMPLICIT_MENTION:
533 $return['implicit_mentions'][] = $prefix . $tag['name'];
537 DBA::close($taglist);
543 * Counts posts for given tag
545 * @param string $search
546 * @param integer $uid
547 * @return integer number of posts
549 public static function countByTag(string $search, int $uid = 0)
551 $condition = ["`name` = ? AND (`uid` = ? OR (`uid` = ? AND NOT `global`))
552 AND (`network` IN (?, ?, ?, ?) OR (`uid` = ? AND `uid` != ?))",
553 $search, 0, $uid, Protocol::ACTIVITYPUB, Protocol::DFRN, Protocol::DIASPORA, Protocol::OSTATUS, $uid, 0];
555 return DBA::count('tag-search-view', $condition);
559 * Search posts for given tag
561 * @param string $search
562 * @param integer $uid
563 * @param integer $start
564 * @param integer $limit
565 * @param integer $last_uriid
566 * @return array with URI-ID
568 public static function getURIIdListByTag(string $search, int $uid = 0, int $start = 0, int $limit = 100, int $last_uriid = 0)
570 $condition = ["`name` = ? AND (`uid` = ? OR (`uid` = ? AND NOT `global`))
571 AND (`network` IN (?, ?, ?, ?) OR (`uid` = ? AND `uid` != ?))",
572 $search, 0, $uid, Protocol::ACTIVITYPUB, Protocol::DFRN, Protocol::DIASPORA, Protocol::OSTATUS, $uid, 0];
574 if (!empty($last_uriid)) {
575 $condition = DBA::mergeConditions($condition, ["`uri-id` < ?", $last_uriid]);
579 'order' => ['uri-id' => true],
580 'limit' => [$start, $limit]
583 $tags = DBA::select('tag-search-view', ['uri-id'], $condition, $params);
586 while ($tag = DBA::fetch($tags)) {
587 $uriids[] = $tag['uri-id'];
595 * Returns a list of the most frequent global hashtags over the given period
597 * @param int $period Period in hours to consider posts
598 * @param int $limit Number of returned tags
602 public static function getGlobalTrendingHashtags(int $period, $limit = 10)
604 $tags = DI::cache()->get('global_trending_tags-' . $period . '-' . $limit);
608 return self::setGlobalTrendingHashtags($period, $limit);
613 * Fetch the blocked tags as SQL
617 private static function getBlockedSQL()
619 $blocked_txt = DI::config()->get('system', 'blocked_tags');
620 if (empty($blocked_txt)) {
624 $blocked = explode(',', $blocked_txt);
625 array_walk($blocked, function(&$value) { $value = "'" . DBA::escape(trim($value)) . "'";});
626 return " AND NOT `name` IN (" . implode(',', $blocked) . ")";
630 * Creates a list of the most frequent global hashtags over the given period
632 * @param int $period Period in hours to consider posts
633 * @param int $limit Number of returned tags
637 public static function setGlobalTrendingHashtags(int $period, int $limit = 10)
639 // Get a uri-id that is at least X hours old.
640 // We use the uri-id in the query for the hash tags since this is much faster
641 $post = Post::selectFirstThread(['uri-id'], ["`uid` = ? AND `received` < ?", 0, DateTimeFormat::utc('now - ' . $period . ' hour')],
642 ['order' => ['received' => true]]);
643 if (empty($post['uri-id'])) {
647 $block_sql = self::getBlockedSQL();
649 $tagsStmt = DBA::p("SELECT `name` AS `term`, COUNT(*) AS `score`, COUNT(DISTINCT(`author-id`)) as `authors`
650 FROM `tag-search-view`
651 WHERE `private` = ? AND `uid` = ? AND `uri-id` > ? $block_sql
652 GROUP BY `term` ORDER BY `authors` DESC, `score` DESC LIMIT ?",
653 Item::PUBLIC, 0, $post['uri-id'], $limit);
655 if (DBA::isResult($tagsStmt)) {
656 $tags = DBA::toArray($tagsStmt);
657 DI::cache()->set('global_trending_tags-' . $period . '-' . $limit, $tags, Duration::DAY);
665 * Returns a list of the most frequent local hashtags over the given period
667 * @param int $period Period in hours to consider posts
668 * @param int $limit Number of returned tags
672 public static function getLocalTrendingHashtags(int $period, $limit = 10)
674 $tags = DI::cache()->get('local_trending_tags-' . $period . '-' . $limit);
678 return self::setLocalTrendingHashtags($period, $limit);
683 * Returns a list of the most frequent local hashtags over the given period
685 * @param int $period Period in hours to consider posts
686 * @param int $limit Number of returned tags
690 public static function setLocalTrendingHashtags(int $period, int $limit = 10)
692 // Get a uri-id that is at least X hours old.
693 // We use the uri-id in the query for the hash tags since this is much faster
694 $post = Post::selectFirstThread(['uri-id'], ["`uid` = ? AND `received` < ?", 0, DateTimeFormat::utc('now - ' . $period . ' hour')],
695 ['order' => ['received' => true]]);
696 if (empty($post['uri-id'])) {
700 $block_sql = self::getBlockedSQL();
702 $tagsStmt = DBA::p("SELECT `name` AS `term`, COUNT(*) AS `score`, COUNT(DISTINCT(`author-id`)) as `authors`
703 FROM `tag-search-view`
704 WHERE `private` = ? AND `wall` AND `origin` AND `uri-id` > ? $block_sql
705 GROUP BY `term` ORDER BY `authors` DESC, `score` DESC LIMIT ?",
706 Item::PUBLIC, $post['uri-id'], $limit);
708 if (DBA::isResult($tagsStmt)) {
709 $tags = DBA::toArray($tagsStmt);
710 DI::cache()->set('local_trending_tags-' . $period . '-' . $limit, $tags, Duration::DAY);
718 * Check if the provided tag is of one of the provided term types.
721 * @param int ...$types
724 public static function isType($tag, ...$types)
727 foreach ($types as $type) {
728 if (array_key_exists($type, self::TAG_CHARACTER)) {
729 $tag_chars[] = self::TAG_CHARACTER[$type];
733 return Strings::startsWithChars($tag, $tag_chars);
737 * Fetch user who subscribed to the given tag
740 * @return array User list
742 private static function getUIDListByTag(string $tag)
745 $searches = DBA::select('search', ['uid'], ['term' => $tag]);
746 while ($search = DBA::fetch($searches)) {
747 $uids[] = $search['uid'];
749 DBA::close($searches);
755 * Fetch user who subscribed to the tags of the given item
757 * @param integer $uri_id
758 * @return array User list
760 public static function getUIDListByURIId(int $uri_id)
763 $tags = self::getByURIId($uri_id, [self::HASHTAG]);
765 foreach ($tags as $tag) {
766 $uids = array_merge($uids, self::getUIDListByTag(self::TAG_CHARACTER[self::HASHTAG] . $tag['name']));
769 return array_unique($uids);