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