src/Model/Post/Media.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2022, the Friendica project
   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\Post;
  23
  24 use Friendica\Content\Text\BBCode;
  25 use Friendica\Core\Logger;
  26 use Friendica\Core\Protocol;
  27 use Friendica\Core\System;
  28 use Friendica\Database\Database;
  29 use Friendica\Database\DBA;
  30 use Friendica\DI;
  31 use Friendica\Model\Contact;
  32 use Friendica\Model\Item;
  33 use Friendica\Model\Photo;
  34 use Friendica\Model\Post;
  35 use Friendica\Network\HTTPClient\Client\HttpClientAccept;
  36 use Friendica\Network\HTTPClient\Client\HttpClientOptions;
  37 use Friendica\Util\Images;
  38 use Friendica\Util\Network;
  39 use Friendica\Util\ParseUrl;
  40 use Friendica\Util\Proxy;
  41 use Friendica\Util\Strings;
  42
  43 /**
  44  * Class Media
  45  *
  46  * This Model class handles media interactions.
  47  * This tables stores medias (images, videos, audio files) related to posts.
  48  */
  49 class Media
  50 {
  51         const UNKNOWN     = 0;
  52         const IMAGE       = 1;
  53         const VIDEO       = 2;
  54         const AUDIO       = 3;
  55         const TEXT        = 4;
  56         const APPLICATION = 5;
  57         const TORRENT     = 16;
  58         const HTML        = 17;
  59         const XML         = 18;
  60         const PLAIN       = 19;
  61         const ACTIVITY    = 20;
  62         const ACCOUNT     = 21;
  63         const DOCUMENT    = 128;
  64
  65         /**
  66          * Insert a post-media record
  67          *
  68          * @param array $media
  69          * @return void
  70          */
  71         public static function insert(array $media, bool $force = false)
  72         {
  73                 if (empty($media['url']) || empty($media['uri-id']) || !isset($media['type'])) {
  74                         Logger::warning('Incomplete media data', ['media' => $media]);
  75                         return;
  76                 }
  77
  78                 if (DBA::exists('post-media', ['uri-id' => $media['uri-id'], 'preview' => $media['url']])) {
  79                         Logger::info('Media already exists as preview', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'callstack' => System::callstack()]);
  80                         return;
  81                 }
  82
  83                 // "document" has got the lowest priority. So when the same file is both attached as document
  84                 // and embedded as picture then we only store the picture or replace the document
  85                 $found = DBA::selectFirst('post-media', ['type'], ['uri-id' => $media['uri-id'], 'url' => $media['url']]);
  86                 if (!$force && !empty($found) && (($found['type'] != self::DOCUMENT) || ($media['type'] == self::DOCUMENT))) {
  87                         Logger::info('Media already exists', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'callstack' => System::callstack()]);
  88                         return;
  89                 }
  90
  91                 $media = self::unsetEmptyFields($media);
  92                 $media = DI::dbaDefinition()->truncateFieldsForTable('post-media', $media);
  93
  94                 // We are storing as fast as possible to avoid duplicated network requests
  95                 // when fetching additional information for pictures and other content.
  96                 $result = DBA::insert('post-media', $media, Database::INSERT_UPDATE);
  97                 Logger::info('Stored media', ['result' => $result, 'media' => $media, 'callstack' => System::callstack()]);
  98                 $stored = $media;
  99
 100                 $media = self::fetchAdditionalData($media);
 101                 $media = self::unsetEmptyFields($media);
 102                 $media = DI::dbaDefinition()->truncateFieldsForTable('post-media', $media);
 103
 104                 if (array_diff_assoc($media, $stored)) {
 105                         $result = DBA::insert('post-media', $media, Database::INSERT_UPDATE);
 106                         Logger::info('Updated media', ['result' => $result, 'media' => $media]);
 107                 } else {
 108                         Logger::info('Nothing to update', ['media' => $media]);
 109                 }
 110         }
 111
 112         /**
 113          * Remove empty media fields
 114          *
 115          * @param array $media
 116          * @return array cleaned media array
 117          */
 118         private static function unsetEmptyFields(array $media): array
 119         {
 120                 $fields = ['mimetype', 'height', 'width', 'size', 'preview', 'preview-height', 'preview-width', 'blurhash', 'description'];
 121                 foreach ($fields as $field) {
 122                         if (empty($media[$field])) {
 123                                 unset($media[$field]);
 124                         }
 125                 }
 126                 return $media;
 127         }
 128
 129         /**
 130          * Copy attachments from one uri-id to another
 131          *
 132          * @param integer $from_uri_id
 133          * @param integer $to_uri_id
 134          * @return void
 135          */
 136         public static function copy(int $from_uri_id, int $to_uri_id)
 137         {
 138                 $attachments = self::getByURIId($from_uri_id);
 139                 foreach ($attachments as $attachment) {
 140                         $attachment['uri-id'] = $to_uri_id;
 141                         self::insert($attachment);
 142                 }
 143         }
 144
 145         /**
 146          * Creates the "[attach]" element from the given attributes
 147          *
 148          * @param string $href
 149          * @param integer $length
 150          * @param string $type
 151          * @param string $title
 152          * @return string "[attach]" element
 153          */
 154         public static function getAttachElement(string $href, int $length, string $type, string $title = ''): string
 155         {
 156                 $media = self::fetchAdditionalData(['type' => self::DOCUMENT, 'url' => $href,
 157                         'size' => $length, 'mimetype' => $type, 'description' => $title]);
 158
 159                 return '[attach]href="' . $media['url'] . '" length="' . $media['size'] .
 160                         '" type="' . $media['mimetype'] . '" title="' . $media['description'] . '"[/attach]';
 161         }
 162
 163         /**
 164          * Fetch additional data for the provided media array
 165          *
 166          * @param array $media
 167          * @return array media array with additional data
 168          */
 169         public static function fetchAdditionalData(array $media): array
 170         {
 171                 if (Network::isLocalLink($media['url'])) {
 172                         $media = self::fetchLocalData($media);
 173                 }
 174
 175                 // Fetch the mimetype or size if missing.
 176                 if (empty($media['mimetype']) || empty($media['size'])) {
 177                         $timeout = DI::config()->get('system', 'xrd_timeout');
 178                         $curlResult = DI::httpClient()->head($media['url'], [HttpClientOptions::TIMEOUT => $timeout]);
 179
 180                         // Workaround for systems that can't handle a HEAD request
 181                         if (!$curlResult->isSuccess() && ($curlResult->getReturnCode() == 405)) {
 182                                 $curlResult = DI::httpClient()->get($media['url'], HttpClientAccept::DEFAULT, [HttpClientOptions::TIMEOUT => $timeout]);
 183                         }
 184
 185                         if ($curlResult->isSuccess()) {
 186                                 if (empty($media['mimetype'])) {
 187                                         $media['mimetype'] = $curlResult->getHeader('Content-Type')[0] ?? '';
 188                                 }
 189                                 if (empty($media['size'])) {
 190                                         $media['size'] = (int)($curlResult->getHeader('Content-Length')[0] ?? 0);
 191                                 }
 192                         } else {
 193                                 Logger::notice('Could not fetch head', ['media' => $media]);
 194                         }
 195                 }
 196
 197                 $filetype = !empty($media['mimetype']) ? strtolower(current(explode('/', $media['mimetype']))) : '';
 198
 199                 if (($media['type'] == self::IMAGE) || ($filetype == 'image')) {
 200                         $imagedata = Images::getInfoFromURLCached($media['url']);
 201                         if ($imagedata) {
 202                                 $media['mimetype'] = $imagedata['mime'];
 203                                 $media['size'] = $imagedata['size'];
 204                                 $media['width'] = $imagedata[0];
 205                                 $media['height'] = $imagedata[1];
 206                                 $media['blurhash'] = $imagedata['blurhash'] ?? null;
 207                         } else {
 208                                 Logger::notice('No image data', ['media' => $media]);
 209                         }
 210                         if (!empty($media['preview'])) {
 211                                 $imagedata = Images::getInfoFromURLCached($media['preview']);
 212                                 if ($imagedata) {
 213                                         $media['preview-width'] = $imagedata[0];
 214                                         $media['preview-height'] = $imagedata[1];
 215                                 }
 216                         }
 217                 }
 218
 219                 if ($media['type'] != self::DOCUMENT) {
 220                         $media = self::addType($media);
 221                 }
 222
 223                 if (in_array($media['type'], [self::TEXT, self::APPLICATION, self::HTML, self::XML, self::PLAIN])) {
 224                         $media = self::addActivity($media);
 225                 }
 226
 227                 if (in_array($media['type'], [self::TEXT, self::APPLICATION, self::HTML, self::XML, self::PLAIN])) {
 228                         $media = self::addAccount($media);
 229                 }
 230
 231                 if ($media['type'] == self::HTML) {
 232                         $data = ParseUrl::getSiteinfoCached($media['url'], false);
 233                         $media['preview'] = $data['images'][0]['src'] ?? null;
 234                         $media['preview-height'] = $data['images'][0]['height'] ?? null;
 235                         $media['preview-width'] = $data['images'][0]['width'] ?? null;
 236                         $media['blurhash'] = $data['images'][0]['blurhash'] ?? null;
 237                         $media['description'] = $data['text'] ?? null;
 238                         $media['name'] = $data['title'] ?? null;
 239                         $media['author-url'] = $data['author_url'] ?? null;
 240                         $media['author-name'] = $data['author_name'] ?? null;
 241                         $media['author-image'] = $data['author_img'] ?? null;
 242                         $media['publisher-url'] = $data['publisher_url'] ?? null;
 243                         $media['publisher-name'] = $data['publisher_name'] ?? null;
 244                         $media['publisher-image'] = $data['publisher_img'] ?? null;
 245                 }
 246                 return $media;
 247         }
 248
 249         /**
 250          * Adds the activity type if the media entry is linked to an activity
 251          *
 252          * @param array $media
 253          * @return array
 254          */
 255         private static function addActivity(array $media): array
 256         {
 257                 $id = Item::fetchByLink($media['url']);
 258                 if (empty($id)) {
 259                         return $media;
 260                 }
 261
 262                 $item = Post::selectFirst([], ['id' => $id, 'network' => Protocol::FEDERATED]);
 263                 if (empty($item['id'])) {
 264                         Logger::debug('Not a federated activity', ['id' => $id, 'uri-id' => $media['uri-id'], 'url' => $media['url']]);
 265                         return $media;
 266                 }
 267
 268                 if (!empty($item['plink']) && Strings::compareLink($item['plink'], $media['url']) &&
 269                         parse_url($item['plink'], PHP_URL_HOST) != parse_url($item['uri'], PHP_URL_HOST)) {
 270                         Logger::debug('Not a link to an activity', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'plink' => $item['plink'], 'uri' => $item['uri']]);
 271                         return $media;
 272                 }
 273
 274                 if (in_array($item['network'], [Protocol::ACTIVITYPUB, Protocol::DFRN])) {
 275                         $media['mimetype'] = 'application/activity+json';
 276                 } elseif ($item['network'] == Protocol::DIASPORA) {
 277                         $media['mimetype'] = 'application/xml';
 278                 }
 279
 280                 $contact = Contact::getById($item['author-id'], ['avatar', 'gsid']);
 281                 if (!empty($contact['gsid'])) {
 282                         $gserver = DBA::selectFirst('gserver', ['url', 'site_name'], ['id' => $contact['gsid']]);
 283                 }
 284
 285                 $media['type'] = self::ACTIVITY;
 286                 $media['media-uri-id'] = $item['uri-id'];
 287                 $media['height'] = null;
 288                 $media['width'] = null;
 289                 $media['preview'] = null;
 290                 $media['preview-height'] = null;
 291                 $media['preview-width'] = null;
 292                 $media['blurhash'] = null;
 293                 $media['description'] = $item['body'];
 294                 $media['name'] = $item['title'];
 295                 $media['author-url'] = $item['author-link'];
 296                 $media['author-name'] = $item['author-name'];
 297                 $media['author-image'] = $contact['avatar'] ?? $item['author-avatar'];
 298                 $media['publisher-url'] = $gserver['url'] ?? null;
 299                 $media['publisher-name'] = $gserver['site_name'] ?? null;
 300                 $media['publisher-image'] = null;
 301
 302                 Logger::debug('Activity detected', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'plink' => $item['plink'], 'uri' => $item['uri']]);
 303                 return $media;
 304         }
 305
 306         /**
 307          * Adds the account type if the media entry is linked to an account
 308          *
 309          * @param array $media
 310          * @return array
 311          */
 312         private static function addAccount(array $media): array
 313         {
 314                 $contact = Contact::getByURL($media['url'], false);
 315                 if (empty($contact) || ($contact['network'] == Protocol::PHANTOM)) {
 316                         return $media;
 317                 }
 318
 319                 if (in_array($contact['network'], [Protocol::ACTIVITYPUB, Protocol::DFRN])) {
 320                         $media['mimetype'] = 'application/activity+json';
 321                 }
 322
 323                 if (!empty($contact['gsid'])) {
 324                         $gserver = DBA::selectFirst('gserver', ['url', 'site_name'], ['id' => $contact['gsid']]);
 325                 }
 326
 327                 $media['type'] = self::ACCOUNT;
 328                 $media['media-uri-id'] = $contact['uri-id'];
 329                 $media['height'] = null;
 330                 $media['width'] = null;
 331                 $media['preview'] = null;
 332                 $media['preview-height'] = null;
 333                 $media['preview-width'] = null;
 334                 $media['blurhash'] = null;
 335                 $media['description'] = $contact['about'];
 336                 $media['name'] = $contact['name'];
 337                 $media['author-url'] = $contact['url'];
 338                 $media['author-name'] = $contact['name'];
 339                 $media['author-image'] = $contact['avatar'];
 340                 $media['publisher-url'] = $gserver['url'] ?? null;
 341                 $media['publisher-name'] = $gserver['site_name'] ?? null;
 342                 $media['publisher-image'] = null;
 343
 344                 Logger::debug('Account detected', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'uri' => $contact['url']]);
 345                 return $media;
 346         }
 347
 348         /**
 349          * Fetch media data from local resources
 350          * @param array $media
 351          * @return array media with added data
 352          */
 353         private static function fetchLocalData(array $media): array
 354         {
 355                 if (!preg_match('|.*?/photo/(.*[a-fA-F0-9])\-(.*[0-9])\..*[\w]|', $media['url'] ?? '', $matches)) {
 356                         return $media;
 357                 }
 358                 $photo = Photo::selectFirst([], ['resource-id' => $matches[1], 'scale' => $matches[2]]);
 359                 if (!empty($photo)) {
 360                         $media['mimetype'] = $photo['type'];
 361                         $media['size'] = $photo['datasize'];
 362                         $media['width'] = $photo['width'];
 363                         $media['height'] = $photo['height'];
 364                         $media['blurhash'] = $photo['blurhash'];
 365                 }
 366
 367                 if (!preg_match('|.*?/photo/(.*[a-fA-F0-9])\-(.*[0-9])\..*[\w]|', $media['preview'] ?? '', $matches)) {
 368                         return $media;
 369                 }
 370                 $photo = Photo::selectFirst([], ['resource-id' => $matches[1], 'scale' => $matches[2]]);
 371                 if (!empty($photo)) {
 372                         $media['preview-width'] = $photo['width'];
 373                         $media['preview-height'] = $photo['height'];
 374                 }
 375
 376                 return $media;
 377         }
 378
 379         /**
 380          * Add the detected type to the media array
 381          *
 382          * @param array $data
 383          * @return array data array with the detected type
 384          */
 385         public static function addType(array $data): array
 386         {
 387                 if (empty($data['mimetype'])) {
 388                         Logger::info('No MimeType provided', ['media' => $data]);
 389                         return $data;
 390                 }
 391
 392                 $type = explode('/', current(explode(';', $data['mimetype'])));
 393                 if (count($type) < 2) {
 394                         Logger::info('Unknown MimeType', ['type' => $type, 'media' => $data]);
 395                         $data['type'] = self::UNKNOWN;
 396                         return $data;
 397                 }
 398
 399                 $filetype = strtolower($type[0]);
 400                 $subtype = strtolower($type[1]);
 401
 402                 if ($filetype == 'image') {
 403                         $data['type'] = self::IMAGE;
 404                 } elseif ($filetype == 'video') {
 405                         $data['type'] = self::VIDEO;
 406                 } elseif ($filetype == 'audio') {
 407                         $data['type'] = self::AUDIO;
 408                 } elseif (($filetype == 'text') && ($subtype == 'html')) {
 409                         $data['type'] = self::HTML;
 410                 } elseif (($filetype == 'text') && ($subtype == 'xml')) {
 411                         $data['type'] = self::XML;
 412                 } elseif (($filetype == 'text') && ($subtype == 'plain')) {
 413                         $data['type'] = self::PLAIN;
 414                 } elseif ($filetype == 'text') {
 415                         $data['type'] = self::TEXT;
 416                 } elseif (($filetype == 'application') && ($subtype == 'x-bittorrent')) {
 417                         $data['type'] = self::TORRENT;
 418                 } elseif ($filetype == 'application') {
 419                         $data['type'] = self::APPLICATION;
 420                 } else {
 421                         $data['type'] = self::UNKNOWN;
 422                         Logger::info('Unknown type', ['filetype' => $filetype, 'subtype' => $subtype, 'media' => $data]);
 423                         return $data;
 424                 }
 425
 426                 Logger::debug('Detected type', ['filetype' => $filetype, 'subtype' => $subtype, 'media' => $data]);
 427                 return $data;
 428         }
 429
 430         /**
 431          * Tests for path patterns that are usef for picture links in Friendica
 432          *
 433          * @param string $page    Link to the image page
 434          * @param string $preview Preview picture
 435          * @return boolean
 436          */
 437         private static function isPictureLink(string $page, string $preview): bool
 438         {
 439                 return preg_match('#/photos/.*/image/#ism', $page) && preg_match('#/photo/.*-1\.#ism', $preview);
 440         }
 441
 442         /**
 443          * Add media links and remove them from the body
 444          *
 445          * @param integer $uriid
 446          * @param string $body
 447          * @return string Body without media links
 448          */
 449         public static function insertFromBody(int $uriid, string $body, bool $endmatch = false): string
 450         {
 451                 $endmatchpattern = $endmatch ? '\z' : '';
 452                 // Simplify image codes
 453                 $unshared_body = $body = preg_replace("/\[img\=([0-9]*)x([0-9]*)\](.*?)\[\/img\]$endmatchpattern/ism", '[img]$3[/img]', $body);
 454
 455                 $attachments = [];
 456                 if (preg_match_all("#\[url=([^\]]+?)\]\s*\[img=([^\[\]]*)\]([^\[\]]*)\[\/img\]\s*\[/url\]$endmatchpattern#ism", $body, $pictures, PREG_SET_ORDER)) {
 457                         foreach ($pictures as $picture) {
 458                                 if (!self::isPictureLink($picture[1], $picture[2])) {
 459                                         continue;
 460                                 }
 461                                 $body = str_replace($picture[0], '', $body);
 462                                 $image = str_replace('-1.', '-0.', $picture[2]);
 463                                 $attachments[$image] = ['uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $image,
 464                                         'preview' => $picture[2], 'description' => $picture[3]];
 465                         }
 466                 }
 467
 468                 if (preg_match_all("/\[img=([^\[\]]*)\]([^\[\]]*)\[\/img\]$endmatchpattern/Usi", $body, $pictures, PREG_SET_ORDER)) {
 469                         foreach ($pictures as $picture) {
 470                                 $body = str_replace($picture[0], '', $body);
 471                                 $attachments[$picture[1]] = ['uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $picture[1], 'description' => $picture[2]];
 472                         }
 473                 }
 474
 475                 if (preg_match_all("#\[url=([^\]]+?)\]\s*\[img\]([^\[]+?)\[/img\]\s*\[/url\]$endmatchpattern#ism", $body, $pictures, PREG_SET_ORDER)) {
 476                         foreach ($pictures as $picture) {
 477                                 if (!self::isPictureLink($picture[1], $picture[2])) {
 478                                         continue;
 479                                 }
 480                                 $body = str_replace($picture[0], '', $body);
 481                                 $image = str_replace('-1.', '-0.', $picture[2]);
 482                                 $attachments[$image] = ['uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $image,
 483                                         'preview' => $picture[2], 'description' => null];
 484                         }
 485                 }
 486
 487                 if (preg_match_all("/\[img\]([^\[\]]*)\[\/img\]$endmatchpattern/ism", $body, $pictures, PREG_SET_ORDER)) {
 488                         foreach ($pictures as $picture) {
 489                                 $body = str_replace($picture[0], '', $body);
 490                                 $attachments[$picture[1]] = ['uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $picture[1]];
 491                         }
 492                 }
 493
 494                 if (preg_match_all("/\[audio\]([^\[\]]*)\[\/audio\]$endmatchpattern/ism", $body, $audios, PREG_SET_ORDER)) {
 495                         foreach ($audios as $audio) {
 496                                 $body = str_replace($audio[0], '', $body);
 497                                 $attachments[$audio[1]] = ['uri-id' => $uriid, 'type' => self::AUDIO, 'url' => $audio[1]];
 498                         }
 499                 }
 500
 501                 if (preg_match_all("/\[video\]([^\[\]]*)\[\/video\]$endmatchpattern/ism", $body, $videos, PREG_SET_ORDER)) {
 502                         foreach ($videos as $video) {
 503                                 $body = str_replace($video[0], '', $body);
 504                                 $attachments[$video[1]] = ['uri-id' => $uriid, 'type' => self::VIDEO, 'url' => $video[1]];
 505                         }
 506                 }
 507
 508                 if ($uriid != 0) {
 509                         foreach ($attachments as $attachment) {
 510                                 if (Post\Link::exists($uriid, $attachment['preview'] ?? $attachment['url'])) {
 511                                         continue;
 512                                 }
 513
 514                                 // Only store attachments that are part of the unshared body
 515                                 if (Item::containsLink($unshared_body, $attachment['preview'] ?? $attachment['url'], $attachment['type'])) {
 516                                         self::insert($attachment);
 517                                 }
 518                         }
 519                 }
 520
 521                 return trim($body);
 522         }
 523
 524         /**
 525          * Remove media that is at the end of the body
 526          *
 527          * @param string $body
 528          * @return string
 529          */
 530         public static function removeFromEndOfBody(string $body): string
 531         {
 532                 do {
 533                         $prebody = $body;
 534                         $body = self::insertFromBody(0, $body, true);
 535                 } while ($prebody != $body);
 536                 return $body;
 537         }
 538
 539         /**
 540          * Add media links from a relevant url in the body
 541          *
 542          * @param integer $uriid
 543          * @param string $body
 544          * @return void
 545          */
 546         public static function insertFromRelevantUrl(int $uriid, string $body)
 547         {
 548                 // Remove all hashtags and mentions
 549                 $body = preg_replace("/([#@!])\[url\=(.*?)\](.*?)\[\/url\]/ism", '', $body);
 550
 551                 // Search for pure links
 552                 if (preg_match_all("/\[url\](https?:.*?)\[\/url\]/ism", $body, $matches)) {
 553                         foreach ($matches[1] as $url) {
 554                                 Logger::info('Got page url (link without description)', ['uri-id' => $uriid, 'url' => $url]);
 555                                 self::insert(['uri-id' => $uriid, 'type' => self::UNKNOWN, 'url' => $url]);
 556                         }
 557                 }
 558
 559                 // Search for links with descriptions
 560                 if (preg_match_all("/\[url\=(https?:.*?)\].*?\[\/url\]/ism", $body, $matches)) {
 561                         foreach ($matches[1] as $url) {
 562                                 Logger::info('Got page url (link with description)', ['uri-id' => $uriid, 'url' => $url]);
 563                                 self::insert(['uri-id' => $uriid, 'type' => self::UNKNOWN, 'url' => $url]);
 564                         }
 565                 }
 566         }
 567
 568         /**
 569          * Add media links from the attachment field
 570          *
 571          * @param integer $uriid
 572          * @param string $body
 573          * @return void
 574          */
 575         public static function insertFromAttachmentData(int $uriid, string $body)
 576         {
 577                 $data = BBCode::getAttachmentData($body);
 578                 if (empty($data))  {
 579                         return;
 580                 }
 581
 582                 Logger::info('Adding attachment data', ['data' => $data]);
 583                 $attachment = [
 584                         'uri-id' => $uriid,
 585                         'type' => self::HTML,
 586                         'url' => $data['url'],
 587                         'preview' => $data['preview'] ?? null,
 588                         'description' => $data['description'] ?? null,
 589                         'name' => $data['title'] ?? null,
 590                         'author-url' => $data['author_url'] ?? null,
 591                         'author-name' => $data['author_name'] ?? null,
 592                         'publisher-url' => $data['provider_url'] ?? null,
 593                         'publisher-name' => $data['provider_name'] ?? null,
 594                 ];
 595                 if (!empty($data['image'])) {
 596                         $attachment['preview'] = $data['image'];
 597                 }
 598                 self::insert($attachment);
 599         }
 600
 601         /**
 602          * Add media links from the attach field
 603          *
 604          * @param integer $uriid
 605          * @param string $attach
 606          * @return void
 607          */
 608         public static function insertFromAttachment(int $uriid, string $attach)
 609         {
 610                 if (!preg_match_all('|\[attach\]href=\"(.*?)\" length=\"(.*?)\" type=\"(.*?)\"(?: title=\"(.*?)\")?|', $attach, $matches, PREG_SET_ORDER)) {
 611                         return;
 612                 }
 613
 614                 foreach ($matches as $attachment) {
 615                         $media['type'] = self::DOCUMENT;
 616                         $media['uri-id'] = $uriid;
 617                         $media['url'] = $attachment[1];
 618                         $media['size'] = $attachment[2];
 619                         $media['mimetype'] = $attachment[3];
 620                         $media['description'] = $attachment[4] ?? '';
 621
 622                         self::insert($media);
 623                 }
 624         }
 625
 626         /**
 627          * Retrieves the media attachments associated with the provided item ID.
 628          *
 629          * @param int $uri_id URI id
 630          * @param array $types Media types
 631          * @return array|bool Array on success, false on error
 632          * @throws \Exception
 633          */
 634         public static function getByURIId(int $uri_id, array $types = [])
 635         {
 636                 $condition = ['uri-id' => $uri_id];
 637
 638                 if (!empty($types)) {
 639                         $condition = DBA::mergeConditions($condition, ['type' => $types]);
 640                 }
 641
 642                 return DBA::selectToArray('post-media', [], $condition, ['order' => ['id']]);
 643         }
 644
 645         /**
 646          * Checks if media attachments are associated with the provided item ID.
 647          *
 648          * @param int $uri_id URI id
 649          * @param array $types Media types
 650          * @return bool Whether media attachment exists
 651          * @throws \Exception
 652          */
 653         public static function existsByURIId(int $uri_id, array $types = []): bool
 654         {
 655                 $condition = ['uri-id' => $uri_id];
 656
 657                 if (!empty($types)) {
 658                         $condition = DBA::mergeConditions($condition, ['type' => $types]);
 659                 }
 660
 661                 return DBA::exists('post-media', $condition);
 662         }
 663
 664         /**
 665          * Split the attachment media in the three segments "visual", "link" and "additional"
 666          *
 667          * @param int    $uri_id URI id
 668          * @param array  $links list of links that shouldn't be added
 669          * @param bool   $has_media
 670          * @return array attachments
 671          */
 672         public static function splitAttachments(int $uri_id, array $links = [], bool $has_media = true): array
 673         {
 674                 $attachments = ['visual' => [], 'link' => [], 'additional' => []];
 675
 676                 if (!$has_media) {
 677                         return $attachments;
 678                 }
 679
 680                 $media = self::getByURIId($uri_id);
 681                 if (empty($media)) {
 682                         return $attachments;
 683                 }
 684
 685                 $heights = [];
 686                 $selected = '';
 687                 $previews = [];
 688
 689                 foreach ($media as $medium) {
 690                         foreach ($links as $link) {
 691                                 if (Strings::compareLink($link, $medium['url'])) {
 692                                         continue 2;
 693                                 }
 694                         }
 695
 696                         // Avoid adding separate media entries for previews
 697                         foreach ($previews as $preview) {
 698                                 if (Strings::compareLink($preview, $medium['url'])) {
 699                                         continue 2;
 700                                 }
 701                         }
 702
 703                         // Currently these two types are ignored here.
 704                         // Posts are added differently and contacts are not displayed as attachments.
 705                         if (in_array($medium['type'], [self::ACCOUNT, self::ACTIVITY])) {
 706                                 continue;
 707                         }
 708
 709                         if (!empty($medium['preview'])) {
 710                                 $previews[] = $medium['preview'];
 711                         }
 712
 713                         $type = explode('/', explode(';', $medium['mimetype'] ?? '')[0]);
 714                         if (count($type) < 2) {
 715                                 Logger::info('Unknown MimeType', ['type' => $type, 'media' => $medium]);
 716                                 $filetype = 'unkn';
 717                                 $subtype = 'unkn';
 718                         } else {
 719                                 $filetype = strtolower($type[0]);
 720                                 $subtype = strtolower($type[1]);
 721                         }
 722
 723                         $medium['filetype'] = $filetype;
 724                         $medium['subtype'] = $subtype;
 725
 726                         if ($medium['type'] == self::HTML || (($filetype == 'text') && ($subtype == 'html'))) {
 727                                 $attachments['link'][] = $medium;
 728                                 continue;
 729                         }
 730
 731                         if (in_array($medium['type'], [self::AUDIO, self::IMAGE]) ||
 732                                 in_array($filetype, ['audio', 'image'])) {
 733                                 $attachments['visual'][] = $medium;
 734                         } elseif (($medium['type'] == self::VIDEO) || ($filetype == 'video')) {
 735                                 if (!empty($medium['height'])) {
 736                                         // Peertube videos are delivered in many different resolutions. We pick a moderate one.
 737                                         // Since only Peertube provides a "height" parameter, this wouldn't be executed
 738                                         // when someone for example on Mastodon was sharing multiple videos in a single post.
 739                                         $heights[$medium['height']] = $medium['url'];
 740                                         $video[$medium['url']] = $medium;
 741                                 } else {
 742                                         $attachments['visual'][] = $medium;
 743                                 }
 744                         } else {
 745                                 $attachments['additional'][] = $medium;
 746                         }
 747                 }
 748
 749                 if (!empty($heights)) {
 750                         ksort($heights);
 751                         foreach ($heights as $height => $url) {
 752                                 if (empty($selected) || $height <= 480) {
 753                                         $selected = $url;
 754                                 }
 755                         }
 756
 757                         if (!empty($selected)) {
 758                                 $attachments['visual'][] = $video[$selected];
 759                                 unset($video[$selected]);
 760                                 foreach ($video as $element) {
 761                                         $attachments['additional'][] = $element;
 762                                 }
 763                         }
 764                 }
 765
 766                 return $attachments;
 767         }
 768
 769         /**
 770          * Add media attachments to the body
 771          *
 772          * @param int    $uriid
 773          * @param string $body
 774          * @param array  $types
 775          *
 776          * @return string body
 777          */
 778         public static function addAttachmentsToBody(int $uriid, string $body = '', array $types = [self::IMAGE, self::AUDIO, self::VIDEO]): string
 779         {
 780                 if (empty($body)) {
 781                         $item = Post::selectFirst(['body'], ['uri-id' => $uriid]);
 782                         if (!DBA::isResult($item)) {
 783                                 return '';
 784                         }
 785                         $body = $item['body'];
 786                 }
 787                 $original_body = $body;
 788
 789                 $body = preg_replace("/\s*\[attachment .*?\].*?\[\/attachment\]\s*/ism", '', $body);
 790
 791                 foreach (self::getByURIId($uriid, $types) as $media) {
 792                         if (Item::containsLink($body, $media['preview'] ?? $media['url'], $media['type'])) {
 793                                 continue;
 794                         }
 795
 796                         if ($media['type'] == self::IMAGE) {
 797                                 if (!empty($media['preview'])) {
 798                                         if (!empty($media['description'])) {
 799                                                 $body .= "\n[url=" . $media['url'] . "][img=" . $media['preview'] . ']' . $media['description'] .'[/img][/url]';
 800                                         } else {
 801                                                 $body .= "\n[url=" . $media['url'] . "][img]" . $media['preview'] .'[/img][/url]';
 802                                         }
 803                                 } else {
 804                                         if (!empty($media['description'])) {
 805                                                 $body .= "\n[img=" . $media['url'] . ']' . $media['description'] .'[/img]';
 806                                         } else {
 807                                                 $body .= "\n[img]" . $media['url'] .'[/img]';
 808                                         }
 809                                 }
 810                         } elseif ($media['type'] == self::AUDIO) {
 811                                 $body .= "\n[audio]" . $media['url'] . "[/audio]\n";
 812                         } elseif ($media['type'] == self::VIDEO) {
 813                                 $body .= "\n[video]" . $media['url'] . "[/video]\n";
 814                         }
 815                 }
 816
 817                 if (preg_match("/.*(\[attachment.*?\].*?\[\/attachment\]).*/ism", $original_body, $match)) {
 818                         $body .= "\n" . $match[1];
 819                 }
 820
 821                 return $body;
 822         }
 823
 824         /**
 825          * Get preview link for given media id
 826          *
 827          * @param integer $id   media id
 828          * @param string  $size One of the Proxy::SIZE_* constants
 829          * @return string preview link
 830          */
 831         public static function getPreviewUrlForId(int $id, string $size = ''): string
 832         {
 833                 $url = DI::baseUrl() . '/photo/preview/';
 834                 switch ($size) {
 835                         case Proxy::SIZE_MICRO:
 836                                 $url .= Proxy::PIXEL_MICRO . '/';
 837                                 break;
 838                         case Proxy::SIZE_THUMB:
 839                                 $url .= Proxy::PIXEL_THUMB . '/';
 840                                 break;
 841                         case Proxy::SIZE_SMALL:
 842                                 $url .= Proxy::PIXEL_SMALL . '/';
 843                                 break;
 844                         case Proxy::SIZE_MEDIUM:
 845                                 $url .= Proxy::PIXEL_MEDIUM . '/';
 846                                 break;
 847                         case Proxy::SIZE_LARGE:
 848                                 $url .= Proxy::PIXEL_LARGE . '/';
 849                                 break;
 850                 }
 851                 return $url . $id;
 852         }
 853
 854         /**
 855          * Get media link for given media id
 856          *
 857          * @param integer $id   media id
 858          * @param string  $size One of the Proxy::SIZE_* constants
 859          * @return string media link
 860          */
 861         public static function getUrlForId(int $id, string $size = ''): string
 862         {
 863                 $url = DI::baseUrl() . '/photo/media/';
 864                 switch ($size) {
 865                         case Proxy::SIZE_MICRO:
 866                                 $url .= Proxy::PIXEL_MICRO . '/';
 867                                 break;
 868                         case Proxy::SIZE_THUMB:
 869                                 $url .= Proxy::PIXEL_THUMB . '/';
 870                                 break;
 871                         case Proxy::SIZE_SMALL:
 872                                 $url .= Proxy::PIXEL_SMALL . '/';
 873                                 break;
 874                         case Proxy::SIZE_MEDIUM:
 875                                 $url .= Proxy::PIXEL_MEDIUM . '/';
 876                                 break;
 877                         case Proxy::SIZE_LARGE:
 878                                 $url .= Proxy::PIXEL_LARGE . '/';
 879                                 break;
 880                 }
 881                 return $url . $id;
 882         }
 883 }