X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;f=src%2FModel%2FTag.php;h=0ea19f55d9e11f5e8bc0f11585fd5f287e6480bf;hb=55db2670d1662b32c6faa820c60c44e16cffab20;hp=40f2f8d6a1600ffe0622ca81ec6d09fa0259c482;hpb=cc5e5be931bc21e44d0cb26778e8cc15be924e41;p=friendica.git
diff --git a/src/Model/Tag.php b/src/Model/Tag.php
index 40f2f8d6a1..0ea19f55d9 100644
--- a/src/Model/Tag.php
+++ b/src/Model/Tag.php
@@ -1,6 +1,6 @@
'#',
self::MENTION => '@',
- self::IMPLICIT_MENTION => '%',
self::EXCLUSIVE_MENTION => '!',
+ self::IMPLICIT_MENTION => '%',
];
/**
@@ -63,18 +77,18 @@ class Tag
* @param integer $type
* @param string $name
* @param string $url
- * @param boolean $probing
+ * @param integer $target
*/
- 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 = '', int $target = null)
{
if ($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;
}
@@ -87,7 +101,7 @@ class Tag
$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;
@@ -97,44 +111,32 @@ class Tag
Logger::notice('Wrong scheme in url', ['url' => $url, 'callstack' => System::callstack(20)]);
}
- if (!$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]);
- }
+ $cid = Contact::getIdForURL($url, 0, false);
+ Logger::debug('Got id for contact', ['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' => $url]);
+ if (empty($cid)) {
+ $tag = DBA::selectFirst('tag', ['name', 'type'], ['url' => $url]);
+ if (!empty($tag)) {
+ if ($tag['name'] != substr($name, 0, 96)) {
+ DBA::update('tag', ['name' => substr($name, 0, 96)], ['url' => $url]);
+ }
+ if (!empty($target) && ($tag['type'] != $target)) {
+ DBA::update('tag', ['type' => $target], ['url' => $url]);
}
}
- } else {
- $cid = Contact::getIdForURL($url, 0, false);
- Logger::info('Got id by probing', ['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' => $url]);
}
}
if (empty($cid)) {
- if (($type != self::HASHTAG) && !empty($url) && ($url != $name)) {
- $url = strtolower($url);
- } else {
- $url = '';
+ if (!in_array($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);
+ $tagid = self::getID($name, $url, $target);
if (empty($tagid)) {
return;
}
@@ -151,28 +153,87 @@ class Tag
}
}
- 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)]);
}
+ /**
+ * Fetch the target type for the given url
+ *
+ * @param string $url
+ * @param bool $fetch Fetch information via network operations
+ * @return null|int
+ */
+ public static function getTargetType(string $url, bool $fetch = true)
+ {
+ $target = null;
+
+ if (empty($url)) {
+ return $target;
+ }
+
+ $tag = DBA::selectFirst('tag', ['url', 'type'], ['url' => $url]);
+ if (!empty($tag['type'])) {
+ $target = $tag['type'];
+ if ($target != self::GENERAL_COLLECTION) {
+ Logger::debug('Found existing type', ['type' => $tag['type'], 'url' => $url]);
+ return $target;
+ }
+ }
+
+ if ($url == ActivityPub::PUBLIC_COLLECTION) {
+ $target = self::PUBLIC_COLLECTION;
+ Logger::debug('Public collection', ['url' => $url]);
+ } else {
+ if (DBA::exists('apcontact', ['followers' => $url])) {
+ $target = self::FOLLOWER_COLLECTION;
+ Logger::debug('Found collection via existing apcontact', ['url' => $url]);
+ } elseif (Contact::getIdForURL($url, 0, $fetch ? null : false)) {
+ $target = self::ACCOUNT;
+ Logger::debug('URL is an account', ['url' => $url]);
+ } elseif ($fetch && ($target != self::GENERAL_COLLECTION)) {
+ $content = ActivityPub::fetchContent($url);
+ if (!empty($content['type']) && ($content['type'] == 'OrderedCollection')) {
+ $target = self::GENERAL_COLLECTION;
+ Logger::debug('URL is an ordered collection', ['url' => $url]);
+ }
+ }
+ }
+
+ if (!empty($target) && !empty($tag['url']) && ($tag['type'] != $target)) {
+ DBA::update('tag', ['type' => $target], ['url' => $url]);
+ }
+
+ if (empty($target)) {
+ Logger::debug('No type could be detected', ['url' => $url]);
+ }
+
+ return $target;
+ }
+
/**
* Get a tag id for a given tag name and url
*
* @param string $name
* @param string $url
+ * @param int $type
* @return void
*/
- public static function getID(string $name, string $url = '')
+ public static function getID(string $name, string $url = '', int $type = null)
{
$fields = ['name' => substr($name, 0, 96), 'url' => $url];
+ if (!empty($type)) {
+ $fields['type'] = $type;
+ }
+
$tag = DBA::selectFirst('tag', ['id'], $fields);
if (DBA::isResult($tag)) {
return $tag['id'];
}
- DBA::insert('tag', $fields, true);
+ DBA::insert('tag', $fields, Database::INSERT_IGNORE);
$tid = DBA::lastInsertId();
if (!empty($tid)) {
return $tid;
@@ -202,22 +263,40 @@ class Tag
}
/**
- * Store tags and mentions from the body
- *
- * @param integer $uriid URI-Id
+ * Get tags and mentions from the body
+ *
* @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;
}
@@ -232,7 +311,7 @@ class Tag
* Store raw tags (not encapsulated in links) from the body
* This function is needed in the intermediate phase.
* Later we can call item::setHashtags in advance to have all tags converted.
- *
+ *
* @param integer $uriid URI-Id
* @param string $body Body of the post
*/
@@ -263,7 +342,7 @@ class Tag
*/
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]]);
}
/**
@@ -338,14 +417,14 @@ class Tag
public static function createImplicitMentions(int $uri_id, int $parent_uri_id)
{
// Always mention the direct parent author
- $parent = Item::selectFirst(['author-link', 'author-name'], ['uri-id' => $parent_uri_id]);
+ $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]);
+ $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']);
}
@@ -360,10 +439,10 @@ class Tag
* @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];
- return DBA::selectToArray('tag-view', ['type', 'name', 'url'], $condition);
+ return DBA::selectToArray('tag-view', ['type', 'name', 'url', 'tag-type'], $condition);
}
/**
@@ -374,7 +453,7 @@ class Tag
* @return string tags and mentions
* @throws \Exception
*/
- public static function getCSVByURIId(int $uri_id, array $type = [self::HASHTAG, self::MENTION, self::IMPLICIT_MENTION, self::EXCLUSIVE_MENTION])
+ public static function getCSVByURIId(int $uri_id, array $type = [self::HASHTAG, self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION])
{
$tag_list = [];
$tags = self::getByURIId($uri_id, $type);
@@ -405,7 +484,7 @@ class Tag
$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'] == '') {
@@ -421,14 +500,18 @@ class Tag
$item['body'] = str_replace($orig_tag, $tag['url'], $item['body']);
}
- $return['hashtags'][] = $prefix . '' . htmlspecialchars($tag['name']) . '';
- $return['tags'][] = $prefix . '' . htmlspecialchars($tag['name']) . '';
+ $return['hashtags'][] = '' . $prefix . '' . htmlspecialchars($tag['name']) . '';
+ $return['tags'][] = '' . $prefix . '' . htmlspecialchars($tag['name']) . '';
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'][] = $prefix . '' . htmlspecialchars($tag['name']) . '';
- $return['tags'][] = $prefix . '' . htmlspecialchars($tag['name']) . '';
+ }
+ $return['mentions'][] = '' . $prefix . '' . htmlspecialchars($tag['name']) . '';
+ $return['tags'][] = '' . $prefix . '' . htmlspecialchars($tag['name']) . '';
break;
case self::IMPLICIT_MENTION:
$return['implicit_mentions'][] = $prefix . $tag['name'];
@@ -449,10 +532,11 @@ class Tag
*/
public static function countByTag(string $search, int $uid = 0)
{
- $condition = ["`name` = ? AND (NOT `private` OR (`private` AND `uid` = ?))", $search, $uid];
- $params = ['group_by' => ['uri-id']];
+ $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, $params);
+ return DBA::count('tag-search-view', $condition);
}
/**
@@ -462,14 +546,21 @@ class Tag
* @param integer $uid
* @param integer $start
* @param integer $limit
+ * @param integer $last_uriid
* @return array with URI-ID
*/
- public static function getURIIdListByTag(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]
];
@@ -488,54 +579,123 @@ class Tag
* 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 $tags ?: [];
+ return [];
}
/**