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