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 isPictureLink(string $page, string $preview): bool
 465         {
 466                 return (preg_match('#/photo/.*-0\.#ism', $page) || preg_match('#/photos/.*/image/#ism', $page)) && preg_match('#/photo/.*-[01]\.#ism', $preview);
 467         }
 468
 469         /**
 470          * Add media links and remove them from the body
 471          *
 472          * @param integer $uriid
 473          * @param string $body
 474          * @return string Body without media links
 475          */
 476         public static function insertFromBody(int $uriid, string $body, bool $endmatch = false): string
 477         {
 478                 $endmatchpattern = $endmatch ? '\z' : '';
 479                 // Simplify image codes
 480                 $unshared_body = $body = preg_replace("/\[img\=([0-9]*)x([0-9]*)\](.*?)\[\/img\]$endmatchpattern/ism", '[img]$3[/img]', $body);
 481
 482                 $attachments = [];
 483                 if (preg_match_all("#\[url=([^\]]+?)\]\s*\[img=([^\[\]]*)\]([^\[\]]*)\[\/img\]\s*\[/url\]$endmatchpattern#ism", $body, $pictures, PREG_SET_ORDER)) {
 484                         foreach ($pictures as $picture) {
 485                                 if (self::isPictureLink($picture[1], $picture[2])) {
 486                                         $body = str_replace($picture[0], '', $body);
 487                                         $image = str_replace('-1.', '-0.', $picture[2]);
 488                                         $attachments[$image] = [
 489                                                 'uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $image,
 490                                                 'preview' => $picture[2], 'description' => $picture[3]
 491                                         ];
 492                                 } else {
 493                                         $body = str_replace($picture[0], '', $body);
 494                                         $attachments[$picture[1]] = [
 495                                                 'uri-id' => $uriid, 'type' => self::UNKNOWN, 'url' => $picture[1],
 496                                                 'preview' => $picture[2], 'description' => $picture[3]
 497                                         ];
 498                                 }
 499                         }
 500                 }
 501
 502                 if (preg_match_all("/\[img=([^\[\]]*)\]([^\[\]]*)\[\/img\]$endmatchpattern/Usi", $body, $pictures, PREG_SET_ORDER)) {
 503                         foreach ($pictures as $picture) {
 504                                 $body = str_replace($picture[0], '', $body);
 505                                 $attachments[$picture[1]] = ['uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $picture[1], 'description' => $picture[2]];
 506                         }
 507                 }
 508
 509                 if (preg_match_all("#\[url=([^\]]+?)\]\s*\[img\]([^\[]+?)\[/img\]\s*\[/url\]$endmatchpattern#ism", $body, $pictures, PREG_SET_ORDER)) {
 510                         foreach ($pictures as $picture) {
 511                                 if (self::isPictureLink($picture[1], $picture[2])) {
 512                                         $body = str_replace($picture[0], '', $body);
 513                                         $image = str_replace('-1.', '-0.', $picture[2]);
 514                                         $attachments[$image] = [
 515                                                 'uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $image,
 516                                                 'preview' => $picture[2], 'description' => null
 517                                         ];
 518                                 } else {
 519                                         $body = str_replace($picture[0], '', $body);
 520                                         $attachments[$picture[1]] = [
 521                                                 'uri-id' => $uriid, 'type' => self::UNKNOWN, 'url' => $picture[1],
 522                                                 'preview' => $picture[2], 'description' => null
 523                                         ];
 524                                 }
 525                         }
 526                 }
 527
 528                 if (preg_match_all("/\[img\]([^\[\]]*)\[\/img\]$endmatchpattern/ism", $body, $pictures, PREG_SET_ORDER)) {
 529                         foreach ($pictures as $picture) {
 530                                 $body = str_replace($picture[0], '', $body);
 531                                 $attachments[$picture[1]] = ['uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $picture[1]];
 532                         }
 533                 }
 534
 535                 if (preg_match_all("/\[audio\]([^\[\]]*)\[\/audio\]$endmatchpattern/ism", $body, $audios, PREG_SET_ORDER)) {
 536                         foreach ($audios as $audio) {
 537                                 $body = str_replace($audio[0], '', $body);
 538                                 $attachments[$audio[1]] = ['uri-id' => $uriid, 'type' => self::AUDIO, 'url' => $audio[1]];
 539                         }
 540                 }
 541
 542                 if (preg_match_all("/\[video\]([^\[\]]*)\[\/video\]$endmatchpattern/ism", $body, $videos, PREG_SET_ORDER)) {
 543                         foreach ($videos as $video) {
 544                                 $body = str_replace($video[0], '', $body);
 545                                 $attachments[$video[1]] = ['uri-id' => $uriid, 'type' => self::VIDEO, 'url' => $video[1]];
 546                         }
 547                 }
 548
 549                 if ($uriid != 0) {
 550                         foreach ($attachments as $attachment) {
 551                                 if (Post\Link::exists($uriid, $attachment['preview'] ?? $attachment['url'])) {
 552                                         continue;
 553                                 }
 554
 555                                 // Only store attachments that are part of the unshared body
 556                                 if (Item::containsLink($unshared_body, $attachment['preview'] ?? $attachment['url'], $attachment['type'])) {
 557                                         self::insert($attachment);
 558                                 }
 559                         }
 560                 }
 561
 562                 return trim($body);
 563         }
 564
 565         /**
 566          * Remove media that is at the end of the body
 567          *
 568          * @param string $body
 569          * @return string
 570          */
 571         public static function removeFromEndOfBody(string $body): string
 572         {
 573                 do {
 574                         $prebody = $body;
 575                         $body = self::insertFromBody(0, $body, true);
 576                 } while ($prebody != $body);
 577                 return $body;
 578         }
 579
 580         /**
 581          * Remove media from the body
 582          *
 583          * @param string $body
 584          * @return string
 585          */
 586         public static function removeFromBody(string $body): string
 587         {
 588                 do {
 589                         $prebody = $body;
 590                         $body = self::insertFromBody(0, $body);
 591                 } while ($prebody != $body);
 592                 return $body;
 593         }
 594
 595         /**
 596          * Add media links from a relevant url in the body
 597          *
 598          * @param integer $uriid
 599          * @param string $body
 600          * @return void
 601          */
 602         public static function insertFromRelevantUrl(int $uriid, string $body, string $fullbody, string $network)
 603         {
 604                 // Remove all hashtags and mentions
 605                 $body = preg_replace("/([#@!])\[url\=(.*?)\](.*?)\[\/url\]/ism", '', $body);
 606
 607                 // Search for pure links
 608                 if (preg_match_all("/\[url\](https?:.*?)\[\/url\]/ism", $body, $matches)) {
 609                         foreach ($matches[1] as $url) {
 610                                 Logger::info('Got page url (link without description)', ['uri-id' => $uriid, 'url' => $url]);
 611                                 $result = self::insert(['uri-id' => $uriid, 'type' => self::UNKNOWN, 'url' => $url], false, $network);
 612                                 if ($result && ($network == Protocol::DFRN)) {
 613                                         self::revertHTMLType($uriid, $url, $fullbody);
 614                                         Logger::debug('Revert HTML type', ['uri-id' => $uriid, 'url' => $url]);
 615                                 } elseif ($result) {
 616                                         Logger::debug('Media had been added', ['uri-id' => $uriid, 'url' => $url]);
 617                                 } else {
 618                                         Logger::debug('Media had not been added', ['uri-id' => $uriid, 'url' => $url]);
 619                                 }
 620                         }
 621                 }
 622
 623                 // Search for links with descriptions
 624                 if (preg_match_all("/\[url\=(https?:.*?)\].*?\[\/url\]/ism", $body, $matches)) {
 625                         foreach ($matches[1] as $url) {
 626                                 Logger::info('Got page url (link with description)', ['uri-id' => $uriid, 'url' => $url]);
 627                                 $result = self::insert(['uri-id' => $uriid, 'type' => self::UNKNOWN, 'url' => $url], false, $network);
 628                                 if ($result && ($network == Protocol::DFRN)) {
 629                                         self::revertHTMLType($uriid, $url, $fullbody);
 630                                         Logger::debug('Revert HTML type', ['uri-id' => $uriid, 'url' => $url]);
 631                                 } elseif ($result) {
 632                                         Logger::debug('Media has been added', ['uri-id' => $uriid, 'url' => $url]);
 633                                 } else {
 634                                         Logger::debug('Media has not been added', ['uri-id' => $uriid, 'url' => $url]);
 635                                 }
 636                         }
 637                 }
 638         }
 639
 640         /**
 641          * Revert the media type of links to UNKNOWN for DFRN posts when they aren't attached
 642          *
 643          * @param integer $uriid
 644          * @param string $url
 645          * @param string $body
 646          * @return void
 647          */
 648         private static function revertHTMLType(int $uriid, string $url, string $body)
 649         {
 650                 $attachment = BBCode::getAttachmentData($body);
 651                 if (!empty($attachment['url']) && Network::getUrlMatch($attachment['url'], $url)) {
 652                         return;
 653                 }
 654                 DBA::update('post-media', ['type' => self::UNKNOWN], ['uri-id' => $uriid, 'type' => self::HTML, 'url' => $url]);
 655         }
 656
 657         /**
 658          * Add media links from the attachment field
 659          *
 660          * @param integer $uriid
 661          * @param string $body
 662          * @return void
 663          */
 664         public static function insertFromAttachmentData(int $uriid, string $body)
 665         {
 666                 $data = BBCode::getAttachmentData($body);
 667                 if (empty($data)) {
 668                         return;
 669                 }
 670
 671                 Logger::info('Adding attachment data', ['data' => $data]);
 672                 $attachment = [
 673                         'uri-id' => $uriid,
 674                         'type' => self::HTML,
 675                         'url' => $data['url'],
 676                         'preview' => $data['preview'] ?? null,
 677                         'description' => $data['description'] ?? null,
 678                         'name' => $data['title'] ?? null,
 679                         'author-url' => $data['author_url'] ?? null,
 680                         'author-name' => $data['author_name'] ?? null,
 681                         'publisher-url' => $data['provider_url'] ?? null,
 682                         'publisher-name' => $data['provider_name'] ?? null,
 683                 ];
 684                 if (!empty($data['image'])) {
 685                         $attachment['preview'] = $data['image'];
 686                 }
 687                 self::insert($attachment);
 688         }
 689
 690         /**
 691          * Add media links from the attach field
 692          *
 693          * @param integer $uriid
 694          * @param string $attach
 695          * @return void
 696          */
 697         public static function insertFromAttachment(int $uriid, string $attach)
 698         {
 699                 if (!preg_match_all('|\[attach\]href=\"(.*?)\" length=\"(.*?)\" type=\"(.*?)\"(?: title=\"(.*?)\")?|', $attach, $matches, PREG_SET_ORDER)) {
 700                         return;
 701                 }
 702
 703                 foreach ($matches as $attachment) {
 704                         $media['type'] = self::DOCUMENT;
 705                         $media['uri-id'] = $uriid;
 706                         $media['url'] = $attachment[1];
 707                         $media['size'] = $attachment[2];
 708                         $media['mimetype'] = $attachment[3];
 709                         $media['description'] = $attachment[4] ?? '';
 710
 711                         self::insert($media);
 712                 }
 713         }
 714
 715         /**
 716          * Retrieves the media attachments associated with the provided item ID.
 717          *
 718          * @param int $uri_id URI id
 719          * @param array $types Media types
 720          * @return array|bool Array on success, false on error
 721          * @throws \Exception
 722          */
 723         public static function getByURIId(int $uri_id, array $types = [])
 724         {
 725                 $condition = ["`uri-id` = ? AND `type` != ?", $uri_id, self::UNKNOWN];
 726
 727                 if (!empty($types)) {
 728                         $condition = DBA::mergeConditions($condition, ['type' => $types]);
 729                 }
 730
 731                 return DBA::selectToArray('post-media', [], $condition, ['order' => ['id']]);
 732         }
 733
 734         /**
 735          * Checks if media attachments are associated with the provided item ID.
 736          *
 737          * @param int $uri_id URI id
 738          * @param array $types Media types
 739          * @return bool Whether media attachment exists
 740          * @throws \Exception
 741          */
 742         public static function existsByURIId(int $uri_id, array $types = []): bool
 743         {
 744                 $condition = ["`uri-id` = ? AND `type` != ?", $uri_id, self::UNKNOWN];
 745
 746                 if (!empty($types)) {
 747                         $condition = DBA::mergeConditions($condition, ['type' => $types]);
 748                 }
 749
 750                 return DBA::exists('post-media', $condition);
 751         }
 752
 753         /**
 754          * Delete media by uri-id and media type
 755          *
 756          * @param int $uri_id URI id
 757          * @param array $types Media types
 758          * @return bool Whether media attachment exists
 759          * @throws \Exception
 760          */
 761         public static function deleteByURIId(int $uri_id, array $types = []): bool
 762         {
 763                 $condition = ['uri-id' => $uri_id];
 764
 765                 if (!empty($types)) {
 766                         $condition = DBA::mergeConditions($condition, ['type' => $types]);
 767                 }
 768
 769                 return DBA::delete('post-media', $condition);
 770         }
 771
 772         /**
 773          * Split the attachment media in the three segments "visual", "link" and "additional"
 774          *
 775          * @param int    $uri_id URI id
 776          * @param array  $links list of links that shouldn't be added
 777          * @param bool   $has_media
 778          * @return array attachments
 779          */
 780         public static function splitAttachments(int $uri_id, array $links = [], bool $has_media = true): array
 781         {
 782                 $attachments = ['visual' => [], 'link' => [], 'additional' => []];
 783
 784                 if (!$has_media) {
 785                         return $attachments;
 786                 }
 787
 788                 $media = self::getByURIId($uri_id);
 789                 if (empty($media)) {
 790                         return $attachments;
 791                 }
 792
 793                 $heights = [];
 794                 $selected = '';
 795                 $previews = [];
 796
 797                 foreach ($media as $medium) {
 798                         foreach ($links as $link) {
 799                                 if (Strings::compareLink($link, $medium['url'])) {
 800                                         continue 2;
 801                                 }
 802                         }
 803
 804                         // Avoid adding separate media entries for previews
 805                         foreach ($previews as $preview) {
 806                                 if (Strings::compareLink($preview, $medium['url'])) {
 807                                         continue 2;
 808                                 }
 809                         }
 810
 811                         // Currently these two types are ignored here.
 812                         // Posts are added differently and contacts are not displayed as attachments.
 813                         if (in_array($medium['type'], [self::ACCOUNT, self::ACTIVITY])) {
 814                                 continue;
 815                         }
 816
 817                         if (!empty($medium['preview'])) {
 818                                 $previews[] = $medium['preview'];
 819                         }
 820
 821                         $type = explode('/', explode(';', $medium['mimetype'] ?? '')[0]);
 822                         if (count($type) < 2) {
 823                                 Logger::info('Unknown MimeType', ['type' => $type, 'media' => $medium]);
 824                                 $filetype = 'unkn';
 825                                 $subtype = 'unkn';
 826                         } else {
 827                                 $filetype = strtolower($type[0]);
 828                                 $subtype = strtolower($type[1]);
 829                         }
 830
 831                         $medium['filetype'] = $filetype;
 832                         $medium['subtype'] = $subtype;
 833
 834                         if ($medium['type'] == self::HTML || (($filetype == 'text') && ($subtype == 'html'))) {
 835                                 $attachments['link'][] = $medium;
 836                                 continue;
 837                         }
 838
 839                         if (
 840                                 in_array($medium['type'], [self::AUDIO, self::IMAGE]) ||
 841                                 in_array($filetype, ['audio', 'image'])
 842                         ) {
 843                                 $attachments['visual'][] = $medium;
 844                         } elseif (($medium['type'] == self::VIDEO) || ($filetype == 'video')) {
 845                                 if (!empty($medium['height'])) {
 846                                         // Peertube videos are delivered in many different resolutions. We pick a moderate one.
 847                                         // Since only Peertube provides a "height" parameter, this wouldn't be executed
 848                                         // when someone for example on Mastodon was sharing multiple videos in a single post.
 849                                         $heights[$medium['height']] = $medium['url'];
 850                                         $video[$medium['url']] = $medium;
 851                                 } else {
 852                                         $attachments['visual'][] = $medium;
 853                                 }
 854                         } else {
 855                                 $attachments['additional'][] = $medium;
 856                         }
 857                 }
 858
 859                 if (!empty($heights)) {
 860                         ksort($heights);
 861                         foreach ($heights as $height => $url) {
 862                                 if (empty($selected) || $height <= 480) {
 863                                         $selected = $url;
 864                                 }
 865                         }
 866
 867                         if (!empty($selected)) {
 868                                 $attachments['visual'][] = $video[$selected];
 869                                 unset($video[$selected]);
 870                                 foreach ($video as $element) {
 871                                         $attachments['additional'][] = $element;
 872                                 }
 873                         }
 874                 }
 875
 876                 return $attachments;
 877         }
 878
 879         /**
 880          * Add media attachments to the body
 881          *
 882          * @param int    $uriid
 883          * @param string $body
 884          * @param array  $types
 885          *
 886          * @return string body
 887          */
 888         public static function addAttachmentsToBody(int $uriid, string $body = '', array $types = [self::IMAGE, self::AUDIO, self::VIDEO]): string
 889         {
 890                 if (empty($body)) {
 891                         $item = Post::selectFirst(['body'], ['uri-id' => $uriid]);
 892                         if (!DBA::isResult($item)) {
 893                                 return '';
 894                         }
 895                         $body = $item['body'];
 896                 }
 897                 $original_body = $body;
 898
 899                 $body = BBCode::removeAttachment($body);
 900
 901                 foreach (self::getByURIId($uriid, $types) as $media) {
 902                         if (Item::containsLink($body, $media['preview'] ?? $media['url'], $media['type'])) {
 903                                 continue;
 904                         }
 905
 906                         if ($media['type'] == self::IMAGE) {
 907                                 if (!empty($media['preview'])) {
 908                                         if (!empty($media['description'])) {
 909                                                 $body .= "\n[url=" . $media['url'] . "][img=" . $media['preview'] . ']' . $media['description'] . '[/img][/url]';
 910                                         } else {
 911                                                 $body .= "\n[url=" . $media['url'] . "][img]" . $media['preview'] . '[/img][/url]';
 912                                         }
 913                                 } else {
 914                                         if (!empty($media['description'])) {
 915                                                 $body .= "\n[img=" . $media['url'] . ']' . $media['description'] . '[/img]';
 916                                         } else {
 917                                                 $body .= "\n[img]" . $media['url'] . '[/img]';
 918                                         }
 919                                 }
 920                         } elseif ($media['type'] == self::AUDIO) {
 921                                 $body .= "\n[audio]" . $media['url'] . "[/audio]\n";
 922                         } elseif ($media['type'] == self::VIDEO) {
 923                                 $body .= "\n[video]" . $media['url'] . "[/video]\n";
 924                         }
 925                 }
 926
 927                 if (preg_match("/.*(\[attachment.*?\].*?\[\/attachment\]).*/ism", $original_body, $match)) {
 928                         $body .= "\n" . $match[1];
 929                 }
 930
 931                 return $body;
 932         }
 933
 934         /**
 935          * Add an [attachment] element to the body for a given uri-id with a HTML media element
 936          *
 937          * @param integer $uriid
 938          * @param string $body
 939          * @return string
 940          */
 941         public static function addHTMLAttachmentToBody(int $uriid, string $body): string
 942         {
 943                 if (preg_match("/.*(\[attachment.*?\].*?\[\/attachment\]).*/ism", $body, $match)) {
 944                         return $body;
 945                 }
 946
 947                 $links = self::getByURIId($uriid, [self::HTML]);
 948                 if (empty($links)) {
 949                         return $body;
 950                 }
 951
 952                 $data = [
 953                         'type' => 'link',
 954                         'url'  => $links[0]['url'],
 955                         'title' => $links[0]['name'],
 956                         'text' => $links[0]['description'],
 957                         'publisher_name' => $links[0]['publisher-name'],
 958                         'publisher_url' => $links[0]['publisher-url'],
 959                         'publisher_img' => $links[0]['publisher-image'],
 960                         'author_name' => $links[0]['author-name'],
 961                         'author_url' => $links[0]['author-url'],
 962                         'author_img' => $links[0]['author-image'],
 963                         'images' => [[
 964                                 'src' => $links[0]['preview'],
 965                                 'height' => $links[0]['preview-height'],
 966                                 'width' => $links[0]['preview-width'],
 967                         ]]
 968                 ];
 969                 $body .= "\n" . PageInfo::getFooterFromData($data);
 970
 971                 return $body;
 972         }
 973
 974         /**
 975          * Add a link to the body for a given uri-id with a HTML media element
 976          *
 977          * @param integer $uriid
 978          * @param string $body
 979          * @return string
 980          */
 981         public static function addHTMLLinkToBody(int $uriid, string $body): string
 982         {
 983                 $links = self::getByURIId($uriid, [self::HTML]);
 984                 if (empty($links)) {
 985                         return $body;
 986                 }
 987
 988                 if (strpos($body, $links[0]['url'])) {
 989                         return $body;
 990                 }
 991
 992                 if (!empty($links[0]['name']) && ($links[0]['name'] != $links[0]['url'])) {
 993                         return $body . "\n[url=" . $links[0]['url'] . ']' . $links[0]['name'] . "[/url]";
 994                 } else {
 995                         return $body . "\n[url]" . $links[0]['url'] . "[/url]";
 996                 }
 997         }
 998
 999         /**
1000          * Add an [attachment] element to the body and a link to raw-body for a given uri-id with a HTML media element
1001          *
1002          * @param array $item
1003          * @return array
1004          */
1005         public static function addHTMLAttachmentToItem(array $item): array
1006         {
1007                 if (($item['gravity'] == Item::GRAVITY_ACTIVITY) || empty($item['uri-id'])) {
1008                         return $item;
1009                 }
1010
1011                 $item['body'] = self::addHTMLAttachmentToBody($item['uri-id'], $item['body']);
1012
1013                 if (!empty($item['raw-body'])) {
1014                         $item['raw-body'] = self::addHTMLLinkToBody($item['uri-id'], $item['raw-body']);
1015                 }
1016
1017                 return $item;
1018         }
1019
1020         /**
1021          * Get preview link for given media id
1022          *
1023          * @param integer $id   media id
1024          * @param string  $size One of the Proxy::SIZE_* constants
1025          * @return string preview link
1026          */
1027         public static function getPreviewUrlForId(int $id, string $size = ''): string
1028         {
1029                 $url = DI::baseUrl() . '/photo/preview/';
1030                 switch ($size) {
1031                         case Proxy::SIZE_MICRO:
1032                                 $url .= Proxy::PIXEL_MICRO . '/';
1033                                 break;
1034                         case Proxy::SIZE_THUMB:
1035                                 $url .= Proxy::PIXEL_THUMB . '/';
1036                                 break;
1037                         case Proxy::SIZE_SMALL:
1038                                 $url .= Proxy::PIXEL_SMALL . '/';
1039                                 break;
1040                         case Proxy::SIZE_MEDIUM:
1041                                 $url .= Proxy::PIXEL_MEDIUM . '/';
1042                                 break;
1043                         case Proxy::SIZE_LARGE:
1044                                 $url .= Proxy::PIXEL_LARGE . '/';
1045                                 break;
1046                 }
1047                 return $url . $id;
1048         }
1049
1050         /**
1051          * Get media link for given media id
1052          *
1053          * @param integer $id   media id
1054          * @param string  $size One of the Proxy::SIZE_* constants
1055          * @return string media link
1056          */
1057         public static function getUrlForId(int $id, string $size = ''): string
1058         {
1059                 $url = DI::baseUrl() . '/photo/media/';
1060                 switch ($size) {
1061                         case Proxy::SIZE_MICRO:
1062                                 $url .= Proxy::PIXEL_MICRO . '/';
1063                                 break;
1064                         case Proxy::SIZE_THUMB:
1065                                 $url .= Proxy::PIXEL_THUMB . '/';
1066                                 break;
1067                         case Proxy::SIZE_SMALL:
1068                                 $url .= Proxy::PIXEL_SMALL . '/';
1069                                 break;
1070                         case Proxy::SIZE_MEDIUM:
1071                                 $url .= Proxy::PIXEL_MEDIUM . '/';
1072                                 break;
1073                         case Proxy::SIZE_LARGE:
1074                                 $url .= Proxy::PIXEL_LARGE . '/';
1075                                 break;
1076                 }
1077                 return $url . $id;
1078         }
1079 }