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