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