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