src/Content/Text/NPF.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\Content\Text;
  23
  24 use DOMDocument;
  25 use DOMElement;
  26 use Friendica\Model\Photo;
  27 use Friendica\Model\Post;
  28
  29 /**
  30  * Tumblr Neue Post Format
  31  * @see https://www.tumblr.com/docs/npf
  32  */
  33 class NPF
  34 {
  35         static public function fromBBCode(string $bbcode, int $uri_id): array
  36         {
  37                 $npf = [];
  38
  39                 $bbcode = self::prepareBody($bbcode);
  40
  41                 $html = BBCode::convert($bbcode, false, BBCode::CONNECTORS);
  42                 if (empty($html)) {
  43                         return [];
  44                 }
  45
  46                 $doc = new DOMDocument();
  47                 if (!@$doc->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'))) {
  48                         return [];
  49                 }
  50
  51                 $element = $doc->getElementsByTagName('body')->item(0);
  52                 $npf = self::routeChildren($element, $uri_id, 0, $npf);
  53
  54                 return self::addLinkBlock($uri_id, 0, $npf);
  55         }
  56
  57         static private function prepareBody(string $body): string
  58         {
  59                 $shared = BBCode::fetchShareAttributes($body);
  60                 if (!empty($shared)) {
  61                         $body = $shared['shared'];
  62                 }
  63
  64                 $body = BBCode::removeAttachment($body);
  65
  66                 $body = preg_replace("/\[img\=([0-9]*)x([0-9]*)\](.*?)\[\/img\]/ism", '[img]$3[/img]', $body);
  67
  68                 if (preg_match_all("#\[url=([^\]]+?)\]\s*\[img=([^\[\]]*)\]([^\[\]]*)\[\/img\]\s*\[/url\]#ism", $body, $pictures, PREG_SET_ORDER)) {
  69                         foreach ($pictures as $picture) {
  70                                 if (preg_match('#/photo/.*-[01]\.#ism', $picture[2]) && (preg_match('#/photo/.*-0\.#ism', $picture[1]) || preg_match('#/photos/.*/image/#ism', $picture[1]))) {
  71                                         $body = str_replace($picture[0], "\n\n[img=" . str_replace('-1.', '-0.', $picture[2]) . "]" . $picture[3] . "[/img]\n\n", $body);
  72                                 }
  73                         }
  74                 }
  75
  76                 $body = preg_replace("/\[img\=(.*?)\](.*?)\[\/img\]/ism", "\n\n[img=$1]$2[/img]\n\n", $body);
  77
  78                 if (preg_match_all("#\[url=([^\]]+?)\]\s*\[img\]([^\[]+?)\[/img\]\s*\[/url\]#ism", $body, $pictures, PREG_SET_ORDER)) {
  79                         foreach ($pictures as $picture) {
  80                                 if (preg_match('#/photo/.*-[01]\.#ism', $picture[2]) && (preg_match('#/photo/.*-0\.#ism', $picture[1]) || preg_match('#/photos/.*/image/#ism', $picture[1]))) {
  81                                         $body = str_replace($picture[0], "\n\n[img]" . str_replace('-1.', '-0.', $picture[2]) . "[/img]\n\n", $body);
  82                                 }
  83                         }
  84                 }
  85
  86                 $body = preg_replace("/\[img\](.*?)\[\/img\]/ism", "\n\n[img]$1[/img]\n\n", $body);
  87                 $body = preg_replace("/\[audio\](.*?)\[\/audio\]/ism", "\n\n[audio]$1[/audio]\n\n", $body);
  88                 $body = preg_replace("/\[video\](.*?)\[\/video\]/ism", "\n\n[video]$1[/video]\n\n", $body);
  89
  90                 do {
  91                         $oldbody = $body;
  92                         $body = str_replace(["\n\n\n"], ["\n\n"], $body);
  93                 } while ($oldbody != $body);
  94
  95                 return trim($body);
  96         }
  97
  98         static private function routeChildren(DOMElement $element, int $uri_id, int $level, array $npf): array
  99         {
 100                 $text       = '';
 101                 $formatting = [];
 102
 103                 foreach ($element->childNodes as $child) {
 104                         switch ($child->nodeName) {
 105                                 case 'blockquote':
 106                                         $npf = self::addText($text, $formatting, $npf);
 107                                         $npf = self::addQuoteBlock($child, $uri_id, $level, $npf);
 108                                         break;
 109
 110                                 case 'h1':
 111                                         $npf = self::addText($text, $formatting, $npf);
 112                                         $npf = self::addTextBlock($child, $uri_id, $level, $npf, 'heading1');
 113                                         break;
 114
 115                                 case 'h2':
 116                                         $npf = self::addText($text, $formatting, $npf);
 117                                         $npf = self::addTextBlock($child, $uri_id, $level, $npf, 'heading1');
 118                                         break;
 119
 120                                 case 'h3':
 121                                         $npf = self::addText($text, $formatting, $npf);
 122                                         $npf = self::addTextBlock($child, $uri_id, $level, $npf, 'heading1');
 123                                         break;
 124
 125                                 case 'h4':
 126                                         $npf = self::addText($text, $formatting, $npf);
 127                                         $npf = self::addTextBlock($child, $uri_id, $level, $npf, 'heading2');
 128                                         break;
 129
 130                                 case 'h5':
 131                                         $npf = self::addText($text, $formatting, $npf);
 132                                         $npf = self::addTextBlock($child, $uri_id, $level, $npf, 'heading2');
 133                                         break;
 134
 135                                 case 'h6':
 136                                         $npf = self::addText($text, $formatting, $npf);
 137                                         $npf = self::addTextBlock($child, $uri_id, $level, $npf, 'heading2');
 138                                         break;
 139
 140                                 case 'ul':
 141                                         $npf = self::addText($text, $formatting, $npf);
 142                                         $npf = self::addListBlock($child, $uri_id, $level, $npf, false, 0);
 143                                         break;
 144
 145                                 case 'ol':
 146                                         $npf = self::addText($text, $formatting, $npf);
 147                                         $npf = self::addListBlock($child, $uri_id, $level, $npf, true, 0);
 148                                         break;
 149
 150                                 case 'hr':
 151                                 case 'br':
 152                                         $text .= "\n";
 153                                         break;
 154
 155                                 case 'pre':
 156                                 case 'code':
 157                                         $npf = self::addText($text, $formatting, $npf);
 158                                         $npf = self::addTextBlock($child, $uri_id, $level, $npf, 'indented');
 159                                         break;
 160
 161                                 case 'a':
 162                                         $npf = self::addText($text, $formatting, $npf);
 163                                         $npf = self::addMediaBlock($child, $uri_id, $level, $npf);
 164                                         break;
 165
 166                                 case 'table':
 167                                         // Unsupported
 168                                         // $child->ownerDocument->saveHTML($child)
 169                                         break;
 170
 171                                 case 'img':
 172                                         $npf = self::addText($text, $formatting, $npf);
 173                                         $npf = self::addImageBlock($child, $uri_id, $level, $npf);
 174                                         break;
 175
 176                                 case 'p':
 177                                 case 'div':
 178                                         $npf = self::addText($text, $formatting, $npf);
 179                                         $npf = self::addTextBlock($child, $uri_id, $level, $npf);
 180                                         break;
 181
 182                                 default:
 183                                         $text .= $child->textContent;
 184                                         break;
 185                         }
 186                 }
 187                 return $npf;
 188         }
 189
 190         static private function addText(string $text, array $formatting, array $npf): array
 191         {
 192                 if (empty($text)) {
 193                         return $npf;
 194                 }
 195                 $block = [
 196                         'type' => 'text',
 197                         'text' => $text,
 198                 ];
 199
 200                 if (!empty($formatting)) {
 201                         $block['formatting'] = $formatting;
 202                 }
 203
 204                 $npf[] = $block;
 205
 206                 return $npf;
 207         }
 208
 209         static private function routeElement(DOMElement $element, int $uri_id, int $level, array $npf): array
 210         {
 211                 switch ($element->nodeName) {
 212                         case 'blockquote':
 213                                 $npf = self::addQuoteBlock($element, $uri_id, $level, $npf);
 214                                 break;
 215
 216                         case 'h1':
 217                                 $npf = self::addTextBlock($element, $uri_id, $level, $npf, 'heading1');
 218                                 break;
 219
 220                         case 'h2':
 221                                 $npf = self::addTextBlock($element, $uri_id, $level, $npf, 'heading1');
 222                                 break;
 223
 224                         case 'h3':
 225                                 $npf = self::addTextBlock($element, $uri_id, $level, $npf, 'heading1');
 226                                 break;
 227
 228                         case 'h4':
 229                                 $npf = self::addTextBlock($element, $uri_id, $level, $npf, 'heading2');
 230                                 break;
 231
 232                         case 'h5':
 233                                 $npf = self::addTextBlock($element, $uri_id, $level, $npf, 'heading2');
 234                                 break;
 235
 236                         case 'h6':
 237                                 $npf = self::addTextBlock($element, $uri_id, $level, $npf, 'heading2');
 238                                 break;
 239
 240                         case 'ul':
 241                                 $npf = self::addListBlock($element, $uri_id, $level, $npf, false, 0);
 242                                 break;
 243
 244                         case 'ol':
 245                                 $npf = self::addListBlock($element, $uri_id, $level, $npf, true, 0);
 246                                 break;
 247
 248                         case 'hr':
 249                         case 'br':
 250                                 break;
 251
 252                         case 'pre':
 253                         case 'code':
 254                                 $npf = self::addTextBlock($element, $uri_id, $level, $npf, 'indented');
 255                                 break;
 256
 257                         case 'a':
 258                                 $npf = self::addMediaBlock($element, $uri_id, $level, $npf);
 259                                 break;
 260
 261                         case 'table':
 262                                 // Unsupported
 263                                 // $element->ownerDocument->saveHTML($element)
 264                                 break;
 265
 266                         case 'img':
 267                                 $npf = self::addImageBlock($element, $uri_id, $level, $npf);
 268                                 break;
 269
 270                         default:
 271                                 $npf = self::addTextBlock($element, $uri_id, $level, $npf);
 272                                 break;
 273                 }
 274                 return $npf;
 275         }
 276
 277         static private function addImageBlock(DOMElement $element, int $uri_id, int $level, array $npf): array
 278         {
 279                 $attributes = [];
 280                 foreach ($element->attributes as $key => $attribute) {
 281                         $attributes[$key] = $attribute->value;
 282                 }
 283                 if (empty($attributes['src'])) {
 284                         return $npf;
 285                 }
 286
 287                 $block = [
 288                         'type'  => 'image',
 289                         'media' => [],
 290                 ];
 291
 292                 if (!empty($attributes['alt'])) {
 293                         $block['alt_text'] = $attributes['alt'];
 294                 }
 295
 296                 if (!empty($attributes['title']) && ($attributes['alt'] ?? '' != $attributes['title'])) {
 297                         $block['caption'] = $attributes['title'];
 298                 }
 299
 300                 $rid = Photo::ridFromURI($attributes['src']);
 301                 if (!empty($rid)) {
 302                         $photos = Photo::selectToArray([], ['resource-id' => $rid]);
 303                         foreach ($photos as $photo) {
 304                                 $block['media'][] = [
 305                                         'type'   => $photo['type'],
 306                                         'url'    => str_replace('-0.', '-' . $photo['scale'] . '.', $attributes['src']),
 307                                         'width'  => $photo['width'],
 308                                         'height' => $photo['height'],
 309                                 ];
 310                         }
 311                         if (empty($attributes['alt']) && !empty($photos[0]['desc'])) {
 312                                 $block['alt_text'] = $photos[0]['desc'];
 313                         }
 314                 } elseif ($media = Post\Media::getByURL($uri_id, $attributes['src'], [Post\Media::IMAGE])) {
 315                         $block['media'][] = [
 316                                 'type'   => $media['mimetype'],
 317                                 'url'    => $media['url'],
 318                                 'width'  => $media['width'],
 319                                 'height' => $media['height'],
 320                         ];
 321                         if (empty($attributes['alt']) && !empty($media['description'])) {
 322                                 $block['alt_text'] = $media['description'];
 323                         }
 324                 } else {
 325                         $block['media'][] = ['url' => $attributes['src']];
 326                 }
 327
 328                 if ($level > 0) {
 329                         $block['indent_level'] = $level;
 330                 }
 331
 332                 $npf[] = $block;
 333
 334                 return $npf;
 335         }
 336
 337         static private function addMediaBlock(DOMElement $element, int $uri_id, int $level, array $npf): array
 338         {
 339                 $attributes = [];
 340                 foreach ($element->attributes as $key => $attribute) {
 341                         $attributes[$key] = $attribute->value;
 342                 }
 343                 if (empty($attributes['href'])) {
 344                         return $npf;
 345                 }
 346
 347                 $media = Post\Media::getByURL($uri_id, $attributes['href'], [Post\Media::AUDIO, Post\Media::VIDEO]);
 348                 if (!empty($media)) {
 349                         switch ($media['type']) {
 350                                 case Post\Media::AUDIO:
 351                                         $block = [
 352                                                 'type' => 'audio',
 353                                                 'media' => [
 354                                                         'type' => $media['mimetype'],
 355                                                         'url'  => $media['url'],
 356                                                 ]
 357                                         ];
 358
 359                                         if (!empty($media['name'])) {
 360                                                 $block['title'] = $media['name'];
 361                                         } elseif (!empty($media['description'])) {
 362                                                 $block['title'] = $media['description'];
 363                                         }
 364
 365                                         $block = self::addPoster($media, $block);
 366                                         break;
 367
 368                                 case Post\Media::VIDEO:
 369                                         $block = [
 370                                                 'type' => 'video',
 371                                                 'media' => [
 372                                                         'type' => $media['mimetype'],
 373                                                         'url'  => $media['url'],
 374                                                 ]
 375                                         ];
 376
 377                                         $block = self::addPoster($media, $block);
 378                                         break;
 379                         }
 380                 } else {
 381                         $block = [
 382                                 'type' => 'text',
 383                                 'text' => $element->textContent,
 384                                 'formatting' => [
 385                                         'start' => 0,
 386                                         'end'   => strlen($element->textContent),
 387                                         'type'  => 'link',
 388                                         'url'   => $attributes['href']
 389                                 ]
 390                         ];
 391                 }
 392
 393                 if ($level > 0) {
 394                         $block['indent_level'] = $level;
 395                 }
 396
 397                 $npf[] = $block;
 398
 399                 return $npf;
 400         }
 401
 402         static private function addPoster(array $media, array $block): array
 403         {
 404                 $poster = [];
 405                 if (!empty($media['preview'])) {
 406                         $poster['url'] = $media['preview'];
 407                 }
 408                 if (!empty($media['preview-width'])) {
 409                         $poster['width'] = $media['preview-width'];
 410                 }
 411                 if (!empty($media['preview-height'])) {
 412                         $poster['height'] = $media['preview-height'];
 413                 }
 414                 if (!empty($poster)) {
 415                         $block['poster'] = $poster;
 416                 }
 417                 return $block;
 418         }
 419
 420         static private function getTypeForNodeName(string $nodename): string
 421         {
 422                 switch ($nodename) {
 423                         case 'b':
 424                         case 'strong':
 425                                 return 'bold';
 426
 427                         case 'i':
 428                         case 'em':
 429                                 return 'italic';
 430
 431                         case 's':
 432                                 return 'strikethrough';
 433                 }
 434                 return '';
 435         }
 436
 437         static private function fetchText(DOMElement $element, array $text = ['text' => '', 'formatting' => []]): array
 438         {
 439                 foreach ($element->childNodes as $child) {
 440                         $start = strlen($text['text']);
 441
 442                         $type = self::getTypeForNodeName($child->nodeName);
 443
 444                         if ($child->nodeName == 'br') {
 445                                 $text['text'] .= "\n";
 446                         } elseif (($type != '') || in_array($child->nodeName, ['#text', 'code', 'a', 'p', 'span', 'u', 'img', 'summary', 'ul', 'blockquote', 'h3', 'ol'])) {
 447                                 $text['text'] .= $child->textContent;
 448                         } else {
 449                                 echo $element->ownerDocument->saveHTML($element) . "\n";
 450                                 die($child->nodeName . "\n");
 451                         }
 452                         if (!empty($type)) {
 453                                 $text['formatting'][] = ['start' => $start, 'end' => strlen($text['text']), 'type' => $type];
 454                         }
 455                 }
 456                 return $text;
 457         }
 458
 459         static private function addQuoteBlock(DOMElement $element, int $uri_id, int $level, array $npf): array
 460         {
 461                 $block = ['type' => 'text', 'subtype' => 'indented'];
 462
 463                 if ($level > 0) {
 464                         $block['indent_level'] = $level;
 465                 }
 466
 467                 $npf[] = $block;
 468
 469                 $npf = self::routeChildren($element, $uri_id, 0, $npf);
 470
 471                 return $npf;
 472         }
 473
 474         static private function addTextBlock(DOMElement $element, int $uri_id, int $level, array $npf, string $subtype = ''): array
 475         {
 476                 if (empty($subtype) && ($element->textContent == $element->firstChild->textContent) && ($element->firstChild->nodeName != '#text')) {
 477                         return self::routeElement($element->firstChild, $uri_id, $level, $npf);
 478                 }
 479
 480                 $block = ['type' => 'text'];
 481
 482                 if (!empty($subtype)) {
 483                         $block['subtype'] = $subtype;
 484                 }
 485
 486                 $text = self::fetchText($element);
 487
 488                 $block['text']       = $text['text'];
 489                 $block['formatting'] = $text['formatting'];
 490
 491                 if (empty($subtype)) {
 492                         $type = self::getTypeForNodeName($element->nodeName);
 493                         if (!empty($type)) {
 494                                 $block['formatting'][] = ['start' => 0, 'end' => strlen($block['text']), 'type' => $type];
 495                         }
 496                 }
 497
 498                 if (empty($block['formatting'])) {
 499                         unset($block['formatting']);
 500                 }
 501
 502                 if ($level > 0) {
 503                         $block['indent_level'] = $level;
 504                 }
 505
 506                 $npf[] = $block;
 507
 508                 return $npf;
 509         }
 510
 511         static private function addListBlock(DOMElement $element, int $uri_id, int $level, array $npf, bool $ordered): array
 512         {
 513                 foreach ($element->childNodes as $child) {
 514                         switch ($child->nodeName) {
 515                                 case 'ul':
 516                                         $npf = self::addListBlock($child, $uri_id, $level++, $npf, false);
 517                                 case 'ol':
 518                                         $npf = self::addListBlock($child, $uri_id, $level++, $npf, true);
 519                                 case 'li':
 520                                         $text = self::fetchText($child);
 521
 522                                         $block = [
 523                                                 'type'    => 'text',
 524                                                 'subtype' => $ordered ? 'ordered-list-item' : 'unordered-list-item',
 525                                                 'text'    => $text['text']
 526                                         ];
 527                                         if ($level > 0) {
 528                                                 $block['indent_level'] = $level;
 529                                         }
 530                                         if (!empty($text['formatting'])) {
 531                                                 $block['formatting'] = $text['formatting'];
 532                                         }
 533                                         $npf[] = $block;
 534                         }
 535                 }
 536
 537                 return $npf;
 538         }
 539
 540         static private function addLinkBlock(int $uri_id, int $level, array $npf): array
 541         {
 542                 foreach (Post\Media::getByURIId($uri_id, [Post\Media::HTML]) as $link) {
 543                         $host = parse_url($link['url'], PHP_URL_HOST);
 544                         if (in_array($host, ['www.youtube.com', 'youtu.be'])) {
 545                                 $block = [
 546                                         'type'     => 'video',
 547                                         'provider' => 'youtube',
 548                                         'url'      => $link['url'],
 549                                 ];
 550                         } elseif (in_array($host, ['vimeo.com'])) {
 551                                 $block = [
 552                                         'type'     => 'video',
 553                                         'provider' => 'vimeo',
 554                                         'url'      => $link['url'],
 555                                 ];
 556                         } elseif (in_array($host, ['open.spotify.com'])) {
 557                                 $block = [
 558                                         'type'     => 'audio',
 559                                         'provider' => 'spotify',
 560                                         'url'      => $link['url'],
 561                                 ];
 562                         } else {
 563                                 $block = [
 564                                         'type' => 'link',
 565                                         'url'  => $link['url'],
 566                                 ];
 567                                 if (!empty($link['name'])) {
 568                                         $block['title'] = $link['name'];
 569                                 }
 570                                 if (!empty($link['description'])) {
 571                                         $block['description'] = $link['description'];
 572                                 }
 573                                 if (!empty($link['author-name'])) {
 574                                         $block['author'] = $link['author-name'];
 575                                 }
 576                                 if (!empty($link['publisher-name'])) {
 577                                         $block['site_name'] = $link['publisher-name'];
 578                                 }
 579                         }
 580
 581                         if ($level > 0) {
 582                                 $block['indent_level'] = $level;
 583                         }
 584
 585                         $npf[] = self::addPoster($link, $block);
 586                 }
 587                 return $npf;
 588         }
 589 }