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                 $level = 0;
 189                 foreach ($callstack as $entry) {
 190                         if (in_array($entry, ['ol', 'ul', 'blockquote'])) {
 191                                 ++$level;
 192                         }
 193                 }
 194                 return max(0, $level - 1);
 195         }
 196
 197         static private function getSubTypeByCallstack($callstack, string $text): string
 198         {
 199                 $subtype = '';
 200                 foreach ($callstack as $entry) {
 201                         switch ($entry) {
 202                                 case 'ol':
 203                                         $subtype = 'ordered-list-item';
 204                                         break;
 205
 206                                 case 'ul':
 207                                         $subtype = 'unordered-list-item';
 208                                         break;
 209
 210                                 case 'h1':
 211                                         $subtype = self::$heading_subtype[$entry];
 212                                         break;
 213
 214                                 case 'h2':
 215                                         $subtype = self::$heading_subtype[$entry];
 216                                         break;
 217
 218                                 case 'h3':
 219                                         $subtype = self::$heading_subtype[$entry];
 220                                         break;
 221
 222                                 case 'h4':
 223                                         $subtype = self::$heading_subtype[$entry];
 224                                         break;
 225
 226                                 case 'h5':
 227                                         $subtype = self::$heading_subtype[$entry];
 228                                         break;
 229
 230                                 case 'h6':
 231                                         $subtype = self::$heading_subtype[$entry];
 232                                         break;
 233
 234                                 case 'blockquote':
 235                                         $subtype = mb_strlen($text) < 100 ? 'quote' : 'indented';
 236                                         break;
 237
 238                                 case 'pre':
 239                                         $subtype = 'indented';
 240                                         break;
 241
 242                                 case 'code':
 243                                         $subtype = 'chat';
 244                                         break;
 245                         }
 246                 }
 247                 return $subtype;
 248         }
 249
 250         static private function addFormatting(DOMElement $element, int $uri_id, string $type, array $callstack, array $npf, string $text, array $formatting): array
 251         {
 252                 $start = mb_strlen($text);
 253                 list($npf, $text, $formatting) = self::routeChildren($element, $uri_id, false, $callstack, $npf, $text, $formatting);
 254
 255                 if (!empty($type)) {
 256                         $formatting[] = [
 257                                 'start' => $start,
 258                                 'end'   => mb_strlen($text),
 259                                 'type'  => $type
 260                         ];
 261                 }
 262                 return [$npf, $text, $formatting];
 263         }
 264
 265         static private function addInlineLink(DOMElement $element, int $uri_id, array $callstack, array $npf, string $text, array $formatting): array
 266         {
 267                 $start = mb_strlen($text);
 268                 list($npf, $text, $formatting) = self::routeChildren($element, $uri_id, false, $callstack, $npf, $text, $formatting);
 269
 270                 $attributes = [];
 271                 foreach ($element->attributes as $key => $attribute) {
 272                         $attributes[$key] = trim($attribute->value);
 273                 }
 274                 if (!empty($attributes['href'])) {
 275                         $formatting[] = [
 276                                 'start' => $start,
 277                                 'end'   => mb_strlen($text),
 278                                 'type'  => 'link',
 279                                 'url'   => $attributes['href']
 280                         ];
 281                 }
 282                 return [$npf, $text, $formatting];
 283         }
 284
 285         static private function addBlock(string $text, array $formatting, array $npf, array $callstack): array
 286         {
 287                 $block = [
 288                         'type'      => 'text',
 289                         'subtype'   => '',
 290                         'text'      => $text,
 291                 ];
 292
 293                 if (!empty($formatting)) {
 294                         $block['formatting'] = $formatting;
 295                 }
 296
 297                 $level = self::getLevelByCallstack($callstack);
 298                 if ($level > 0) {
 299                         $block['indent_level'] = $level;
 300                 }
 301
 302                 $subtype = self::getSubTypeByCallstack($callstack, $text);
 303                 if ($subtype) {
 304                         $block['subtype'] = $subtype;
 305                 } else {
 306                         unset($block['subtype']);
 307                 }
 308
 309                 $npf[] = $block;
 310                 $text = '';
 311                 $formatting = [];
 312                 return [$npf, $text, $formatting];
 313         }
 314
 315         static private function addPoster(array $media, array $block): array
 316         {
 317                 $poster = [];
 318                 if (!empty($media['preview'])) {
 319                         $poster['url'] = $media['preview'];
 320                 }
 321                 if (!empty($media['preview-width'])) {
 322                         $poster['width'] = $media['preview-width'];
 323                 }
 324                 if (!empty($media['preview-height'])) {
 325                         $poster['height'] = $media['preview-height'];
 326                 }
 327                 if (!empty($poster)) {
 328                         $block['poster'] = [$poster];
 329                 }
 330                 return $block;
 331         }
 332
 333         static private function addLinkBlockForUriId(int $uri_id, int $level, array $npf): array
 334         {
 335                 foreach (Post\Media::getByURIId($uri_id, [Post\Media::HTML]) as $link) {
 336                         $host = parse_url($link['url'], PHP_URL_HOST);
 337                         if (in_array($host, ['www.youtube.com', 'youtu.be'])) {
 338                                 $block = [
 339                                         'type'     => 'video',
 340                                         'provider' => 'youtube',
 341                                         'url'      => $link['url'],
 342                                 ];
 343                         } elseif (in_array($host, ['vimeo.com'])) {
 344                                 $block = [
 345                                         'type'     => 'video',
 346                                         'provider' => 'vimeo',
 347                                         'url'      => $link['url'],
 348                                 ];
 349                         } elseif (in_array($host, ['open.spotify.com'])) {
 350                                 $block = [
 351                                         'type'     => 'audio',
 352                                         'provider' => 'spotify',
 353                                         'url'      => $link['url'],
 354                                 ];
 355                         } else {
 356                                 $block = [
 357                                         'type' => 'link',
 358                                         'url'  => $link['url'],
 359                                 ];
 360                                 if (!empty($link['name'])) {
 361                                         $block['title'] = $link['name'];
 362                                 }
 363                                 if (!empty($link['description'])) {
 364                                         $block['description'] = $link['description'];
 365                                 }
 366                                 if (!empty($link['author-name'])) {
 367                                         $block['author'] = $link['author-name'];
 368                                 }
 369                                 if (!empty($link['publisher-name'])) {
 370                                         $block['site_name'] = $link['publisher-name'];
 371                                 }
 372                         }
 373
 374                         if ($level > 0) {
 375                                 $block['indent_level'] = $level;
 376                         }
 377
 378                         $npf[] = self::addPoster($link, $block);
 379                 }
 380                 return $npf;
 381         }
 382
 383         static private function addImageBlock(DOMElement $element, int $uri_id, int $level, array $npf): array
 384         {
 385                 $attributes = [];
 386                 foreach ($element->attributes as $key => $attribute) {
 387                         $attributes[$key] = trim($attribute->value);
 388                 }
 389                 if (empty($attributes['src'])) {
 390                         return $npf;
 391                 }
 392
 393                 $block = [
 394                         'type'  => 'image',
 395                         'media' => [],
 396                 ];
 397
 398                 if (!empty($attributes['alt'])) {
 399                         $block['alt_text'] = $attributes['alt'];
 400                 }
 401
 402                 if (!empty($attributes['title']) && (($attributes['alt'] ?? '') != $attributes['title'])) {
 403                         $block['caption'] = $attributes['title'];
 404                 }
 405
 406                 $rid = Photo::ridFromURI($attributes['src']);
 407                 if (!empty($rid)) {
 408                         $photos = Photo::selectToArray([], ['resource-id' => $rid]);
 409                         foreach ($photos as $photo) {
 410                                 $block['media'][] = [
 411                                         'type'   => $photo['type'],
 412                                         'url'    => str_replace('-0.', '-' . $photo['scale'] . '.', $attributes['src']),
 413                                         'width'  => $photo['width'],
 414                                         'height' => $photo['height'],
 415                                 ];
 416                         }
 417                         if (empty($attributes['alt']) && !empty($photos[0]['desc'])) {
 418                                 $block['alt_text'] = $photos[0]['desc'];
 419                         }
 420                 } elseif ($media = Post\Media::getByURL($uri_id, $attributes['src'], [Post\Media::IMAGE])) {
 421                         $block['media'][] = [
 422                                 'type'   => $media['mimetype'],
 423                                 'url'    => $media['url'],
 424                                 'width'  => $media['width'],
 425                                 'height' => $media['height'],
 426                         ];
 427                         if (empty($attributes['alt']) && !empty($media['description'])) {
 428                                 $block['alt_text'] = $media['description'];
 429                         }
 430                 } else {
 431                         $block['media'][] = ['url' => $attributes['src']];
 432                 }
 433
 434                 if ($level > 0) {
 435                         $block['indent_level'] = $level;
 436                 }
 437
 438                 $npf[] = $block;
 439
 440                 return $npf;
 441         }
 442
 443         static private function addLinkBlock(DOMElement $element, int $uri_id, int $level, array $npf): array
 444         {
 445                 $attributes = [];
 446                 foreach ($element->attributes as $key => $attribute) {
 447                         $attributes[$key] = trim($attribute->value);
 448                 }
 449                 if (empty($attributes['href'])) {
 450                         return $npf;
 451                 }
 452
 453                 $media = Post\Media::getByURL($uri_id, $attributes['href'], [Post\Media::AUDIO, Post\Media::VIDEO]);
 454                 if (!empty($media)) {
 455                         switch ($media['type']) {
 456                                 case Post\Media::AUDIO:
 457                                         $block = [
 458                                                 'type' => 'audio',
 459                                                 'media' => [
 460                                                         'type' => $media['mimetype'],
 461                                                         'url'  => $media['url'],
 462                                                 ]
 463                                         ];
 464
 465                                         if (!empty($media['name'])) {
 466                                                 $block['title'] = $media['name'];
 467                                         } elseif (!empty($media['description'])) {
 468                                                 $block['title'] = $media['description'];
 469                                         }
 470
 471                                         $block = self::addPoster($media, $block);
 472                                         break;
 473
 474                                 case Post\Media::VIDEO:
 475                                         $block = [
 476                                                 'type' => 'video',
 477                                                 'media' => [
 478                                                         'type' => $media['mimetype'],
 479                                                         'url'  => $media['url'],
 480                                                 ]
 481                                         ];
 482
 483                                         $block = self::addPoster($media, $block);
 484                                         break;
 485                         }
 486                 } else {
 487                         $block = [
 488                                 'type' => 'text',
 489                                 'text' => $element->textContent,
 490                                 'formatting' => [
 491                                         [
 492                                                 'start' => 0,
 493                                                 'end'   => mb_strlen($element->textContent),
 494                                                 'type'  => 'link',
 495                                                 'url'   => $attributes['href']
 496                                         ]
 497                                 ]
 498                         ];
 499                 }
 500
 501                 if ($level > 0) {
 502                         $block['indent_level'] = $level;
 503                 }
 504
 505                 $npf[] = $block;
 506
 507                 return $npf;
 508         }
 509 }