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\Text\BBCode;
  25 use Friendica\Core\Logger;
  26 use Friendica\Core\Protocol;
  27 use Friendica\Core\System;
  28 use Friendica\Database\Database;
  29 use Friendica\Database\DBA;
  30 use Friendica\DI;
  31 use Friendica\Model\Contact;
  32 use Friendica\Model\Item;
  33 use Friendica\Model\ItemURI;
  34 use Friendica\Model\Photo;
  35 use Friendica\Model\Post;
  36 use Friendica\Network\HTTPClient\Client\HttpClientAccept;
  37 use Friendica\Network\HTTPClient\Client\HttpClientOptions;
  38 use Friendica\Util\Images;
  39 use Friendica\Util\Network;
  40 use Friendica\Util\ParseUrl;
  41 use Friendica\Util\Proxy;
  42 use Friendica\Util\Strings;
  43
  44 /**
  45  * Class Media
  46  *
  47  * This Model class handles media interactions.
  48  * This tables stores medias (images, videos, audio files) related to posts.
  49  */
  50 class Media
  51 {
  52         const UNKNOWN     = 0;
  53         const IMAGE       = 1;
  54         const VIDEO       = 2;
  55         const AUDIO       = 3;
  56         const TEXT        = 4;
  57         const APPLICATION = 5;
  58         const TORRENT     = 16;
  59         const HTML        = 17;
  60         const XML         = 18;
  61         const PLAIN       = 19;
  62         const ACTIVITY    = 20;
  63         const ACCOUNT     = 21;
  64         const DOCUMENT    = 128;
  65
  66         /**
  67          * Insert a post-media record
  68          *
  69          * @param array $media
  70          * @param bool  $force
  71          * @return void
  72          */
  73         public static function insert(array $media, bool $force = false)
  74         {
  75                 if (empty($media['url']) || empty($media['uri-id']) || !isset($media['type'])) {
  76                         Logger::warning('Incomplete media data', ['media' => $media]);
  77                         return;
  78                 }
  79
  80                 if (DBA::exists('post-media', ['uri-id' => $media['uri-id'], 'preview' => $media['url']])) {
  81                         Logger::info('Media already exists as preview', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'callstack' => System::callstack()]);
  82                         return;
  83                 }
  84
  85                 // "document" has got the lowest priority. So when the same file is both attached as document
  86                 // and embedded as picture then we only store the picture or replace the document
  87                 $found = DBA::selectFirst('post-media', ['type'], ['uri-id' => $media['uri-id'], 'url' => $media['url']]);
  88                 if (!$force && !empty($found) && (($found['type'] != self::DOCUMENT) || ($media['type'] == self::DOCUMENT))) {
  89                         Logger::info('Media already exists', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'callstack' => System::callstack()]);
  90                         return;
  91                 }
  92
  93                 if (!ItemURI::exists($media['uri-id'])) {
  94                         Logger::info('Media referenced URI ID not found', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'callstack' => System::callstack()]);
  95                         return;
  96                 }
  97
  98                 $media = self::unsetEmptyFields($media);
  99                 $media = DI::dbaDefinition()->truncateFieldsForTable('post-media', $media);
 100
 101                 // We are storing as fast as possible to avoid duplicated network requests
 102                 // when fetching additional information for pictures and other content.
 103                 $result = DBA::insert('post-media', $media, Database::INSERT_UPDATE);
 104                 Logger::info('Stored media', ['result' => $result, 'media' => $media, 'callstack' => System::callstack()]);
 105                 $stored = $media;
 106
 107                 $media = self::fetchAdditionalData($media);
 108                 $media = self::unsetEmptyFields($media);
 109                 $media = DI::dbaDefinition()->truncateFieldsForTable('post-media', $media);
 110
 111                 if (array_diff_assoc($media, $stored)) {
 112                         $result = DBA::insert('post-media', $media, Database::INSERT_UPDATE);
 113                         Logger::info('Updated media', ['result' => $result, 'media' => $media]);
 114                 } else {
 115                         Logger::info('Nothing to update', ['media' => $media]);
 116                 }
 117         }
 118
 119         /**
 120          * Remove empty media fields
 121          *
 122          * @param array $media
 123          * @return array cleaned media array
 124          */
 125         private static function unsetEmptyFields(array $media): array
 126         {
 127                 $fields = ['mimetype', 'height', 'width', 'size', 'preview', 'preview-height', 'preview-width', 'blurhash', 'description'];
 128                 foreach ($fields as $field) {
 129                         if (empty($media[$field])) {
 130                                 unset($media[$field]);
 131                         }
 132                 }
 133                 return $media;
 134         }
 135
 136         /**
 137          * Copy attachments from one uri-id to another
 138          *
 139          * @param integer $from_uri_id
 140          * @param integer $to_uri_id
 141          * @return void
 142          */
 143         public static function copy(int $from_uri_id, int $to_uri_id)
 144         {
 145                 $attachments = self::getByURIId($from_uri_id);
 146                 foreach ($attachments as $attachment) {
 147                         $attachment['uri-id'] = $to_uri_id;
 148                         self::insert($attachment);
 149                 }
 150         }
 151
 152         /**
 153          * Creates the "[attach]" element from the given attributes
 154          *
 155          * @param string $href
 156          * @param integer $length
 157          * @param string $type
 158          * @param string $title
 159          * @return string "[attach]" element
 160          */
 161         public static function getAttachElement(string $href, int $length, string $type, string $title = ''): string
 162         {
 163                 $media = self::fetchAdditionalData(['type' => self::DOCUMENT, 'url' => $href,
 164                         'size' => $length, 'mimetype' => $type, 'description' => $title]);
 165
 166                 return '[attach]href="' . $media['url'] . '" length="' . $media['size'] .
 167                         '" type="' . $media['mimetype'] . '" title="' . $media['description'] . '"[/attach]';
 168         }
 169
 170         /**
 171          * Fetch additional data for the provided media array
 172          *
 173          * @param array $media
 174          * @return array media array with additional data
 175          */
 176         public static function fetchAdditionalData(array $media): array
 177         {
 178                 if (Network::isLocalLink($media['url'])) {
 179                         $media = self::fetchLocalData($media);
 180                 }
 181
 182                 // Fetch the mimetype or size if missing.
 183                 if (Network::isValidHttpUrl($media['url']) && (empty($media['mimetype']) || empty($media['size']))) {
 184                         $timeout = DI::config()->get('system', 'xrd_timeout');
 185                         $curlResult = DI::httpClient()->head($media['url'], [HttpClientOptions::TIMEOUT => $timeout]);
 186
 187                         // Workaround for systems that can't handle a HEAD request
 188                         if (!$curlResult->isSuccess() && ($curlResult->getReturnCode() == 405)) {
 189                                 $curlResult = DI::httpClient()->get($media['url'], HttpClientAccept::DEFAULT, [HttpClientOptions::TIMEOUT => $timeout]);
 190                         }
 191
 192                         if ($curlResult->isSuccess()) {
 193                                 if (empty($media['mimetype'])) {
 194                                         $media['mimetype'] = $curlResult->getHeader('Content-Type')[0] ?? '';
 195                                 }
 196                                 if (empty($media['size'])) {
 197                                         $media['size'] = (int)($curlResult->getHeader('Content-Length')[0] ?? 0);
 198                                 }
 199                         } else {
 200                                 Logger::notice('Could not fetch head', ['media' => $media]);
 201                         }
 202                 }
 203
 204                 $filetype = !empty($media['mimetype']) ? strtolower(current(explode('/', $media['mimetype']))) : '';
 205
 206                 if (($media['type'] == self::IMAGE) || ($filetype == 'image')) {
 207                         $imagedata = Images::getInfoFromURLCached($media['url']);
 208                         if ($imagedata) {
 209                                 $media['mimetype'] = $imagedata['mime'];
 210                                 $media['size'] = $imagedata['size'];
 211                                 $media['width'] = $imagedata[0];
 212                                 $media['height'] = $imagedata[1];
 213                                 $media['blurhash'] = $imagedata['blurhash'] ?? null;
 214                         } else {
 215                                 Logger::notice('No image data', ['media' => $media]);
 216                         }
 217                         if (!empty($media['preview'])) {
 218                                 $imagedata = Images::getInfoFromURLCached($media['preview']);
 219                                 if ($imagedata) {
 220                                         $media['preview-width'] = $imagedata[0];
 221                                         $media['preview-height'] = $imagedata[1];
 222                                 }
 223                         }
 224                 }
 225
 226                 if ($media['type'] != self::DOCUMENT) {
 227                         $media = self::addType($media);
 228                 }
 229
 230                 if (in_array($media['type'], [self::TEXT, self::APPLICATION, self::HTML, self::XML, self::PLAIN])) {
 231                         $media = self::addActivity($media);
 232                 }
 233
 234                 if (in_array($media['type'], [self::TEXT, self::APPLICATION, self::HTML, self::XML, self::PLAIN])) {
 235                         $media = self::addAccount($media);
 236                 }
 237
 238                 if ($media['type'] == self::HTML) {
 239                         $media = self::addPage($media);
 240                 }
 241
 242                 return $media;
 243         }
 244
 245         /**
 246          * Adds the activity type if the media entry is linked to an activity
 247          *
 248          * @param array $media
 249          * @return array
 250          */
 251         private static function addActivity(array $media): array
 252         {
 253                 $id = Item::fetchByLink($media['url']);
 254                 if (empty($id)) {
 255                         return $media;
 256                 }
 257
 258                 $item = Post::selectFirst([], ['id' => $id, 'network' => Protocol::FEDERATED]);
 259                 if (empty($item['id'])) {
 260                         Logger::debug('Not a federated activity', ['id' => $id, 'uri-id' => $media['uri-id'], 'url' => $media['url']]);
 261                         return $media;
 262                 }
 263
 264                 if (!empty($item['plink']) && Strings::compareLink($item['plink'], $media['url']) &&
 265                         parse_url($item['plink'], PHP_URL_HOST) != parse_url($item['uri'], PHP_URL_HOST)) {
 266                         Logger::debug('Not a link to an activity', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'plink' => $item['plink'], 'uri' => $item['uri']]);
 267                         return $media;
 268                 }
 269
 270                 if (in_array($item['network'], [Protocol::ACTIVITYPUB, Protocol::DFRN])) {
 271                         $media['mimetype'] = 'application/activity+json';
 272                 } elseif ($item['network'] == Protocol::DIASPORA) {
 273                         $media['mimetype'] = 'application/xml';
 274                 }
 275
 276                 $contact = Contact::getById($item['author-id'], ['avatar', 'gsid']);
 277                 if (!empty($contact['gsid'])) {
 278                         $gserver = DBA::selectFirst('gserver', ['url', 'site_name'], ['id' => $contact['gsid']]);
 279                 }
 280
 281                 $media['type'] = self::ACTIVITY;
 282                 $media['media-uri-id'] = $item['uri-id'];
 283                 $media['height'] = null;
 284                 $media['width'] = null;
 285                 $media['preview'] = null;
 286                 $media['preview-height'] = null;
 287                 $media['preview-width'] = null;
 288                 $media['blurhash'] = null;
 289                 $media['description'] = $item['body'];
 290                 $media['name'] = $item['title'];
 291                 $media['author-url'] = $item['author-link'];
 292                 $media['author-name'] = $item['author-name'];
 293                 $media['author-image'] = $contact['avatar'] ?? $item['author-avatar'];
 294                 $media['publisher-url'] = $gserver['url'] ?? null;
 295                 $media['publisher-name'] = $gserver['site_name'] ?? null;
 296                 $media['publisher-image'] = null;
 297
 298                 Logger::debug('Activity detected', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'plink' => $item['plink'], 'uri' => $item['uri']]);
 299                 return $media;
 300         }
 301
 302         /**
 303          * Adds the account type if the media entry is linked to an account
 304          *
 305          * @param array $media
 306          * @return array
 307          */
 308         private static function addAccount(array $media): array
 309         {
 310                 $contact = Contact::getByURL($media['url'], false);
 311                 if (empty($contact) || ($contact['network'] == Protocol::PHANTOM)) {
 312                         return $media;
 313                 }
 314
 315                 if (in_array($contact['network'], [Protocol::ACTIVITYPUB, Protocol::DFRN])) {
 316                         $media['mimetype'] = 'application/activity+json';
 317                 }
 318
 319                 if (!empty($contact['gsid'])) {
 320                         $gserver = DBA::selectFirst('gserver', ['url', 'site_name'], ['id' => $contact['gsid']]);
 321                 }
 322
 323                 $media['type'] = self::ACCOUNT;
 324                 $media['media-uri-id'] = $contact['uri-id'];
 325                 $media['height'] = null;
 326                 $media['width'] = null;
 327                 $media['preview'] = null;
 328                 $media['preview-height'] = null;
 329                 $media['preview-width'] = null;
 330                 $media['blurhash'] = null;
 331                 $media['description'] = $contact['about'];
 332                 $media['name'] = $contact['name'];
 333                 $media['author-url'] = $contact['url'];
 334                 $media['author-name'] = $contact['name'];
 335                 $media['author-image'] = $contact['avatar'];
 336                 $media['publisher-url'] = $gserver['url'] ?? null;
 337                 $media['publisher-name'] = $gserver['site_name'] ?? null;
 338                 $media['publisher-image'] = null;
 339
 340                 Logger::debug('Account detected', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'uri' => $contact['url']]);
 341                 return $media;
 342         }
 343
 344         /**
 345          * Add page infos for HTML entries
 346          *
 347          * @param array $media
 348          * @return array
 349          */
 350         private static function addPage(array $media): array
 351         {
 352                 $data = ParseUrl::getSiteinfoCached($media['url'], false);
 353                 $media['preview'] = $data['images'][0]['src'] ?? null;
 354                 $media['preview-height'] = $data['images'][0]['height'] ?? null;
 355                 $media['preview-width'] = $data['images'][0]['width'] ?? null;
 356                 $media['blurhash'] = $data['images'][0]['blurhash'] ?? null;
 357                 $media['description'] = $data['text'] ?? null;
 358                 $media['name'] = $data['title'] ?? null;
 359                 $media['author-url'] = $data['author_url'] ?? null;
 360                 $media['author-name'] = $data['author_name'] ?? null;
 361                 $media['author-image'] = $data['author_img'] ?? null;
 362                 $media['publisher-url'] = $data['publisher_url'] ?? null;
 363                 $media['publisher-name'] = $data['publisher_name'] ?? null;
 364                 $media['publisher-image'] = $data['publisher_img'] ?? null;
 365
 366                 return $media;
 367         }
 368
 369         /**
 370          * Fetch media data from local resources
 371          * @param array $media
 372          * @return array media with added data
 373          */
 374         private static function fetchLocalData(array $media): array
 375         {
 376                 if (!preg_match('|.*?/photo/(.*[a-fA-F0-9])\-(.*[0-9])\..*[\w]|', $media['url'] ?? '', $matches)) {
 377                         return $media;
 378                 }
 379                 $photo = Photo::selectFirst([], ['resource-id' => $matches[1], 'scale' => $matches[2]]);
 380                 if (!empty($photo)) {
 381                         $media['mimetype'] = $photo['type'];
 382                         $media['size'] = $photo['datasize'];
 383                         $media['width'] = $photo['width'];
 384                         $media['height'] = $photo['height'];
 385                         $media['blurhash'] = $photo['blurhash'];
 386                 }
 387
 388                 if (!preg_match('|.*?/photo/(.*[a-fA-F0-9])\-(.*[0-9])\..*[\w]|', $media['preview'] ?? '', $matches)) {
 389                         return $media;
 390                 }
 391                 $photo = Photo::selectFirst([], ['resource-id' => $matches[1], 'scale' => $matches[2]]);
 392                 if (!empty($photo)) {
 393                         $media['preview-width'] = $photo['width'];
 394                         $media['preview-height'] = $photo['height'];
 395                 }
 396
 397                 return $media;
 398         }
 399
 400         /**
 401          * Add the detected type to the media array
 402          *
 403          * @param array $data
 404          * @return array data array with the detected type
 405          */
 406         public static function addType(array $data): array
 407         {
 408                 if (empty($data['mimetype'])) {
 409                         Logger::info('No MimeType provided', ['media' => $data]);
 410                         return $data;
 411                 }
 412
 413                 $type = explode('/', current(explode(';', $data['mimetype'])));
 414                 if (count($type) < 2) {
 415                         Logger::info('Unknown MimeType', ['type' => $type, 'media' => $data]);
 416                         $data['type'] = self::UNKNOWN;
 417                         return $data;
 418                 }
 419
 420                 $filetype = strtolower($type[0]);
 421                 $subtype = strtolower($type[1]);
 422
 423                 if ($filetype == 'image') {
 424                         $data['type'] = self::IMAGE;
 425                 } elseif ($filetype == 'video') {
 426                         $data['type'] = self::VIDEO;
 427                 } elseif ($filetype == 'audio') {
 428                         $data['type'] = self::AUDIO;
 429                 } elseif (($filetype == 'text') && ($subtype == 'html')) {
 430                         $data['type'] = self::HTML;
 431                 } elseif (($filetype == 'text') && ($subtype == 'xml')) {
 432                         $data['type'] = self::XML;
 433                 } elseif (($filetype == 'text') && ($subtype == 'plain')) {
 434                         $data['type'] = self::PLAIN;
 435                 } elseif ($filetype == 'text') {
 436                         $data['type'] = self::TEXT;
 437                 } elseif (($filetype == 'application') && ($subtype == 'x-bittorrent')) {
 438                         $data['type'] = self::TORRENT;
 439                 } elseif ($filetype == 'application') {
 440                         $data['type'] = self::APPLICATION;
 441                 } else {
 442                         $data['type'] = self::UNKNOWN;
 443                         Logger::info('Unknown type', ['filetype' => $filetype, 'subtype' => $subtype, 'media' => $data]);
 444                         return $data;
 445                 }
 446
 447                 Logger::debug('Detected type', ['filetype' => $filetype, 'subtype' => $subtype, 'media' => $data]);
 448                 return $data;
 449         }
 450
 451         /**
 452          * Tests for path patterns that are usef for picture links in Friendica
 453          *
 454          * @param string $page    Link to the image page
 455          * @param string $preview Preview picture
 456          * @return boolean
 457          */
 458         private static function isPictureLink(string $page, string $preview): bool
 459         {
 460                 return preg_match('#/photos/.*/image/#ism', $page) && preg_match('#/photo/.*-1\.#ism', $preview);
 461         }
 462
 463         /**
 464          * Add media links and remove them from the body
 465          *
 466          * @param integer $uriid
 467          * @param string $body
 468          * @return string Body without media links
 469          */
 470         public static function insertFromBody(int $uriid, string $body, bool $endmatch = false): string
 471         {
 472                 $endmatchpattern = $endmatch ? '\z' : '';
 473                 // Simplify image codes
 474                 $unshared_body = $body = preg_replace("/\[img\=([0-9]*)x([0-9]*)\](.*?)\[\/img\]$endmatchpattern/ism", '[img]$3[/img]', $body);
 475
 476                 $attachments = [];
 477                 if (preg_match_all("#\[url=([^\]]+?)\]\s*\[img=([^\[\]]*)\]([^\[\]]*)\[\/img\]\s*\[/url\]$endmatchpattern#ism", $body, $pictures, PREG_SET_ORDER)) {
 478                         foreach ($pictures as $picture) {
 479                                 if (!self::isPictureLink($picture[1], $picture[2])) {
 480                                         continue;
 481                                 }
 482                                 $body = str_replace($picture[0], '', $body);
 483                                 $image = str_replace('-1.', '-0.', $picture[2]);
 484                                 $attachments[$image] = ['uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $image,
 485                                         'preview' => $picture[2], 'description' => $picture[3]];
 486                         }
 487                 }
 488
 489                 if (preg_match_all("/\[img=([^\[\]]*)\]([^\[\]]*)\[\/img\]$endmatchpattern/Usi", $body, $pictures, PREG_SET_ORDER)) {
 490                         foreach ($pictures as $picture) {
 491                                 $body = str_replace($picture[0], '', $body);
 492                                 $attachments[$picture[1]] = ['uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $picture[1], 'description' => $picture[2]];
 493                         }
 494                 }
 495
 496                 if (preg_match_all("#\[url=([^\]]+?)\]\s*\[img\]([^\[]+?)\[/img\]\s*\[/url\]$endmatchpattern#ism", $body, $pictures, PREG_SET_ORDER)) {
 497                         foreach ($pictures as $picture) {
 498                                 if (!self::isPictureLink($picture[1], $picture[2])) {
 499                                         continue;
 500                                 }
 501                                 $body = str_replace($picture[0], '', $body);
 502                                 $image = str_replace('-1.', '-0.', $picture[2]);
 503                                 $attachments[$image] = ['uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $image,
 504                                         'preview' => $picture[2], 'description' => null];
 505                         }
 506                 }
 507
 508                 if (preg_match_all("/\[img\]([^\[\]]*)\[\/img\]$endmatchpattern/ism", $body, $pictures, PREG_SET_ORDER)) {
 509                         foreach ($pictures as $picture) {
 510                                 $body = str_replace($picture[0], '', $body);
 511                                 $attachments[$picture[1]] = ['uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $picture[1]];
 512                         }
 513                 }
 514
 515                 if (preg_match_all("/\[audio\]([^\[\]]*)\[\/audio\]$endmatchpattern/ism", $body, $audios, PREG_SET_ORDER)) {
 516                         foreach ($audios as $audio) {
 517                                 $body = str_replace($audio[0], '', $body);
 518                                 $attachments[$audio[1]] = ['uri-id' => $uriid, 'type' => self::AUDIO, 'url' => $audio[1]];
 519                         }
 520                 }
 521
 522                 if (preg_match_all("/\[video\]([^\[\]]*)\[\/video\]$endmatchpattern/ism", $body, $videos, PREG_SET_ORDER)) {
 523                         foreach ($videos as $video) {
 524                                 $body = str_replace($video[0], '', $body);
 525                                 $attachments[$video[1]] = ['uri-id' => $uriid, 'type' => self::VIDEO, 'url' => $video[1]];
 526                         }
 527                 }
 528
 529                 if ($uriid != 0) {
 530                         foreach ($attachments as $attachment) {
 531                                 if (Post\Link::exists($uriid, $attachment['preview'] ?? $attachment['url'])) {
 532                                         continue;
 533                                 }
 534
 535                                 // Only store attachments that are part of the unshared body
 536                                 if (Item::containsLink($unshared_body, $attachment['preview'] ?? $attachment['url'], $attachment['type'])) {
 537                                         self::insert($attachment);
 538                                 }
 539                         }
 540                 }
 541
 542                 return trim($body);
 543         }
 544
 545         /**
 546          * Remove media that is at the end of the body
 547          *
 548          * @param string $body
 549          * @return string
 550          */
 551         public static function removeFromEndOfBody(string $body): string
 552         {
 553                 do {
 554                         $prebody = $body;
 555                         $body = self::insertFromBody(0, $body, true);
 556                 } while ($prebody != $body);
 557                 return $body;
 558         }
 559
 560         /**
 561          * Add media links from a relevant url in the body
 562          *
 563          * @param integer $uriid
 564          * @param string $body
 565          * @return void
 566          */
 567         public static function insertFromRelevantUrl(int $uriid, string $body, string $fullbody, string $network)
 568         {
 569                 // Remove all hashtags and mentions
 570                 $body = preg_replace("/([#@!])\[url\=(.*?)\](.*?)\[\/url\]/ism", '', $body);
 571
 572                 // Search for pure links
 573                 if (preg_match_all("/\[url\](https?:.*?)\[\/url\]/ism", $body, $matches)) {
 574                         foreach ($matches[1] as $url) {
 575                                 Logger::info('Got page url (link without description)', ['uri-id' => $uriid, 'url' => $url]);
 576                                 self::insert(['uri-id' => $uriid, 'type' => self::UNKNOWN, 'url' => $url], false, $network);
 577                                 if ($network == Protocol::DFRN) {
 578                                         self::revertHTMLType($uriid, $url, $fullbody);
 579                                 }
 580                         }
 581                 }
 582
 583                 // Search for links with descriptions
 584                 if (preg_match_all("/\[url\=(https?:.*?)\].*?\[\/url\]/ism", $body, $matches)) {
 585                         foreach ($matches[1] as $url) {
 586                                 Logger::info('Got page url (link with description)', ['uri-id' => $uriid, 'url' => $url]);
 587                                 self::insert(['uri-id' => $uriid, 'type' => self::UNKNOWN, 'url' => $url], false, $network);
 588                                 if ($network == Protocol::DFRN) {
 589                                         self::revertHTMLType($uriid, $url, $fullbody);
 590                                 }
 591                         }
 592                 }
 593         }
 594
 595         /**
 596          * Revert the media type of links to UNKNOWN for DFRN posts when they aren't attached
 597          *
 598          * @param integer $uriid
 599          * @param string $url
 600          * @param string $body
 601          * @return void
 602          */
 603         private static function revertHTMLType(int $uriid, string $url, string $body)
 604         {
 605                 $attachment = BBCode::getAttachmentData($body);
 606                 if (!empty($attachment['url']) && Network::getUrlMatch($attachment['url'], $url)) {
 607                         return;
 608                 }
 609                 DBA::update('post-media', ['type' => self::UNKNOWN], ['uri-id' => $uriid, 'type' => self::HTML, 'url' => $url]);
 610         }
 611
 612         /**
 613          * Add media links from the attachment field
 614          *
 615          * @param integer $uriid
 616          * @param string $body
 617          * @return void
 618          */
 619         public static function insertFromAttachmentData(int $uriid, string $body)
 620         {
 621                 $data = BBCode::getAttachmentData($body);
 622                 if (empty($data))  {
 623                         return;
 624                 }
 625
 626                 Logger::info('Adding attachment data', ['data' => $data]);
 627                 $attachment = [
 628                         'uri-id' => $uriid,
 629                         'type' => self::HTML,
 630                         'url' => $data['url'],
 631                         'preview' => $data['preview'] ?? null,
 632                         'description' => $data['description'] ?? null,
 633                         'name' => $data['title'] ?? null,
 634                         'author-url' => $data['author_url'] ?? null,
 635                         'author-name' => $data['author_name'] ?? null,
 636                         'publisher-url' => $data['provider_url'] ?? null,
 637                         'publisher-name' => $data['provider_name'] ?? null,
 638                 ];
 639                 if (!empty($data['image'])) {
 640                         $attachment['preview'] = $data['image'];
 641                 }
 642                 self::insert($attachment);
 643         }
 644
 645         /**
 646          * Add media links from the attach field
 647          *
 648          * @param integer $uriid
 649          * @param string $attach
 650          * @return void
 651          */
 652         public static function insertFromAttachment(int $uriid, string $attach)
 653         {
 654                 if (!preg_match_all('|\[attach\]href=\"(.*?)\" length=\"(.*?)\" type=\"(.*?)\"(?: title=\"(.*?)\")?|', $attach, $matches, PREG_SET_ORDER)) {
 655                         return;
 656                 }
 657
 658                 foreach ($matches as $attachment) {
 659                         $media['type'] = self::DOCUMENT;
 660                         $media['uri-id'] = $uriid;
 661                         $media['url'] = $attachment[1];
 662                         $media['size'] = $attachment[2];
 663                         $media['mimetype'] = $attachment[3];
 664                         $media['description'] = $attachment[4] ?? '';
 665
 666                         self::insert($media);
 667                 }
 668         }
 669
 670         /**
 671          * Retrieves the media attachments associated with the provided item ID.
 672          *
 673          * @param int $uri_id URI id
 674          * @param array $types Media types
 675          * @return array|bool Array on success, false on error
 676          * @throws \Exception
 677          */
 678         public static function getByURIId(int $uri_id, array $types = [])
 679         {
 680                 $condition = ["`uri-id` = ? AND `type` != ?", $uri_id, self::UNKNOWN];
 681
 682                 if (!empty($types)) {
 683                         $condition = DBA::mergeConditions($condition, ['type' => $types]);
 684                 }
 685
 686                 return DBA::selectToArray('post-media', [], $condition, ['order' => ['id']]);
 687         }
 688
 689         /**
 690          * Checks if media attachments are associated with the provided item ID.
 691          *
 692          * @param int $uri_id URI id
 693          * @param array $types Media types
 694          * @return bool Whether media attachment exists
 695          * @throws \Exception
 696          */
 697         public static function existsByURIId(int $uri_id, array $types = []): bool
 698         {
 699                 $condition = ["`uri-id` = ? AND `type` != ?", $uri_id, self::UNKNOWN];
 700
 701                 if (!empty($types)) {
 702                         $condition = DBA::mergeConditions($condition, ['type' => $types]);
 703                 }
 704
 705                 return DBA::exists('post-media', $condition);
 706         }
 707
 708         /**
 709          * Split the attachment media in the three segments "visual", "link" and "additional"
 710          *
 711          * @param int    $uri_id URI id
 712          * @param array  $links list of links that shouldn't be added
 713          * @param bool   $has_media
 714          * @return array attachments
 715          */
 716         public static function splitAttachments(int $uri_id, array $links = [], bool $has_media = true): array
 717         {
 718                 $attachments = ['visual' => [], 'link' => [], 'additional' => []];
 719
 720                 if (!$has_media) {
 721                         return $attachments;
 722                 }
 723
 724                 $media = self::getByURIId($uri_id);
 725                 if (empty($media)) {
 726                         return $attachments;
 727                 }
 728
 729                 $heights = [];
 730                 $selected = '';
 731                 $previews = [];
 732
 733                 foreach ($media as $medium) {
 734                         foreach ($links as $link) {
 735                                 if (Strings::compareLink($link, $medium['url'])) {
 736                                         continue 2;
 737                                 }
 738                         }
 739
 740                         // Avoid adding separate media entries for previews
 741                         foreach ($previews as $preview) {
 742                                 if (Strings::compareLink($preview, $medium['url'])) {
 743                                         continue 2;
 744                                 }
 745                         }
 746
 747                         // Currently these two types are ignored here.
 748                         // Posts are added differently and contacts are not displayed as attachments.
 749                         if (in_array($medium['type'], [self::ACCOUNT, self::ACTIVITY])) {
 750                                 continue;
 751                         }
 752
 753                         if (!empty($medium['preview'])) {
 754                                 $previews[] = $medium['preview'];
 755                         }
 756
 757                         $type = explode('/', explode(';', $medium['mimetype'] ?? '')[0]);
 758                         if (count($type) < 2) {
 759                                 Logger::info('Unknown MimeType', ['type' => $type, 'media' => $medium]);
 760                                 $filetype = 'unkn';
 761                                 $subtype = 'unkn';
 762                         } else {
 763                                 $filetype = strtolower($type[0]);
 764                                 $subtype = strtolower($type[1]);
 765                         }
 766
 767                         $medium['filetype'] = $filetype;
 768                         $medium['subtype'] = $subtype;
 769
 770                         if ($medium['type'] == self::HTML || (($filetype == 'text') && ($subtype == 'html'))) {
 771                                 $attachments['link'][] = $medium;
 772                                 continue;
 773                         }
 774
 775                         if (in_array($medium['type'], [self::AUDIO, self::IMAGE]) ||
 776                                 in_array($filetype, ['audio', 'image'])) {
 777                                 $attachments['visual'][] = $medium;
 778                         } elseif (($medium['type'] == self::VIDEO) || ($filetype == 'video')) {
 779                                 if (!empty($medium['height'])) {
 780                                         // Peertube videos are delivered in many different resolutions. We pick a moderate one.
 781                                         // Since only Peertube provides a "height" parameter, this wouldn't be executed
 782                                         // when someone for example on Mastodon was sharing multiple videos in a single post.
 783                                         $heights[$medium['height']] = $medium['url'];
 784                                         $video[$medium['url']] = $medium;
 785                                 } else {
 786                                         $attachments['visual'][] = $medium;
 787                                 }
 788                         } else {
 789                                 $attachments['additional'][] = $medium;
 790                         }
 791                 }
 792
 793                 if (!empty($heights)) {
 794                         ksort($heights);
 795                         foreach ($heights as $height => $url) {
 796                                 if (empty($selected) || $height <= 480) {
 797                                         $selected = $url;
 798                                 }
 799                         }
 800
 801                         if (!empty($selected)) {
 802                                 $attachments['visual'][] = $video[$selected];
 803                                 unset($video[$selected]);
 804                                 foreach ($video as $element) {
 805                                         $attachments['additional'][] = $element;
 806                                 }
 807                         }
 808                 }
 809
 810                 return $attachments;
 811         }
 812
 813         /**
 814          * Add media attachments to the body
 815          *
 816          * @param int    $uriid
 817          * @param string $body
 818          * @param array  $types
 819          *
 820          * @return string body
 821          */
 822         public static function addAttachmentsToBody(int $uriid, string $body = '', array $types = [self::IMAGE, self::AUDIO, self::VIDEO]): string
 823         {
 824                 if (empty($body)) {
 825                         $item = Post::selectFirst(['body'], ['uri-id' => $uriid]);
 826                         if (!DBA::isResult($item)) {
 827                                 return '';
 828                         }
 829                         $body = $item['body'];
 830                 }
 831                 $original_body = $body;
 832
 833                 $body = preg_replace("/\s*\[attachment .*?\].*?\[\/attachment\]\s*/ism", '', $body);
 834
 835                 foreach (self::getByURIId($uriid, $types) as $media) {
 836                         if (Item::containsLink($body, $media['preview'] ?? $media['url'], $media['type'])) {
 837                                 continue;
 838                         }
 839
 840                         if ($media['type'] == self::IMAGE) {
 841                                 if (!empty($media['preview'])) {
 842                                         if (!empty($media['description'])) {
 843                                                 $body .= "\n[url=" . $media['url'] . "][img=" . $media['preview'] . ']' . $media['description'] .'[/img][/url]';
 844                                         } else {
 845                                                 $body .= "\n[url=" . $media['url'] . "][img]" . $media['preview'] .'[/img][/url]';
 846                                         }
 847                                 } else {
 848                                         if (!empty($media['description'])) {
 849                                                 $body .= "\n[img=" . $media['url'] . ']' . $media['description'] .'[/img]';
 850                                         } else {
 851                                                 $body .= "\n[img]" . $media['url'] .'[/img]';
 852                                         }
 853                                 }
 854                         } elseif ($media['type'] == self::AUDIO) {
 855                                 $body .= "\n[audio]" . $media['url'] . "[/audio]\n";
 856                         } elseif ($media['type'] == self::VIDEO) {
 857                                 $body .= "\n[video]" . $media['url'] . "[/video]\n";
 858                         }
 859                 }
 860
 861                 if (preg_match("/.*(\[attachment.*?\].*?\[\/attachment\]).*/ism", $original_body, $match)) {
 862                         $body .= "\n" . $match[1];
 863                 }
 864
 865                 return $body;
 866         }
 867
 868         /**
 869          * Get preview link for given media id
 870          *
 871          * @param integer $id   media id
 872          * @param string  $size One of the Proxy::SIZE_* constants
 873          * @return string preview link
 874          */
 875         public static function getPreviewUrlForId(int $id, string $size = ''): string
 876         {
 877                 $url = DI::baseUrl() . '/photo/preview/';
 878                 switch ($size) {
 879                         case Proxy::SIZE_MICRO:
 880                                 $url .= Proxy::PIXEL_MICRO . '/';
 881                                 break;
 882                         case Proxy::SIZE_THUMB:
 883                                 $url .= Proxy::PIXEL_THUMB . '/';
 884                                 break;
 885                         case Proxy::SIZE_SMALL:
 886                                 $url .= Proxy::PIXEL_SMALL . '/';
 887                                 break;
 888                         case Proxy::SIZE_MEDIUM:
 889                                 $url .= Proxy::PIXEL_MEDIUM . '/';
 890                                 break;
 891                         case Proxy::SIZE_LARGE:
 892                                 $url .= Proxy::PIXEL_LARGE . '/';
 893                                 break;
 894                 }
 895                 return $url . $id;
 896         }
 897
 898         /**
 899          * Get media link for given media id
 900          *
 901          * @param integer $id   media id
 902          * @param string  $size One of the Proxy::SIZE_* constants
 903          * @return string media link
 904          */
 905         public static function getUrlForId(int $id, string $size = ''): string
 906         {
 907                 $url = DI::baseUrl() . '/photo/media/';
 908                 switch ($size) {
 909                         case Proxy::SIZE_MICRO:
 910                                 $url .= Proxy::PIXEL_MICRO . '/';
 911                                 break;
 912                         case Proxy::SIZE_THUMB:
 913                                 $url .= Proxy::PIXEL_THUMB . '/';
 914                                 break;
 915                         case Proxy::SIZE_SMALL:
 916                                 $url .= Proxy::PIXEL_SMALL . '/';
 917                                 break;
 918                         case Proxy::SIZE_MEDIUM:
 919                                 $url .= Proxy::PIXEL_MEDIUM . '/';
 920                                 break;
 921                         case Proxy::SIZE_LARGE:
 922                                 $url .= Proxy::PIXEL_LARGE . '/';
 923                                 break;
 924                 }
 925                 return $url . $id;
 926         }
 927 }