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 PCATEGORY         = 4;
  44     const FILE              = 5;
  45     const SAVEDSEARCH       = 6;
  46     const CONVERSATION      = 7;
  47         /**
  48          * An implicit mention is a mention in a comment body that is redundant with the threading information.
  49          */
  50     const IMPLICIT_MENTION  = 8;
  51         /**
  52          * An exclusive mention transfers the ownership of the post to the target account, usually a forum.
  53          */
  54     const EXCLUSIVE_MENTION = 9;
  55
  56     const TAG_CHARACTER = [
  57         self::HASHTAG           => '#',
  58         self::MENTION           => '@',
  59         self::IMPLICIT_MENTION  => '%',
  60         self::EXCLUSIVE_MENTION => '!',
  61     ];
  62
  63     const OBJECT_TYPE_POST  = 1;
  64     const OBJECT_TYPE_PHOTO = 2;
  65
  66         /**
  67          * Returns a list of the most frequent global hashtags over the given period
  68          *
  69          * @param int $period Period in hours to consider posts
  70          * @return array
  71          * @throws \Exception
  72          */
  73         public static function getGlobalTrendingHashtags(int $period, $limit = 10)
  74         {
  75                 $tags = DI::cache()->get('global_trending_tags');
  76
  77                 if (!$tags) {
  78                         $tagsStmt = DBA::p("SELECT t.`term`, COUNT(*) AS `score`
  79                                 FROM `term` t
  80                                  JOIN `item` i ON i.`id` = t.`oid` AND i.`uid` = t.`uid`
  81                                  JOIN `thread` ON `thread`.`iid` = i.`id`
  82                                 WHERE `thread`.`visible`
  83                                   AND NOT `thread`.`deleted`
  84                                   AND NOT `thread`.`moderated`
  85                                   AND NOT `thread`.`private`
  86                                   AND t.`uid` = 0
  87                                   AND t.`otype` = ?
  88                                   AND t.`type` = ?
  89                                   AND t.`term` != ''
  90                                   AND i.`received` > DATE_SUB(NOW(), INTERVAL ? HOUR)
  91                                 GROUP BY `term`
  92                                 ORDER BY `score` DESC
  93                                 LIMIT ?",
  94                                 Term::OBJECT_TYPE_POST,
  95                                 Term::HASHTAG,
  96                                 $period,
  97                                 $limit
  98                         );
  99
 100                         if (DBA::isResult($tagsStmt)) {
 101                                 $tags = DBA::toArray($tagsStmt);
 102                                 DI::cache()->set('global_trending_tags', $tags, Duration::HOUR);
 103                         }
 104                 }
 105
 106                 return $tags ?: [];
 107         }
 108
 109         /**
 110          * Returns a list of the most frequent local hashtags over the given period
 111          *
 112          * @param int $period Period in hours to consider posts
 113          * @return array
 114          * @throws \Exception
 115          */
 116         public static function getLocalTrendingHashtags(int $period, $limit = 10)
 117         {
 118                 $tags = DI::cache()->get('local_trending_tags');
 119
 120                 if (!$tags) {
 121                         $tagsStmt = DBA::p("SELECT t.`term`, COUNT(*) AS `score`
 122                                 FROM `term` t
 123                                 JOIN `item` i ON i.`id` = t.`oid` AND i.`uid` = t.`uid`
 124                                 JOIN `thread` ON `thread`.`iid` = i.`id`
 125                                 JOIN `user` ON `user`.`uid` = `thread`.`uid` AND NOT `user`.`hidewall`
 126                                 WHERE `thread`.`visible`
 127                                   AND NOT `thread`.`deleted`
 128                                   AND NOT `thread`.`moderated`
 129                                   AND NOT `thread`.`private`
 130                                   AND `thread`.`wall`
 131                                   AND `thread`.`origin`
 132                                   AND t.`otype` = ?
 133                                   AND t.`type` = ?
 134                                   AND t.`term` != ''
 135                                   AND i.`received` > DATE_SUB(NOW(), INTERVAL ? HOUR)
 136                                 GROUP BY `term`
 137                                 ORDER BY `score` DESC
 138                                 LIMIT ?",
 139                                 Term::OBJECT_TYPE_POST,
 140                                 Term::HASHTAG,
 141                                 $period,
 142                                 $limit
 143                         );
 144
 145                         if (DBA::isResult($tagsStmt)) {
 146                                 $tags = DBA::toArray($tagsStmt);
 147                                 DI::cache()->set('local_trending_tags', $tags, Duration::HOUR);
 148                         }
 149                 }
 150
 151                 return $tags ?: [];
 152         }
 153
 154         /**
 155          * Generates the legacy item.tag field comma-separated BBCode string from an item ID.
 156          * Includes only hashtags, implicit and explicit mentions.
 157          *
 158          * @param int $item_id
 159          * @return string
 160          * @throws \Exception
 161          */
 162         public static function tagTextFromItemId($item_id)
 163         {
 164                 $tag_list = [];
 165                 $tags = self::tagArrayFromItemId($item_id, [self::HASHTAG, self::MENTION, self::IMPLICIT_MENTION]);
 166                 foreach ($tags as $tag) {
 167                         $tag_list[] = self::TAG_CHARACTER[$tag['type']] . '[url=' . $tag['url'] . ']' . $tag['term'] . '[/url]';
 168                 }
 169
 170                 return implode(',', $tag_list);
 171         }
 172
 173         /**
 174          * Retrieves the terms from the provided type(s) associated with the provided item ID.
 175          *
 176          * @param int       $item_id
 177          * @param int|array $type
 178          * @return array
 179          * @throws \Exception
 180          */
 181         public static function tagArrayFromItemId($item_id, $type = [self::HASHTAG, self::MENTION])
 182         {
 183                 $condition = ['otype' => self::OBJECT_TYPE_POST, 'oid' => $item_id, 'type' => $type];
 184                 $tags = DBA::select('term', ['type', 'term', 'url'], $condition);
 185                 if (!DBA::isResult($tags)) {
 186                         return [];
 187                 }
 188
 189                 return DBA::toArray($tags);
 190         }
 191
 192         /**
 193          * Generates the legacy item.file field string from an item ID.
 194          * Includes only file and category terms.
 195          *
 196          * @param int $item_id
 197          * @return string
 198          * @throws \Exception
 199          */
 200         public static function fileTextFromItemId($item_id)
 201         {
 202                 $file_text = '';
 203                 $tags = self::tagArrayFromItemId($item_id, [self::FILE, self::CATEGORY]);
 204                 foreach ($tags as $tag) {
 205                         if ($tag['type'] == self::CATEGORY) {
 206                                 $file_text .= '<' . $tag['term'] . '>';
 207                         } else {
 208                                 $file_text .= '[' . $tag['term'] . ']';
 209                         }
 210                 }
 211
 212                 return $file_text;
 213         }
 214
 215         /**
 216          * Inserts new terms for the provided item ID based on the legacy item.tag field BBCode content.
 217          * Deletes all previous tag terms for the same item ID.
 218          * Sets both the item.mention and thread.mentions field flags if a mention concerning the item UID is found.
 219          *
 220          * @param int    $item_id
 221          * @param string $tag_str
 222          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 223          */
 224         public static function insertFromTagFieldByItemId($item_id, $tag_str)
 225         {
 226                 $profile_base = DI::baseUrl();
 227                 $profile_data = parse_url($profile_base);
 228                 $profile_path = $profile_data['path'] ?? '';
 229                 $profile_base_friendica = $profile_data['host'] . $profile_path . '/profile/';
 230                 $profile_base_diaspora = $profile_data['host'] . $profile_path . '/u/';
 231
 232                 $fields = ['guid', 'uid', 'id', 'edited', 'deleted', 'created', 'received', 'title', 'body', 'parent'];
 233                 $item = Item::selectFirst($fields, ['id' => $item_id]);
 234                 if (!DBA::isResult($item)) {
 235                         return;
 236                 }
 237
 238                 $item['tag'] = $tag_str;
 239
 240                 // Clean up all tags
 241                 self::deleteByItemId($item_id);
 242
 243                 if ($item['deleted']) {
 244                         return;
 245                 }
 246
 247                 $taglist = explode(',', $item['tag']);
 248
 249                 $tags_string = '';
 250                 foreach ($taglist as $tag) {
 251                         if (Strings::startsWith($tag, self::TAG_CHARACTER)) {
 252                                 $tags_string .= ' ' . trim($tag);
 253                         } else {
 254                                 $tags_string .= ' #' . trim($tag);
 255                         }
 256                 }
 257
 258                 $data = ' ' . $item['title'] . ' ' . $item['body'] . ' ' . $tags_string . ' ';
 259
 260                 // ignore anything in a code block
 261                 $data = preg_replace('/\[code\](.*?)\[\/code\]/sm', '', $data);
 262
 263                 $tags = [];
 264
 265                 $pattern = '/\W\#([^\[].*?)[\s\'".,:;\?!\[\]\/]/ism';
 266                 if (preg_match_all($pattern, $data, $matches)) {
 267                         foreach ($matches[1] as $match) {
 268                                 $tags['#' . $match] = '';
 269                         }
 270                 }
 271
 272                 $pattern = '/\W([\#@!%])\[url\=(.*?)\](.*?)\[\/url\]/ism';
 273                 if (preg_match_all($pattern, $data, $matches, PREG_SET_ORDER)) {
 274                         foreach ($matches as $match) {
 275
 276                                 if (in_array($match[1], [
 277                                         self::TAG_CHARACTER[self::MENTION],
 278                                         self::TAG_CHARACTER[self::IMPLICIT_MENTION],
 279                                         self::TAG_CHARACTER[self::EXCLUSIVE_MENTION]
 280                                 ])) {
 281                                         $contact = Contact::getDetailsByURL($match[2], 0);
 282                                         if (!empty($contact['addr'])) {
 283                                                 $match[3] = $contact['addr'];
 284                                         }
 285
 286                                         if (!empty($contact['url'])) {
 287                                                 $match[2] = $contact['url'];
 288                                         }
 289                                 }
 290
 291                                 $tags[$match[2]] = $match[1] . trim($match[3], ',.:;[]/\"?!');
 292                         }
 293                 }
 294
 295                 foreach ($tags as $link => $tag) {
 296                         if (self::isType($tag, self::HASHTAG)) {
 297                                 // try to ignore #039 or #1 or anything like that
 298                                 if (ctype_digit(substr(trim($tag), 1))) {
 299                                         continue;
 300                                 }
 301
 302                                 // try to ignore html hex escapes, e.g. #x2317
 303                                 if ((substr(trim($tag), 1, 1) == 'x' || substr(trim($tag), 1, 1) == 'X') && ctype_digit(substr(trim($tag), 2))) {
 304                                         continue;
 305                                 }
 306
 307                                 $type = self::HASHTAG;
 308                                 $term = substr($tag, 1);
 309                                 $link = '';
 310                         } elseif (self::isType($tag, self::MENTION, self::EXCLUSIVE_MENTION, self::IMPLICIT_MENTION)) {
 311                                 if (self::isType($tag, self::MENTION, self::EXCLUSIVE_MENTION)) {
 312                                         $type = self::MENTION;
 313                                 } else {
 314                                         $type = self::IMPLICIT_MENTION;
 315                                 }
 316
 317                                 $contact = Contact::getDetailsByURL($link, 0);
 318                                 if (!empty($contact['name'])) {
 319                                         $term = $contact['name'];
 320                                 } else {
 321                                         $term = substr($tag, 1);
 322                                 }
 323                         } else { // This shouldn't happen
 324                                 $type = self::HASHTAG;
 325                                 $term = $tag;
 326                                 $link = '';
 327
 328                                 Logger::notice('Unknown term type', ['tag' => $tag]);
 329                         }
 330
 331                         if (DBA::exists('term', ['uid' => $item['uid'], 'otype' => self::OBJECT_TYPE_POST, 'oid' => $item_id, 'term' => $term, 'type' => $type])) {
 332                                 continue;
 333                         }
 334
 335                         if ($item['uid'] == 0) {
 336                                 $global = true;
 337                                 DBA::update('term', ['global' => true], ['otype' => self::OBJECT_TYPE_POST, 'guid' => $item['guid']]);
 338                         } else {
 339                                 $global = DBA::exists('term', ['uid' => 0, 'otype' => self::OBJECT_TYPE_POST, 'guid' => $item['guid']]);
 340                         }
 341
 342                         DBA::insert('term', [
 343                                 'uid'      => $item['uid'],
 344                                 'oid'      => $item_id,
 345                                 'otype'    => self::OBJECT_TYPE_POST,
 346                                 'type'     => $type,
 347                                 'term'     => $term,
 348                                 'url'      => $link,
 349                                 'guid'     => $item['guid'],
 350                                 'created'  => $item['created'],
 351                                 'received' => $item['received'],
 352                                 'global'   => $global
 353                         ]);
 354
 355                         // Search for mentions
 356                         if (self::isType($tag, self::MENTION, self::EXCLUSIVE_MENTION)
 357                                 && (
 358                                         strpos($link, $profile_base_friendica) !== false
 359                                         || strpos($link, $profile_base_diaspora) !== false
 360                                 )
 361                         ) {
 362                                 $users_stmt = DBA::p("SELECT `uid` FROM `contact` WHERE self AND (`url` = ? OR `nurl` = ?)", $link, $link);
 363                                 $users = DBA::toArray($users_stmt);
 364                                 foreach ($users AS $user) {
 365                                         if ($user['uid'] == $item['uid']) {
 366                                                 /// @todo This function is called from Item::update - so we mustn't call that function here
 367                                                 DBA::update('item', ['mention' => true], ['id' => $item_id]);
 368                                                 DBA::update('thread', ['mention' => true], ['iid' => $item['parent']]);
 369                                         }
 370                                 }
 371                         }
 372                 }
 373         }
 374
 375         /**
 376          * Inserts new terms for the provided item ID based on the legacy item.file field BBCode content.
 377          * Deletes all previous file terms for the same item ID.
 378          *
 379          * @param integer $item_id item id
 380          * @param         $files
 381          * @return void
 382          * @throws \Exception
 383          */
 384         public static function insertFromFileFieldByItemId($item_id, $files)
 385         {
 386                 $message = Item::selectFirst(['uid', 'deleted'], ['id' => $item_id]);
 387                 if (!DBA::isResult($message)) {
 388                         return;
 389                 }
 390
 391                 // Clean up all tags
 392                 DBA::delete('term', ['otype' => self::OBJECT_TYPE_POST, 'oid' => $item_id, 'type' => [self::FILE, self::CATEGORY]]);
 393
 394                 if ($message["deleted"]) {
 395                         return;
 396                 }
 397
 398                 $message['file'] = $files;
 399
 400                 if (preg_match_all("/\[(.*?)\]/ism", $message["file"], $files)) {
 401                         foreach ($files[1] as $file) {
 402                                 DBA::insert('term', [
 403                                         'uid' => $message["uid"],
 404                                         'oid' => $item_id,
 405                                         'otype' => self::OBJECT_TYPE_POST,
 406                                         'type' => self::FILE,
 407                                         'term' => $file
 408                                 ]);
 409                         }
 410                 }
 411
 412                 if (preg_match_all("/\<(.*?)\>/ism", $message["file"], $files)) {
 413                         foreach ($files[1] as $file) {
 414                                 DBA::insert('term', [
 415                                         'uid' => $message["uid"],
 416                                         'oid' => $item_id,
 417                                         'otype' => self::OBJECT_TYPE_POST,
 418                                         'type' => self::CATEGORY,
 419                                         'term' => $file
 420                                 ]);
 421                         }
 422                 }
 423         }
 424
 425         /**
 426          * Sorts an item's tags into mentions, hashtags and other tags. Generate personalized URLs by user and modify the
 427          * provided item's body with them.
 428          *
 429          * @param array $item
 430          * @return array
 431          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 432          * @throws \ImagickException
 433          */
 434         public static function populateTagsFromItem(&$item)
 435         {
 436                 $return = [
 437                         'tags' => [],
 438                         'hashtags' => [],
 439                         'mentions' => [],
 440                         'implicit_mentions' => [],
 441                 ];
 442
 443                 $searchpath = DI::baseUrl() . "/search?tag=";
 444
 445                 $taglist = DBA::select(
 446                         'term',
 447                         ['type', 'term', 'url'],
 448                         ['otype' => self::OBJECT_TYPE_POST, 'oid' => $item['id'], 'type' => [self::HASHTAG, self::MENTION, self::IMPLICIT_MENTION]],
 449                         ['order' => ['tid']]
 450                 );
 451                 while ($tag = DBA::fetch($taglist)) {
 452                         if ($tag['url'] == '') {
 453                                 $tag['url'] = $searchpath . rawurlencode($tag['term']);
 454                         }
 455
 456                         $orig_tag = $tag['url'];
 457
 458                         $prefix = self::TAG_CHARACTER[$tag['type']];
 459                         switch($tag['type']) {
 460                                 case self::HASHTAG:
 461                                         if ($orig_tag != $tag['url']) {
 462                                                 $item['body'] = str_replace($orig_tag, $tag['url'], $item['body']);
 463                                         }
 464
 465                                         $return['hashtags'][] = $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['term']) . '</a>';
 466                                         $return['tags'][] = $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['term']) . '</a>';
 467                                         break;
 468                                 case self::MENTION:
 469                                         $tag['url'] = Contact::magicLink($tag['url']);
 470                                         $return['mentions'][] = $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['term']) . '</a>';
 471                                         $return['tags'][] = $prefix . '<a href="' . $tag['url'] . '" target="_blank" rel="noopener noreferrer">' . htmlspecialchars($tag['term']) . '</a>';
 472                                         break;
 473                                 case self::IMPLICIT_MENTION:
 474                                         $return['implicit_mentions'][] = $prefix . $tag['term'];
 475                                         break;
 476                         }
 477                 }
 478                 DBA::close($taglist);
 479
 480                 return $return;
 481         }
 482
 483         /**
 484          * Delete tags of the specific type(s) from an item
 485          *
 486          * @param int       $item_id
 487          * @param int|array $type
 488          * @throws \Exception
 489          */
 490         public static function deleteByItemId($item_id, $type = [self::HASHTAG, self::MENTION, self::IMPLICIT_MENTION])
 491         {
 492                 if (empty($item_id)) {
 493                         return;
 494                 }
 495
 496                 // Clean up all tags
 497                 DBA::delete('term', ['otype' => self::OBJECT_TYPE_POST, 'oid' => $item_id, 'type' => $type]);
 498         }
 499
 500         /**
 501          * Check if the provided tag is of one of the provided term types.
 502          *
 503          * @param string $tag
 504          * @param int    ...$types
 505          * @return bool
 506          */
 507         public static function isType($tag, ...$types)
 508         {
 509                 $tag_chars = [];
 510                 foreach ($types as $type) {
 511                         if (array_key_exists($type, self::TAG_CHARACTER)) {
 512                                 $tag_chars[] = self::TAG_CHARACTER[$type];
 513                         }
 514                 }
 515
 516                 return Strings::startsWith($tag, $tag_chars);
 517         }
 518 }