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\Model\Item;
  32 use Friendica\Model\Post;
  33 use Friendica\Util\Images;
  34 use Friendica\Util\ParseUrl;
  35 use Friendica\Util\Strings;
  36
  37 /**
  38  * Class Media
  39  *
  40  * This Model class handles media interactions.
  41  * This tables stores medias (images, videos, audio files) related to posts.
  42  */
  43 class Media
  44 {
  45         const UNKNOWN     = 0;
  46         const IMAGE       = 1;
  47         const VIDEO       = 2;
  48         const AUDIO       = 3;
  49         const TEXT        = 4;
  50         const APPLICATION = 5;
  51         const TORRENT     = 16;
  52         const HTML        = 17;
  53         const XML         = 18;
  54         const PLAIN       = 19;
  55         const DOCUMENT    = 128;
  56
  57         /**
  58          * Insert a post-media record
  59          *
  60          * @param array $media
  61          * @return void
  62          */
  63         public static function insert(array $media, bool $force = false)
  64         {
  65                 if (empty($media['url']) || empty($media['uri-id']) || !isset($media['type'])) {
  66                         Logger::warning('Incomplete media data', ['media' => $media]);
  67                         return;
  68                 }
  69
  70                 // "document" has got the lowest priority. So when the same file is both attached as document
  71                 // and embedded as picture then we only store the picture or replace the document
  72                 $found = DBA::selectFirst('post-media', ['type'], ['uri-id' => $media['uri-id'], 'url' => $media['url']]);
  73                 if (!$force && !empty($found) && (($found['type'] != self::DOCUMENT) || ($media['type'] == self::DOCUMENT))) {
  74                         Logger::info('Media already exists', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'callstack' => System::callstack()]);
  75                         return;
  76                 }
  77
  78                 $media = self::unsetEmptyFields($media);
  79
  80                 // We are storing as fast as possible to avoid duplicated network requests
  81                 // when fetching additional information for pictures and other content.
  82                 $result = DBA::insert('post-media', $media, Database::INSERT_UPDATE);
  83                 Logger::info('Stored media', ['result' => $result, 'media' => $media, 'callstack' => System::callstack()]);
  84                 $stored = $media;
  85
  86                 $media = self::fetchAdditionalData($media);
  87                 $media = self::unsetEmptyFields($media);
  88
  89                 if (array_diff_assoc($media, $stored)) {
  90                         $result = DBA::insert('post-media', $media, Database::INSERT_UPDATE);
  91                         Logger::info('Updated media', ['result' => $result, 'media' => $media]);
  92                 } else {
  93                         Logger::info('Nothing to update', ['media' => $media]);
  94                 }
  95         }
  96
  97         /**
  98          * Remove empty media fields
  99          *
 100          * @param array $media
 101          * @return array cleaned media array
 102          */
 103         private static function unsetEmptyFields(array $media)
 104         {
 105                 $fields = ['mimetype', 'height', 'width', 'size', 'preview', 'preview-height', 'preview-width', 'description'];
 106                 foreach ($fields as $field) {
 107                         if (empty($media[$field])) {
 108                                 unset($media[$field]);
 109                         }
 110                 }
 111                 return $media;
 112         }
 113
 114         /**
 115          * Copy attachments from one uri-id to another
 116          *
 117          * @param integer $from_uri_id
 118          * @param integer $to_uri_id
 119          * @return void
 120          */
 121         public static function copy(int $from_uri_id, int $to_uri_id)
 122         {
 123                 $attachments = self::getByURIId($from_uri_id);
 124                 foreach ($attachments as $attachment) {
 125                         $attachment['uri-id'] = $to_uri_id;
 126                         self::insert($attachment);
 127                 }
 128         }
 129
 130         /**
 131          * Creates the "[attach]" element from the given attributes
 132          *
 133          * @param string $href
 134          * @param integer $length
 135          * @param string $type
 136          * @param string $title
 137          * @return string "[attach]" element
 138          */
 139         public static function getAttachElement(string $href, int $length, string $type, string $title = '')
 140         {
 141                 $media = self::fetchAdditionalData(['type' => self::DOCUMENT, 'url' => $href,
 142                         'size' => $length, 'mimetype' => $type, 'description' => $title]);
 143
 144                 return '[attach]href="' . $media['url'] . '" length="' . $media['size'] .
 145                         '" type="' . $media['mimetype'] . '" title="' . $media['description'] . '"[/attach]';
 146         }
 147
 148         /**
 149          * Fetch additional data for the provided media array
 150          *
 151          * @param array $media
 152          * @return array media array with additional data
 153          */
 154         public static function fetchAdditionalData(array $media)
 155         {
 156                 // Fetch the mimetype or size if missing.
 157                 if (empty($media['mimetype']) || empty($media['size'])) {
 158                         $timeout = DI::config()->get('system', 'xrd_timeout');
 159                         $curlResult = DI::httpRequest()->head($media['url'], ['timeout' => $timeout]);
 160                         if ($curlResult->isSuccess()) {
 161                                 if (empty($media['mimetype'])) {
 162                                         $media['mimetype'] = $curlResult->getHeader('Content-Type');
 163                                 }
 164                                 if (empty($media['size'])) {
 165                                         $media['size'] = (int)$curlResult->getHeader('Content-Length');
 166                                 }
 167                         } else {
 168                                 Logger::notice('Could not fetch head', ['media' => $media]);
 169                         }
 170                 }
 171
 172                 $filetype = !empty($media['mimetype']) ? strtolower(current(explode('/', $media['mimetype']))) : '';
 173
 174                 if (($media['type'] == self::IMAGE) || ($filetype == 'image')) {
 175                         $imagedata = Images::getInfoFromURLCached($media['url']);
 176                         if (!empty($imagedata)) {
 177                                 $media['mimetype'] = $imagedata['mime'];
 178                                 $media['size'] = $imagedata['size'];
 179                                 $media['width'] = $imagedata[0];
 180                                 $media['height'] = $imagedata[1];
 181                         } else {
 182                                 Logger::notice('No image data', ['media' => $media]);
 183                         }
 184                         if (!empty($media['preview'])) {
 185                                 $imagedata = Images::getInfoFromURLCached($media['preview']);
 186                                 if (!empty($imagedata)) {
 187                                         $media['preview-width'] = $imagedata[0];
 188                                         $media['preview-height'] = $imagedata[1];
 189                                 }
 190                         }
 191                 }
 192
 193                 if ($media['type'] != self::DOCUMENT) {
 194                         $media = self::addType($media);
 195                 }
 196
 197                 if ($media['type'] == self::HTML) {
 198                         $data = ParseUrl::getSiteinfoCached($media['url'], false);
 199                         $media['preview'] = $data['images'][0]['src'] ?? null;
 200                         $media['preview-height'] = $data['images'][0]['height'] ?? null;
 201                         $media['preview-width'] = $data['images'][0]['width'] ?? null;
 202                         $media['description'] = $data['text'] ?? null;
 203                         $media['name'] = $data['title'] ?? null;
 204                         $media['author-url'] = $data['author_url'] ?? null;
 205                         $media['author-name'] = $data['author_name'] ?? null;
 206                         $media['author-image'] = $data['author_img'] ?? null;
 207                         $media['publisher-url'] = $data['publisher_url'] ?? null;
 208                         $media['publisher-name'] = $data['publisher_name'] ?? null;
 209                         $media['publisher-image'] = $data['publisher_img'] ?? null;
 210                 }
 211                 return $media;
 212         }
 213
 214         /**
 215          * Add the detected type to the media array
 216          *
 217          * @param array $data
 218          * @return array data array with the detected type
 219          */
 220         public static function addType(array $data)
 221         {
 222                 if (empty($data['mimetype'])) {
 223                         Logger::info('No MimeType provided', ['media' => $data]);
 224                         return $data;
 225                 }
 226
 227                 $type = explode('/', current(explode(';', $data['mimetype'])));
 228                 if (count($type) < 2) {
 229                         Logger::info('Unknown MimeType', ['type' => $type, 'media' => $data]);
 230                         $data['type'] = self::UNKNOWN;
 231                         return $data;
 232                 }
 233
 234                 $filetype = strtolower($type[0]);
 235                 $subtype = strtolower($type[1]);
 236
 237                 if ($filetype == 'image') {
 238                         $data['type'] = self::IMAGE;
 239                 } elseif ($filetype == 'video') {
 240                         $data['type'] = self::VIDEO;
 241                 } elseif ($filetype == 'audio') {
 242                         $data['type'] = self::AUDIO;
 243                 } elseif (($filetype == 'text') && ($subtype == 'html')) {
 244                         $data['type'] = self::HTML;
 245                 } elseif (($filetype == 'text') && ($subtype == 'xml')) {
 246                         $data['type'] = self::XML;
 247                 } elseif (($filetype == 'text') && ($subtype == 'plain')) {
 248                         $data['type'] = self::PLAIN;
 249                 } elseif ($filetype == 'text') {
 250                         $data['type'] = self::TEXT;
 251                 } elseif (($filetype == 'application') && ($subtype == 'x-bittorrent')) {
 252                         $data['type'] = self::TORRENT;
 253                 } elseif ($filetype == 'application') {
 254                         $data['type'] = self::APPLICATION;
 255                 } else {
 256                         $data['type'] = self::UNKNOWN;
 257                         Logger::info('Unknown type', ['filetype' => $filetype, 'subtype' => $subtype, 'media' => $data]);
 258                         return $data;
 259                 }
 260
 261                 Logger::debug('Detected type', ['filetype' => $filetype, 'subtype' => $subtype, 'media' => $data]);
 262                 return $data;
 263         }
 264
 265         /**
 266          * Tests for path patterns that are usef for picture links in Friendica
 267          *
 268          * @param string $page    Link to the image page
 269          * @param string $preview Preview picture
 270          * @return boolean
 271          */
 272         private static function isPictureLink(string $page, string $preview)
 273         {
 274                 return preg_match('#/photos/.*/image/#ism', $page) && preg_match('#/photo/.*-1\.#ism', $preview);
 275         }
 276
 277         /**
 278          * Add media links and remove them from the body
 279          *
 280          * @param integer $uriid
 281          * @param string $body
 282          * @return string Body without media links
 283          */
 284         public static function insertFromBody(int $uriid, string $body)
 285         {
 286                 // Simplify image codes
 287                 $body = preg_replace("/\[img\=([0-9]*)x([0-9]*)\](.*?)\[\/img\]/ism", '[img]$3[/img]', $body);
 288
 289                 $unshared_body = preg_replace("/\s*\[share .*?\].*?\[\/share\]\s*/ism", '', $body);
 290
 291                 $attachments = [];
 292                 if (preg_match_all("#\[url=([^\]]+?)\]\s*\[img=([^\[\]]*)\]([^\[\]]*)\[\/img\]\s*\[/url\]#ism", $body, $pictures, PREG_SET_ORDER)) {
 293                         foreach ($pictures as $picture) {
 294                                 if (!self::isPictureLink($picture[1], $picture[2])) {
 295                                         continue;
 296                                 }
 297                                 $body = str_replace($picture[0], '', $body);
 298                                 $image = str_replace('-1.', '-0.', $picture[2]);
 299                                 $attachments[$image] = ['uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $image,
 300                                         'preview' => $picture[2], 'description' => $picture[3]];
 301                         }
 302                 }
 303
 304                 if (preg_match_all("/\[img=([^\[\]]*)\]([^\[\]]*)\[\/img\]/Usi", $body, $pictures, PREG_SET_ORDER)) {
 305                         foreach ($pictures as $picture) {
 306                                 $body = str_replace($picture[0], '', $body);
 307                                 $attachments[$picture[1]] = ['uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $picture[1], 'description' => $picture[2]];
 308                         }
 309                 }
 310
 311                 if (preg_match_all("#\[url=([^\]]+?)\]\s*\[img\]([^\[]+?)\[/img\]\s*\[/url\]#ism", $body, $pictures, PREG_SET_ORDER)) {
 312                         foreach ($pictures as $picture) {
 313                                 if (!self::isPictureLink($picture[1], $picture[2])) {
 314                                         continue;
 315                                 }
 316                                 $body = str_replace($picture[0], '', $body);
 317                                 $image = str_replace('-1.', '-0.', $picture[2]);
 318                                 $attachments[$image] = ['uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $image,
 319                                         'preview' => $picture[2], 'description' => null];
 320                         }
 321                 }
 322
 323                 if (preg_match_all("/\[img\]([^\[\]]*)\[\/img\]/ism", $body, $pictures, PREG_SET_ORDER)) {
 324                         foreach ($pictures as $picture) {
 325                                 $body = str_replace($picture[0], '', $body);
 326                                 $attachments[$picture[1]] = ['uri-id' => $uriid, 'type' => self::IMAGE, 'url' => $picture[1]];
 327                         }
 328                 }
 329
 330                 if (preg_match_all("/\[audio\]([^\[\]]*)\[\/audio\]/ism", $body, $audios, PREG_SET_ORDER)) {
 331                         foreach ($audios as $audio) {
 332                                 $body = str_replace($audio[0], '', $body);
 333                                 $attachments[$audio[1]] = ['uri-id' => $uriid, 'type' => self::AUDIO, 'url' => $audio[1]];
 334                         }
 335                 }
 336
 337                 if (preg_match_all("/\[video\]([^\[\]]*)\[\/video\]/ism", $body, $videos, PREG_SET_ORDER)) {
 338                         foreach ($videos as $video) {
 339                                 $body = str_replace($video[0], '', $body);
 340                                 $attachments[$video[1]] = ['uri-id' => $uriid, 'type' => self::VIDEO, 'url' => $video[1]];
 341                         }
 342                 }
 343
 344                 $url = PageInfo::getRelevantUrlFromBody($body);
 345                 if (!empty($url)) {
 346                         Logger::debug('Got page url', ['url' => $url]);
 347                         $attachments[$url] = ['uri-id' => $uriid, 'type' => self::UNKNOWN, 'url' => $url];
 348                 }
 349
 350                 foreach ($attachments as $attachment) {
 351                         // Only store attachments that are part of the unshared body
 352                         if (strpos($unshared_body, $attachment['url']) !== false) {
 353                                 self::insert($attachment);
 354                         }
 355                 }
 356
 357                 return trim($body);
 358         }
 359
 360         /**
 361          * Add media links from the attachment field
 362          *
 363          * @param integer $uriid
 364          * @param string $body
 365          */
 366         public static function insertFromAttachmentData(int $uriid, string $body)
 367         {
 368                 // Don't look at the shared content
 369                 $body = preg_replace("/\s*\[share .*?\].*?\[\/share\]\s*/ism", '', $body);
 370
 371                 $data = BBCode::getAttachmentData($body);
 372                 if (empty($data))  {
 373                         return;
 374                 }
 375
 376                 Logger::info('Adding attachment data', ['data' => $data]);
 377                 $attachment = [
 378                         'uri-id' => $uriid,
 379                         'type' => self::HTML,
 380                         'url' => $data['url'],
 381                         'preview' => $data['preview'] ?? null,
 382                         'description' => $data['description'] ?? null,
 383                         'name' => $data['title'] ?? null,
 384                         'author-url' => $data['author_url'] ?? null,
 385                         'author-name' => $data['author_name'] ?? null,
 386                         'publisher-url' => $data['provider_url'] ?? null,
 387                         'publisher-name' => $data['provider_name'] ?? null,
 388                 ];
 389                 if (!empty($data['image'])) {
 390                         $attachment['preview'] = $data['image'];
 391                 }
 392                 self::insert($attachment);
 393         }
 394
 395         /**
 396          * Add media links from the attach field
 397          *
 398          * @param integer $uriid
 399          * @param string $attach
 400          * @return void
 401          */
 402         public static function insertFromAttachment(int $uriid, string $attach)
 403         {
 404                 if (!preg_match_all('|\[attach\]href=\"(.*?)\" length=\"(.*?)\" type=\"(.*?)\"(?: title=\"(.*?)\")?|', $attach, $matches, PREG_SET_ORDER)) {
 405                         return;
 406                 }
 407
 408                 foreach ($matches as $attachment) {
 409                         $media['type'] = self::DOCUMENT;
 410                         $media['uri-id'] = $uriid;
 411                         $media['url'] = $attachment[1];
 412                         $media['size'] = $attachment[2];
 413                         $media['mimetype'] = $attachment[3];
 414                         $media['description'] = $attachment[4] ?? '';
 415
 416                         self::insert($media);
 417                 }
 418         }
 419
 420         /**
 421          * Retrieves the media attachments associated with the provided item ID.
 422          *
 423          * @param int $uri_id
 424          * @param array $types
 425          * @return array
 426          * @throws \Exception
 427          */
 428         public static function getByURIId(int $uri_id, array $types = [])
 429         {
 430                 $condition = ['uri-id' => $uri_id];
 431
 432                 if (!empty($types)) {
 433                         $condition = DBA::mergeConditions($condition, ['type' => $types]);
 434                 }
 435
 436                 return DBA::selectToArray('post-media', [], $condition);
 437         }
 438
 439         /**
 440          * Checks if media attachments are associated with the provided item ID.
 441          *
 442          * @param int $uri_id
 443          * @param array $types
 444          * @return array
 445          * @throws \Exception
 446          */
 447         public static function existsByURIId(int $uri_id, array $types = [])
 448         {
 449                 $condition = ['uri-id' => $uri_id];
 450
 451                 if (!empty($types)) {
 452                         $condition = DBA::mergeConditions($condition, ['type' => $types]);
 453                 }
 454
 455                 return DBA::exists('post-media', $condition);
 456         }
 457
 458         /**
 459          * Split the attachment media in the three segments "visual", "link" and "additional"
 460          *
 461          * @param int    $uri_id
 462          * @param string $guid
 463          * @param array  $links ist of links that shouldn't be added
 464          * @return array attachments
 465          */
 466         public static function splitAttachments(int $uri_id, string $guid = '', array $links = [])
 467         {
 468                 $attachments = ['visual' => [], 'link' => [], 'additional' => []];
 469
 470                 $media = self::getByURIId($uri_id);
 471                 if (empty($media)) {
 472                         return $attachments;
 473                 }
 474
 475                 $height = 0;
 476                 $selected = '';
 477
 478                 foreach ($media as $medium) {
 479                         foreach ($links as $link) {
 480                                 if (Strings::compareLink($link, $medium['url'])) {
 481                                         continue 2;
 482                                 }
 483                         }
 484
 485                         $type = explode('/', current(explode(';', $medium['mimetype'])));
 486                         if (count($type) < 2) {
 487                                 Logger::info('Unknown MimeType', ['type' => $type, 'media' => $medium]);
 488                                 $filetype = 'unkn';
 489                                 $subtype = 'unkn';
 490                         } else {
 491                                 $filetype = strtolower($type[0]);
 492                                 $subtype = strtolower($type[1]);
 493                         }
 494
 495                         $medium['filetype'] = $filetype;
 496                         $medium['subtype'] = $subtype;
 497
 498                         if ($medium['type'] == self::HTML || (($filetype == 'text') && ($subtype == 'html'))) {
 499                                 $attachments['link'][] = $medium;
 500                                 continue;
 501                         }
 502
 503                         if (in_array($medium['type'], [self::AUDIO, self::IMAGE]) ||
 504                                 in_array($filetype, ['audio', 'image'])) {
 505                                 $attachments['visual'][] = $medium;
 506                         } elseif (($medium['type'] == self::VIDEO) || ($filetype == 'video')) {
 507                                 if (strpos($medium['url'], $guid) !== false) {
 508                                         // Peertube videos are delivered in many different resolutions. We pick a moderate one.
 509                                         // By checking against the GUID we also ensure to only work this way on Peertube posts.
 510                                         // This wouldn't be executed when someone for example on Mastodon was sharing multiple videos in a single post.
 511                                         if (empty($height) || ($height > $medium['height']) && ($medium['height'] >= 480)) {
 512                                                 $height = $medium['height'];
 513                                                 $selected = $medium['url'];
 514                                         }
 515                                         $video[$medium['url']] = $medium;
 516                                 } else {
 517                                         $attachments['visual'][] = $medium;
 518                                 }
 519                         } else {
 520                                 $attachments['additional'][] = $medium;
 521                         }
 522                 }
 523                 if (!empty($selected)) {
 524                         $attachments['visual'][] = $video[$selected];
 525                         unset($video[$selected]);
 526                         foreach ($video as $element) {
 527                                 $attachments['additional'][] = $element;
 528                         }
 529                 }
 530                 return $attachments;
 531         }
 532
 533         /**
 534          * Add media attachments to the body
 535          *
 536          * @param int $uriid
 537          * @param string $body
 538          * @return string body
 539          */
 540         public static function addAttachmentsToBody(int $uriid, string $body = '')
 541         {
 542                 if (empty($body)) {
 543                         $item = Post::selectFirst(['body'], ['uri-id' => $uriid]);
 544                         if (!DBA::isResult($item)) {
 545                                 return '';
 546                         }
 547                         $body = $item['body'];
 548                 }
 549                 $original_body = $body;
 550
 551                 $body = preg_replace("/\s*\[attachment .*?\].*?\[\/attachment\]\s*/ism", '', $body);
 552
 553                 foreach (self::getByURIId($uriid, [self::IMAGE, self::AUDIO, self::VIDEO]) as $media) {
 554                         if (Item::containsLink($body, $media['url'])) {
 555                                 continue;
 556                         }
 557
 558                         if ($media['type'] == self::IMAGE) {
 559                                 if (!empty($media['preview'])) {
 560                                         if (!empty($media['description'])) {
 561                                                 $body .= "\n[url=" . $media['url'] . "][img=" . $media['preview'] . ']' . $media['description'] .'[/img][/url]';
 562                                         } else {
 563                                                 $body .= "\n[url=" . $media['url'] . "][img]" . $media['preview'] .'[/img][/url]';
 564                                         }
 565                                 } else {
 566                                         if (!empty($media['description'])) {
 567                                                 $body .= "\n[img=" . $media['url'] . ']' . $media['description'] .'[/img]';
 568                                         } else {
 569                                                 $body .= "\n[img]" . $media['url'] .'[/img]';
 570                                         }
 571                                 }
 572                         } elseif ($media['type'] == self::AUDIO) {
 573                                 $body .= "\n[audio]" . $media['url'] . "[/audio]\n";
 574                         } elseif ($media['type'] == self::VIDEO) {
 575                                 $body .= "\n[video]" . $media['url'] . "[/video]\n";
 576                         }
 577                 }
 578
 579                 if (preg_match("/.*(\[attachment.*?\].*?\[\/attachment\]).*/ism", $original_body, $match)) {
 580                         $body .= "\n" . $match[1];
 581                 }
 582
 583                 return $body;
 584         }
 585 }