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