src/Model/Term.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\Core\Cache\Duration;
  25 use Friendica\Core\Logger;
  26 use Friendica\Database\DBA;
  27 use Friendica\DI;
  28 use Friendica\Util\Strings;
  29
  30 /**
  31  * Class Term
  32  *
  33  * This Model class handles term table interactions.
  34  * This tables stores relevant terms related to posts, photos and searches, like hashtags, mentions and
  35  * user-applied categories.
  36  */
  37 class Term
  38 {
  39     const UNKNOWN           = 0;
  40     const HASHTAG           = 1;
  41     const MENTION           = 2;
  42     const CATEGORY          = 3;
  43     const FILE              = 5;
  44         /**
  45          * An implicit mention is a mention in a comment body that is redundant with the threading information.
  46          */
  47     const IMPLICIT_MENTION  = 8;
  48         /**
  49          * An exclusive mention transfers the ownership of the post to the target account, usually a forum.
  50          */
  51     const EXCLUSIVE_MENTION = 9;
  52
  53     const TAG_CHARACTER = [
  54         self::HASHTAG           => '#',
  55         self::MENTION           => '@',
  56         self::IMPLICIT_MENTION  => '%',
  57         self::EXCLUSIVE_MENTION => '!',
  58     ];
  59
  60     const OBJECT_TYPE_POST  = 1;
  61     const OBJECT_TYPE_PHOTO = 2;
  62
  63         /**
  64          * Returns a list of the most frequent global hashtags over the given period
  65          *
  66          * @param int $period Period in hours to consider posts
  67          * @return array
  68          * @throws \Exception
  69          */
  70         public static function getGlobalTrendingHashtags(int $period, $limit = 10)
  71         {
  72                 $tags = DI::cache()->get('global_trending_tags');
  73
  74                 if (!$tags) {
  75                         $tagsStmt = DBA::p("SELECT t.`term`, COUNT(*) AS `score`
  76                                 FROM `term` t
  77                                  JOIN `item` i ON i.`id` = t.`oid` AND i.`uid` = t.`uid`
  78                                  JOIN `thread` ON `thread`.`iid` = i.`id`
  79                                 WHERE `thread`.`visible`
  80                                   AND NOT `thread`.`deleted`
  81                                   AND NOT `thread`.`moderated`
  82                                   AND `thread`.`private` = ?
  83                                   AND t.`uid` = 0
  84                                   AND t.`otype` = ?
  85                                   AND t.`type` = ?
  86                                   AND t.`term` != ''
  87                                   AND i.`received` > DATE_SUB(NOW(), INTERVAL ? HOUR)
  88                                 GROUP BY `term`
  89                                 ORDER BY `score` DESC
  90                                 LIMIT ?",
  91                                 Item::PUBLIC,
  92                                 Term::OBJECT_TYPE_POST,
  93                                 Term::HASHTAG,
  94                                 $period,
  95                                 $limit
  96                         );
  97
  98                         if (DBA::isResult($tagsStmt)) {
  99                                 $tags = DBA::toArray($tagsStmt);
 100                                 DI::cache()->set('global_trending_tags', $tags, Duration::HOUR);
 101                         }
 102                 }
 103
 104                 return $tags ?: [];
 105         }
 106
 107         /**
 108          * Returns a list of the most frequent local hashtags over the given period
 109          *
 110          * @param int $period Period in hours to consider posts
 111          * @return array
 112          * @throws \Exception
 113          */
 114         public static function getLocalTrendingHashtags(int $period, $limit = 10)
 115         {
 116                 $tags = DI::cache()->get('local_trending_tags');
 117
 118                 if (!$tags) {
 119                         $tagsStmt = DBA::p("SELECT t.`term`, COUNT(*) AS `score`
 120                                 FROM `term` t
 121                                 JOIN `item` i ON i.`id` = t.`oid` AND i.`uid` = t.`uid`
 122                                 JOIN `thread` ON `thread`.`iid` = i.`id`
 123                                 WHERE `thread`.`visible`
 124                                   AND NOT `thread`.`deleted`
 125                                   AND NOT `thread`.`moderated`
 126                                   AND `thread`.`private` = ?
 127                                   AND `thread`.`wall`
 128                                   AND `thread`.`origin`
 129                                   AND t.`otype` = ?
 130                                   AND t.`type` = ?
 131                                   AND t.`term` != ''
 132                                   AND i.`received` > DATE_SUB(NOW(), INTERVAL ? HOUR)
 133                                 GROUP BY `term`
 134                                 ORDER BY `score` DESC
 135                                 LIMIT ?",
 136                                 Item::PUBLIC,
 137                                 Term::OBJECT_TYPE_POST,
 138                                 Term::HASHTAG,
 139                                 $period,
 140                                 $limit
 141                         );
 142
 143                         if (DBA::isResult($tagsStmt)) {
 144                                 $tags = DBA::toArray($tagsStmt);
 145                                 DI::cache()->set('local_trending_tags', $tags, Duration::HOUR);
 146                         }
 147                 }
 148
 149                 return $tags ?: [];
 150         }
 151
 152         /**
 153          * Generates the legacy item.tag field comma-separated BBCode string from an item ID.
 154          * Includes only hashtags, implicit and explicit mentions.
 155          *
 156          * @param int $item_id
 157          * @return string
 158          * @throws \Exception
 159          */
 160         public static function tagTextFromItemId($item_id)
 161         {
 162                 $tag_list = [];
 163                 $tags = self::tagArrayFromItemId($item_id, [self::HASHTAG, self::MENTION, self::IMPLICIT_MENTION]);
 164                 foreach ($tags as $tag) {
 165                         $tag_list[] = self::TAG_CHARACTER[$tag['type']] . '[url=' . $tag['url'] . ']' . $tag['term'] . '[/url]';
 166                 }
 167
 168                 return implode(',', $tag_list);
 169         }
 170
 171         /**
 172          * Retrieves the terms from the provided type(s) associated with the provided item ID.
 173          *
 174          * @param int       $item_id
 175          * @param int|array $type
 176          * @return array
 177          * @throws \Exception
 178          */
 179         public static function tagArrayFromItemId($item_id, $type = [self::HASHTAG, self::MENTION])
 180         {
 181                 $condition = ['otype' => self::OBJECT_TYPE_POST, 'oid' => $item_id, 'type' => $type];
 182                 $tags = DBA::select('term', ['type', 'term', 'url'], $condition);
 183                 if (!DBA::isResult($tags)) {
 184                         return [];
 185                 }
 186
 187                 return DBA::toArray($tags);
 188         }
 189
 190         /**
 191          * Generates the legacy item.file field string from an item ID.
 192          * Includes only file and category terms.
 193          *
 194          * @param int $item_id
 195          * @return string
 196          * @throws \Exception
 197          */
 198         public static function fileTextFromItemId($item_id)
 199         {
 200                 $file_text = '';
 201                 $tags = self::tagArrayFromItemId($item_id, [self::FILE, self::CATEGORY]);
 202                 foreach ($tags as $tag) {
 203                         if ($tag['type'] == self::CATEGORY) {
 204                                 $file_text .= '<' . $tag['term'] . '>';
 205                         } else {
 206                                 $file_text .= '[' . $tag['term'] . ']';
 207                         }
 208                 }
 209
 210                 return $file_text;
 211         }
 212
 213         /**
 214          * Inserts new terms for the provided item ID based on the legacy item.tag field BBCode content.
 215          * Deletes all previous tag terms for the same item ID.
 216          * Sets both the item.mention and thread.mentions field flags if a mention concerning the item UID is found.
 217          *
 218          * @param int    $item_id
 219          * @param string $tag_str
 220          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 221          */
 222         public static function insertFromTagFieldByItemId($item_id, $tag_str)
 223         {
 224                 $profile_base = DI::baseUrl();
 225                 $profile_data = parse_url($profile_base);
 226                 $profile_path = $profile_data['path'] ?? '';
 227                 $profile_base_friendica = $profile_data['host'] . $profile_path . '/profile/';
 228                 $profile_base_diaspora = $profile_data['host'] . $profile_path . '/u/';
 229
 230                 $fields = ['guid', 'uid', 'id', 'edited', 'deleted', 'created', 'received', 'title', 'body', 'parent'];
 231                 $item = Item::selectFirst($fields, ['id' => $item_id]);
 232                 if (!DBA::isResult($item)) {
 233                         return;
 234                 }
 235
 236                 $item['tag'] = $tag_str;
 237
 238                 // Clean up all tags
 239                 self::deleteByItemId($item_id);
 240
 241                 if ($item['deleted']) {
 242                         return;
 243                 }
 244
 245                 $taglist = explode(',', $item['tag']);
 246
 247                 $tags_string = '';
 248                 foreach ($taglist as $tag) {
 249                         if (Strings::startsWith($tag, self::TAG_CHARACTER)) {
 250                                 $tags_string .= ' ' . trim($tag);
 251                         } else {
 252                                 $tags_string .= ' #' . trim($tag);
 253                         }
 254                 }
 255
 256                 $data = ' ' . $item['title'] . ' ' . $item['body'] . ' ' . $tags_string . ' ';
 257
 258                 // ignore anything in a code block
 259                 $data = preg_replace('/\[code\](.*?)\[\/code\]/sm', '', $data);
 260
 261                 $tags = [];
 262
 263                 $pattern = '/\W\#([^\[].*?)[\s\'".,:;\?!\[\]\/]/ism';
 264                 if (preg_match_all($pattern, $data, $matches)) {
 265                         foreach ($matches[1] as $match) {
 266                                 $tags['#' . $match] = '';
 267                         }
 268                 }
 269
 270                 $pattern = '/\W([\#@!%])\[url\=(.*?)\](.*?)\[\/url\]/ism';
 271                 if (preg_match_all($pattern, $data, $matches, PREG_SET_ORDER)) {
 272                         foreach ($matches as $match) {
 273
 274                                 if (in_array($match[1], [
 275                                         self::TAG_CHARACTER[self::MENTION],
 276                                         self::TAG_CHARACTER[self::IMPLICIT_MENTION],
 277                                         self::TAG_CHARACTER[self::EXCLUSIVE_MENTION]
 278                                 ])) {
 279                                         $contact = Contact::getDetailsByURL($match[2], 0);
 280                                         if (!empty($contact['addr'])) {
 281                                                 $match[3] = $contact['addr'];
 282                                         }
 283
 284                                         if (!empty($contact['url'])) {
 285                                                 $match[2] = $contact['url'];
 286                                         }
 287                                 }
 288
 289                                 $tags[$match[2]] = $match[1] . trim($match[3], ',.:;[]/\"?!');
 290                         }
 291                 }
 292
 293                 foreach ($tags as $link => $tag) {
 294                         if (self::isType($tag, self::HASHTAG)) {
 295                                 // try to ignore #039 or #1 or anything like that
 296                                 if (ctype_digit(substr(trim($tag), 1))) {
 297                                         continue;
 298                                 }
 299
 300                                 // try to ignore html hex escapes, e.g. #x2317
 301                                 if ((substr(trim($tag), 1, 1) == 'x' || substr(trim($tag), 1, 1) == 'X') && ctype_digit(substr(trim($tag), 2))) {
 302                                         continue;
 303                                 }
 304
 305                                 $type = self::HASHTAG;
 306                                 $term = substr($tag, 1);
 307                                 $link = '';
 308                         } elseif (self::isType($tag, self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION)) {
 309                                 if (self::isType($tag, self::MENTION, self::EXCLUSIVE_MENTION)) {
 310                                         $type = self::MENTION;
 311                                 } else {
 312                                         $type = self::IMPLICIT_MENTION;
 313                                 }
 314
 315                                 $contact = Contact::getDetailsByURL($link, 0);
 316                                 if (!empty($contact['name'])) {
 317                                         $term = $contact['name'];
 318                                 } else {
 319                                         $term = substr($tag, 1);
 320                                 }
 321                         } else { // This shouldn't happen
 322                                 $type = self::HASHTAG;
 323                                 $term = $tag;
 324                                 $link = '';
 325
 326                                 Logger::notice('Unknown term type', ['tag' => $tag]);
 327                         }
 328
 329                         if (DBA::exists('term', ['uid' => $item['uid'], 'otype' => self::OBJECT_TYPE_POST, 'oid' => $item_id, 'term' => $term, 'type' => $type])) {
 330                                 continue;
 331                         }
 332
 333                         if (empty($term)) {
 334                                 continue;
 335                         }
 336
 337                         if ($item['uid'] == 0) {
 338                                 $global = true;
 339                                 DBA::update('term', ['global' => true], ['otype' => self::OBJECT_TYPE_POST, 'guid' => $item['guid']]);
 340                         } else {
 341                                 $global = DBA::exists('term', ['uid' => 0, 'otype' => self::OBJECT_TYPE_POST, 'guid' => $item['guid']]);
 342                         }
 343
 344                         DBA::insert('term', [
 345                                 'uid'      => $item['uid'],
 346                                 'oid'      => $item_id,
 347                                 'otype'    => self::OBJECT_TYPE_POST,
 348                                 'type'     => $type,
 349                                 'term'     => substr($term, 0, 255),
 350                                 'url'      => $link,
 351                                 'guid'     => $item['guid'],
 352                                 'created'  => $item['created'],
 353                                 'received' => $item['received'],
 354                                 'global'   => $global
 355                         ]);
 356
 357                         // Search for mentions
 358                         if (self::isType($tag, self::MENTION, self::EXCLUSIVE_MENTION)
 359                                 && (
 360                                         strpos($link, $profile_base_friendica) !== false
 361                                         || strpos($link, $profile_base_diaspora) !== false
 362                                 )
 363                         ) {
 364                                 $users_stmt = DBA::p("SELECT `uid` FROM `contact` WHERE self AND (`url` = ? OR `nurl` = ?)", $link, $link);
 365                                 $users = DBA::toArray($users_stmt);
 366                                 foreach ($users AS $user) {
 367                                         if ($user['uid'] == $item['uid']) {
 368                                                 /// @todo This function is called from Item::update - so we mustn't call that function here
 369                                                 DBA::update('item', ['mention' => true], ['id' => $item_id]);
 370                                                 DBA::update('thread', ['mention' => true], ['iid' => $item['parent']]);
 371                                         }
 372                                 }
 373                         }
 374                 }
 375         }
 376
 377         /**
 378          * Inserts new terms for the provided item ID based on the legacy item.file field BBCode content.
 379          * Deletes all previous file terms for the same item ID.
 380          *
 381          * @param integer $item_id item id
 382          * @param         $files
 383          * @return void
 384          * @throws \Exception
 385          */
 386         public static function insertFromFileFieldByItemId($item_id, $files)
 387         {
 388                 $message = Item::selectFirst(['uid', 'deleted'], ['id' => $item_id]);
 389                 if (!DBA::isResult($message)) {
 390                         return;
 391                 }
 392
 393                 // Clean up all tags
 394                 DBA::delete('term', ['otype' => self::OBJECT_TYPE_POST, 'oid' => $item_id, 'type' => [self::FILE, self::CATEGORY]]);
 395
 396                 if ($message["deleted"]) {
 397                         return;
 398                 }
 399
 400                 $message['file'] = $files;
 401
 402                 if (preg_match_all("/\[(.*?)\]/ism", $message["file"], $files)) {
 403                         foreach ($files[1] as $file) {
 404                                 DBA::insert('term', [
 405                                         'uid' => $message["uid"],
 406                                         'oid' => $item_id,
 407                                         'otype' => self::OBJECT_TYPE_POST,
 408                                         'type' => self::FILE,
 409                                         'term' => $file
 410                                 ]);
 411                         }
 412                 }
 413
 414                 if (preg_match_all("/\<(.*?)\>/ism", $message["file"], $files)) {
 415                         foreach ($files[1] as $file) {
 416                                 DBA::insert('term', [
 417                                         'uid' => $message["uid"],
 418                                         'oid' => $item_id,
 419                                         'otype' => self::OBJECT_TYPE_POST,
 420                                         'type' => self::CATEGORY,
 421                                         'term' => $file
 422                                 ]);
 423                         }
 424                 }
 425         }
 426
 427         /**
 428          * Sorts an item's tags into mentions, hashtags and other tags. Generate personalized URLs by user and modify the
 429          * provided item's body with them.
 430          *
 431          * @param array $item
 432          * @return array
 433          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 434          * @throws \ImagickException
 435          */
 436         public static function populateTagsFromItem(&$item)
 437         {
 438                 $return = [
 439                         'tags' => [],
 440                         'hashtags' => [],
 441                         'mentions' => [],
 442                         'implicit_mentions' => [],
 443                 ];
 444
 445                 $searchpath = DI::baseUrl() . "/search?tag=";
 446
 447                 $taglist = DBA::select(
 448                         'term',
 449                         ['type', 'term', 'url'],
 450                         ['otype' => self::OBJECT_TYPE_POST, 'oid' => $item['id'], 'type' => [self::HASHTAG, self::MENTION, self::IMPLICIT_MENTION]],
 451                         ['order' => ['tid']]
 452                 );
 453                 while ($tag = DBA::fetch($taglist)) {
 454                         if ($tag['url'] == '') {
 455                                 $tag['url'] = $searchpath . rawurlencode($tag['term']);
 456                         }
 457
 458                         $orig_tag = $tag['url'];
 459
 460                         $prefix = self::TAG_CHARACTER[$tag['type']];
 461                         switch($tag['type']) {
 462                                 case self::HASHTAG:
 463                                         if ($orig_tag != $tag['url']) {
 464                                                 $item['body'] = str_replace($orig_tag, $tag['url'], $item['body']);
 465                                         }
 466
 467                                         $return['hashtags'][] = $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['term']) . '</a>';
 468                                         $return['tags'][] = $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['term']) . '</a>';
 469                                         break;
 470                                 case self::MENTION:
 471                                         $tag['url'] = Contact::magicLink($tag['url']);
 472                                         $return['mentions'][] = $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['term']) . '</a>';
 473                                         $return['tags'][] = $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['term']) . '</a>';
 474                                         break;
 475                                 case self::IMPLICIT_MENTION:
 476                                         $return['implicit_mentions'][] = $prefix . $tag['term'];
 477                                         break;
 478                         }
 479                 }
 480                 DBA::close($taglist);
 481
 482                 return $return;
 483         }
 484
 485         /**
 486          * Delete tags of the specific type(s) from an item
 487          *
 488          * @param int       $item_id
 489          * @param int|array $type
 490          * @throws \Exception
 491          */
 492         public static function deleteByItemId($item_id, $type = [self::HASHTAG, self::MENTION, self::IMPLICIT_MENTION])
 493         {
 494                 if (empty($item_id)) {
 495                         return;
 496                 }
 497
 498                 // Clean up all tags
 499                 DBA::delete('term', ['otype' => self::OBJECT_TYPE_POST, 'oid' => $item_id, 'type' => $type]);
 500         }
 501
 502         /**
 503          * Check if the provided tag is of one of the provided term types.
 504          *
 505          * @param string $tag
 506          * @param int    ...$types
 507          * @return bool
 508          */
 509         public static function isType($tag, ...$types)
 510         {
 511                 $tag_chars = [];
 512                 foreach ($types as $type) {
 513                         if (array_key_exists($type, self::TAG_CHARACTER)) {
 514                                 $tag_chars[] = self::TAG_CHARACTER[$type];
 515                         }
 516                 }
 517
 518                 return Strings::startsWith($tag, $tag_chars);
 519         }
 520 }