src/Model/Tag.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2020, Friendica
   4  *
   5  * @license GNU AGPL version 3 or any later version
   6  *
   7  * This program is free software: you can redistribute it and/or modify
   8  * it under the terms of the GNU Affero General Public License as
   9  * published by the Free Software Foundation, either version 3 of the
  10  * License, or (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU Affero General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Affero General Public License
  18  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  19  *
  20  */
  21
  22 namespace Friendica\Model;
  23
  24 use Friendica\Content\Text\BBCode;
  25 use Friendica\Core\Cache\Duration;
  26 use Friendica\Core\Logger;
  27 use Friendica\Core\System;
  28 use Friendica\Database\DBA;
  29 use Friendica\DI;
  30 use Friendica\Util\Strings;
  31
  32 /**
  33  * Class Tag
  34  *
  35  * This Model class handles tag table interactions.
  36  * This tables stores relevant tags related to posts, like hashtags and mentions.
  37  */
  38 class Tag
  39 {
  40         const UNKNOWN  = 0;
  41         const HASHTAG  = 1;
  42         const MENTION  = 2;
  43         /**
  44          * An implicit mention is a mention in a comment body that is redundant with the threading information.
  45          */
  46         const IMPLICIT_MENTION  = 8;
  47         /**
  48          * An exclusive mention transfers the ownership of the post to the target account, usually a forum.
  49          */
  50         const EXCLUSIVE_MENTION = 9;
  51
  52         const TAG_CHARACTER = [
  53                 self::HASHTAG           => '#',
  54                 self::MENTION           => '@',
  55                 self::IMPLICIT_MENTION  => '%',
  56                 self::EXCLUSIVE_MENTION => '!',
  57         ];
  58
  59         /**
  60          * Store tag/mention elements
  61          *
  62          * @param integer $uriid
  63          * @param integer $type
  64          * @param string  $name
  65          * @param string  $url
  66          * @param boolean $probing
  67          */
  68         public static function store(int $uriid, int $type, string $name, string $url = '', $probing = true)
  69         {
  70                 if ($type == self::HASHTAG) {
  71                         // Remove some common "garbarge" from tags
  72                         $name = trim($name, "\x00..\x20\xFF#!@,;.:'/?!^°$%".'"');
  73
  74                         $tags = explode(self::TAG_CHARACTER[self::HASHTAG], $name);
  75                         if (count($tags) > 1) {
  76                                 foreach ($tags as $tag) {
  77                                         self::store($uriid, $type, $tag, $url, $probing);
  78                                 }
  79                                 return;
  80                         }
  81                 }
  82
  83                 if (empty($name)) {
  84                         return;
  85                 }
  86
  87                 $cid = 0;
  88                 $tagid = 0;
  89
  90                 if (in_array($type, [self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION])) {
  91                         if (empty($url)) {
  92                                 // No mention without a contact url
  93                                 return;
  94                         }
  95
  96                         if (!$probing) {
  97                                 $condition = ['nurl' => Strings::normaliseLink($url), 'uid' => 0, 'deleted' => false];
  98                                 $contact = DBA::selectFirst('contact', ['id'], $condition, ['order' => ['id']]);
  99                                 if (DBA::isResult($contact)) {
 100                                         $cid = $contact['id'];
 101                                         Logger::info('Got id for contact url', ['cid' => $cid, 'url' => $url]);
 102                                 }
 103
 104                                 if (empty($cid)) {
 105                                         $ssl_url = str_replace('http://', 'https://', $url);
 106                                         $condition = ['`alias` IN (?, ?, ?) AND `uid` = ? AND NOT `deleted`', $url, Strings::normaliseLink($url), $ssl_url, 0];
 107                                         $contact = DBA::selectFirst('contact', ['id'], $condition, ['order' => ['id']]);
 108                                         if (DBA::isResult($contact)) {
 109                                                 $cid = $contact['id'];
 110                                                 Logger::info('Got id for contact alias', ['cid' => $cid, 'url' => $url]);
 111                                         }
 112                                 }
 113                         } else {
 114                                 $cid = Contact::getIdForURL($url, 0, true);
 115                                 Logger::info('Got id by probing', ['cid' => $cid, 'url' => $url]);
 116                         }
 117
 118                         if (empty($cid)) {
 119                                 // The contact wasn't found in the system (most likely some dead account)
 120                                 // We ensure that we only store a single entry by overwriting the previous name
 121                                 Logger::info('Contact not found, updating tag', ['url' => $url, 'name' => $name]);
 122                                 DBA::update('tag', ['name' => substr($name, 0, 96)], ['url' => $url]);
 123                         }
 124                 }
 125
 126                 if (empty($cid)) {
 127                         $fields = ['name' => substr($name, 0, 96), 'url' => ''];
 128
 129                         if (($type != self::HASHTAG) && !empty($url) && ($url != $name)) {
 130                                 $fields['url'] = strtolower($url);
 131                         }
 132
 133                         $tag = DBA::selectFirst('tag', ['id'], $fields);
 134                         if (!DBA::isResult($tag)) {
 135                                 DBA::insert('tag', $fields, true);
 136                                 $tagid = DBA::lastInsertId();
 137                         } else {
 138                                 $tagid = $tag['id'];
 139                         }
 140
 141                         if (empty($tagid)) {
 142                                 Logger::error('No tag id created', $fields);
 143                                 return;
 144                         }
 145                 }
 146
 147                 $fields = ['uri-id' => $uriid, 'type' => $type, 'tid' => $tagid, 'cid' => $cid];
 148
 149                 if (in_array($type, [self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION])) {
 150                         $condition = $fields;
 151                         $condition['type'] = [self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION];
 152                         if (DBA::exists('post-tag', $condition)) {
 153                                 Logger::info('Tag already exists', $fields);
 154                                 return;
 155                         }
 156                 }
 157
 158                 DBA::insert('post-tag', $fields, true);
 159
 160                 Logger::info('Stored tag/mention', ['uri-id' => $uriid, 'tag-id' => $tagid, 'contact-id' => $cid, 'name' => $name, 'type' => $type, 'callstack' => System::callstack(8)]);
 161         }
 162
 163         /**
 164          * Store tag/mention elements
 165          *
 166          * @param integer $uriid
 167          * @param string $hash
 168          * @param string $name
 169          * @param string $url
 170          * @param boolean $probing
 171          */
 172         public static function storeByHash(int $uriid, string $hash, string $name, string $url = '', $probing = true)
 173         {
 174                 $type = self::getTypeForHash($hash);
 175                 if ($type == self::UNKNOWN) {
 176                         return;
 177                 }
 178
 179                 self::store($uriid, $type, $name, $url, $probing);
 180         }
 181
 182         /**
 183          * Store tags and mentions from the body
 184          *
 185          * @param integer $uriid   URI-Id
 186          * @param string  $body    Body of the post
 187          * @param string  $tags    Accepted tags
 188          * @param boolean $probing Perform a probing for contacts, adding them if needed
 189          */
 190         public static function storeFromBody(int $uriid, string $body, string $tags = null, $probing = true)
 191         {
 192                 if (is_null($tags)) {
 193                         $tags =  self::TAG_CHARACTER[self::HASHTAG] . self::TAG_CHARACTER[self::MENTION] . self::TAG_CHARACTER[self::EXCLUSIVE_MENTION];
 194                 }
 195
 196                 Logger::info('Check for tags', ['uri-id' => $uriid, 'hash' => $tags, 'callstack' => System::callstack()]);
 197
 198                 if (!preg_match_all("/([" . $tags . "])\[url\=([^\[\]]*)\]([^\[\]]*)\[\/url\]/ism", $body, $result, PREG_SET_ORDER)) {
 199                         return;
 200                 }
 201
 202                 Logger::info('Found tags', ['uri-id' => $uriid, 'hash' => $tags, 'result' => $result]);
 203
 204                 foreach ($result as $tag) {
 205                         self::storeByHash($uriid, $tag[1], $tag[3], $tag[2], $probing);
 206                 }
 207         }
 208
 209         /**
 210          * Store raw tags (not encapsulated in links) from the body
 211          * This function is needed in the intermediate phase.
 212          * Later we can call item::setHashtags in advance to have all tags converted.
 213          *
 214          * @param integer $uriid URI-Id
 215          * @param string  $body   Body of the post
 216          */
 217         public static function storeRawTagsFromBody(int $uriid, string $body)
 218         {
 219                 Logger::info('Check for tags', ['uri-id' => $uriid, 'callstack' => System::callstack()]);
 220
 221                 $result = BBCode::getTags($body);
 222                 if (empty($result)) {
 223                         return;
 224                 }
 225
 226                 Logger::info('Found tags', ['uri-id' => $uriid, 'result' => $result]);
 227
 228                 foreach ($result as $tag) {
 229                         if (substr($tag, 0, 1) != self::TAG_CHARACTER[self::HASHTAG]) {
 230                                 continue;
 231                         }
 232                         self::storeByHash($uriid, substr($tag, 0, 1), substr($tag, 1));
 233                 }
 234         }
 235
 236         /**
 237          * Checks for stored hashtags and mentions for the given post
 238          *
 239          * @param integer $uriid
 240          * @return bool
 241          */
 242         public static function existsForPost(int $uriid)
 243         {
 244                 return DBA::exists('post-tag', ['uri-id' => $uriid, 'type' => [self::HASHTAG, self::MENTION, self::IMPLICIT_MENTION, self::EXCLUSIVE_MENTION]]);
 245         }
 246
 247         /**
 248          * Remove tag/mention
 249          *
 250          * @param integer $uriid
 251          * @param integer $type
 252          * @param string $name
 253          * @param string $url
 254          */
 255         public static function remove(int $uriid, int $type, string $name, string $url = '')
 256         {
 257                 $condition = ['uri-id' => $uriid, 'type' => $type, 'url' => $url];
 258                 if ($type == self::HASHTAG) {
 259                         $condition['name'] = $name;
 260                 }
 261
 262                 $tag = DBA::selectFirst('tag-view', ['tid', 'cid'], $condition);
 263                 if (!DBA::isResult($tag)) {
 264                         return;
 265                 }
 266
 267                 Logger::info('Removing tag/mention', ['uri-id' => $uriid, 'tid' => $tag['tid'], 'name' => $name, 'url' => $url, 'callstack' => System::callstack(8)]);
 268                 DBA::delete('post-tag', ['uri-id' => $uriid, 'type' => $type, 'tid' => $tag['tid'], 'cid' => $tag['cid']]);
 269         }
 270
 271         /**
 272          * Remove tag/mention
 273          *
 274          * @param integer $uriid
 275          * @param string $hash
 276          * @param string $name
 277          * @param string $url
 278          */
 279         public static function removeByHash(int $uriid, string $hash, string $name, string $url = '')
 280         {
 281                 $type = self::getTypeForHash($hash);
 282                 if ($type == self::UNKNOWN) {
 283                         return;
 284                 }
 285
 286                 self::remove($uriid, $type, $name, $url);
 287         }
 288
 289         /**
 290          * Get the type for the given hash
 291          *
 292          * @param string $hash
 293          * @return integer type
 294          */
 295         private static function getTypeForHash(string $hash)
 296         {
 297                 if ($hash == self::TAG_CHARACTER[self::MENTION]) {
 298                         return self::MENTION;
 299                 } elseif ($hash == self::TAG_CHARACTER[self::EXCLUSIVE_MENTION]) {
 300                         return self::EXCLUSIVE_MENTION;
 301                 } elseif ($hash == self::TAG_CHARACTER[self::IMPLICIT_MENTION]) {
 302                         return self::IMPLICIT_MENTION;
 303                 } elseif ($hash == self::TAG_CHARACTER[self::HASHTAG]) {
 304                         return self::HASHTAG;
 305                 } else {
 306                         return self::UNKNOWN;
 307                 }
 308         }
 309
 310         /**
 311          * Retrieves the terms from the provided type(s) associated with the provided item ID.
 312          *
 313          * @param int       $item_id
 314          * @param int|array $type
 315          * @return array
 316          * @throws \Exception
 317          */
 318         public static function getByURIId(int $uri_id, array $type = [self::HASHTAG, self::MENTION, self::IMPLICIT_MENTION, self::EXCLUSIVE_MENTION])
 319         {
 320                 $condition = ['uri-id' => $uri_id, 'type' => $type];
 321                 return DBA::selectToArray('tag-view', ['type', 'name', 'url'], $condition);
 322         }
 323
 324         /**
 325          * Sorts an item's tags into mentions, hashtags and other tags. Generate personalized URLs by user and modify the
 326          * provided item's body with them.
 327          *
 328          * @param array $item
 329          * @return array
 330          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 331          * @throws \ImagickException
 332          */
 333         public static function populateFromItem(&$item)
 334         {
 335                 $return = [
 336                         'tags' => [],
 337                         'hashtags' => [],
 338                         'mentions' => [],
 339                         'implicit_mentions' => [],
 340                 ];
 341
 342                 $searchpath = DI::baseUrl() . "/search?tag=";
 343
 344                 $taglist = DBA::select('tag-view', ['type', 'name', 'url'],
 345                         ['uri-id' => $item['uri-id'], 'type' => [self::HASHTAG, self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION]]);
 346                 while ($tag = DBA::fetch($taglist)) {
 347                         if ($tag['url'] == '') {
 348                                 $tag['url'] = $searchpath . rawurlencode($tag['name']);
 349                         }
 350
 351                         $orig_tag = $tag['url'];
 352
 353                         $prefix = self::TAG_CHARACTER[$tag['type']];
 354                         switch($tag['type']) {
 355                                 case self::HASHTAG:
 356                                         if ($orig_tag != $tag['url']) {
 357                                                 $item['body'] = str_replace($orig_tag, $tag['url'], $item['body']);
 358                                         }
 359
 360                                         $return['hashtags'][] = $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['name']) . '</a>';
 361                                         $return['tags'][] = $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['name']) . '</a>';
 362                                         break;
 363                                 case self::MENTION:
 364                                 case self::EXCLUSIVE_MENTION:
 365                                                 $tag['url'] = Contact::magicLink($tag['url']);
 366                                         $return['mentions'][] = $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['name']) . '</a>';
 367                                         $return['tags'][] = $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['name']) . '</a>';
 368                                         break;
 369                                 case self::IMPLICIT_MENTION:
 370                                         $return['implicit_mentions'][] = $prefix . $tag['name'];
 371                                         break;
 372                         }
 373                 }
 374                 DBA::close($taglist);
 375
 376                 return $return;
 377         }
 378
 379         /**
 380          * Search posts for given tag
 381          *
 382          * @param string $search
 383          * @param integer $uid
 384          * @param integer $start
 385          * @param integer $limit
 386          * @return array with URI-ID
 387          */
 388         public static function getURIIdListByTag(string $search, int $uid = 0, int $start = 0, int $limit = 100)
 389         {
 390                 $condition = ["`name` = ? AND (NOT `private` OR (`private` AND `uid` = ?))", $search, $uid];
 391                 $params = [
 392                         'order' => ['uri-id' => true],
 393                         'group_by' => ['uri-id'],
 394                         'limit' => [$start, $limit]
 395                 ];
 396
 397                 $tags = DBA::select('tag-search-view', ['uri-id'], $condition, $params);
 398
 399                 $uriids = [];
 400                 while ($tag = DBA::fetch($tags)) {
 401                         $uriids[] = $tag['uri-id'];
 402                 }
 403                 DBA::close($tags);
 404
 405                 return $uriids;
 406         }
 407
 408         /**
 409          * Returns a list of the most frequent global hashtags over the given period
 410          *
 411          * @param int $period Period in hours to consider posts
 412          * @return array
 413          * @throws \Exception
 414          */
 415         public static function getGlobalTrendingHashtags(int $period, $limit = 10)
 416         {
 417                 $tags = DI::cache()->get('global_trending_tags');
 418
 419                 if (empty($tags)) {
 420                         $tagsStmt = DBA::p("SELECT `name` AS `term`, COUNT(*) AS `score`
 421                                 FROM `tag-search-view`
 422                                 WHERE `private` = ? AND `received` > DATE_SUB(NOW(), INTERVAL ? HOUR)
 423                                 GROUP BY `term` ORDER BY `score` DESC LIMIT ?",
 424                                 Item::PUBLIC, $period, $limit);
 425
 426                         if (DBA::isResult($tagsStmt)) {
 427                                 $tags = DBA::toArray($tagsStmt);
 428                                 DI::cache()->set('global_trending_tags', $tags, Duration::HOUR);
 429                         }
 430                 }
 431
 432                 return $tags ?: [];
 433         }
 434
 435         /**
 436          * Returns a list of the most frequent local hashtags over the given period
 437          *
 438          * @param int $period Period in hours to consider posts
 439          * @return array
 440          * @throws \Exception
 441          */
 442         public static function getLocalTrendingHashtags(int $period, $limit = 10)
 443         {
 444                 $tags = DI::cache()->get('local_trending_tags');
 445
 446                 if (empty($tags)) {
 447                         $tagsStmt = DBA::p("SELECT `name` AS `term`, COUNT(*) AS `score`
 448                                 FROM `tag-search-view`
 449                                 WHERE `private` = ? AND `wall` AND `origin` AND `received` > DATE_SUB(NOW(), INTERVAL ? HOUR)
 450                                 GROUP BY `term` ORDER BY `score` DESC LIMIT ?",
 451                                 Item::PUBLIC, $period, $limit);
 452
 453                         if (DBA::isResult($tagsStmt)) {
 454                                 $tags = DBA::toArray($tagsStmt);
 455                                 DI::cache()->set('local_trending_tags', $tags, Duration::HOUR);
 456                         }
 457                 }
 458
 459                 return $tags ?: [];
 460         }
 461
 462         /**
 463          * Check if the provided tag is of one of the provided term types.
 464          *
 465          * @param string $tag
 466          * @param int    ...$types
 467          * @return bool
 468          */
 469         public static function isType($tag, ...$types)
 470         {
 471                 $tag_chars = [];
 472                 foreach ($types as $type) {
 473                         if (array_key_exists($type, self::TAG_CHARACTER)) {
 474                                 $tag_chars[] = self::TAG_CHARACTER[$type];
 475                         }
 476                 }
 477
 478                 return Strings::startsWith($tag, $tag_chars);
 479         }
 480 }