src/Model/Post/Media.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2023, 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\PageInfo;
  25 use Friendica\Content\Text\BBCode;
  26 use Friendica\Core\Logger;
  27 use Friendica\Core\Protocol;
  28 use Friendica\Core\System;
  29 use Friendica\Database\Database;
  30 use Friendica\Database\DBA;
  31 use Friendica\DI;
  32 use Friendica\Model\Contact;
  33 use Friendica\Model\Item;
  34 use Friendica\Model\ItemURI;
  35 use Friendica\Model\Photo;
  36 use Friendica\Model\Post;
  37 use Friendica\Network\HTTPClient\Client\HttpClientAccept;
  38 use Friendica\Network\HTTPClient\Client\HttpClientOptions;
  39 use Friendica\Util\Images;
  40 use Friendica\Util\Network;
  41 use Friendica\Util\ParseUrl;
  42 use Friendica\Util\Proxy;
  43 use Friendica\Util\Strings;
  44
  45 /**
  46  * Class Media
  47  *
  48  * This Model class handles media interactions.
  49  * This tables stores medias (images, videos, audio files) related to posts.
  50  */
  51 class Media
  52 {
  53         const UNKNOWN     = 0;
  54         const IMAGE       = 1;
  55         const VIDEO       = 2;
  56         const AUDIO       = 3;
  57         const TEXT        = 4;
  58         const APPLICATION = 5;
  59         const TORRENT     = 16;
  60         const HTML        = 17;
  61         const XML         = 18;
  62         const PLAIN       = 19;
  63         const ACTIVITY    = 20;
  64         const ACCOUNT     = 21;
  65         const DOCUMENT    = 128;
  66
  67         /**
  68          * Insert a post-media record
  69          *
  70          * @param array $media
  71          * @param bool  $force
  72          * @return bool
  73          */
  74         public static function insert(array $media, bool $force = false): bool
  75         {
  76                 if (empty($media['url']) || empty($media['uri-id']) || !isset($media['type'])) {
  77                         Logger::warning('Incomplete media data', ['media' => $media]);
  78                         return false;
  79                 }
  80
  81                 if (DBA::exists('post-media', ['uri-id' => $media['uri-id'], 'preview' => $media['url']])) {
  82                         Logger::info('Media already exists as preview', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'callstack' => System::callstack()]);
  83                         return false;
  84                 }
  85
  86                 // "document" has got the lowest priority. So when the same file is both attached as document
  87                 // and embedded as picture then we only store the picture or replace the document
  88                 $found = DBA::selectFirst('post-media', ['type'], ['uri-id' => $media['uri-id'], 'url' => $media['url']]);
  89                 if (!$force && !empty($found) && (($found['type'] != self::DOCUMENT) || ($media['type'] == self::DOCUMENT))) {
  90                         Logger::info('Media already exists', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'callstack' => System::callstack()]);
  91                         return false;
  92                 }
  93
  94                 if (!ItemURI::exists($media['uri-id'])) {
  95                         Logger::info('Media referenced URI ID not found', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'callstack' => System::callstack()]);
  96                         return false;
  97                 }
  98
  99                 $media = self::unsetEmptyFields($media);
 100                 $media = DI::dbaDefinition()->truncateFieldsForTable('post-media', $media);
 101
 102                 // We are storing as fast as possible to avoid duplicated network requests
 103                 // when fetching additional information for pictures and other content.
 104                 $result = DBA::insert('post-media', $media, Database::INSERT_UPDATE);
 105                 Logger::info('Stored media', ['result' => $result, 'media' => $media, 'callstack' => System::callstack()]);
 106                 $stored = $media;
 107
 108                 $media = self::fetchAdditionalData($media);
 109                 $media = self::unsetEmptyFields($media);
 110                 $media = DI::dbaDefinition()->truncateFieldsForTable('post-media', $media);
 111
 112                 if (array_diff_assoc($media, $stored)) {
 113                         $result = DBA::insert('post-media', $media, Database::INSERT_UPDATE);
 114                         Logger::info('Updated media', ['result' => $result, 'media' => $media]);
 115                 } else {
 116                         Logger::info('Nothing to update', ['media' => $media]);
 117                 }
 118                 return $result;
 119         }
 120
 121         /**
 122          * Remove empty media fields
 123          *
 124          * @param array $media
 125          * @return array cleaned media array
 126          */
 127         private static function unsetEmptyFields(array $media): array
 128         {
 129                 $fields = ['mimetype', 'height', 'width', 'size', 'preview', 'preview-height', 'preview-width', 'blurhash', 'description'];
 130                 foreach ($fields as $field) {
 131                         if (empty($media[$field])) {
 132                                 unset($media[$field]);
 133                         }
 134                 }
 135                 return $media;
 136         }
 137
 138         /**
 139          * Copy attachments from one uri-id to another
 140          *
 141          * @param integer $from_uri_id
 142          * @param integer $to_uri_id
 143          * @return void
 144          */
 145         public static function copy(int $from_uri_id, int $to_uri_id)
 146         {
 147                 $attachments = self::getByURIId($from_uri_id);
 148                 foreach ($attachments as $attachment) {
 149                         $attachment['uri-id'] = $to_uri_id;
 150                         self::insert($attachment);
 151                 }
 152         }
 153
 154         /**
 155          * Creates the "[attach]" element from the given attributes
 156          *
 157          * @param string $href
 158          * @param integer $length
 159          * @param string $type
 160          * @param string $title
 161          * @return string "[attach]" element
 162          */
 163         public static function getAttachElement(string $href, int $length, string $type, string $title = ''): string
 164         {
 165                 $media = self::fetchAdditionalData([
 166                         'type' => self::DOCUMENT, 'url' => $href,
 167                         'size' => $length, 'mimetype' => $type, 'description' => $title
 168                 ]);
 169
 170                 return '[attach]href="' . $media['url'] . '" length="' . $media['size'] .
 171                         '" type="' . $media['mimetype'] . '" title="' . $media['description'] . '"[/attach]';
 172         }
 173
 174         /**
 175          * Fetch additional data for the provided media array
 176          *
 177          * @param array $media
 178          * @return array media array with additional data
 179          */
 180         public static function fetchAdditionalData(array $media): array
 181         {
 182                 if (Network::isLocalLink($media['url'])) {
 183                         $media = self::fetchLocalData($media);
 184                 }
 185
 186                 // Fetch the mimetype or size if missing.
 187                 if (Network::isValidHttpUrl($media['url']) && (empty($media['mimetype']) || empty($media['size']))) {
 188                         $timeout = DI::config()->get('system', 'xrd_timeout');
 189                         $curlResult = DI::httpClient()->head($media['url'], [HttpClientOptions::TIMEOUT => $timeout]);
 190
 191                         // Workaround for systems that can't handle a HEAD request
 192                         if (!$curlResult->isSuccess() && ($curlResult->getReturnCode() == 405)) {
 193                                 $curlResult = DI::httpClient()->get($media['url'], HttpClientAccept::DEFAULT, [HttpClientOptions::TIMEOUT => $timeout]);
 194                         }
 195
 196                         if ($curlResult->isSuccess()) {
 197                                 if (empty($media['mimetype'])) {
 198                                         $media['mimetype'] = $curlResult->getHeader('Content-Type')[0] ?? '';
 199                                 }
 200                                 if (empty($media['size'])) {
 201                                         $media['size'] = (int)($curlResult->getHeader('Content-Length')[0] ?? 0);
 202                                 }
 203                         } else {
 204                                 Logger::notice('Could not fetch head', ['media' => $media]);
 205                         }
 206                 }
 207
 208                 $filetype = !empty($media['mimetype']) ? strtolower(current(explode('/', $media['mimetype']))) : '';
 209
 210                 if (($media['type'] == self::IMAGE) || ($filetype == 'image')) {
 211                         $imagedata = Images::getInfoFromURLCached($media['url']);
 212                         if ($imagedata) {
 213                                 $media['mimetype'] = $imagedata['mime'];
 214                                 $media['size'] = $imagedata['size'];
 215                                 $media['width'] = $imagedata[0];
 216                                 $media['height'] = $imagedata[1];
 217                                 $media['blurhash'] = $imagedata['blurhash'] ?? null;
 218                         } else {
 219                                 Logger::notice('No image data', ['media' => $media]);
 220                         }
 221                         if (!empty($media['preview'])) {
 222                                 $imagedata = Images::getInfoFromURLCached($media['preview']);
 223                                 if ($imagedata) {
 224                                         $media['preview-width'] = $imagedata[0];
 225                                         $media['preview-height'] = $imagedata[1];
 226                                 }
 227                         }
 228                 }
 229
 230                 if ($media['type'] != self::DOCUMENT) {
 231                         $media = self::addType($media);
 232                 }
 233
 234                 if (in_array($media['type'], [self::TEXT, self::APPLICATION, self::HTML, self::XML, self::PLAIN])) {
 235                         $media = self::addActivity($media);
 236                 }
 237
 238                 if (in_array($media['type'], [self::TEXT, self::APPLICATION, self::HTML, self::XML, self::PLAIN])) {
 239                         $media = self::addAccount($media);
 240                 }
 241
 242                 if ($media['type'] == self::HTML) {
 243                         $media = self::addPage($media);
 244                 }
 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 $media;
 266                 }
 267
 268                 if (
 269                         !empty($item['plink']) && Strings::compareLink($item['plink'], $media['url']) &&
 270                         parse_url($item['plink'], PHP_URL_HOST) != parse_url($item['uri'], PHP_URL_HOST)
 271                 ) {
 272                         Logger::debug('Not a link to an activity', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'plink' => $item['plink'], 'uri' => $item['uri']]);
 273                         return $media;
 274                 }
 275
 276                 if (in_array($item['network'], [Protocol::ACTIVITYPUB, Protocol::DFRN])) {
 277                         $media['mimetype'] = 'application/activity+json';
 278                 } elseif ($item['network'] == Protocol::DIASPORA) {
 279                         $media['mimetype'] = 'application/xml';
 280                 }
 281
 282                 $contact = Contact::getById($item['author-id'], ['avatar', 'gsid']);
 283                 if (!empty($contact['gsid'])) {
 284                         $gserver = DBA::selectFirst('gserver', ['url', 'site_name'], ['id' => $contact['gsid']]);
 285                 }
 286
 287                 $media['type'] = self::ACTIVITY;
 288                 $media['media-uri-id'] = $item['uri-id'];
 289                 $media['height'] = null;
 290                 $media['width'] = null;
 291                 $media['preview'] = null;
 292                 $media['preview-height'] = null;
 293                 $media['preview-width'] = null;
 294                 $media['blurhash'] = null;
 295                 $media['description'] = $item['body'];
 296                 $media['name'] = $item['title'];
 297                 $media['author-url'] = $item['author-link'];
 298                 $media['author-name'] = $item['author-name'];
 299                 $media['author-image'] = $contact['avatar'] ?? $item['author-avatar'];
 300                 $media['publisher-url'] = $gserver['url'] ?? null;
 301                 $media['publisher-name'] = $gserver['site_name'] ?? null;
 302                 $media['publisher-image'] = null;
 303
 304                 Logger::debug('Activity detected', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'plink' => $item['plink'], 'uri' => $item['uri']]);
 305                 return $media;
 306         }
 307
 308         /**
 309          * Adds the account type if the media entry is linked to an account
 310          *
 311          * @param array $media
 312          * @return array
 313          */
 314         private static function addAccount(array $media): array
 315         {
 316                 $contact = Contact::getByURL($media['url'], false);
 317                 if (empty($contact) || ($contact['network'] == Protocol::PHANTOM)) {
 318                         return $media;
 319                 }
 320
 321                 if (in_array($contact['network'], [Protocol::ACTIVITYPUB, Protocol::DFRN])) {
 322                         $media['mimetype'] = 'application/activity+json';
 323                 }
 324
 325                 if (!empty($contact['gsid'])) {
 326                         $gserver = DBA::selectFirst('gserver', ['url', 'site_name'], ['id' => $contact['gsid']]);
 327                 }
 328
 329                 $media['type'] = self::ACCOUNT;
 330                 $media['media-uri-id'] = $contact['uri-id'];
 331                 $media['height'] = null;
 332                 $media['width'] = null;
 333                 $media['preview'] = null;
 334                 $media['preview-height'] = null;
 335                 $media['preview-width'] = null;
 336                 $media['blurhash'] = null;
 337                 $media['description'] = $contact['about'];
 338                 $media['name'] = $contact['name'];
 339                 $media['author-url'] = $contact['url'];
 340                 $media['author-name'] = $contact['name'];
 341                 $media['author-image'] = $contact['avatar'];
 342                 $media['publisher-url'] = $gserver['url'] ?? null;
 343                 $media['publisher-name'] = $gserver['site_name'] ?? null;
 344                 $media['publisher-image'] = null;
 345
 346                 Logger::debug('Account detected', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'uri' => $contact['url']]);
 347                 return $media;
 348         }
 349
 350         /**
 351          * Add page infos for HTML entries
 352          *
 353          * @param array $media
 354          * @return array
 355          */
 356         private static function addPage(array $media): array
 357         {
 358                 $data = ParseUrl::getSiteinfoCached($media['url'], false);
 359                 $media['preview'] = $data['images'][0]['src'] ?? null;
 360                 $media['preview-height'] = $data['images'][0]['height'] ?? null;
 361                 $media['preview-width'] = $data['images'][0]['width'] ?? null;
 362                 $media['blurhash'] = $data['images'][0]['blurhash'] ?? null;
 363                 $media['description'] = $data['text'] ?? null;
 364                 $media['name'] = $data['title'] ?? null;
 365                 $media['author-url'] = $data['author_url'] ?? null;
 366                 $media['author-name'] = $data['author_name'] ?? null;
 367                 $media['author-image'] = $data['author_img'] ?? null;
 368                 $media['publisher-url'] = $data['publisher_url'] ?? null;
 369                 $media['publisher-name'] = $data['publisher_name'] ?? null;
 370                 $media['publisher-image'] = $data['publisher_img'] ?? null;
 371
 372                 return $media;
 373         }
 374
 375         /**
 376          * Fetch media data from local resources
 377          * @param array $media
 378          * @return array media with added data
 379          */
 380         private static function fetchLocalData(array $media): array
 381         {
 382                 if (!preg_match('|.*?/photo/(.*[a-fA-F0-9])\-(.*[0-9])\..*[\w]|', $media['url'] ?? '', $matches)) {
 383                         return $media;
 384                 }
 385                 $photo = Photo::selectFirst([], ['resource-id' => $matches[1], 'scale' => $matches[2]]);
 386                 if (!empty($photo)) {
 387                         $media['mimetype'] = $photo['type'];
 388                         $media['size'] = $photo['datasize'];
 389                         $media['width'] = $photo['width'];
 390                         $media['height'] = $photo['height'];
 391                         $media['blurhash'] = $photo['blurhash'];
 392                 }
 393
 394                 if (!preg_match('|.*?/photo/(.*[a-fA-F0-9])\-(.*[0-9])\..*[\w]|', $media['preview'] ?? '', $matches)) {
 395                         return $media;
 396                 }
 397                 $photo = Photo::selectFirst([], ['resource-id' => $matches[1], 'scale' => $matches[2]]);
 398                 if (!empty($photo)) {
 399                         $media['preview-width'] = $photo['width'];
 400                         $media['preview-height'] = $photo['height'];
 401                 }
 402
 403                 return $media;
 404         }
 405
 406         /**
 407          * Add the detected type to the media array
 408          *
 409          * @param array $data
 410          * @return array data array with the detected type
 411          */
 412         public static function addType(array $data): array
 413         {
 414                 if (empty($data['mimetype'])) {
 415                         Logger::info('No MimeType provided', ['media' => $data]);
 416                         return $data;
 417                 }
 418
 419                 $type = explode('/', current(explode(';', $data['mimetype'])));
 420                 if (count($type) < 2) {
 421                         Logger::info('Unknown MimeType', ['type' => $type, 'media' => $data]);
 422                         $data['type'] = self::UNKNOWN;
 423                         return $data;
 424                 }
 425
 426                 $filetype = strtolower($type[0]);
 427                 $subtype = strtolower($type[1]);
 428
 429                 if ($filetype == 'image') {
 430                         $data['type'] = self::IMAGE;
 431                 } elseif ($filetype == 'video') {
 432                         $data['type'] = self::VIDEO;
 433                 } elseif ($filetype == 'audio') {
 434                         $data['type'] = self::AUDIO;
 435                 } elseif (($filetype == 'text') && ($subtype == 'html')) {
 436                         $data['type'] = self::HTML;
 437                 } elseif (($filetype == 'text') && ($subtype == 'xml')) {
 438                         $data['type'] = self::XML;
 439                 } elseif (($filetype == 'text') && ($subtype == 'plain')) {
 440                         $data['type'] = self::PLAIN;
 441                 } elseif ($filetype == 'text') {
 442                         $data['type'] = self::TEXT;
 443                 } elseif (($filetype == 'application') && ($subtype == 'x-bittorrent')) {
 444                         $data['type'] = self::TORRENT;
 445                 } elseif ($filetype == 'application') {
 446                         $data['type'] = self::APPLICATION;
 447                 } else {
 448                         $data['type'] = self::UNKNOWN;
 449                         Logger::info('Unknown type', ['filetype' => $filetype, 'subtype' => $subtype, 'media' => $data]);
 450                         return $data;
 451                 }
 452
 453                 Logger::debug('Detected type', ['filetype' => $filetype, 'subtype' => $subtype, 'media' => $data]);
 454                 return $data;
 455         }
 456
 457         /**
 458          * Tests for path patterns that are usef for picture links in Friendica
 459          *
 460          * @param string $page    Link to the image page
 461          * @param string $preview Preview picture
 462          * @return boolean
 463          */
 464         private static function isLinkToPhoto(string $page, string $preview): bool
 465         {
 466                 return preg_match('#/photo/.*-0\.#ism', $page) && preg_match('#/photo/.*-[01]\.#ism', $preview);
 467         }
 468
 469         /**
 470          * Tests for path patterns that are usef for picture links in Friendica
 471          *
 472          * @param string $page    Link to the image page
 473          * @param string $preview Preview picture
 474          * @return boolean
 475          */
 476         private static function isLinkToImagePage(string $page, string $preview): bool
 477         {
 478                 return preg_match('#/photos/.*/image/#ism', $page) && preg_match('#/photo/.*-[01]\.#ism', $preview);
 479         }
 480
 481         /**
 482          * Add media links and remove them from the body
 483          *
 484          * @param integer $uriid
 485          * @param string  $body
 486          * @param bool    $endmatch
 487          * @param bool    $removepicturelinks
 488          * @return string Body without media links
 489          */
 490         public static function insertFromBody(int $uriid, string $body, bool $endmatch = false, bool $removepicturelinks = false): string
 491         {
 492                 $endmatchpattern = $endmatch ? '\z' : '';
 493                 // Simplify image codes
 494                 $unshared_body = $body = preg_replace("/\[img\=([0-9]*)x([0-9]*)\](.*?)\[\/img\]$endmatchpattern/ism", '[img]$3[/img]', $body);
 495
 496                 $attachments = [];
 497                 if (preg_match_all("#\[url=([^\]]+?)\]\s*\[img=([^\[\]]*)\]([^\[\]]*)\[\/img\]\s*\[/url\]$endmatchpattern#ism", $body, $pictures, PREG_SET_ORDER)) {
 498                         foreach ($pictures as $picture) {
 499                                 if (self::isLinkToImagePage($picture[1], $picture[2])) {
 500                                         $body = str_replace($picture[0], '', $body);
 501                                         $image = str_replace('-1.', '-0.', $picture[2]);
 502                                         $attachments[$image] = [
 503                                                 'uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $image,
 504                                                 'preview' => $picture[2], 'description' => $picture[3]
 505                                         ];
 506                                 } elseif (self::isLinkToPhoto($picture[1], $picture[2])) {
 507                                         $body = str_replace($picture[0], '', $body);
 508                                         $attachments[$picture[1]] = [
 509                                                 'uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $picture[1],
 510                                                 'preview' => $picture[2], 'description' => $picture[3]
 511                                         ];
 512                                 } elseif ($removepicturelinks) {
 513                                         $body = str_replace($picture[0], '', $body);
 514                                         $attachments[$picture[1]] = [
 515                                                 'uri-id' => $uriid, 'type' => self::UNKNOWN, 'url' => $picture[1],
 516                                                 'preview' => $picture[2], 'description' => $picture[3]
 517                                         ];
 518                                 }
 519                         }
 520                 }
 521
 522                 if (preg_match_all("/\[img=([^\[\]]*)\]([^\[\]]*)\[\/img\]$endmatchpattern/Usi", $body, $pictures, PREG_SET_ORDER)) {
 523                         foreach ($pictures as $picture) {
 524                                 $body = str_replace($picture[0], '', $body);
 525                                 $attachments[$picture[1]] = ['uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $picture[1], 'description' => $picture[2]];
 526                         }
 527                 }
 528
 529                 if (preg_match_all("#\[url=([^\]]+?)\]\s*\[img\]([^\[]+?)\[/img\]\s*\[/url\]$endmatchpattern#ism", $body, $pictures, PREG_SET_ORDER)) {
 530                         foreach ($pictures as $picture) {
 531                                 if (self::isLinkToImagePage($picture[1], $picture[2])) {
 532                                         $body = str_replace($picture[0], '', $body);
 533                                         $image = str_replace('-1.', '-0.', $picture[2]);
 534                                         $attachments[$image] = [
 535                                                 'uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $image,
 536                                                 'preview' => $picture[2], 'description' => null
 537                                         ];
 538                                 } elseif (self::isLinkToPhoto($picture[1], $picture[2])) {
 539                                         $body = str_replace($picture[0], '', $body);
 540                                         $attachments[$picture[1]] = [
 541                                                 'uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $picture[1],
 542                                                 'preview' => $picture[2], 'description' => null
 543                                         ];
 544                                 } elseif ($removepicturelinks) {
 545                                         $body = str_replace($picture[0], '', $body);
 546                                         $attachments[$picture[1]] = [
 547                                                 'uri-id' => $uriid, 'type' => self::UNKNOWN, 'url' => $picture[1],
 548                                                 'preview' => $picture[2], 'description' => null
 549                                         ];
 550                                 }
 551                         }
 552                 }
 553
 554                 if (preg_match_all("/\[img\]([^\[\]]*)\[\/img\]$endmatchpattern/ism", $body, $pictures, PREG_SET_ORDER)) {
 555                         foreach ($pictures as $picture) {
 556                                 $body = str_replace($picture[0], '', $body);
 557                                 $attachments[$picture[1]] = ['uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $picture[1]];
 558                         }
 559                 }
 560
 561                 if (preg_match_all("/\[audio\]([^\[\]]*)\[\/audio\]$endmatchpattern/ism", $body, $audios, PREG_SET_ORDER)) {
 562                         foreach ($audios as $audio) {
 563                                 $body = str_replace($audio[0], '', $body);
 564                                 $attachments[$audio[1]] = ['uri-id' => $uriid, 'type' => self::AUDIO, 'url' => $audio[1]];
 565                         }
 566                 }
 567
 568                 if (preg_match_all("/\[video\]([^\[\]]*)\[\/video\]$endmatchpattern/ism", $body, $videos, PREG_SET_ORDER)) {
 569                         foreach ($videos as $video) {
 570                                 $body = str_replace($video[0], '', $body);
 571                                 $attachments[$video[1]] = ['uri-id' => $uriid, 'type' => self::VIDEO, 'url' => $video[1]];
 572                         }
 573                 }
 574
 575                 if ($uriid != 0) {
 576                         foreach ($attachments as $attachment) {
 577                                 if (Post\Link::exists($uriid, $attachment['preview'] ?? $attachment['url'])) {
 578                                         continue;
 579                                 }
 580
 581                                 // Only store attachments that are part of the unshared body
 582                                 if (Item::containsLink($unshared_body, $attachment['preview'] ?? $attachment['url'], $attachment['type'])) {
 583                                         self::insert($attachment);
 584                                 }
 585                         }
 586                 }
 587
 588                 return trim($body);
 589         }
 590
 591         /**
 592          * Remove media that is at the end of the body
 593          *
 594          * @param string $body
 595          * @return string
 596          */
 597         public static function removeFromEndOfBody(string $body): string
 598         {
 599                 do {
 600                         $prebody = $body;
 601                         $body = self::insertFromBody(0, $body, true);
 602                 } while ($prebody != $body);
 603                 return $body;
 604         }
 605
 606         /**
 607          * Remove media from the body
 608          *
 609          * @param string $body
 610          * @return string
 611          */
 612         public static function removeFromBody(string $body): string
 613         {
 614                 do {
 615                         $prebody = $body;
 616                         $body = self::insertFromBody(0, $body, false, true);
 617                 } while ($prebody != $body);
 618                 return $body;
 619         }
 620
 621         /**
 622          * Add media links from a relevant url in the body
 623          *
 624          * @param integer $uriid
 625          * @param string $body
 626          * @return void
 627          */
 628         public static function insertFromRelevantUrl(int $uriid, string $body, string $fullbody, string $network)
 629         {
 630                 // Remove all hashtags and mentions
 631                 $body = preg_replace("/([#@!])\[url\=(.*?)\](.*?)\[\/url\]/ism", '', $body);
 632
 633                 // Search for pure links
 634                 if (preg_match_all("/\[url\](https?:.*?)\[\/url\]/ism", $body, $matches)) {
 635                         foreach ($matches[1] as $url) {
 636                                 Logger::info('Got page url (link without description)', ['uri-id' => $uriid, 'url' => $url]);
 637                                 $result = self::insert(['uri-id' => $uriid, 'type' => self::UNKNOWN, 'url' => $url], false, $network);
 638                                 if ($result && ($network == Protocol::DFRN)) {
 639                                         self::revertHTMLType($uriid, $url, $fullbody);
 640                                         Logger::debug('Revert HTML type', ['uri-id' => $uriid, 'url' => $url]);
 641                                 } elseif ($result) {
 642                                         Logger::debug('Media had been added', ['uri-id' => $uriid, 'url' => $url]);
 643                                 } else {
 644                                         Logger::debug('Media had not been added', ['uri-id' => $uriid, 'url' => $url]);
 645                                 }
 646                         }
 647                 }
 648
 649                 // Search for links with descriptions
 650                 if (preg_match_all("/\[url\=(https?:.*?)\].*?\[\/url\]/ism", $body, $matches)) {
 651                         foreach ($matches[1] as $url) {
 652                                 Logger::info('Got page url (link with description)', ['uri-id' => $uriid, 'url' => $url]);
 653                                 $result = self::insert(['uri-id' => $uriid, 'type' => self::UNKNOWN, 'url' => $url], false, $network);
 654                                 if ($result && ($network == Protocol::DFRN)) {
 655                                         self::revertHTMLType($uriid, $url, $fullbody);
 656                                         Logger::debug('Revert HTML type', ['uri-id' => $uriid, 'url' => $url]);
 657                                 } elseif ($result) {
 658                                         Logger::debug('Media has been added', ['uri-id' => $uriid, 'url' => $url]);
 659                                 } else {
 660                                         Logger::debug('Media has not been added', ['uri-id' => $uriid, 'url' => $url]);
 661                                 }
 662                         }
 663                 }
 664         }
 665
 666         /**
 667          * Revert the media type of links to UNKNOWN for DFRN posts when they aren't attached
 668          *
 669          * @param integer $uriid
 670          * @param string $url
 671          * @param string $body
 672          * @return void
 673          */
 674         private static function revertHTMLType(int $uriid, string $url, string $body)
 675         {
 676                 $attachment = BBCode::getAttachmentData($body);
 677                 if (!empty($attachment['url']) && Network::getUrlMatch($attachment['url'], $url)) {
 678                         return;
 679                 }
 680                 DBA::update('post-media', ['type' => self::UNKNOWN], ['uri-id' => $uriid, 'type' => self::HTML, 'url' => $url]);
 681         }
 682
 683         /**
 684          * Add media links from the attachment field
 685          *
 686          * @param integer $uriid
 687          * @param string $body
 688          * @return void
 689          */
 690         public static function insertFromAttachmentData(int $uriid, string $body)
 691         {
 692                 $data = BBCode::getAttachmentData($body);
 693                 if (empty($data)) {
 694                         return;
 695                 }
 696
 697                 Logger::info('Adding attachment data', ['data' => $data]);
 698                 $attachment = [
 699                         'uri-id' => $uriid,
 700                         'type' => self::HTML,
 701                         'url' => $data['url'],
 702                         'preview' => $data['preview'] ?? null,
 703                         'description' => $data['description'] ?? null,
 704                         'name' => $data['title'] ?? null,
 705                         'author-url' => $data['author_url'] ?? null,
 706                         'author-name' => $data['author_name'] ?? null,
 707                         'publisher-url' => $data['provider_url'] ?? null,
 708                         'publisher-name' => $data['provider_name'] ?? null,
 709                 ];
 710                 if (!empty($data['image'])) {
 711                         $attachment['preview'] = $data['image'];
 712                 }
 713                 self::insert($attachment);
 714         }
 715
 716         /**
 717          * Add media links from the attach field
 718          *
 719          * @param integer $uriid
 720          * @param string $attach
 721          * @return void
 722          */
 723         public static function insertFromAttachment(int $uriid, string $attach)
 724         {
 725                 if (!preg_match_all('|\[attach\]href=\"(.*?)\" length=\"(.*?)\" type=\"(.*?)\"(?: title=\"(.*?)\")?|', $attach, $matches, PREG_SET_ORDER)) {
 726                         return;
 727                 }
 728
 729                 foreach ($matches as $attachment) {
 730                         $media['type'] = self::DOCUMENT;
 731                         $media['uri-id'] = $uriid;
 732                         $media['url'] = $attachment[1];
 733                         $media['size'] = $attachment[2];
 734                         $media['mimetype'] = $attachment[3];
 735                         $media['description'] = $attachment[4] ?? '';
 736
 737                         self::insert($media);
 738                 }
 739         }
 740
 741         /**
 742          * Retrieves the media attachments associated with the provided item ID.
 743          *
 744          * @param int $uri_id URI id
 745          * @param array $types Media types
 746          * @return array|bool Array on success, false on error
 747          * @throws \Exception
 748          */
 749         public static function getByURIId(int $uri_id, array $types = [])
 750         {
 751                 $condition = ["`uri-id` = ? AND `type` != ?", $uri_id, self::UNKNOWN];
 752
 753                 if (!empty($types)) {
 754                         $condition = DBA::mergeConditions($condition, ['type' => $types]);
 755                 }
 756
 757                 return DBA::selectToArray('post-media', [], $condition, ['order' => ['id']]);
 758         }
 759
 760         /**
 761          * Retrieves the media attachment with the provided media id.
 762          *
 763          * @param int $id  id
 764          * @return array|bool Array on success, false on error
 765          * @throws \Exception
 766          */
 767         public static function getById(int $id)
 768         {
 769                 return DBA::selectFirst('post-media', [], ['id' => $id]);
 770         }
 771
 772         /**
 773          * Update post-media entries
 774          *
 775          * @param array $fields
 776          * @param int $id
 777          * @return bool
 778          */
 779         public static function updateById(array $fields, int $id): bool
 780         {
 781                 return DBA::update('post-media', $fields, ['id' => $id]);
 782         }
 783
 784         /**
 785          * Checks if media attachments are associated with the provided item ID.
 786          *
 787          * @param int $uri_id URI id
 788          * @param array $types Media types
 789          * @return bool Whether media attachment exists
 790          * @throws \Exception
 791          */
 792         public static function existsByURIId(int $uri_id, array $types = []): bool
 793         {
 794                 $condition = ["`uri-id` = ? AND `type` != ?", $uri_id, self::UNKNOWN];
 795
 796                 if (!empty($types)) {
 797                         $condition = DBA::mergeConditions($condition, ['type' => $types]);
 798                 }
 799
 800                 return DBA::exists('post-media', $condition);
 801         }
 802
 803         /**
 804          * Delete media by uri-id and media type
 805          *
 806          * @param int $uri_id URI id
 807          * @param array $types Media types
 808          * @return bool result of deletion
 809          * @throws \Exception
 810          */
 811         public static function deleteByURIId(int $uri_id, array $types = []): bool
 812         {
 813                 $condition = ['uri-id' => $uri_id];
 814
 815                 if (!empty($types)) {
 816                         $condition = DBA::mergeConditions($condition, ['type' => $types]);
 817                 }
 818
 819                 return DBA::delete('post-media', $condition);
 820         }
 821
 822         /**
 823          * Delete media by id
 824          *
 825          * @param int $id media id
 826          * @return bool result of deletion
 827          * @throws \Exception
 828          */
 829         public static function deleteById(int $id): bool
 830         {
 831                 return DBA::delete('post-media', ['id' => $id]);
 832         }
 833
 834         /**
 835          * Split the attachment media in the three segments "visual", "link" and "additional"
 836          *
 837          * @param int    $uri_id URI id
 838          * @param array  $links list of links that shouldn't be added
 839          * @param bool   $has_media
 840          * @return array attachments
 841          */
 842         public static function splitAttachments(int $uri_id, array $links = [], bool $has_media = true): array
 843         {
 844                 $attachments = ['visual' => [], 'link' => [], 'additional' => []];
 845
 846                 if (!$has_media) {
 847                         return $attachments;
 848                 }
 849
 850                 $media = self::getByURIId($uri_id);
 851                 if (empty($media)) {
 852                         return $attachments;
 853                 }
 854
 855                 $heights = [];
 856                 $selected = '';
 857                 $previews = [];
 858
 859                 foreach ($media as $medium) {
 860                         foreach ($links as $link) {
 861                                 if (Strings::compareLink($link, $medium['url'])) {
 862                                         continue 2;
 863                                 }
 864                         }
 865
 866                         // Avoid adding separate media entries for previews
 867                         foreach ($previews as $preview) {
 868                                 if (Strings::compareLink($preview, $medium['url'])) {
 869                                         continue 2;
 870                                 }
 871                         }
 872
 873                         // Currently these two types are ignored here.
 874                         // Posts are added differently and contacts are not displayed as attachments.
 875                         if (in_array($medium['type'], [self::ACCOUNT, self::ACTIVITY])) {
 876                                 continue;
 877                         }
 878
 879                         if (!empty($medium['preview'])) {
 880                                 $previews[] = $medium['preview'];
 881                         }
 882
 883                         $type = explode('/', explode(';', $medium['mimetype'] ?? '')[0]);
 884                         if (count($type) < 2) {
 885                                 Logger::info('Unknown MimeType', ['type' => $type, 'media' => $medium]);
 886                                 $filetype = 'unkn';
 887                                 $subtype = 'unkn';
 888                         } else {
 889                                 $filetype = strtolower($type[0]);
 890                                 $subtype = strtolower($type[1]);
 891                         }
 892
 893                         $medium['filetype'] = $filetype;
 894                         $medium['subtype'] = $subtype;
 895
 896                         if ($medium['type'] == self::HTML || (($filetype == 'text') && ($subtype == 'html'))) {
 897                                 $attachments['link'][] = $medium;
 898                                 continue;
 899                         }
 900
 901                         if (
 902                                 in_array($medium['type'], [self::AUDIO, self::IMAGE]) ||
 903                                 in_array($filetype, ['audio', 'image'])
 904                         ) {
 905                                 $attachments['visual'][] = $medium;
 906                         } elseif (($medium['type'] == self::VIDEO) || ($filetype == 'video')) {
 907                                 if (!empty($medium['height'])) {
 908                                         // Peertube videos are delivered in many different resolutions. We pick a moderate one.
 909                                         // Since only Peertube provides a "height" parameter, this wouldn't be executed
 910                                         // when someone for example on Mastodon was sharing multiple videos in a single post.
 911                                         $heights[$medium['height']] = $medium['url'];
 912                                         $video[$medium['url']] = $medium;
 913                                 } else {
 914                                         $attachments['visual'][] = $medium;
 915                                 }
 916                         } else {
 917                                 $attachments['additional'][] = $medium;
 918                         }
 919                 }
 920
 921                 if (!empty($heights)) {
 922                         ksort($heights);
 923                         foreach ($heights as $height => $url) {
 924                                 if (empty($selected) || $height <= 480) {
 925                                         $selected = $url;
 926                                 }
 927                         }
 928
 929                         if (!empty($selected)) {
 930                                 $attachments['visual'][] = $video[$selected];
 931                                 unset($video[$selected]);
 932                                 foreach ($video as $element) {
 933                                         $attachments['additional'][] = $element;
 934                                 }
 935                         }
 936                 }
 937
 938                 return $attachments;
 939         }
 940
 941         /**
 942          * Add media attachments to the body
 943          *
 944          * @param int    $uriid
 945          * @param string $body
 946          * @param array  $types
 947          *
 948          * @return string body
 949          */
 950         public static function addAttachmentsToBody(int $uriid, string $body = '', array $types = [self::IMAGE, self::AUDIO, self::VIDEO]): string
 951         {
 952                 if (empty($body)) {
 953                         $item = Post::selectFirst(['body'], ['uri-id' => $uriid]);
 954                         if (!DBA::isResult($item)) {
 955                                 return '';
 956                         }
 957                         $body = $item['body'];
 958                 }
 959                 $original_body = $body;
 960
 961                 $body = BBCode::removeAttachment($body);
 962
 963                 foreach (self::getByURIId($uriid, $types) as $media) {
 964                         if (Item::containsLink($body, $media['preview'] ?? $media['url'], $media['type'])) {
 965                                 continue;
 966                         }
 967
 968                         if ($media['type'] == self::IMAGE) {
 969                                 if (!empty($media['preview'])) {
 970                                         if (!empty($media['description'])) {
 971                                                 $body .= "\n[url=" . $media['url'] . "][img=" . $media['preview'] . ']' . $media['description'] . '[/img][/url]';
 972                                         } else {
 973                                                 $body .= "\n[url=" . $media['url'] . "][img]" . $media['preview'] . '[/img][/url]';
 974                                         }
 975                                 } else {
 976                                         if (!empty($media['description'])) {
 977                                                 $body .= "\n[img=" . $media['url'] . ']' . $media['description'] . '[/img]';
 978                                         } else {
 979                                                 $body .= "\n[img]" . $media['url'] . '[/img]';
 980                                         }
 981                                 }
 982                         } elseif ($media['type'] == self::AUDIO) {
 983                                 $body .= "\n[audio]" . $media['url'] . "[/audio]\n";
 984                         } elseif ($media['type'] == self::VIDEO) {
 985                                 $body .= "\n[video]" . $media['url'] . "[/video]\n";
 986                         }
 987                 }
 988
 989                 if (preg_match("/.*(\[attachment.*?\].*?\[\/attachment\]).*/ism", $original_body, $match)) {
 990                         $body .= "\n" . $match[1];
 991                 }
 992
 993                 return $body;
 994         }
 995
 996         /**
 997          * Add an [attachment] element to the body for a given uri-id with a HTML media element
 998          *
 999          * @param integer $uriid
1000          * @param string $body
1001          * @return string
1002          */
1003         public static function addHTMLAttachmentToBody(int $uriid, string $body): string
1004         {
1005                 if (preg_match("/.*(\[attachment.*?\].*?\[\/attachment\]).*/ism", $body, $match)) {
1006                         return $body;
1007                 }
1008
1009                 $links = self::getByURIId($uriid, [self::HTML]);
1010                 if (empty($links)) {
1011                         return $body;
1012                 }
1013
1014                 $data = [
1015                         'type' => 'link',
1016                         'url'  => $links[0]['url'],
1017                         'title' => $links[0]['name'],
1018                         'text' => $links[0]['description'],
1019                         'publisher_name' => $links[0]['publisher-name'],
1020                         'publisher_url' => $links[0]['publisher-url'],
1021                         'publisher_img' => $links[0]['publisher-image'],
1022                         'author_name' => $links[0]['author-name'],
1023                         'author_url' => $links[0]['author-url'],
1024                         'author_img' => $links[0]['author-image'],
1025                         'images' => [[
1026                                 'src' => $links[0]['preview'],
1027                                 'height' => $links[0]['preview-height'],
1028                                 'width' => $links[0]['preview-width'],
1029                         ]]
1030                 ];
1031                 $body .= "\n" . PageInfo::getFooterFromData($data);
1032
1033                 return $body;
1034         }
1035
1036         /**
1037          * Add a link to the body for a given uri-id with a HTML media element
1038          *
1039          * @param integer $uriid
1040          * @param string $body
1041          * @return string
1042          */
1043         public static function addHTMLLinkToBody(int $uriid, string $body): string
1044         {
1045                 $links = self::getByURIId($uriid, [self::HTML]);
1046                 if (empty($links)) {
1047                         return $body;
1048                 }
1049
1050                 if (strpos($body, $links[0]['url'])) {
1051                         return $body;
1052                 }
1053
1054                 if (!empty($links[0]['name']) && ($links[0]['name'] != $links[0]['url'])) {
1055                         return $body . "\n[url=" . $links[0]['url'] . ']' . $links[0]['name'] . "[/url]";
1056                 } else {
1057                         return $body . "\n[url]" . $links[0]['url'] . "[/url]";
1058                 }
1059         }
1060
1061         /**
1062          * Add an [attachment] element to the body and a link to raw-body for a given uri-id with a HTML media element
1063          *
1064          * @param array $item
1065          * @return array
1066          */
1067         public static function addHTMLAttachmentToItem(array $item): array
1068         {
1069                 if (($item['gravity'] == Item::GRAVITY_ACTIVITY) || empty($item['uri-id'])) {
1070                         return $item;
1071                 }
1072
1073                 $item['body'] = self::addHTMLAttachmentToBody($item['uri-id'], $item['body']);
1074
1075                 if (!empty($item['raw-body'])) {
1076                         $item['raw-body'] = self::addHTMLLinkToBody($item['uri-id'], $item['raw-body']);
1077                 }
1078
1079                 return $item;
1080         }
1081
1082         /**
1083          * Get preview link for given media id
1084          *
1085          * @param integer $id   media id
1086          * @param string  $size One of the Proxy::SIZE_* constants
1087          * @return string preview link
1088          */
1089         public static function getPreviewUrlForId(int $id, string $size = ''): string
1090         {
1091                 $url = DI::baseUrl() . '/photo/preview/';
1092                 switch ($size) {
1093                         case Proxy::SIZE_MICRO:
1094                                 $url .= Proxy::PIXEL_MICRO . '/';
1095                                 break;
1096                         case Proxy::SIZE_THUMB:
1097                                 $url .= Proxy::PIXEL_THUMB . '/';
1098                                 break;
1099                         case Proxy::SIZE_SMALL:
1100                                 $url .= Proxy::PIXEL_SMALL . '/';
1101                                 break;
1102                         case Proxy::SIZE_MEDIUM:
1103                                 $url .= Proxy::PIXEL_MEDIUM . '/';
1104                                 break;
1105                         case Proxy::SIZE_LARGE:
1106                                 $url .= Proxy::PIXEL_LARGE . '/';
1107                                 break;
1108                 }
1109                 return $url . $id;
1110         }
1111
1112         /**
1113          * Get media link for given media id
1114          *
1115          * @param integer $id   media id
1116          * @param string  $size One of the Proxy::SIZE_* constants
1117          * @return string media link
1118          */
1119         public static function getUrlForId(int $id, string $size = ''): string
1120         {
1121                 $url = DI::baseUrl() . '/photo/media/';
1122                 switch ($size) {
1123                         case Proxy::SIZE_MICRO:
1124                                 $url .= Proxy::PIXEL_MICRO . '/';
1125                                 break;
1126                         case Proxy::SIZE_THUMB:
1127                                 $url .= Proxy::PIXEL_THUMB . '/';
1128                                 break;
1129                         case Proxy::SIZE_SMALL:
1130                                 $url .= Proxy::PIXEL_SMALL . '/';
1131                                 break;
1132                         case Proxy::SIZE_MEDIUM:
1133                                 $url .= Proxy::PIXEL_MEDIUM . '/';
1134                                 break;
1135                         case Proxy::SIZE_LARGE:
1136                                 $url .= Proxy::PIXEL_LARGE . '/';
1137                                 break;
1138                 }
1139                 return $url . $id;
1140         }
1141 }