<?php
/**
- * @copyright Copyright (C) 2020, Friendica
+ * @copyright Copyright (C) 2010-2022, the Friendica project
*
* @license GNU AGPL version 3 or any later version
*
namespace Friendica\Model;
use Friendica\Content\Text\BBCode;
-use Friendica\Core\Cache\Duration;
+use Friendica\Core\Cache\Enum\Duration;
use Friendica\Core\Logger;
+use Friendica\Core\Protocol;
use Friendica\Core\System;
+use Friendica\Database\Database;
use Friendica\Database\DBA;
use Friendica\DI;
+use Friendica\Util\DateTimeFormat;
use Friendica\Util\Strings;
/**
const UNKNOWN = 0;
const HASHTAG = 1;
const MENTION = 2;
- const CATEGORY = 3;
- const FILE = 5;
/**
* An implicit mention is a mention in a comment body that is redundant with the threading information.
*/
const IMPLICIT_MENTION = 8;
/**
- * An exclusive mention transfers the ownership of the post to the target account, usually a forum.
+ * An exclusive mention transmits the post only to the target account without transmitting it to the followers, usually a forum.
*/
const EXCLUSIVE_MENTION = 9;
+ const TO = 10;
+ const CC = 11;
+ const BTO = 12;
+ const BCC = 13;
+
const TAG_CHARACTER = [
self::HASHTAG => '#',
self::MENTION => '@',
- self::IMPLICIT_MENTION => '%',
self::EXCLUSIVE_MENTION => '!',
+ self::IMPLICIT_MENTION => '%',
];
/**
* @param integer $type
* @param string $name
* @param string $url
- * @param boolean $probing
*/
- public static function store(int $uriid, int $type, string $name, string $url = '', $probing = true)
+ public static function store(int $uriid, int $type, string $name, string $url = '')
{
if ($type == self::HASHTAG) {
- // Remove some common "garbarge" from tags
- $name = trim($name, "\x00..\x20\xFF#!@,;.:'/?!^°$%".'"');
+ // Trim Unicode non-word characters
+ $name = preg_replace('/(^\W+)|(\W+$)/us', '', $name);
$tags = explode(self::TAG_CHARACTER[self::HASHTAG], $name);
if (count($tags) > 1) {
foreach ($tags as $tag) {
- self::store($uriid, $type, $tag, $url, $probing);
+ self::store($uriid, $type, $tag, $url);
}
return;
}
$cid = 0;
$tagid = 0;
- if (in_array($type, [self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION])) {
+ if (in_array($type, [self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION, self::TO, self::CC, self::BTO, self::BCC])) {
if (empty($url)) {
// No mention without a contact url
return;
}
- if (!$probing) {
- $condition = ['nurl' => Strings::normaliseLink($url), 'uid' => 0, 'deleted' => false];
- $contact = DBA::selectFirst('contact', ['id'], $condition, ['order' => ['id']]);
- if (DBA::isResult($contact)) {
- $cid = $contact['id'];
- Logger::info('Got id for contact url', ['cid' => $cid, 'url' => $url]);
- }
-
- if (empty($cid)) {
- $ssl_url = str_replace('http://', 'https://', $url);
- $condition = ['`alias` IN (?, ?, ?) AND `uid` = ? AND NOT `deleted`', $url, Strings::normaliseLink($url), $ssl_url, 0];
- $contact = DBA::selectFirst('contact', ['id'], $condition, ['order' => ['id']]);
- if (DBA::isResult($contact)) {
- $cid = $contact['id'];
- Logger::info('Got id for contact alias', ['cid' => $cid, 'url' => $url]);
- }
- }
- } else {
- $cid = Contact::getIdForURL($url, 0, true);
- Logger::info('Got id by probing', ['cid' => $cid, 'url' => $url]);
+ if ((substr($url, 0, 7) == 'https//') || (substr($url, 0, 6) == 'http//')) {
+ Logger::notice('Wrong scheme in url', ['url' => $url, 'callstack' => System::callstack(20)]);
}
+ $cid = Contact::getIdForURL($url, 0, false);
+ Logger::debug('Got id for contact', ['cid' => $cid, 'url' => $url]);
+
if (empty($cid)) {
// The contact wasn't found in the system (most likely some dead account)
// We ensure that we only store a single entry by overwriting the previous name
- Logger::info('Contact not found, updating tag', ['url' => $url, 'name' => $name]);
- DBA::update('tag', ['name' => substr($name, 0, 96)], ['url' => $url]);
+ Logger::info('URL is not a known contact, updating tag', ['url' => $url, 'name' => $name]);
+ if (!DBA::exists('tag', ['name' => substr($name, 0, 96), 'url' => $url])) {
+ DBA::update('tag', ['name' => substr($name, 0, 96)], ['url' => $url]);
+ }
}
}
if (empty($cid)) {
- $fields = ['name' => substr($name, 0, 96), 'url' => ''];
-
- if (($type != self::HASHTAG) && !empty($url) && ($url != $name)) {
- $fields['url'] = strtolower($url);
- }
-
- $tag = DBA::selectFirst('tag', ['id'], $fields);
- if (!DBA::isResult($tag)) {
- DBA::insert('tag', $fields, true);
- $tagid = DBA::lastInsertId();
- } else {
- $tagid = $tag['id'];
+ if (!in_array($type, [self::TO, self::CC, self::BTO, self::BCC])) {
+ if (($type != self::HASHTAG) && !empty($url) && ($url != $name)) {
+ $url = strtolower($url);
+ } else {
+ $url = '';
+ }
}
+ $tagid = self::getID($name, $url);
if (empty($tagid)) {
- Logger::error('No tag id created', $fields);
return;
}
}
}
}
- DBA::insert('post-tag', $fields, true);
+ DBA::insert('post-tag', $fields, Database::INSERT_IGNORE);
Logger::info('Stored tag/mention', ['uri-id' => $uriid, 'tag-id' => $tagid, 'contact-id' => $cid, 'name' => $name, 'type' => $type, 'callstack' => System::callstack(8)]);
}
+ /**
+ * Get a tag id for a given tag name and url
+ *
+ * @param string $name
+ * @param string $url
+ * @return void
+ */
+ public static function getID(string $name, string $url = '')
+ {
+ $fields = ['name' => substr($name, 0, 96), 'url' => $url];
+
+ $tag = DBA::selectFirst('tag', ['id'], $fields);
+ if (DBA::isResult($tag)) {
+ return $tag['id'];
+ }
+
+ DBA::insert('tag', $fields, Database::INSERT_IGNORE);
+ $tid = DBA::lastInsertId();
+ if (!empty($tid)) {
+ return $tid;
+ }
+
+ Logger::error('No tag id created', $fields);
+ return 0;
+ }
+
/**
* Store tag/mention elements
*
}
/**
- * Store tags and mentions from the body
+ * Get tags and mentions from the body
*
- * @param integer $uriid URI-Id
* @param string $body Body of the post
* @param string $tags Accepted tags
- * @param boolean $probing Perform a probing for contacts, adding them if needed
+ *
+ * @return array Tag list
*/
- public static function storeFromBody(int $uriid, string $body, string $tags = null, $probing = true)
+ public static function getFromBody(string $body, string $tags = null)
{
if (is_null($tags)) {
$tags = self::TAG_CHARACTER[self::HASHTAG] . self::TAG_CHARACTER[self::MENTION] . self::TAG_CHARACTER[self::EXCLUSIVE_MENTION];
}
+ if (!preg_match_all("/([" . $tags . "])\[url\=([^\[\]]*)\]([^\[\]]*)\[\/url\]/ism", $body, $result, PREG_SET_ORDER)) {
+ return [];
+ }
+
+ return $result;
+ }
+
+ /**
+ * Store tags and mentions from the body
+ *
+ * @param integer $uriid URI-Id
+ * @param string $body Body of the post
+ * @param string $tags Accepted tags
+ * @param boolean $probing Perform a probing for contacts, adding them if needed
+ */
+ public static function storeFromBody(int $uriid, string $body, string $tags = null, $probing = true)
+ {
Logger::info('Check for tags', ['uri-id' => $uriid, 'hash' => $tags, 'callstack' => System::callstack()]);
- if (!preg_match_all("/([" . $tags . "])\[url\=([^\[\]]*)\]([^\[\]]*)\[\/url\]/ism", $body, $result, PREG_SET_ORDER)) {
+ $result = self::getFromBody($body, $tags);
+ if (empty($result)) {
return;
}
*/
public static function existsForPost(int $uriid)
{
- return DBA::exists('post-tag', ['uri-id' => $uriid, 'type' => [self::HASHTAG, self::MENTION, self::IMPLICIT_MENTION, self::EXCLUSIVE_MENTION]]);
+ return DBA::exists('post-tag', ['uri-id' => $uriid, 'type' => [self::HASHTAG, self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION]]);
}
/**
}
}
+ /**
+ * Create implicit mentions for a given post
+ *
+ * @param integer $uri_id
+ * @param integer $parent_uri_id
+ */
+ public static function createImplicitMentions(int $uri_id, int $parent_uri_id)
+ {
+ // Always mention the direct parent author
+ $parent = Post::selectFirst(['author-link', 'author-name'], ['uri-id' => $parent_uri_id]);
+ self::store($uri_id, self::IMPLICIT_MENTION, $parent['author-name'], $parent['author-link']);
+
+ if (DI::config()->get('system', 'disable_implicit_mentions')) {
+ return;
+ }
+
+ $tags = DBA::select('tag-view', ['name', 'url'], ['uri-id' => $parent_uri_id, 'type' => [self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION]]);
+ while ($tag = DBA::fetch($tags)) {
+ self::store($uri_id, self::IMPLICIT_MENTION, $tag['name'], $tag['url']);
+ }
+ DBA::close($tags);
+ }
+
/**
* Retrieves the terms from the provided type(s) associated with the provided item ID.
*
* @return array
* @throws \Exception
*/
- public static function getByURIId(int $uri_id, array $type = [self::HASHTAG, self::MENTION, self::IMPLICIT_MENTION, self::EXCLUSIVE_MENTION])
+ public static function getByURIId(int $uri_id, array $type = [self::HASHTAG, self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION])
{
$condition = ['uri-id' => $uri_id, 'type' => $type];
- $tags = DBA::select('tag-view', ['type', 'name', 'url'], $condition);
- if (!DBA::isResult($tags)) {
- return [];
- }
+ return DBA::selectToArray('tag-view', ['type', 'name', 'url'], $condition);
+ }
+ /**
+ * Return a string with all tags and mentions
+ *
+ * @param integer $uri_id
+ * @param array $type
+ * @return string tags and mentions
+ * @throws \Exception
+ */
+ public static function getCSVByURIId(int $uri_id, array $type = [self::HASHTAG, self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION])
+ {
$tag_list = [];
- while ($tag = DBA::fetch($tags)) {
- $tag['term'] = $tag['name']; /// @todo Remove this line when all occurrences of "term" had been replaced with "name"
- $tag_list[] = $tag;
+ $tags = self::getByURIId($uri_id, $type);
+ foreach ($tags as $tag) {
+ $tag_list[] = self::TAG_CHARACTER[$tag['type']] . '[url=' . $tag['url'] . ']' . $tag['name'] . '[/url]';
}
- return $tag_list;
+ return implode(',', $tag_list);
}
/**
$searchpath = DI::baseUrl() . "/search?tag=";
- $taglist = DBA::select('tag-view', ['type', 'name', 'url'],
+ $taglist = DBA::select('tag-view', ['type', 'name', 'url', 'cid'],
['uri-id' => $item['uri-id'], 'type' => [self::HASHTAG, self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION]]);
while ($tag = DBA::fetch($taglist)) {
if ($tag['url'] == '') {
$item['body'] = str_replace($orig_tag, $tag['url'], $item['body']);
}
- $return['hashtags'][] = $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['name']) . '</a>';
- $return['tags'][] = $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['name']) . '</a>';
+ $return['hashtags'][] = '<bdi>' . $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['name']) . '</a></bdi>';
+ $return['tags'][] = '<bdi>' . $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['name']) . '</a></bdi>';
break;
case self::MENTION:
case self::EXCLUSIVE_MENTION:
+ if (!empty($tag['cid'])) {
+ $tag['url'] = Contact::magicLinkById($tag['cid']);
+ } else {
$tag['url'] = Contact::magicLink($tag['url']);
- $return['mentions'][] = $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['name']) . '</a>';
- $return['tags'][] = $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['name']) . '</a>';
+ }
+ $return['mentions'][] = '<bdi>' . $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['name']) . '</a></bdi>';
+ $return['tags'][] = '<bdi>' . $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['name']) . '</a></bdi>';
break;
case self::IMPLICIT_MENTION:
$return['implicit_mentions'][] = $prefix . $tag['name'];
return $return;
}
+ /**
+ * Counts posts for given tag
+ *
+ * @param string $search
+ * @param integer $uid
+ * @return integer number of posts
+ */
+ public static function countByTag(string $search, int $uid = 0)
+ {
+ $condition = ["`name` = ? AND (`uid` = ? OR (`uid` = ? AND NOT `global`))
+ AND (`network` IN (?, ?, ?, ?) OR (`uid` = ? AND `uid` != ?))",
+ $search, 0, $uid, Protocol::ACTIVITYPUB, Protocol::DFRN, Protocol::DIASPORA, Protocol::OSTATUS, $uid, 0];
+
+ return DBA::count('tag-search-view', $condition);
+ }
+
/**
* Search posts for given tag
*
* @param integer $uid
* @param integer $start
* @param integer $limit
+ * @param integer $last_uriid
* @return array with URI-ID
*/
- public static function getURIIdListForTag(string $search, int $uid = 0, int $start = 0, int $limit = 100)
+ public static function getURIIdListByTag(string $search, int $uid = 0, int $start = 0, int $limit = 100, int $last_uriid = 0)
{
- $condition = ["`name` = ? AND (NOT `private` OR (`private` AND `uid` = ?))", $search, $uid];
+ $condition = ["`name` = ? AND (`uid` = ? OR (`uid` = ? AND NOT `global`))
+ AND (`network` IN (?, ?, ?, ?) OR (`uid` = ? AND `uid` != ?))",
+ $search, 0, $uid, Protocol::ACTIVITYPUB, Protocol::DFRN, Protocol::DIASPORA, Protocol::OSTATUS, $uid, 0];
+
+ if (!empty($last_uriid)) {
+ $condition = DBA::mergeConditions($condition, ["`uri-id` < ?", $last_uriid]);
+ }
+
$params = [
'order' => ['uri-id' => true],
- 'group_by' => ['uri-id'],
'limit' => [$start, $limit]
];
* Returns a list of the most frequent global hashtags over the given period
*
* @param int $period Period in hours to consider posts
+ * @param int $limit Number of returned tags
* @return array
* @throws \Exception
*/
public static function getGlobalTrendingHashtags(int $period, $limit = 10)
{
- $tags = DI::cache()->get('global_trending_tags');
-
- if (empty($tags)) {
- $tagsStmt = DBA::p("SELECT `name` AS `term`, COUNT(*) AS `score`
- FROM `tag-search-view`
- WHERE `private` = ? AND `received` > DATE_SUB(NOW(), INTERVAL ? HOUR)
- GROUP BY `term` ORDER BY `score` DESC LIMIT ?",
- Item::PUBLIC, $period, $limit);
-
- if (DBA::isResult($tagsStmt)) {
- $tags = DBA::toArray($tagsStmt);
- DI::cache()->set('global_trending_tags', $tags, Duration::HOUR);
- }
+ $tags = DI::cache()->get('global_trending_tags-' . $period . '-' . $limit);
+ if (!empty($tags)) {
+ return $tags;
+ } else {
+ return self::setGlobalTrendingHashtags($period, $limit);
+ }
+ }
+
+ /**
+ * Fetch the blocked tags as SQL
+ *
+ * @return string
+ */
+ private static function getBlockedSQL()
+ {
+ $blocked_txt = DI::config()->get('system', 'blocked_tags');
+ if (empty($blocked_txt)) {
+ return '';
}
- return $tags ?: [];
+ $blocked = explode(',', $blocked_txt);
+ array_walk($blocked, function(&$value) { $value = "'" . DBA::escape(trim($value)) . "'";});
+ return " AND NOT `name` IN (" . implode(',', $blocked) . ")";
+ }
+
+ /**
+ * Creates a list of the most frequent global hashtags over the given period
+ *
+ * @param int $period Period in hours to consider posts
+ * @param int $limit Number of returned tags
+ * @return array
+ * @throws \Exception
+ */
+ public static function setGlobalTrendingHashtags(int $period, int $limit = 10)
+ {
+ // Get a uri-id that is at least X hours old.
+ // We use the uri-id in the query for the hash tags since this is much faster
+ $post = Post::selectFirstThread(['uri-id'], ["`uid` = ? AND `received` < ?", 0, DateTimeFormat::utc('now - ' . $period . ' hour')],
+ ['order' => ['received' => true]]);
+ if (empty($post['uri-id'])) {
+ return [];
+ }
+
+ $block_sql = self::getBlockedSQL();
+
+ $tagsStmt = DBA::p("SELECT `name` AS `term`, COUNT(*) AS `score`, COUNT(DISTINCT(`author-id`)) as `authors`
+ FROM `tag-search-view`
+ WHERE `private` = ? AND `uid` = ? AND `uri-id` > ? $block_sql
+ GROUP BY `term` ORDER BY `authors` DESC, `score` DESC LIMIT ?",
+ Item::PUBLIC, 0, $post['uri-id'], $limit);
+
+ if (DBA::isResult($tagsStmt)) {
+ $tags = DBA::toArray($tagsStmt);
+ DI::cache()->set('global_trending_tags-' . $period . '-' . $limit, $tags, Duration::DAY);
+ return $tags;
+ }
+
+ return [];
}
/**
* Returns a list of the most frequent local hashtags over the given period
*
* @param int $period Period in hours to consider posts
+ * @param int $limit Number of returned tags
* @return array
* @throws \Exception
*/
public static function getLocalTrendingHashtags(int $period, $limit = 10)
{
- $tags = DI::cache()->get('local_trending_tags');
-
- if (empty($tags)) {
- $tagsStmt = DBA::p("SELECT `name` AS `term`, COUNT(*) AS `score`
- FROM `tag-search-view`
- WHERE `private` = ? AND `wall` AND `origin` AND `received` > DATE_SUB(NOW(), INTERVAL ? HOUR)
- GROUP BY `term` ORDER BY `score` DESC LIMIT ?",
- Item::PUBLIC, $period, $limit);
-
- if (DBA::isResult($tagsStmt)) {
- $tags = DBA::toArray($tagsStmt);
- DI::cache()->set('local_trending_tags', $tags, Duration::HOUR);
+ $tags = DI::cache()->get('local_trending_tags-' . $period . '-' . $limit);
+ if (!empty($tags)) {
+ return $tags;
+ } else {
+ return self::setLocalTrendingHashtags($period, $limit);
+ }
+ }
+
+ /**
+ * Returns a list of the most frequent local hashtags over the given period
+ *
+ * @param int $period Period in hours to consider posts
+ * @param int $limit Number of returned tags
+ * @return array
+ * @throws \Exception
+ */
+ public static function setLocalTrendingHashtags(int $period, int $limit = 10)
+ {
+ // Get a uri-id that is at least X hours old.
+ // We use the uri-id in the query for the hash tags since this is much faster
+ $post = Post::selectFirstThread(['uri-id'], ["`uid` = ? AND `received` < ?", 0, DateTimeFormat::utc('now - ' . $period . ' hour')],
+ ['order' => ['received' => true]]);
+ if (empty($post['uri-id'])) {
+ return [];
+ }
+
+ $block_sql = self::getBlockedSQL();
+
+ $tagsStmt = DBA::p("SELECT `name` AS `term`, COUNT(*) AS `score`, COUNT(DISTINCT(`author-id`)) as `authors`
+ FROM `tag-search-view`
+ WHERE `private` = ? AND `wall` AND `origin` AND `uri-id` > ? $block_sql
+ GROUP BY `term` ORDER BY `authors` DESC, `score` DESC LIMIT ?",
+ Item::PUBLIC, $post['uri-id'], $limit);
+
+ if (DBA::isResult($tagsStmt)) {
+ $tags = DBA::toArray($tagsStmt);
+ DI::cache()->set('local_trending_tags-' . $period . '-' . $limit, $tags, Duration::DAY);
+ return $tags;
+ }
+
+ return [];
+ }
+
+ /**
+ * Check if the provided tag is of one of the provided term types.
+ *
+ * @param string $tag
+ * @param int ...$types
+ * @return bool
+ */
+ public static function isType($tag, ...$types)
+ {
+ $tag_chars = [];
+ foreach ($types as $type) {
+ if (array_key_exists($type, self::TAG_CHARACTER)) {
+ $tag_chars[] = self::TAG_CHARACTER[$type];
}
}
- return $tags ?: [];
+ return Strings::startsWithChars($tag, $tag_chars);
+ }
+
+ /**
+ * Fetch user who subscribed to the given tag
+ *
+ * @param string $tag
+ * @return array User list
+ */
+ private static function getUIDListByTag(string $tag)
+ {
+ $uids = [];
+ $searches = DBA::select('search', ['uid'], ['term' => $tag]);
+ while ($search = DBA::fetch($searches)) {
+ $uids[] = $search['uid'];
+ }
+ DBA::close($searches);
+
+ return $uids;
+ }
+
+ /**
+ * Fetch user who subscribed to the tags of the given item
+ *
+ * @param integer $uri_id
+ * @return array User list
+ */
+ public static function getUIDListByURIId(int $uri_id)
+ {
+ $uids = [];
+ $tags = self::getByURIId($uri_id, [self::HASHTAG]);
+
+ foreach ($tags as $tag) {
+ $uids = array_merge($uids, self::getUIDListByTag(self::TAG_CHARACTER[self::HASHTAG] . $tag['name']));
+ }
+
+ return array_unique($uids);
}
}
projects / friendica.git / blobdiff