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