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