src/Content/Text/HTML.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 DOMXPath;
  26 use Friendica\Protocol\HTTP\MediaType;
  27 use Friendica\Content\Widget\ContactBlock;
  28 use Friendica\Core\Hook;
  29 use Friendica\Core\Renderer;
  30 use Friendica\Core\Search;
  31 use Friendica\DI;
  32 use Friendica\Model\Contact;
  33 use Friendica\Util\Network;
  34 use Friendica\Util\Strings;
  35 use Friendica\Util\XML;
  36 use League\HTMLToMarkdown\HtmlConverter;
  37 use Psr\Http\Message\UriInterface;
  38
  39 class HTML
  40 {
  41         /**
  42          * Search all instances of a specific HTML tag node in the provided DOM document and replaces them with BBCode text nodes.
  43          *
  44          * @see HTML::tagToBBCodeSub()
  45          */
  46         private static function tagToBBCode(DOMDocument $doc, string $tag, array $attributes, string $startbb, string $endbb, bool $ignoreChildren = false)
  47         {
  48                 do {
  49                         $done = self::tagToBBCodeSub($doc, $tag, $attributes, $startbb, $endbb, $ignoreChildren);
  50                 } while ($done);
  51         }
  52
  53         /**
  54          * Search the first specific HTML tag node in the provided DOM document and replaces it with BBCode text nodes.
  55          *
  56          * @param DOMDocument $doc
  57          * @param string      $tag            HTML tag name
  58          * @param array       $attributes     Array of attributes to match and optionally use the value from
  59          * @param string      $startbb        BBCode tag opening
  60          * @param string      $endbb          BBCode tag closing
  61          * @param bool        $ignoreChildren If set to false, the HTML tag children will be appended as text inside the BBCode tag
  62          *                                    Otherwise, they will be entirely ignored. Useful for simple BBCode that draw their
  63          *                                    inner value from an attribute value and disregard the tag children.
  64          * @return bool Whether a replacement was done
  65          */
  66         private static function tagToBBCodeSub(DOMDocument $doc, string $tag, array $attributes, string $startbb, string $endbb, bool $ignoreChildren = false): bool
  67         {
  68                 $savestart = str_replace('$', '\x01', $startbb);
  69                 $replace = false;
  70
  71                 $xpath = new DOMXPath($doc);
  72
  73                 /** @var \DOMNode[] $list */
  74                 $list = $xpath->query("//" . $tag);
  75                 foreach ($list as $node) {
  76                         $attr = [];
  77                         if ($node->attributes->length) {
  78                                 foreach ($node->attributes as $attribute) {
  79                                         $attr[$attribute->name] = $attribute->value;
  80                                 }
  81                         }
  82
  83                         $replace = true;
  84
  85                         $startbb = $savestart;
  86
  87                         $i = 0;
  88
  89                         foreach ($attributes as $attribute => $value) {
  90                                 $startbb = str_replace('\x01' . ++$i, '$1', $startbb);
  91                                 if (strpos('*' . $startbb, '$1') > 0) {
  92                                         if ($replace && (@$attr[$attribute] != '')) {
  93                                                 $startbb = preg_replace($value, $startbb, $attr[$attribute], -1, $count);
  94
  95                                                 // If nothing could be changed
  96                                                 if ($count == 0) {
  97                                                         $replace = false;
  98                                                 }
  99                                         } else {
 100                                                 $replace = false;
 101                                         }
 102                                 } else {
 103                                         if (@$attr[$attribute] != $value) {
 104                                                 $replace = false;
 105                                         }
 106                                 }
 107                         }
 108
 109                         if ($replace) {
 110                                 $StartCode = $doc->createTextNode($startbb);
 111                                 $EndCode = $doc->createTextNode($endbb);
 112
 113                                 $node->parentNode->insertBefore($StartCode, $node);
 114
 115                                 if (!$ignoreChildren && $node->hasChildNodes()) {
 116                                         /** @var \DOMNode $child */
 117                                         foreach ($node->childNodes as $key => $child) {
 118                                                 /* Remove empty text nodes at the start or at the end of the children list */
 119                                                 if ($key > 0 && $key < $node->childNodes->length - 1 || $child->nodeName != '#text' || trim($child->nodeValue) !== '') {
 120                                                         $newNode = $child->cloneNode(true);
 121                                                         $node->parentNode->insertBefore($newNode, $node);
 122                                                 }
 123                                         }
 124                                 }
 125
 126                                 $node->parentNode->insertBefore($EndCode, $node);
 127                                 $node->parentNode->removeChild($node);
 128                         }
 129                 }
 130
 131                 return $replace;
 132         }
 133
 134         /**
 135          * Converter for HTML to BBCode
 136          *
 137          * Made by: ike@piratenpartei.de
 138          * Originally made for the syncom project: http://wiki.piratenpartei.de/Syncom
 139          *                    https://github.com/annando/Syncom
 140          *
 141          * @param string $message
 142          * @param string $basepath
 143          * @return string
 144          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 145          */
 146         public static function toBBCode(string $message, string $basepath = ''): string
 147         {
 148                 /*
 149                  * Check if message is empty to prevent a lot of code below from being executed
 150                  * for just an empty message.
 151                  */
 152                 if ($message === '') {
 153                         return '';
 154                 }
 155
 156                 DI::profiler()->startRecording('rendering');
 157                 $message = str_replace("\r", "", $message);
 158
 159                 $message = Strings::performWithEscapedBlocks($message, '#<pre><code.*</code></pre>#iUs', function ($message) {
 160                         $message = str_replace(
 161                                 [
 162                                         "<li><p>",
 163                                         "</p></li>",
 164                                 ],
 165                                 [
 166                                         "<li>",
 167                                         "</li>",
 168                                 ],
 169                                 $message
 170                         );
 171
 172                         // remove namespaces
 173                         $message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
 174                         $message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
 175
 176                         $doc = new DOMDocument();
 177                         $doc->preserveWhiteSpace = false;
 178
 179                         $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 180
 181                         if (empty($message)) {
 182                                 return '';
 183                         }
 184
 185                         @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 186
 187                         XML::deleteNode($doc, 'style');
 188                         XML::deleteNode($doc, 'head');
 189                         XML::deleteNode($doc, 'title');
 190                         XML::deleteNode($doc, 'meta');
 191                         XML::deleteNode($doc, 'xml');
 192                         XML::deleteNode($doc, 'removeme');
 193
 194                         $xpath = new DomXPath($doc);
 195                         $list = $xpath->query("//pre");
 196                         foreach ($list as $node) {
 197                                 // Ensure to escape unescaped & - they will otherwise raise a warning
 198                                 $safe_value = preg_replace('/&(?!\w+;)/', '&amp;', $node->nodeValue);
 199                                 $node->nodeValue = str_replace("\n", "\r", $safe_value);
 200                         }
 201
 202                         $message = $doc->saveHTML();
 203                         $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br />", " ", ""], $message);
 204                         $message = preg_replace('= [\s]*=i', " ", $message);
 205
 206                         if (empty($message)) {
 207                                 return '';
 208                         }
 209
 210                         @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 211
 212                         self::tagToBBCode($doc, 'html', [], "", "");
 213                         self::tagToBBCode($doc, 'body', [], "", "");
 214
 215                         // Outlook-Quote - Variant 1
 216                         self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal', 'style' => 'margin-left:35.4pt'], '[quote]', '[/quote]');
 217
 218                         // Outlook-Quote - Variant 2
 219                         self::tagToBBCode(
 220                                 $doc,
 221                                 'div',
 222                                 ['style' => 'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'],
 223                                 '[quote]',
 224                                 '[/quote]'
 225                         );
 226
 227                         // MyBB-Stuff
 228                         self::tagToBBCode($doc, 'span', ['style' => 'text-decoration: underline;'], '[u]', '[/u]');
 229                         self::tagToBBCode($doc, 'span', ['style' => 'font-style: italic;'], '[i]', '[/i]');
 230                         self::tagToBBCode($doc, 'span', ['style' => 'font-weight: bold;'], '[b]', '[/b]');
 231
 232                         /* self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
 233                           self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
 234                           self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
 235                           self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
 236                           self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
 237                           self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
 238                           self::node2BBCode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
 239                          */
 240                         // Untested
 241                         //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
 242                         //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
 243                         //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
 244
 245                         self::tagToBBCode($doc, 'span', ['style' => '/.*color:\s*(.+?)[,;].*/'], '[color="$1"]', '[/color]');
 246
 247                         //self::node2BBCode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 248                         //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
 249                         //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
 250                         //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 251                         // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
 252                         // Test
 253                         //self::node2BBCode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
 254                         self::tagToBBCode($doc, 'span', ['class' => 'type-link'], '[class=type-link]', '[/class]');
 255                         self::tagToBBCode($doc, 'span', ['class' => 'type-video'], '[class=type-video]', '[/class]');
 256
 257                         self::tagToBBCode($doc, 'strong', [], '[b]', '[/b]');
 258                         self::tagToBBCode($doc, 'em', [], '[i]', '[/i]');
 259                         self::tagToBBCode($doc, 'b', [], '[b]', '[/b]');
 260                         self::tagToBBCode($doc, 'i', [], '[i]', '[/i]');
 261                         self::tagToBBCode($doc, 'u', [], '[u]', '[/u]');
 262                         self::tagToBBCode($doc, 's', [], '[s]', '[/s]');
 263                         self::tagToBBCode($doc, 'del', [], '[s]', '[/s]');
 264                         self::tagToBBCode($doc, 'strike', [], '[s]', '[/s]');
 265
 266                         self::tagToBBCode($doc, 'big', [], "[size=large]", "[/size]");
 267                         self::tagToBBCode($doc, 'small', [], "[size=small]", "[/size]");
 268
 269                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', '[/quote]');
 270
 271                         self::tagToBBCode($doc, 'br', [], "\n", '');
 272
 273                         self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal'], "\n", "");
 274                         self::tagToBBCode($doc, 'div', ['class' => 'MsoNormal'], "\r", "");
 275
 276                         self::tagToBBCode($doc, 'span', [], "", "");
 277
 278                         self::tagToBBCode($doc, 'span', [], "", "");
 279                         self::tagToBBCode($doc, 'pre', [], "", "");
 280
 281                         self::tagToBBCode($doc, 'div', [], "\r", "\r");
 282                         self::tagToBBCode($doc, 'p', [], "\n", "\n");
 283
 284                         self::tagToBBCode($doc, 'ul', [], "[list]", "[/list]");
 285                         self::tagToBBCode($doc, 'ol', [], "[list=1]", "[/list]");
 286                         self::tagToBBCode($doc, 'li', [], "[*]", "");
 287
 288                         self::tagToBBCode($doc, 'hr', [], "[hr]", "");
 289
 290                         self::tagToBBCode($doc, 'table', [], "[table]", "[/table]");
 291                         self::tagToBBCode($doc, 'th', [], "[th]", "[/th]");
 292                         self::tagToBBCode($doc, 'tr', [], "[tr]", "[/tr]");
 293                         self::tagToBBCode($doc, 'td', [], "[td]", "[/td]");
 294
 295                         self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 296                         self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 297                         self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 298                         self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 299                         self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 300                         self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 301
 302                         self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 303                         self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 304
 305                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'alt' => '/(.+)/'], '[img=$1]$2', '[/img]', true);
 306                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]', true);
 307                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]', true);
 308
 309
 310                         self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]', true);
 311                         self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]', true);
 312                         // Backward compatibility, [iframe] support has been removed in version 2020.12
 313                         self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[url]$1', '[/url]', true);
 314
 315                         self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 316                         self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 317
 318                         $message = $doc->saveHTML();
 319
 320                         // I'm removing something really disturbing
 321                         // Don't know exactly what it is
 322                         $message = str_replace(chr(194) . chr(160), ' ', $message);
 323
 324                         $message = str_replace("&nbsp;", " ", $message);
 325
 326                         // removing multiple DIVs
 327                         $message = preg_replace('=\r *\r=i', "\n", $message);
 328                         $message = str_replace("\r", "\n", $message);
 329
 330                         Hook::callAll('html2bbcode', $message);
 331
 332                         $message = strip_tags($message);
 333
 334                         $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 335
 336                         // remove quotes if they don't make sense
 337                         $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 338
 339                         $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 340                         $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 341
 342                         do {
 343                                 $oldmessage = $message;
 344                                 $message = str_replace("\n \n", "\n\n", $message);
 345                         } while ($oldmessage != $message);
 346
 347                         do {
 348                                 $oldmessage = $message;
 349                                 $message = str_replace("\n\n\n", "\n\n", $message);
 350                         } while ($oldmessage != $message);
 351
 352                         do {
 353                                 $oldmessage = $message;
 354                                 $message = str_replace(
 355                                         [
 356                                                 "[/size]\n\n",
 357                                                 "\n[hr]",
 358                                                 "[hr]\n",
 359                                                 "\n[list",
 360                                                 "[/list]\n",
 361                                                 "\n[/",
 362                                                 "[list]\n",
 363                                                 "[list=1]\n",
 364                                                 "\n[*]"],
 365                                         [
 366                                                 "[/size]\n",
 367                                                 "[hr]",
 368                                                 "[hr]",
 369                                                 "[list",
 370                                                 "[/list]",
 371                                                 "[/",
 372                                                 "[list]",
 373                                                 "[list=1]",
 374                                                 "[*]"],
 375                                         $message
 376                                 );
 377                         } while ($message != $oldmessage);
 378
 379                         $message = str_replace(
 380                                 ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 381                                 ['[b]', '[/b]', '[i]', '[/i]'],
 382                                 $message
 383                         );
 384
 385                         // Handling Yahoo style of mails
 386                         $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 387
 388                         return $message;
 389                 });
 390
 391                 $message = preg_replace_callback(
 392                         '#<pre><code(?: class="language-([^"]*)")?>(.*)</code></pre>#iUs',
 393                         function ($matches) {
 394                                 $prefix = '[code]';
 395                                 if ($matches[1] != '') {
 396                                         $prefix = '[code=' . $matches[1] . ']';
 397                                 }
 398
 399                                 return $prefix . "\n" . html_entity_decode($matches[2]) . "\n" . '[/code]';
 400                         },
 401                         $message
 402                 );
 403
 404                 $message = trim($message);
 405
 406                 if ($basepath != '') {
 407                         $message = self::qualifyURLs($message, $basepath);
 408                 }
 409
 410                 DI::profiler()->stopRecording();
 411                 return $message;
 412         }
 413
 414         /**
 415          * Sub function to complete incomplete URL
 416          *
 417          * @param array  $matches  Result of preg_replace_callback
 418          * @param string $basepath Basepath that is used to complete the URL
 419          *
 420          * @return string The expanded URL
 421          */
 422         private static function qualifyURLsSub(array $matches, string $basepath): string
 423         {
 424                 $base = parse_url($basepath);
 425                 unset($base['query']);
 426                 unset($base['fragment']);
 427
 428                 $link = $matches[0];
 429                 $url = $matches[1];
 430
 431                 if (empty($url) || empty(parse_url($url))) {
 432                         return $matches[0];
 433                 }
 434
 435                 $parts = array_merge($base, parse_url($url));
 436                 $url2 = Network::unparseURL($parts);
 437
 438                 return str_replace($url, $url2, $link);
 439         }
 440
 441         /**
 442          * Complete incomplete URLs in BBCode
 443          *
 444          * @param string $body     Body with URLs
 445          * @param string $basepath Base path that is used to complete the URL
 446          *
 447          * @return string Body with expanded URLs
 448          */
 449         private static function qualifyURLs(string $body, string $basepath): string
 450         {
 451                 $URLSearchString = "^\[\]";
 452
 453                 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
 454                         "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
 455                         "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 456                         "/\[img\](.*?)\[\/img\]/ism",
 457                         "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 458                         "/\[zmg\](.*?)\[\/zmg\]/ism",
 459                         "/\[video\](.*?)\[\/video\]/ism",
 460                         "/\[audio\](.*?)\[\/audio\]/ism",
 461                 ];
 462
 463                 foreach ($matches as $match) {
 464                         $body = preg_replace_callback(
 465                                 $match,
 466                                 function ($match) use ($basepath) {
 467                                         return self::qualifyURLsSub($match, $basepath);
 468                                 },
 469                                 $body
 470                         );
 471                 }
 472                 return $body;
 473         }
 474
 475         private static function breakLines(string $line, int $level, int $wraplength = 75): string
 476         {
 477                 if ($wraplength == 0) {
 478                         $wraplength = 2000000;
 479                 }
 480
 481                 $wraplen = $wraplength - $level;
 482
 483                 $newlines = [];
 484
 485                 do {
 486                         $oldline = $line;
 487
 488                         $subline = substr($line, 0, $wraplen);
 489
 490                         $pos = strrpos($subline, ' ');
 491
 492                         if ($pos == 0) {
 493                                 $pos = strpos($line, ' ');
 494                         }
 495
 496                         if (($pos > 0) && strlen($line) > $wraplen) {
 497                                 $newline = trim(substr($line, 0, $pos));
 498                                 if ($level > 0) {
 499                                         $newline = str_repeat(">", $level) . ' ' . $newline;
 500                                 }
 501
 502                                 $newlines[] = $newline . " ";
 503                                 $line = substr($line, $pos + 1);
 504                         }
 505                 } while ((strlen($line) > $wraplen) && !($oldline == $line));
 506
 507                 if ($level > 0) {
 508                         $line = str_repeat(">", $level) . ' ' . $line;
 509                 }
 510
 511                 $newlines[] = $line;
 512
 513                 return implode("\n", $newlines);
 514         }
 515
 516         private static function quoteLevel(string $message, int $wraplength = 75): string
 517         {
 518                 $lines = explode("\n", $message);
 519
 520                 $newlines = [];
 521                 $level = 0;
 522                 foreach ($lines as $line) {
 523                         $line = trim($line);
 524                         $startquote = false;
 525                         while (strpos("*" . $line, '[quote]') > 0) {
 526                                 $level++;
 527                                 $pos = strpos($line, '[quote]');
 528                                 $line = substr($line, 0, $pos) . substr($line, $pos + 7);
 529                                 $startquote = true;
 530                         }
 531
 532                         $currlevel = $level;
 533
 534                         while (strpos("*" . $line, '[/quote]') > 0) {
 535                                 $level--;
 536                                 if ($level < 0) {
 537                                         $level = 0;
 538                                 }
 539
 540                                 $pos = strpos($line, '[/quote]');
 541                                 $line = substr($line, 0, $pos) . substr($line, $pos + 8);
 542                         }
 543
 544                         if (!$startquote || ($line != '')) {
 545                                 $newlines[] = self::breakLines($line, $currlevel, $wraplength);
 546                         }
 547                 }
 548
 549                 return implode("\n", $newlines);
 550         }
 551
 552         private static function collectURLs(string $message): array
 553         {
 554                 $pattern = '/<a.*?href="(.*?)".*?>(.*?)<\/a>/is';
 555                 preg_match_all($pattern, $message, $result, PREG_SET_ORDER);
 556
 557                 $urls = [];
 558                 foreach ($result as $treffer) {
 559                         $ignore = false;
 560
 561                         // A list of some links that should be ignored
 562                         $list = ["/user/", "/tag/", "/group/", "/profile/", "/search?search=", "/search?tag=", "mailto:", "/u/", "/node/",
 563                                 "//plus.google.com/", "//twitter.com/"];
 564                         foreach ($list as $listitem) {
 565                                 if (strpos($treffer[1], $listitem) !== false) {
 566                                         $ignore = true;
 567                                 }
 568                         }
 569
 570                         if ((strpos($treffer[1], "//twitter.com/") !== false) && (strpos($treffer[1], "/status/") !== false)) {
 571                                 $ignore = false;
 572                         }
 573
 574                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/posts") !== false)) {
 575                                 $ignore = false;
 576                         }
 577
 578                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/photos") !== false)) {
 579                                 $ignore = false;
 580                         }
 581
 582                         $ignore = $ignore || strpos($treffer[1], '#') === 0;
 583
 584                         if (!$ignore) {
 585                                 $urls[$treffer[1]] = $treffer[1];
 586                         }
 587                 }
 588
 589                 return $urls;
 590         }
 591
 592         /**
 593          * @param string $html
 594          * @param int    $wraplength Ensures individual lines aren't longer than this many characters. Doesn't break words.
 595          * @param bool   $compact    True: Completely strips image tags; False: Keeps image URLs
 596          * @return string
 597          */
 598         public static function toPlaintext(string $html, int $wraplength = 75, bool $compact = false): string
 599         {
 600                 DI::profiler()->startRecording('rendering');
 601                 $message = str_replace("\r", "", $html);
 602
 603                 $doc = new DOMDocument();
 604                 $doc->preserveWhiteSpace = false;
 605
 606                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 607
 608                 if (empty($message)) {
 609                         DI::profiler()->stopRecording();
 610                         return '';
 611                 }
 612
 613                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 614
 615                 $message = $doc->saveHTML();
 616                 // Remove eventual UTF-8 BOM
 617                 $message = str_replace("\xC3\x82\xC2\xA0", "", $message);
 618
 619                 // Collecting all links
 620                 $urls = self::collectURLs($message);
 621
 622                 if (empty($message)) {
 623                         DI::profiler()->stopRecording();
 624                         return '';
 625                 }
 626
 627                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 628
 629                 self::tagToBBCode($doc, 'html', [], '', '');
 630                 self::tagToBBCode($doc, 'body', [], '', '');
 631
 632                 if ($compact) {
 633                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 634                 } else {
 635                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 636                 }
 637
 638                 self::tagToBBCode($doc, 'br', [], "\n", '');
 639
 640                 self::tagToBBCode($doc, 'span', [], "", "");
 641                 self::tagToBBCode($doc, 'pre', [], "", "");
 642                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 643                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 644
 645                 self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 646
 647                 self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
 648
 649                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 650                 self::tagToBBCode($doc, 'td', [], "\t", "");
 651
 652                 self::tagToBBCode($doc, 'h1', [], "\n\n*", "*\n");
 653                 self::tagToBBCode($doc, 'h2', [], "\n\n*", "*\n");
 654                 self::tagToBBCode($doc, 'h3', [], "\n\n*", "*\n");
 655                 self::tagToBBCode($doc, 'h4', [], "\n\n*", "*\n");
 656                 self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
 657                 self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 658
 659                 if (!$compact) {
 660                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
 661                 } else {
 662                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' ', ' ');
 663                 }
 664
 665                 // Backward compatibility, [iframe] support has been removed in version 2020.12
 666                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], ' $1 ', '');
 667
 668                 $message = $doc->saveHTML();
 669
 670                 if (!$compact) {
 671                         $message = str_replace("[img]", "", $message);
 672                         $message = str_replace("[/img]", "", $message);
 673                 }
 674
 675                 // was ersetze ich da?
 676                 // Irgendein stoerrisches UTF-Zeug
 677                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 678
 679                 $message = str_replace("&nbsp;", " ", $message);
 680
 681                 // Aufeinanderfolgende DIVs
 682                 $message = preg_replace('=\r *\r=i', "\n", $message);
 683                 $message = str_replace("\r", "\n", $message);
 684
 685                 $message = strip_tags($message);
 686
 687                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 688
 689                 if (!$compact && ($message != '')) {
 690                         foreach ($urls as $id => $url) {
 691                                 if ($url != '' && strpos($message, $url) === false) {
 692                                         $message .= "\n" . $url . ' ';
 693                                 }
 694                         }
 695                 }
 696
 697                 $message = str_replace("\n«", "«\n", $message);
 698                 $message = str_replace("»\n", "\n»", $message);
 699
 700                 do {
 701                         $oldmessage = $message;
 702                         $message = str_replace("\n\n\n", "\n\n", $message);
 703                 } while ($oldmessage != $message);
 704
 705                 $message = self::quoteLevel(trim($message), $wraplength);
 706
 707                 DI::profiler()->stopRecording();
 708                 return trim($message);
 709         }
 710
 711         /**
 712          * Converts provided HTML code to Markdown. The hardwrap parameter maximizes
 713          * compatibility with Diaspora in spite of the Markdown standards.
 714          *
 715          * @param string $html
 716          * @return string
 717          */
 718         public static function toMarkdown(string $html): string
 719         {
 720                 DI::profiler()->startRecording('rendering');
 721                 $converter = new HtmlConverter(['hard_break' => true]);
 722                 $markdown = $converter->convert($html);
 723
 724                 DI::profiler()->stopRecording();
 725                 return $markdown;
 726         }
 727
 728         /**
 729          * Convert video HTML to BBCode tags
 730          *
 731          * @param string $s
 732          * @return string
 733          */
 734         public static function toBBCodeVideo(string $s): string
 735         {
 736                 $s = preg_replace(
 737                         '#<object[^>]+>(.*?)https?://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+)(.*?)</object>#ism',
 738                         '[youtube]$2[/youtube]',
 739                         $s
 740                 );
 741
 742                 $s = preg_replace(
 743                         '#<iframe[^>](.*?)https?://www.youtube.com/embed/([A-Za-z0-9\-_=]+)(.*?)</iframe>#ism',
 744                         '[youtube]$2[/youtube]',
 745                         $s
 746                 );
 747
 748                 $s = preg_replace(
 749                         '#<iframe[^>](.*?)https?://player.vimeo.com/video/([0-9]+)(.*?)</iframe>#ism',
 750                         '[vimeo]$2[/vimeo]',
 751                         $s
 752                 );
 753
 754                 return $s;
 755         }
 756
 757         /**
 758          * transform link href and img src from relative to absolute
 759          *
 760          * @param string $text
 761          * @param string $base base url
 762          * @return string
 763          */
 764         public static function relToAbs(string $text, string $base): string
 765         {
 766                 if (empty($base)) {
 767                         return $text;
 768                 }
 769
 770                 $base = rtrim($base, '/');
 771
 772                 $base2 = $base . "/";
 773
 774                 // Replace links
 775                 $pattern = "/<a([^>]*) href=\"(?!http|https|\/)([^\"]*)\"/";
 776                 $replace = "<a\${1} href=\"" . $base2 . "\${2}\"";
 777                 $text = preg_replace($pattern, $replace, $text);
 778
 779                 $pattern = "/<a([^>]*) href=\"(?!http|https)([^\"]*)\"/";
 780                 $replace = "<a\${1} href=\"" . $base . "\${2}\"";
 781                 $text = preg_replace($pattern, $replace, $text);
 782
 783                 // Replace images
 784                 $pattern = "/<img([^>]*) src=\"(?!http|https|\/)([^\"]*)\"/";
 785                 $replace = "<img\${1} src=\"" . $base2 . "\${2}\"";
 786                 $text = preg_replace($pattern, $replace, $text);
 787
 788                 $pattern = "/<img([^>]*) src=\"(?!http|https)([^\"]*)\"/";
 789                 $replace = "<img\${1} src=\"" . $base . "\${2}\"";
 790                 $text = preg_replace($pattern, $replace, $text);
 791
 792
 793                 // Done
 794                 return $text;
 795         }
 796
 797         /**
 798          * Loader for infinite scrolling
 799          *
 800          * @return string html for loader
 801          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 802          */
 803         public static function scrollLoader(): string
 804         {
 805                 $tpl = Renderer::getMarkupTemplate("scroll_loader.tpl");
 806                 return Renderer::replaceMacros($tpl, [
 807                         'wait' => DI::l10n()->t('Loading more entries...'),
 808                         'end' => DI::l10n()->t('The end')
 809                 ]);
 810         }
 811
 812         /**
 813          * Format contacts as picture links or as text links
 814          *
 815          * @param array   $contact  Array with contacts which contains an array with
 816          *                          int 'id' => The ID of the contact
 817          *                          int 'uid' => The user ID of the user who owns this data
 818          *                          string 'name' => The name of the contact
 819          *                          string 'url' => The url to the profile page of the contact
 820          *                          string 'addr' => The webbie of the contact (e.g.) username@friendica.com
 821          *                          string 'network' => The network to which the contact belongs to
 822          *                          string 'thumb' => The contact picture
 823          *                          string 'click' => js code which is performed when clicking on the contact
 824          * @param boolean $redirect If true try to use the redir url if it's possible
 825          * @param string  $class    CSS class for the
 826          * @param boolean $textmode If true display the contacts as text links
 827          *                          if false display the contacts as picture links
 828          * @return string Formatted html
 829          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 830          * @throws \ImagickException
 831          */
 832         public static function micropro(array $contact, bool $redirect = false, string $class = '', bool $textmode = false): string
 833         {
 834                 // Use the contact URL if no address is available
 835                 if (empty($contact['addr'])) {
 836                         $contact["addr"] = $contact["url"];
 837                 }
 838
 839                 $url = $contact['url'];
 840                 $sparkle = '';
 841                 $redir = false;
 842
 843                 if ($redirect) {
 844                         $url = Contact::magicLinkByContact($contact);
 845                         if (strpos($url, 'contact/redir/') === 0) {
 846                                 $sparkle = ' sparkle';
 847                         }
 848                 }
 849
 850                 // If there is some js available we don't need the url
 851                 if (!empty($contact['click'])) {
 852                         $url = '';
 853                 }
 854
 855                 return Renderer::replaceMacros(Renderer::getMarkupTemplate($textmode ? 'micropro_txt.tpl' : 'micropro_img.tpl'), [
 856                         '$click' => $contact['click'] ?? '',
 857                         '$class' => $class,
 858                         '$url' => $url,
 859                         '$photo' => Contact::getThumb($contact),
 860                         '$name' => $contact['name'],
 861                         'title' => $contact['name'] . ' [' . $contact['addr'] . ']',
 862                         '$parkle' => $sparkle,
 863                         '$redir' => $redir
 864                 ]);
 865         }
 866
 867         /**
 868          * Search box.
 869          *
 870          * @param string $s     Search query.
 871          * @param string $id    HTML id
 872          * @param bool   $aside Display the search widgit aside.
 873          *
 874          * @return string Formatted HTML.
 875          * @throws \Exception
 876          */
 877         public static function search(string $s, string $id = 'search-box', bool $aside = true): string
 878         {
 879                 $mode = 'text';
 880
 881                 if (strpos($s, '#') === 0) {
 882                         $mode = 'tag';
 883                 }
 884                 $save_label = $mode === 'text' ? DI::l10n()->t('Save') : DI::l10n()->t('Follow');
 885
 886                 $values = [
 887                         '$s'            => $s,
 888                         '$q'            => urlencode($s),
 889                         '$id'           => $id,
 890                         '$search_label' => DI::l10n()->t('Search'),
 891                         '$save_label'   => $save_label,
 892                         '$search_hint'  => DI::l10n()->t('@name, !forum, #tags, content'),
 893                         '$mode'         => $mode,
 894                         '$return_url'   => urlencode(Search::getSearchPath($s)),
 895                 ];
 896
 897                 if (!$aside) {
 898                         $values['$search_options'] = [
 899                                 'fulltext' => DI::l10n()->t('Full Text'),
 900                                 'tags'     => DI::l10n()->t('Tags'),
 901                                 'contacts' => DI::l10n()->t('Contacts')
 902                         ];
 903
 904                         if (DI::config()->get('system', 'poco_local_search')) {
 905                                 $values['$searchoption']['forums'] = DI::l10n()->t('Forums');
 906                         }
 907                 }
 908
 909                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 910         }
 911
 912         /**
 913          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
 914          *
 915          * Reasons are expected to have been translated already.
 916          *
 917          * @param string $html
 918          * @param array  $reasons
 919          * @return string
 920          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 921          */
 922         public static function applyContentFilter(string $html, array $reasons): string
 923         {
 924                 if (count($reasons)) {
 925                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
 926                         $html = Renderer::replaceMacros($tpl, [
 927                                 '$reasons'   => $reasons,
 928                                 '$rnd'       => Strings::getRandomHex(8),
 929                                 '$openclose' => DI::l10n()->t('Click to open/close'),
 930                                 '$html'      => $html
 931                         ]);
 932                 }
 933
 934                 return $html;
 935         }
 936
 937         /**
 938          * replace html amp entity with amp char
 939          * @param string $s
 940          * @return string
 941          */
 942         public static function unamp(string $s): string
 943         {
 944                 return str_replace('&amp;', '&', $s);
 945         }
 946
 947         /**
 948          * Clean an HTML text for potentially harmful code
 949          *
 950          * @param string $text
 951          * @param array  $allowedIframeDomains List of allowed iframe source domains without the scheme
 952          * @return string
 953          */
 954         public static function purify(string $text, array $allowedIframeDomains = []): string
 955         {
 956                 // Allows cid: URL scheme
 957                 \HTMLPurifier_URISchemeRegistry::instance()->register('cid', new HTMLPurifier_URIScheme_cid());
 958
 959                 $config = \HTMLPurifier_HTML5Config::createDefault();
 960                 $config->set('HTML.Doctype', 'HTML5');
 961
 962                 // Used to remove iframe with src attribute filtered out
 963                 $config->set('AutoFormat.RemoveEmpty', true);
 964
 965                 $config->set('HTML.SafeIframe', true);
 966
 967                 array_walk($allowedIframeDomains, function (&$domain) {
 968                         // Allow the domain and all its eventual sub-domains
 969                         $domain = '(?:(?!-)[A-Za-z0-9-]{1,63}(?<!-)\.)*' . preg_quote(trim($domain, '/'), '%');
 970                 });
 971
 972                 $config->set('URI.SafeIframeRegexp',
 973                         '%^https://(?:
 974                                 ' . implode('|', $allowedIframeDomains) . '
 975                         )
 976                         (?:/|$) # Prevents bogus domains like youtube.com.fake.tld
 977                         %xi'
 978                 );
 979
 980                 $config->set('Attr.AllowedRel', [
 981                         'noreferrer' => true,
 982                         'noopener' => true,
 983                         'tag' => true,
 984                 ]);
 985                 $config->set('Attr.AllowedFrameTargets', [
 986                         '_blank' => true,
 987                 ]);
 988
 989                 $config->set('AutoFormat.RemoveEmpty.Predicate', [
 990                         'colgroup' => [],        // |
 991                         'th'       => [],        // |
 992                         'td'       => [],        // |
 993                         'iframe'   => ['src'],   // ↳ Default HTMLPurify values
 994                         'i'        => ['class'], // Allows forkawesome icons
 995                 ]);
 996
 997                 // Uncomment to debug HTMLPurifier behavior
 998                 //$config->set('Core.CollectErrors', true);
 999                 //$config->set('Core.MaintainLineNumbers', true);
1000
1001                 $HTMLPurifier = new \HTMLPurifier($config);
1002
1003                 $text = $HTMLPurifier->purify($text);
1004
1005                 /** @var \HTMLPurifier_ErrorCollector $errorCollector */
1006                 // Uncomment to debug HTML Purifier behavior
1007                 //$errorCollector = $HTMLPurifier->context->get('ErrorCollector');
1008                 //var_dump($errorCollector->getRaw());
1009
1010                 return $text;
1011         }
1012
1013         /**
1014          * XPath arbitrary string quoting
1015          *
1016          * @see https://stackoverflow.com/a/45228168
1017          * @param string $value
1018          * @return string
1019          */
1020         public static function xpathQuote(string $value): string
1021         {
1022                 if (false === strpos($value, '"')) {
1023                         return '"' . $value . '"';
1024                 }
1025
1026                 if (false === strpos($value, "'")) {
1027                         return "'" . $value . "'";
1028                 }
1029
1030                 // if the value contains both single and double quotes, construct an
1031                 // expression that concatenates all non-double-quote substrings with
1032                 // the quotes, e.g.:
1033                 //
1034                 //    concat("'foo'", '"', "bar")
1035                 return 'concat(' . implode(', \'"\', ', array_map(['self', 'xpathQuote'], explode('"', $value))) . ')';
1036         }
1037
1038         /**
1039          * Checks if the provided URL is present in the DOM document in an element with the rel="me" attribute
1040          *
1041          * XHTML Friends Network http://gmpg.org/xfn/
1042          *
1043          * @param DOMDocument  $doc
1044          * @param UriInterface $meUrl
1045          * @return bool
1046          */
1047         public static function checkRelMeLink(DOMDocument $doc, UriInterface $meUrl): bool
1048         {
1049                 $xpath = new \DOMXpath($doc);
1050
1051                 // This expression checks that "me" is among the space-delimited values of the "rel" attribute.
1052                 // And that the href attribute contains exactly the provided URL
1053                 $expression = "//*[contains(concat(' ', normalize-space(@rel), ' '), ' me ')][@href = " . self::xpathQuote($meUrl) . "]";
1054
1055                 $result = $xpath->query($expression);
1056
1057                 return $result !== false && $result->length > 0;
1058         }
1059
1060         /**
1061          * @param DOMDocument $doc
1062          * @return string|null Lowercase charset
1063          */
1064         public static function extractCharset(DOMDocument $doc): ?string
1065         {
1066                 $xpath = new DOMXPath($doc);
1067
1068                 $expression = "string(//meta[@charset]/@charset)";
1069                 if ($charset = $xpath->evaluate($expression)) {
1070                         return strtolower($charset);
1071                 }
1072
1073                 try {
1074                         // This expression looks for a meta tag with the http-equiv attribute set to "content-type" ignoring case
1075                         // whose content attribute contains a "charset" string and returns its value
1076                         $expression = "string(//meta[@http-equiv][translate(@http-equiv, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz') = 'content-type'][contains(translate(@content, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ', 'abcdefghijklmnopqrstuvwxyz'), 'charset')]/@content)";
1077                         $mediaType = MediaType::fromContentType($xpath->evaluate($expression));
1078                         if (isset($mediaType->parameters['charset'])) {
1079                                 return strtolower($mediaType->parameters['charset']);
1080                         }
1081                 } catch(\InvalidArgumentException $e) {}
1082
1083                 return null;
1084         }
1085 }