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($url)) {
  92                                 // No mention without a contact url
  93                                 return;
  94                         }
  95
  96                         if ((substr($url, 0, 7) == 'https//') || (substr($url, 0, 6) == 'http//')) {
  97                                 Logger::notice('Wrong scheme in url', ['url' => $url, 'callstack' => System::callstack(20)]);
  98                         }
  99
 100                         if (!$probing) {
 101                                 $condition = ['nurl' => Strings::normaliseLink($url), 'uid' => 0, 'deleted' => false];
 102                                 $contact = DBA::selectFirst('contact', ['id'], $condition, ['order' => ['id']]);
 103                                 if (DBA::isResult($contact)) {
 104                                         $cid = $contact['id'];
 105                                         Logger::info('Got id for contact url', ['cid' => $cid, 'url' => $url]);
 106                                 }
 107
 108                                 if (empty($cid)) {
 109                                         $ssl_url = str_replace('http://', 'https://', $url);
 110                                         $condition = ['`alias` IN (?, ?, ?) AND `uid` = ? AND NOT `deleted`', $url, Strings::normaliseLink($url), $ssl_url, 0];
 111                                         $contact = DBA::selectFirst('contact', ['id'], $condition, ['order' => ['id']]);
 112                                         if (DBA::isResult($contact)) {
 113                                                 $cid = $contact['id'];
 114                                                 Logger::info('Got id for contact alias', ['cid' => $cid, 'url' => $url]);
 115                                         }
 116                                 }
 117                         } else {
 118                                 $cid = Contact::getIdForURL($url, 0, false);
 119                                 Logger::info('Got id by probing', ['cid' => $cid, 'url' => $url]);
 120                         }
 121
 122                         if (empty($cid)) {
 123                                 // The contact wasn't found in the system (most likely some dead account)
 124                                 // We ensure that we only store a single entry by overwriting the previous name
 125                                 Logger::info('Contact not found, updating tag', ['url' => $url, 'name' => $name]);
 126                                 DBA::update('tag', ['name' => substr($name, 0, 96)], ['url' => $url]);
 127                         }
 128                 }
 129
 130                 if (empty($cid)) {
 131                         if (($type != self::HASHTAG) && !empty($url) && ($url != $name)) {
 132                                 $url = strtolower($url);
 133                         } else {
 134                                 $url = '';
 135                         }
 136
 137                         $tagid = self::getID($name, $url);
 138                         if (empty($tagid)) {
 139                                 return;
 140                         }
 141                 }
 142
 143                 $fields = ['uri-id' => $uriid, 'type' => $type, 'tid' => $tagid, 'cid' => $cid];
 144
 145                 if (in_array($type, [self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION])) {
 146                         $condition = $fields;
 147                         $condition['type'] = [self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION];
 148                         if (DBA::exists('post-tag', $condition)) {
 149                                 Logger::info('Tag already exists', $fields);
 150                                 return;
 151                         }
 152                 }
 153
 154                 DBA::insert('post-tag', $fields, true);
 155
 156                 Logger::info('Stored tag/mention', ['uri-id' => $uriid, 'tag-id' => $tagid, 'contact-id' => $cid, 'name' => $name, 'type' => $type, 'callstack' => System::callstack(8)]);
 157         }
 158
 159         /**
 160          * Get a tag id for a given tag name and url
 161          *
 162          * @param string $name
 163          * @param string $url
 164          * @return void
 165          */
 166         public static function getID(string $name, string $url = '')
 167         {
 168                 $fields = ['name' => substr($name, 0, 96), 'url' => $url];
 169
 170                 $tag = DBA::selectFirst('tag', ['id'], $fields);
 171                 if (DBA::isResult($tag)) {
 172                         return $tag['id'];
 173                 }
 174
 175                 DBA::insert('tag', $fields, true);
 176                 $tid = DBA::lastInsertId();
 177                 if (!empty($tid)) {
 178                         return $tid;
 179                 }
 180
 181                 Logger::error('No tag id created', $fields);
 182                 return 0;
 183         }
 184
 185         /**
 186          * Store tag/mention elements
 187          *
 188          * @param integer $uriid
 189          * @param string $hash
 190          * @param string $name
 191          * @param string $url
 192          * @param boolean $probing
 193          */
 194         public static function storeByHash(int $uriid, string $hash, string $name, string $url = '', $probing = true)
 195         {
 196                 $type = self::getTypeForHash($hash);
 197                 if ($type == self::UNKNOWN) {
 198                         return;
 199                 }
 200
 201                 self::store($uriid, $type, $name, $url, $probing);
 202         }
 203
 204         /**
 205          * Store tags and mentions from the body
 206          *
 207          * @param integer $uriid   URI-Id
 208          * @param string  $body    Body of the post
 209          * @param string  $tags    Accepted tags
 210          * @param boolean $probing Perform a probing for contacts, adding them if needed
 211          */
 212         public static function storeFromBody(int $uriid, string $body, string $tags = null, $probing = true)
 213         {
 214                 if (is_null($tags)) {
 215                         $tags =  self::TAG_CHARACTER[self::HASHTAG] . self::TAG_CHARACTER[self::MENTION] . self::TAG_CHARACTER[self::EXCLUSIVE_MENTION];
 216                 }
 217
 218                 Logger::info('Check for tags', ['uri-id' => $uriid, 'hash' => $tags, 'callstack' => System::callstack()]);
 219
 220                 if (!preg_match_all("/([" . $tags . "])\[url\=([^\[\]]*)\]([^\[\]]*)\[\/url\]/ism", $body, $result, PREG_SET_ORDER)) {
 221                         return;
 222                 }
 223
 224                 Logger::info('Found tags', ['uri-id' => $uriid, 'hash' => $tags, 'result' => $result]);
 225
 226                 foreach ($result as $tag) {
 227                         self::storeByHash($uriid, $tag[1], $tag[3], $tag[2], $probing);
 228                 }
 229         }
 230
 231         /**
 232          * Store raw tags (not encapsulated in links) from the body
 233          * This function is needed in the intermediate phase.
 234          * Later we can call item::setHashtags in advance to have all tags converted.
 235          *
 236          * @param integer $uriid URI-Id
 237          * @param string  $body   Body of the post
 238          */
 239         public static function storeRawTagsFromBody(int $uriid, string $body)
 240         {
 241                 Logger::info('Check for tags', ['uri-id' => $uriid, 'callstack' => System::callstack()]);
 242
 243                 $result = BBCode::getTags($body);
 244                 if (empty($result)) {
 245                         return;
 246                 }
 247
 248                 Logger::info('Found tags', ['uri-id' => $uriid, 'result' => $result]);
 249
 250                 foreach ($result as $tag) {
 251                         if (substr($tag, 0, 1) != self::TAG_CHARACTER[self::HASHTAG]) {
 252                                 continue;
 253                         }
 254                         self::storeByHash($uriid, substr($tag, 0, 1), substr($tag, 1));
 255                 }
 256         }
 257
 258         /**
 259          * Checks for stored hashtags and mentions for the given post
 260          *
 261          * @param integer $uriid
 262          * @return bool
 263          */
 264         public static function existsForPost(int $uriid)
 265         {
 266                 return DBA::exists('post-tag', ['uri-id' => $uriid, 'type' => [self::HASHTAG, self::MENTION, self::IMPLICIT_MENTION, self::EXCLUSIVE_MENTION]]);
 267         }
 268
 269         /**
 270          * Remove tag/mention
 271          *
 272          * @param integer $uriid
 273          * @param integer $type
 274          * @param string $name
 275          * @param string $url
 276          */
 277         public static function remove(int $uriid, int $type, string $name, string $url = '')
 278         {
 279                 $condition = ['uri-id' => $uriid, 'type' => $type, 'url' => $url];
 280                 if ($type == self::HASHTAG) {
 281                         $condition['name'] = $name;
 282                 }
 283
 284                 $tag = DBA::selectFirst('tag-view', ['tid', 'cid'], $condition);
 285                 if (!DBA::isResult($tag)) {
 286                         return;
 287                 }
 288
 289                 Logger::info('Removing tag/mention', ['uri-id' => $uriid, 'tid' => $tag['tid'], 'name' => $name, 'url' => $url, 'callstack' => System::callstack(8)]);
 290                 DBA::delete('post-tag', ['uri-id' => $uriid, 'type' => $type, 'tid' => $tag['tid'], 'cid' => $tag['cid']]);
 291         }
 292
 293         /**
 294          * Remove tag/mention
 295          *
 296          * @param integer $uriid
 297          * @param string $hash
 298          * @param string $name
 299          * @param string $url
 300          */
 301         public static function removeByHash(int $uriid, string $hash, string $name, string $url = '')
 302         {
 303                 $type = self::getTypeForHash($hash);
 304                 if ($type == self::UNKNOWN) {
 305                         return;
 306                 }
 307
 308                 self::remove($uriid, $type, $name, $url);
 309         }
 310
 311         /**
 312          * Get the type for the given hash
 313          *
 314          * @param string $hash
 315          * @return integer type
 316          */
 317         private static function getTypeForHash(string $hash)
 318         {
 319                 if ($hash == self::TAG_CHARACTER[self::MENTION]) {
 320                         return self::MENTION;
 321                 } elseif ($hash == self::TAG_CHARACTER[self::EXCLUSIVE_MENTION]) {
 322                         return self::EXCLUSIVE_MENTION;
 323                 } elseif ($hash == self::TAG_CHARACTER[self::IMPLICIT_MENTION]) {
 324                         return self::IMPLICIT_MENTION;
 325                 } elseif ($hash == self::TAG_CHARACTER[self::HASHTAG]) {
 326                         return self::HASHTAG;
 327                 } else {
 328                         return self::UNKNOWN;
 329                 }
 330         }
 331
 332         /**
 333          * Create implicit mentions for a given post
 334          *
 335          * @param integer $uri_id
 336          * @param integer $parent_uri_id
 337          */
 338         public static function createImplicitMentions(int $uri_id, int $parent_uri_id)
 339         {
 340                 // Always mention the direct parent author
 341                 $parent = Item::selectFirst(['author-link', 'author-name'], ['uri-id' => $parent_uri_id]);
 342                 self::store($uri_id, self::IMPLICIT_MENTION, $parent['author-name'], $parent['author-link']);
 343
 344                 if (DI::config()->get('system', 'disable_implicit_mentions')) {
 345                         return;
 346                 }
 347
 348                 $tags = DBA::select('tag-view', ['name', 'url'], ['uri-id' => $parent_uri_id]);
 349                 while ($tag = DBA::fetch($tags)) {
 350                         self::store($uri_id, self::IMPLICIT_MENTION, $tag['name'], $tag['url']);
 351                 }
 352                 DBA::close($tags);
 353         }
 354
 355         /**
 356          * Retrieves the terms from the provided type(s) associated with the provided item ID.
 357          *
 358          * @param int       $item_id
 359          * @param int|array $type
 360          * @return array
 361          * @throws \Exception
 362          */
 363         public static function getByURIId(int $uri_id, array $type = [self::HASHTAG, self::MENTION, self::IMPLICIT_MENTION, self::EXCLUSIVE_MENTION])
 364         {
 365                 $condition = ['uri-id' => $uri_id, 'type' => $type];
 366                 return DBA::selectToArray('tag-view', ['type', 'name', 'url'], $condition);
 367         }
 368
 369         /**
 370          * Return a string with all tags and mentions
 371          *
 372          * @param integer $uri_id
 373          * @param array   $type
 374          * @return string tags and mentions
 375          * @throws \Exception
 376          */
 377         public static function getCSVByURIId(int $uri_id, array $type = [self::HASHTAG, self::MENTION, self::IMPLICIT_MENTION, self::EXCLUSIVE_MENTION])
 378         {
 379                 $tag_list = [];
 380                 $tags = self::getByURIId($uri_id, $type);
 381                 foreach ($tags as $tag) {
 382                         $tag_list[] = self::TAG_CHARACTER[$tag['type']] . '[url=' . $tag['url'] . ']' . $tag['name'] . '[/url]';
 383                 }
 384
 385                 return implode(',', $tag_list);
 386         }
 387
 388         /**
 389          * Sorts an item's tags into mentions, hashtags and other tags. Generate personalized URLs by user and modify the
 390          * provided item's body with them.
 391          *
 392          * @param array $item
 393          * @return array
 394          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 395          * @throws \ImagickException
 396          */
 397         public static function populateFromItem(&$item)
 398         {
 399                 $return = [
 400                         'tags' => [],
 401                         'hashtags' => [],
 402                         'mentions' => [],
 403                         'implicit_mentions' => [],
 404                 ];
 405
 406                 $searchpath = DI::baseUrl() . "/search?tag=";
 407
 408                 $taglist = DBA::select('tag-view', ['type', 'name', 'url'],
 409                         ['uri-id' => $item['uri-id'], 'type' => [self::HASHTAG, self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION]]);
 410                 while ($tag = DBA::fetch($taglist)) {
 411                         if ($tag['url'] == '') {
 412                                 $tag['url'] = $searchpath . rawurlencode($tag['name']);
 413                         }
 414
 415                         $orig_tag = $tag['url'];
 416
 417                         $prefix = self::TAG_CHARACTER[$tag['type']];
 418                         switch($tag['type']) {
 419                                 case self::HASHTAG:
 420                                         if ($orig_tag != $tag['url']) {
 421                                                 $item['body'] = str_replace($orig_tag, $tag['url'], $item['body']);
 422                                         }
 423
 424                                         $return['hashtags'][] = $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['name']) . '</a>';
 425                                         $return['tags'][] = $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['name']) . '</a>';
 426                                         break;
 427                                 case self::MENTION:
 428                                 case self::EXCLUSIVE_MENTION:
 429                                                 $tag['url'] = Contact::magicLink($tag['url']);
 430                                         $return['mentions'][] = $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['name']) . '</a>';
 431                                         $return['tags'][] = $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['name']) . '</a>';
 432                                         break;
 433                                 case self::IMPLICIT_MENTION:
 434                                         $return['implicit_mentions'][] = $prefix . $tag['name'];
 435                                         break;
 436                         }
 437                 }
 438                 DBA::close($taglist);
 439
 440                 return $return;
 441         }
 442
 443         /**
 444          * Search posts for given tag
 445          *
 446          * @param string $search
 447          * @param integer $uid
 448          * @param integer $start
 449          * @param integer $limit
 450          * @return array with URI-ID
 451          */
 452         public static function getURIIdListByTag(string $search, int $uid = 0, int $start = 0, int $limit = 100)
 453         {
 454                 $condition = ["`name` = ? AND (NOT `private` OR (`private` AND `uid` = ?))", $search, $uid];
 455                 $params = [
 456                         'order' => ['uri-id' => true],
 457                         'group_by' => ['uri-id'],
 458                         'limit' => [$start, $limit]
 459                 ];
 460
 461                 $tags = DBA::select('tag-search-view', ['uri-id'], $condition, $params);
 462
 463                 $uriids = [];
 464                 while ($tag = DBA::fetch($tags)) {
 465                         $uriids[] = $tag['uri-id'];
 466                 }
 467                 DBA::close($tags);
 468
 469                 return $uriids;
 470         }
 471
 472         /**
 473          * Returns a list of the most frequent global hashtags over the given period
 474          *
 475          * @param int $period Period in hours to consider posts
 476          * @return array
 477          * @throws \Exception
 478          */
 479         public static function getGlobalTrendingHashtags(int $period, $limit = 10)
 480         {
 481                 $tags = DI::cache()->get('global_trending_tags');
 482
 483                 if (empty($tags)) {
 484                         $tagsStmt = DBA::p("SELECT `name` AS `term`, COUNT(*) AS `score`
 485                                 FROM `tag-search-view`
 486                                 WHERE `private` = ? AND `received` > DATE_SUB(NOW(), INTERVAL ? HOUR)
 487                                 GROUP BY `term` ORDER BY `score` DESC LIMIT ?",
 488                                 Item::PUBLIC, $period, $limit);
 489
 490                         if (DBA::isResult($tagsStmt)) {
 491                                 $tags = DBA::toArray($tagsStmt);
 492                                 DI::cache()->set('global_trending_tags', $tags, Duration::HOUR);
 493                         }
 494                 }
 495
 496                 return $tags ?: [];
 497         }
 498
 499         /**
 500          * Returns a list of the most frequent local hashtags over the given period
 501          *
 502          * @param int $period Period in hours to consider posts
 503          * @return array
 504          * @throws \Exception
 505          */
 506         public static function getLocalTrendingHashtags(int $period, $limit = 10)
 507         {
 508                 $tags = DI::cache()->get('local_trending_tags');
 509
 510                 if (empty($tags)) {
 511                         $tagsStmt = DBA::p("SELECT `name` AS `term`, COUNT(*) AS `score`
 512                                 FROM `tag-search-view`
 513                                 WHERE `private` = ? AND `wall` AND `origin` AND `received` > DATE_SUB(NOW(), INTERVAL ? HOUR)
 514                                 GROUP BY `term` ORDER BY `score` DESC LIMIT ?",
 515                                 Item::PUBLIC, $period, $limit);
 516
 517                         if (DBA::isResult($tagsStmt)) {
 518                                 $tags = DBA::toArray($tagsStmt);
 519                                 DI::cache()->set('local_trending_tags', $tags, Duration::HOUR);
 520                         }
 521                 }
 522
 523                 return $tags ?: [];
 524         }
 525
 526         /**
 527          * Check if the provided tag is of one of the provided term types.
 528          *
 529          * @param string $tag
 530          * @param int    ...$types
 531          * @return bool
 532          */
 533         public static function isType($tag, ...$types)
 534         {
 535                 $tag_chars = [];
 536                 foreach ($types as $type) {
 537                         if (array_key_exists($type, self::TAG_CHARACTER)) {
 538                                 $tag_chars[] = self::TAG_CHARACTER[$type];
 539                         }
 540                 }
 541
 542                 return Strings::startsWithChars($tag, $tag_chars);
 543         }
 544
 545         /**
 546          * Fetch user who subscribed to the given tag
 547          *
 548          * @param string $tag
 549          * @return array User list
 550          */
 551         private static function getUIDListByTag(string $tag)
 552         {
 553                 $uids = [];
 554                 $searches = DBA::select('search', ['uid'], ['term' => $tag]);
 555                 while ($search = DBA::fetch($searches)) {
 556                         $uids[] = $search['uid'];
 557                 }
 558                 DBA::close($searches);
 559
 560                 return $uids;
 561         }
 562
 563         /**
 564          * Fetch user who subscribed to the tags of the given item
 565          *
 566          * @param integer $uri_id
 567          * @return array User list
 568          */
 569         public static function getUIDListByURIId(int $uri_id)
 570         {
 571                 $uids = [];
 572                 $tags = self::getByURIId($uri_id, [self::HASHTAG]);
 573
 574                 foreach ($tags as $tag) {
 575                         $uids = array_merge($uids, self::getUIDListByTag(self::TAG_CHARACTER[self::HASHTAG] . $tag['name']));
 576                 }
 577
 578                 return array_unique($uids);
 579         }
 580 }