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