src/Model/Post/Media.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2021, 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\System;
  28 use Friendica\Database\Database;
  29 use Friendica\Database\DBA;
  30 use Friendica\DI;
  31 use Friendica\Util\Images;
  32 use Friendica\Util\ParseUrl;
  33 use Friendica\Util\Strings;
  34
  35 /**
  36  * Class Media
  37  *
  38  * This Model class handles media interactions.
  39  * This tables stores medias (images, videos, audio files) related to posts.
  40  */
  41 class Media
  42 {
  43         const UNKNOWN     = 0;
  44         const IMAGE       = 1;
  45         const VIDEO       = 2;
  46         const AUDIO       = 3;
  47         const TEXT        = 4;
  48         const APPLICATION = 5;
  49         const TORRENT     = 16;
  50         const HTML        = 17;
  51         const XML         = 18;
  52         const PLAIN       = 19;
  53         const DOCUMENT    = 128;
  54
  55         /**
  56          * Insert a post-media record
  57          *
  58          * @param array $media
  59          * @return void
  60          */
  61         public static function insert(array $media, bool $force = false)
  62         {
  63                 if (empty($media['url']) || empty($media['uri-id']) || !isset($media['type'])) {
  64                         Logger::warning('Incomplete media data', ['media' => $media]);
  65                         return;
  66                 }
  67
  68                 // "document" has got the lowest priority. So when the same file is both attached as document
  69                 // and embedded as picture then we only store the picture or replace the document
  70                 $found = DBA::selectFirst('post-media', ['type'], ['uri-id' => $media['uri-id'], 'url' => $media['url']]);
  71                 if (!$force && !empty($found) && (($found['type'] != self::DOCUMENT) || ($media['type'] == self::DOCUMENT))) {
  72                         Logger::info('Media already exists', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'callstack' => System::callstack()]);
  73                         return;
  74                 }
  75
  76                 $media = self::unsetEmptyFields($media);
  77
  78                 // We are storing as fast as possible to avoid duplicated network requests
  79                 // when fetching additional information for pictures and other content.
  80                 $result = DBA::insert('post-media', $media, Database::INSERT_UPDATE);
  81                 Logger::info('Stored media', ['result' => $result, 'media' => $media, 'callstack' => System::callstack()]);
  82                 $stored = $media;
  83
  84                 $media = self::fetchAdditionalData($media);
  85                 $media = self::unsetEmptyFields($media);
  86
  87                 if (array_diff_assoc($media, $stored)) {
  88                         $result = DBA::insert('post-media', $media, Database::INSERT_UPDATE);
  89                         Logger::info('Updated media', ['result' => $result, 'media' => $media]);
  90                 } else {
  91                         Logger::info('Nothing to update', ['media' => $media]);
  92                 }
  93         }
  94
  95         /**
  96          * Remove empty media fields
  97          *
  98          * @param array $media
  99          * @return array cleaned media array
 100          */
 101         private static function unsetEmptyFields(array $media)
 102         {
 103                 $fields = ['mimetype', 'height', 'width', 'size', 'preview', 'preview-height', 'preview-width', 'description'];
 104                 foreach ($fields as $field) {
 105                         if (empty($media[$field])) {
 106                                 unset($media[$field]);
 107                         }
 108                 }
 109                 return $media;
 110         }
 111
 112         /**
 113          * Copy attachments from one uri-id to another
 114          *
 115          * @param integer $from_uri_id
 116          * @param integer $to_uri_id
 117          * @return void
 118          */
 119         public static function copy(int $from_uri_id, int $to_uri_id)
 120         {
 121                 $attachments = self::getByURIId($from_uri_id);
 122                 foreach ($attachments as $attachment) {
 123                         $attachment['uri-id'] = $to_uri_id;
 124                         self::insert($attachment);
 125                 }
 126         }
 127
 128         /**
 129          * Creates the "[attach]" element from the given attributes
 130          *
 131          * @param string $href
 132          * @param integer $length
 133          * @param string $type
 134          * @param string $title
 135          * @return string "[attach]" element
 136          */
 137         public static function getAttachElement(string $href, int $length, string $type, string $title = '')
 138         {
 139                 $media = self::fetchAdditionalData(['type' => self::DOCUMENT, 'url' => $href,
 140                         'size' => $length, 'mimetype' => $type, 'description' => $title]);
 141
 142                 return '[attach]href="' . $media['url'] . '" length="' . $media['size'] .
 143                         '" type="' . $media['mimetype'] . '" title="' . $media['description'] . '"[/attach]';
 144         }
 145
 146         /**
 147          * Fetch additional data for the provided media array
 148          *
 149          * @param array $media
 150          * @return array media array with additional data
 151          */
 152         public static function fetchAdditionalData(array $media)
 153         {
 154                 // Fetch the mimetype or size if missing.
 155                 if (empty($media['mimetype']) || empty($media['size'])) {
 156                         $timeout = DI::config()->get('system', 'xrd_timeout');
 157                         $curlResult = DI::httpRequest()->head($media['url'], ['timeout' => $timeout]);
 158                         if ($curlResult->isSuccess()) {
 159                                 if (empty($media['mimetype'])) {
 160                                         $media['mimetype'] = $curlResult->getHeader('Content-Type');
 161                                 }
 162                                 if (empty($media['size'])) {
 163                                         $media['size'] = (int)$curlResult->getHeader('Content-Length');
 164                                 }
 165                         } else {
 166                                 Logger::notice('Could not fetch head', ['media' => $media]);
 167                         }
 168                 }
 169
 170                 $filetype = !empty($media['mimetype']) ? strtolower(current(explode('/', $media['mimetype']))) : '';
 171
 172                 if (($media['type'] == self::IMAGE) || ($filetype == 'image')) {
 173                         $imagedata = Images::getInfoFromURLCached($media['url']);
 174                         if (!empty($imagedata)) {
 175                                 $media['mimetype'] = $imagedata['mime'];
 176                                 $media['size'] = $imagedata['size'];
 177                                 $media['width'] = $imagedata[0];
 178                                 $media['height'] = $imagedata[1];
 179                         } else {
 180                                 Logger::notice('No image data', ['media' => $media]);
 181                         }
 182                         if (!empty($media['preview'])) {
 183                                 $imagedata = Images::getInfoFromURLCached($media['preview']);
 184                                 if (!empty($imagedata)) {
 185                                         $media['preview-width'] = $imagedata[0];
 186                                         $media['preview-height'] = $imagedata[1];
 187                                 }
 188                         }
 189                 }
 190
 191                 if ($media['type'] != self::DOCUMENT) {
 192                         $media = self::addType($media);
 193                 }
 194
 195                 if ($media['type'] == self::HTML) {
 196                         $data = ParseUrl::getSiteinfoCached($media['url'], false);
 197                         $media['preview'] = $data['images'][0]['src'] ?? null;
 198                         $media['preview-height'] = $data['images'][0]['height'] ?? null;
 199                         $media['preview-width'] = $data['images'][0]['width'] ?? null;
 200                         $media['description'] = $data['text'] ?? null;
 201                         $media['name'] = $data['title'] ?? null;
 202                         $media['author-url'] = $data['author_url'] ?? null;
 203                         $media['author-name'] = $data['author_name'] ?? null;
 204                         $media['author-image'] = $data['author_img'] ?? null;
 205                         $media['publisher-url'] = $data['publisher_url'] ?? null;
 206                         $media['publisher-name'] = $data['publisher_name'] ?? null;
 207                         $media['publisher-image'] = $data['publisher_img'] ?? null;
 208                 }
 209                 return $media;
 210         }
 211
 212         /**
 213          * Add the detected type to the media array
 214          *
 215          * @param array $data
 216          * @return array data array with the detected type
 217          */
 218         public static function addType(array $data)
 219         {
 220                 if (empty($data['mimetype'])) {
 221                         Logger::info('No MimeType provided', ['media' => $data]);
 222                         return $data;
 223                 }
 224
 225                 $type = explode('/', current(explode(';', $data['mimetype'])));
 226                 if (count($type) < 2) {
 227                         Logger::info('Unknown MimeType', ['type' => $type, 'media' => $data]);
 228                         $data['type'] = self::UNKNOWN;
 229                         return $data;
 230                 }
 231
 232                 $filetype = strtolower($type[0]);
 233                 $subtype = strtolower($type[1]);
 234
 235                 if ($filetype == 'image') {
 236                         $data['type'] = self::IMAGE;
 237                 } elseif ($filetype == 'video') {
 238                         $data['type'] = self::VIDEO;
 239                 } elseif ($filetype == 'audio') {
 240                         $data['type'] = self::AUDIO;
 241                 } elseif (($filetype == 'text') && ($subtype == 'html')) {
 242                         $data['type'] = self::HTML;
 243                 } elseif (($filetype == 'text') && ($subtype == 'xml')) {
 244                         $data['type'] = self::XML;
 245                 } elseif (($filetype == 'text') && ($subtype == 'plain')) {
 246                         $data['type'] = self::PLAIN;
 247                 } elseif ($filetype == 'text') {
 248                         $data['type'] = self::TEXT;
 249                 } elseif (($filetype == 'application') && ($subtype == 'x-bittorrent')) {
 250                         $data['type'] = self::TORRENT;
 251                 } elseif ($filetype == 'application') {
 252                         $data['type'] = self::APPLICATION;
 253                 } else {
 254                         $data['type'] = self::UNKNOWN;
 255                         Logger::info('Unknown type', ['filetype' => $filetype, 'subtype' => $subtype, 'media' => $data]);
 256                         return $data;
 257                 }
 258
 259                 Logger::debug('Detected type', ['filetype' => $filetype, 'subtype' => $subtype, 'media' => $data]);
 260                 return $data;
 261         }
 262
 263         /**
 264          * Tests for path patterns that are usef for picture links in Friendica
 265          *
 266          * @param string $page    Link to the image page
 267          * @param string $preview Preview picture
 268          * @return boolean
 269          */
 270         private static function isPictureLink(string $page, string $preview)
 271         {
 272                 return preg_match('#/photos/.*/image/#ism', $page) && preg_match('#/photo/.*-1\.#ism', $preview);
 273         }
 274
 275         /**
 276          * Add media links and remove them from the body
 277          *
 278          * @param integer $uriid
 279          * @param string $body
 280          * @return string Body without media links
 281          */
 282         public static function insertFromBody(int $uriid, string $body)
 283         {
 284                 // Simplify image codes
 285                 $body = preg_replace("/\[img\=([0-9]*)x([0-9]*)\](.*?)\[\/img\]/ism", '[img]$3[/img]', $body);
 286
 287                 $attachments = [];
 288                 if (preg_match_all("#\[url=([^\]]+?)\]\s*\[img=([^\[\]]*)\]([^\[\]]*)\[\/img\]\s*\[/url\]#ism", $body, $pictures, PREG_SET_ORDER)) {
 289                         foreach ($pictures as $picture) {
 290                                 if (!self::isPictureLink($picture[1], $picture[2])) {
 291                                         continue;
 292                                 }
 293                                 $body = str_replace($picture[0], '', $body);
 294                                 $image = str_replace('-1.', '-0.', $picture[2]);
 295                                 $attachments[$image] = ['uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $image,
 296                                         'preview' => $picture[2], 'description' => $picture[3]];
 297                         }
 298                 }
 299
 300                 if (preg_match_all("/\[img=([^\[\]]*)\]([^\[\]]*)\[\/img\]/Usi", $body, $pictures, PREG_SET_ORDER)) {
 301                         foreach ($pictures as $picture) {
 302                                 $body = str_replace($picture[0], '', $body);
 303                                 $attachments[$picture[1]] = ['uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $picture[1], 'description' => $picture[2]];
 304                         }
 305                 }
 306
 307                 if (preg_match_all("#\[url=([^\]]+?)\]\s*\[img\]([^\[]+?)\[/img\]\s*\[/url\]#ism", $body, $pictures, PREG_SET_ORDER)) {
 308                         foreach ($pictures as $picture) {
 309                                 if (!self::isPictureLink($picture[1], $picture[2])) {
 310                                         continue;
 311                                 }
 312                                 $body = str_replace($picture[0], '', $body);
 313                                 $image = str_replace('-1.', '-0.', $picture[2]);
 314                                 $attachments[$image] = ['uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $image,
 315                                         'preview' => $picture[2], 'description' => null];
 316                         }
 317                 }
 318
 319                 if (preg_match_all("/\[img\]([^\[\]]*)\[\/img\]/ism", $body, $pictures, PREG_SET_ORDER)) {
 320                         foreach ($pictures as $picture) {
 321                                 $body = str_replace($picture[0], '', $body);
 322                                 $attachments[$picture[1]] = ['uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $picture[1]];
 323                         }
 324                 }
 325
 326                 if (preg_match_all("/\[audio\]([^\[\]]*)\[\/audio\]/ism", $body, $audios, PREG_SET_ORDER)) {
 327                         foreach ($audios as $audio) {
 328                                 $body = str_replace($audio[0], '', $body);
 329                                 $attachments[$audio[1]] = ['uri-id' => $uriid, 'type' => self::AUDIO, 'url' => $audio[1]];
 330                         }
 331                 }
 332
 333                 if (preg_match_all("/\[video\]([^\[\]]*)\[\/video\]/ism", $body, $videos, PREG_SET_ORDER)) {
 334                         foreach ($videos as $video) {
 335                                 $body = str_replace($video[0], '', $body);
 336                                 $attachments[$video[1]] = ['uri-id' => $uriid, 'type' => self::VIDEO, 'url' => $video[1]];
 337                         }
 338                 }
 339
 340                 $url = PageInfo::getRelevantUrlFromBody($body);
 341                 if (!empty($url)) {
 342                         Logger::debug('Got page url', ['url' => $url]);
 343                         $attachments[$url] = ['uri-id' => $uriid, 'type' => self::UNKNOWN, 'url' => $url];
 344                 }
 345
 346                 foreach ($attachments as $attachment) {
 347                         self::insert($attachment);
 348                 }
 349
 350                 return trim($body);
 351         }
 352
 353         /**
 354          * Add media links from the attachment field
 355          *
 356          * @param integer $uriid
 357          * @param string $body
 358          */
 359         public static function insertFromAttachmentData(int $uriid, string $body)
 360         {
 361                 $data = BBCode::getAttachmentData($body);
 362                 if (empty($data))  {
 363                         return;
 364                 }
 365
 366                 Logger::info('Adding attachment data', ['data' => $data]);
 367                 $attachment = [
 368                         'uri-id' => $uriid,
 369                         'type' => self::HTML,
 370                         'url' => $data['url'],
 371                         'preview' => $data['preview'] ?? null,
 372                         'description' => $data['description'] ?? null,
 373                         'name' => $data['title'] ?? null,
 374                         'author-url' => $data['author_url'] ?? null,
 375                         'author-name' => $data['author_name'] ?? null,
 376                         'publisher-url' => $data['provider_url'] ?? null,
 377                         'publisher-name' => $data['provider_name'] ?? null,
 378                 ];
 379                 if (!empty($data['image'])) {
 380                         $attachment['preview'] = $data['image'];
 381                 }
 382                 self::insert($attachment);
 383         }
 384
 385         /**
 386          * Add media links from the attach field
 387          *
 388          * @param integer $uriid
 389          * @param string $attach
 390          * @return void
 391          */
 392         public static function insertFromAttachment(int $uriid, string $attach)
 393         {
 394                 if (!preg_match_all('|\[attach\]href=\"(.*?)\" length=\"(.*?)\" type=\"(.*?)\"(?: title=\"(.*?)\")?|', $attach, $matches, PREG_SET_ORDER)) {
 395                         return;
 396                 }
 397
 398                 foreach ($matches as $attachment) {
 399                         $media['type'] = self::DOCUMENT;
 400                         $media['uri-id'] = $uriid;
 401                         $media['url'] = $attachment[1];
 402                         $media['size'] = $attachment[2];
 403                         $media['mimetype'] = $attachment[3];
 404                         $media['description'] = $attachment[4] ?? '';
 405
 406                         self::insert($media);
 407                 }
 408         }
 409
 410         /**
 411          * Retrieves the media attachments associated with the provided item ID.
 412          *
 413          * @param int $uri_id
 414          * @param array $types
 415          * @return array
 416          * @throws \Exception
 417          */
 418         public static function getByURIId(int $uri_id, array $types = [])
 419         {
 420                 $condition = ['uri-id' => $uri_id];
 421
 422                 if (!empty($types)) {
 423                         $condition = DBA::mergeConditions($condition, ['type' => $types]);
 424                 }
 425
 426                 return DBA::selectToArray('post-media', [], $condition);
 427         }
 428
 429         /**
 430          * Checks if media attachments are associated with the provided item ID.
 431          *
 432          * @param int $uri_id
 433          * @param array $types
 434          * @return array
 435          * @throws \Exception
 436          */
 437         public static function existsByURIId(int $uri_id, array $types = [])
 438         {
 439                 $condition = ['uri-id' => $uri_id];
 440
 441                 if (!empty($types)) {
 442                         $condition = DBA::mergeConditions($condition, ['type' => $types]);
 443                 }
 444
 445                 return DBA::exists('post-media', $condition);
 446         }
 447
 448         /**
 449          * Split the attachment media in the three segments "visual", "link" and "additional"
 450          *
 451          * @param int    $uri_id
 452          * @param string $guid
 453          * @param array  $links ist of links that shouldn't be added
 454          * @return array attachments
 455          */
 456         public static function splitAttachments(int $uri_id, string $guid = '', array $links = [])
 457         {
 458                 $attachments = ['visual' => [], 'link' => [], 'additional' => []];
 459
 460                 $media = self::getByURIId($uri_id);
 461                 if (empty($media)) {
 462                         return $attachments;
 463                 }
 464
 465                 $height = 0;
 466                 $selected = '';
 467
 468                 foreach ($media as $medium) {
 469                         foreach ($links as $link) {
 470                                 if (Strings::compareLink($link, $medium['url'])) {
 471                                         continue 2;
 472                                 }
 473                         }
 474
 475                         $type = explode('/', current(explode(';', $medium['mimetype'])));
 476                         if (count($type) < 2) {
 477                                 Logger::info('Unknown MimeType', ['type' => $type, 'media' => $medium]);
 478                                 $filetype = 'unkn';
 479                                 $subtype = 'unkn';
 480                         } else {
 481                                 $filetype = strtolower($type[0]);
 482                                 $subtype = strtolower($type[1]);
 483                         }
 484
 485                         $medium['filetype'] = $filetype;
 486                         $medium['subtype'] = $subtype;
 487
 488                         if ($medium['type'] == self::HTML || (($filetype == 'text') && ($subtype == 'html'))) {
 489                                 $attachments['link'][] = $medium;
 490                                 continue;
 491                         }
 492
 493                         if (in_array($medium['type'], [self::AUDIO, self::IMAGE]) ||
 494                                 in_array($filetype, ['audio', 'image'])) {
 495                                 $attachments['visual'][] = $medium;
 496                         } elseif (($medium['type'] == self::VIDEO) || ($filetype == 'video')) {
 497                                 if (strpos($medium['url'], $guid) !== false) {
 498                                         // Peertube videos are delivered in many different resolutions. We pick a moderate one.
 499                                         // By checking against the GUID we also ensure to only work this way on Peertube posts.
 500                                         // This wouldn't be executed when someone for example on Mastodon was sharing multiple videos in a single post.
 501                                         if (empty($height) || ($height > $medium['height']) && ($medium['height'] >= 480)) {
 502                                                 $height = $medium['height'];
 503                                                 $selected = $medium['url'];
 504                                         }
 505                                         $video[$medium['url']] = $medium;
 506                                 } else {
 507                                         $attachments['visual'][] = $medium;
 508                                 }
 509                         } else {
 510                                 $attachments['additional'][] = $medium;
 511                         }
 512                 }
 513                 if (!empty($selected)) {
 514                         $attachments['visual'][] = $video[$selected];
 515                         unset($video[$selected]);
 516                         foreach ($video as $element) {
 517                                 $attachments['additional'][] = $element;
 518                         }
 519                 }
 520                 return $attachments;
 521         }
 522 }