src/Content/Text/HTML.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2022, 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
  37 class HTML
  38 {
  39         /**
  40          * Search all instances of a specific HTML tag node in the provided DOM document and replaces them with BBCode text nodes.
  41          *
  42          * @see HTML::tagToBBCodeSub()
  43          */
  44         private static function tagToBBCode(DOMDocument $doc, string $tag, array $attributes, string $startbb, string $endbb, bool $ignoreChildren = false)
  45         {
  46                 do {
  47                         $done = self::tagToBBCodeSub($doc, $tag, $attributes, $startbb, $endbb, $ignoreChildren);
  48                 } while ($done);
  49         }
  50
  51         /**
  52          * Search the first specific HTML tag node in the provided DOM document and replaces it with BBCode text nodes.
  53          *
  54          * @param DOMDocument $doc
  55          * @param string      $tag            HTML tag name
  56          * @param array       $attributes     Array of attributes to match and optionally use the value from
  57          * @param string      $startbb        BBCode tag opening
  58          * @param string      $endbb          BBCode tag closing
  59          * @param bool        $ignoreChildren If set to false, the HTML tag children will be appended as text inside the BBCode tag
  60          *                                    Otherwise, they will be entirely ignored. Useful for simple BBCode that draw their
  61          *                                    inner value from an attribute value and disregard the tag children.
  62          * @return bool Whether a replacement was done
  63          */
  64         private static function tagToBBCodeSub(DOMDocument $doc, string $tag, array $attributes, string $startbb, string $endbb, bool $ignoreChildren = false)
  65         {
  66                 $savestart = str_replace('$', '\x01', $startbb);
  67                 $replace = false;
  68
  69                 $xpath = new DOMXPath($doc);
  70
  71                 /** @var \DOMNode[] $list */
  72                 $list = $xpath->query("//" . $tag);
  73                 foreach ($list as $node) {
  74                         $attr = [];
  75                         if ($node->attributes->length) {
  76                                 foreach ($node->attributes as $attribute) {
  77                                         $attr[$attribute->name] = $attribute->value;
  78                                 }
  79                         }
  80
  81                         $replace = true;
  82
  83                         $startbb = $savestart;
  84
  85                         $i = 0;
  86
  87                         foreach ($attributes as $attribute => $value) {
  88                                 $startbb = str_replace('\x01' . ++$i, '$1', $startbb);
  89                                 if (strpos('*' . $startbb, '$1') > 0) {
  90                                         if ($replace && (@$attr[$attribute] != '')) {
  91                                                 $startbb = preg_replace($value, $startbb, $attr[$attribute], -1, $count);
  92
  93                                                 // If nothing could be changed
  94                                                 if ($count == 0) {
  95                                                         $replace = false;
  96                                                 }
  97                                         } else {
  98                                                 $replace = false;
  99                                         }
 100                                 } else {
 101                                         if (@$attr[$attribute] != $value) {
 102                                                 $replace = false;
 103                                         }
 104                                 }
 105                         }
 106
 107                         if ($replace) {
 108                                 $StartCode = $doc->createTextNode($startbb);
 109                                 $EndCode = $doc->createTextNode($endbb);
 110
 111                                 $node->parentNode->insertBefore($StartCode, $node);
 112
 113                                 if (!$ignoreChildren && $node->hasChildNodes()) {
 114                                         /** @var \DOMNode $child */
 115                                         foreach ($node->childNodes as $key => $child) {
 116                                                 /* Remove empty text nodes at the start or at the end of the children list */
 117                                                 if ($key > 0 && $key < $node->childNodes->length - 1 || $child->nodeName != '#text' || trim($child->nodeValue)) {
 118                                                         $newNode = $child->cloneNode(true);
 119                                                         $node->parentNode->insertBefore($newNode, $node);
 120                                                 }
 121                                         }
 122                                 }
 123
 124                                 $node->parentNode->insertBefore($EndCode, $node);
 125                                 $node->parentNode->removeChild($node);
 126                         }
 127                 }
 128
 129                 return $replace;
 130         }
 131
 132         /**
 133          * Converter for HTML to BBCode
 134          *
 135          * Made by: ike@piratenpartei.de
 136          * Originally made for the syncom project: http://wiki.piratenpartei.de/Syncom
 137          *                    https://github.com/annando/Syncom
 138          *
 139          * @param string $message
 140          * @param string $basepath
 141          * @return string
 142          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 143          */
 144         public static function toBBCode($message, $basepath = '')
 145         {
 146                 DI::profiler()->startRecording('rendering');
 147                 $message = str_replace("\r", "", $message);
 148
 149                 $message = Strings::performWithEscapedBlocks($message, '#<pre><code.*</code></pre>#iUs', function ($message) {
 150                         $message = str_replace(
 151                                 [
 152                                         "<li><p>",
 153                                         "</p></li>",
 154                                 ],
 155                                 [
 156                                         "<li>",
 157                                         "</li>",
 158                                 ],
 159                                 $message
 160                         );
 161
 162                         // remove namespaces
 163                         $message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
 164                         $message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
 165
 166                         $doc = new DOMDocument();
 167                         $doc->preserveWhiteSpace = false;
 168
 169                         $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 170
 171                         if (empty($message)) {
 172                                 return '';
 173                         }
 174
 175                         @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 176
 177                         XML::deleteNode($doc, 'style');
 178                         XML::deleteNode($doc, 'head');
 179                         XML::deleteNode($doc, 'title');
 180                         XML::deleteNode($doc, 'meta');
 181                         XML::deleteNode($doc, 'xml');
 182                         XML::deleteNode($doc, 'removeme');
 183
 184                         $xpath = new DomXPath($doc);
 185                         $list = $xpath->query("//pre");
 186                         foreach ($list as $node) {
 187                                 // Ensure to escape unescaped & - they will otherwise raise a warning
 188                                 $safe_value = preg_replace('/&(?!\w+;)/', '&amp;', $node->nodeValue);
 189                                 $node->nodeValue = str_replace("\n", "\r", $safe_value);
 190                         }
 191
 192                         $message = $doc->saveHTML();
 193                         $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br />", " ", ""], $message);
 194                         $message = preg_replace('= [\s]*=i', " ", $message);
 195
 196                         if (empty($message)) {
 197                                 return '';
 198                         }
 199
 200                         @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 201
 202                         self::tagToBBCode($doc, 'html', [], "", "");
 203                         self::tagToBBCode($doc, 'body', [], "", "");
 204
 205                         // Outlook-Quote - Variant 1
 206                         self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal', 'style' => 'margin-left:35.4pt'], '[quote]', '[/quote]');
 207
 208                         // Outlook-Quote - Variant 2
 209                         self::tagToBBCode(
 210                                 $doc,
 211                                 'div',
 212                                 ['style' => 'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'],
 213                                 '[quote]',
 214                                 '[/quote]'
 215                         );
 216
 217                         // MyBB-Stuff
 218                         self::tagToBBCode($doc, 'span', ['style' => 'text-decoration: underline;'], '[u]', '[/u]');
 219                         self::tagToBBCode($doc, 'span', ['style' => 'font-style: italic;'], '[i]', '[/i]');
 220                         self::tagToBBCode($doc, 'span', ['style' => 'font-weight: bold;'], '[b]', '[/b]');
 221
 222                         /* self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
 223                           self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
 224                           self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
 225                           self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
 226                           self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
 227                           self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
 228                           self::node2BBCode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
 229                          */
 230                         // Untested
 231                         //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
 232                         //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
 233                         //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
 234
 235                         self::tagToBBCode($doc, 'span', ['style' => '/.*color:\s*(.+?)[,;].*/'], '[color="$1"]', '[/color]');
 236
 237                         //self::node2BBCode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 238                         //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
 239                         //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
 240                         //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 241                         // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
 242                         // Test
 243                         //self::node2BBCode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
 244                         self::tagToBBCode($doc, 'span', ['class' => 'type-link'], '[class=type-link]', '[/class]');
 245                         self::tagToBBCode($doc, 'span', ['class' => 'type-video'], '[class=type-video]', '[/class]');
 246
 247                         self::tagToBBCode($doc, 'strong', [], '[b]', '[/b]');
 248                         self::tagToBBCode($doc, 'em', [], '[i]', '[/i]');
 249                         self::tagToBBCode($doc, 'b', [], '[b]', '[/b]');
 250                         self::tagToBBCode($doc, 'i', [], '[i]', '[/i]');
 251                         self::tagToBBCode($doc, 'u', [], '[u]', '[/u]');
 252                         self::tagToBBCode($doc, 's', [], '[s]', '[/s]');
 253                         self::tagToBBCode($doc, 'del', [], '[s]', '[/s]');
 254                         self::tagToBBCode($doc, 'strike', [], '[s]', '[/s]');
 255
 256                         self::tagToBBCode($doc, 'big', [], "[size=large]", "[/size]");
 257                         self::tagToBBCode($doc, 'small', [], "[size=small]", "[/size]");
 258
 259                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', '[/quote]');
 260
 261                         self::tagToBBCode($doc, 'br', [], "\n", '');
 262
 263                         self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal'], "\n", "");
 264                         self::tagToBBCode($doc, 'div', ['class' => 'MsoNormal'], "\r", "");
 265
 266                         self::tagToBBCode($doc, 'span', [], "", "");
 267
 268                         self::tagToBBCode($doc, 'span', [], "", "");
 269                         self::tagToBBCode($doc, 'pre', [], "", "");
 270
 271                         self::tagToBBCode($doc, 'div', [], "\r", "\r");
 272                         self::tagToBBCode($doc, 'p', [], "\n", "\n");
 273
 274                         self::tagToBBCode($doc, 'ul', [], "[list]", "[/list]");
 275                         self::tagToBBCode($doc, 'ol', [], "[list=1]", "[/list]");
 276                         self::tagToBBCode($doc, 'li', [], "[*]", "");
 277
 278                         self::tagToBBCode($doc, 'hr', [], "[hr]", "");
 279
 280                         self::tagToBBCode($doc, 'table', [], "[table]", "[/table]");
 281                         self::tagToBBCode($doc, 'th', [], "[th]", "[/th]");
 282                         self::tagToBBCode($doc, 'tr', [], "[tr]", "[/tr]");
 283                         self::tagToBBCode($doc, 'td', [], "[td]", "[/td]");
 284
 285                         self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 286                         self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 287                         self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 288                         self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 289                         self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 290                         self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 291
 292                         self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 293                         self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 294
 295                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'alt' => '/(.+)/'], '[img=$1]$2', '[/img]', true);
 296                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]', true);
 297                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]', true);
 298
 299
 300                         self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]', true);
 301                         self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]', true);
 302                         // Backward compatibility, [iframe] support has been removed in version 2020.12
 303                         self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[url]$1', '[/url]', true);
 304
 305                         self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 306                         self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 307
 308                         $message = $doc->saveHTML();
 309
 310                         // I'm removing something really disturbing
 311                         // Don't know exactly what it is
 312                         $message = str_replace(chr(194) . chr(160), ' ', $message);
 313
 314                         $message = str_replace("&nbsp;", " ", $message);
 315
 316                         // removing multiple DIVs
 317                         $message = preg_replace('=\r *\r=i', "\n", $message);
 318                         $message = str_replace("\r", "\n", $message);
 319
 320                         Hook::callAll('html2bbcode', $message);
 321
 322                         $message = strip_tags($message);
 323
 324                         $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 325
 326                         // remove quotes if they don't make sense
 327                         $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 328
 329                         $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 330                         $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 331
 332                         do {
 333                                 $oldmessage = $message;
 334                                 $message = str_replace("\n \n", "\n\n", $message);
 335                         } while ($oldmessage != $message);
 336
 337                         do {
 338                                 $oldmessage = $message;
 339                                 $message = str_replace("\n\n\n", "\n\n", $message);
 340                         } while ($oldmessage != $message);
 341
 342                         do {
 343                                 $oldmessage = $message;
 344                                 $message = str_replace(
 345                                         [
 346                                                 "[/size]\n\n",
 347                                                 "\n[hr]",
 348                                                 "[hr]\n",
 349                                                 "\n[list",
 350                                                 "[/list]\n",
 351                                                 "\n[/",
 352                                                 "[list]\n",
 353                                                 "[list=1]\n",
 354                                                 "\n[*]"],
 355                                         [
 356                                                 "[/size]\n",
 357                                                 "[hr]",
 358                                                 "[hr]",
 359                                                 "[list",
 360                                                 "[/list]",
 361                                                 "[/",
 362                                                 "[list]",
 363                                                 "[list=1]",
 364                                                 "[*]"],
 365                                         $message
 366                                 );
 367                         } while ($message != $oldmessage);
 368
 369                         $message = str_replace(
 370                                 ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 371                                 ['[b]', '[/b]', '[i]', '[/i]'],
 372                                 $message
 373                         );
 374
 375                         // Handling Yahoo style of mails
 376                         $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 377
 378                         return $message;
 379                 });
 380
 381                 $message = preg_replace_callback(
 382                         '#<pre><code(?: class="language-([^"]*)")?>(.*)</code></pre>#iUs',
 383                         function ($matches) {
 384                                 $prefix = '[code]';
 385                                 if ($matches[1] != '') {
 386                                         $prefix = '[code=' . $matches[1] . ']';
 387                                 }
 388
 389                                 return $prefix . "\n" . html_entity_decode($matches[2]) . "\n" . '[/code]';
 390                         },
 391                         $message
 392                 );
 393
 394                 $message = trim($message);
 395
 396                 if ($basepath != '') {
 397                         $message = self::qualifyURLs($message, $basepath);
 398                 }
 399
 400                 DI::profiler()->stopRecording();
 401                 return $message;
 402         }
 403
 404         /**
 405          * Sub function to complete incomplete URL
 406          *
 407          * @param array  $matches  Result of preg_replace_callback
 408          * @param string $basepath Basepath that is used to complete the URL
 409          *
 410          * @return string The expanded URL
 411          */
 412         private static function qualifyURLsSub($matches, $basepath)
 413         {
 414                 $base = parse_url($basepath);
 415                 unset($base['query']);
 416                 unset($base['fragment']);
 417
 418                 $link = $matches[0];
 419                 $url = $matches[1];
 420
 421                 if (empty($url) || empty(parse_url($url))) {
 422                         return $matches[0];
 423                 }
 424
 425                 $parts = array_merge($base, parse_url($url));
 426                 $url2 = Network::unparseURL($parts);
 427
 428                 return str_replace($url, $url2, $link);
 429         }
 430
 431         /**
 432          * Complete incomplete URLs in BBCode
 433          *
 434          * @param string $body     Body with URLs
 435          * @param string $basepath Base path that is used to complete the URL
 436          *
 437          * @return string Body with expanded URLs
 438          */
 439         private static function qualifyURLs($body, $basepath)
 440         {
 441                 $URLSearchString = "^\[\]";
 442
 443                 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
 444                         "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
 445                         "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 446                         "/\[img\](.*?)\[\/img\]/ism",
 447                         "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 448                         "/\[zmg\](.*?)\[\/zmg\]/ism",
 449                         "/\[video\](.*?)\[\/video\]/ism",
 450                         "/\[audio\](.*?)\[\/audio\]/ism",
 451                 ];
 452
 453                 foreach ($matches as $match) {
 454                         $body = preg_replace_callback(
 455                                 $match,
 456                                 function ($match) use ($basepath) {
 457                                         return self::qualifyURLsSub($match, $basepath);
 458                                 },
 459                                 $body
 460                         );
 461                 }
 462                 return $body;
 463         }
 464
 465         private static function breakLines($line, $level, $wraplength = 75)
 466         {
 467                 if ($wraplength == 0) {
 468                         $wraplength = 2000000;
 469                 }
 470
 471                 $wraplen = $wraplength - $level;
 472
 473                 $newlines = [];
 474
 475                 do {
 476                         $oldline = $line;
 477
 478                         $subline = substr($line, 0, $wraplen);
 479
 480                         $pos = strrpos($subline, ' ');
 481
 482                         if ($pos == 0) {
 483                                 $pos = strpos($line, ' ');
 484                         }
 485
 486                         if (($pos > 0) && strlen($line) > $wraplen) {
 487                                 $newline = trim(substr($line, 0, $pos));
 488                                 if ($level > 0) {
 489                                         $newline = str_repeat(">", $level) . ' ' . $newline;
 490                                 }
 491
 492                                 $newlines[] = $newline . " ";
 493                                 $line = substr($line, $pos + 1);
 494                         }
 495                 } while ((strlen($line) > $wraplen) && !($oldline == $line));
 496
 497                 if ($level > 0) {
 498                         $line = str_repeat(">", $level) . ' ' . $line;
 499                 }
 500
 501                 $newlines[] = $line;
 502
 503                 return implode("\n", $newlines);
 504         }
 505
 506         private static function quoteLevel($message, $wraplength = 75)
 507         {
 508                 $lines = explode("\n", $message);
 509
 510                 $newlines = [];
 511                 $level = 0;
 512                 foreach ($lines as $line) {
 513                         $line = trim($line);
 514                         $startquote = false;
 515                         while (strpos("*" . $line, '[quote]') > 0) {
 516                                 $level++;
 517                                 $pos = strpos($line, '[quote]');
 518                                 $line = substr($line, 0, $pos) . substr($line, $pos + 7);
 519                                 $startquote = true;
 520                         }
 521
 522                         $currlevel = $level;
 523
 524                         while (strpos("*" . $line, '[/quote]') > 0) {
 525                                 $level--;
 526                                 if ($level < 0) {
 527                                         $level = 0;
 528                                 }
 529
 530                                 $pos = strpos($line, '[/quote]');
 531                                 $line = substr($line, 0, $pos) . substr($line, $pos + 8);
 532                         }
 533
 534                         if (!$startquote || ($line != '')) {
 535                                 $newlines[] = self::breakLines($line, $currlevel, $wraplength);
 536                         }
 537                 }
 538
 539                 return implode("\n", $newlines);
 540         }
 541
 542         private static function collectURLs($message)
 543         {
 544                 $pattern = '/<a.*?href="(.*?)".*?>(.*?)<\/a>/is';
 545                 preg_match_all($pattern, $message, $result, PREG_SET_ORDER);
 546
 547                 $urls = [];
 548                 foreach ($result as $treffer) {
 549                         $ignore = false;
 550
 551                         // A list of some links that should be ignored
 552                         $list = ["/user/", "/tag/", "/group/", "/profile/", "/search?search=", "/search?tag=", "mailto:", "/u/", "/node/",
 553                                 "//plus.google.com/", "//twitter.com/"];
 554                         foreach ($list as $listitem) {
 555                                 if (strpos($treffer[1], $listitem) !== false) {
 556                                         $ignore = true;
 557                                 }
 558                         }
 559
 560                         if ((strpos($treffer[1], "//twitter.com/") !== false) && (strpos($treffer[1], "/status/") !== false)) {
 561                                 $ignore = false;
 562                         }
 563
 564                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/posts") !== false)) {
 565                                 $ignore = false;
 566                         }
 567
 568                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/photos") !== false)) {
 569                                 $ignore = false;
 570                         }
 571
 572                         $ignore = $ignore || strpos($treffer[1], '#') === 0;
 573
 574                         if (!$ignore) {
 575                                 $urls[$treffer[1]] = $treffer[1];
 576                         }
 577                 }
 578
 579                 return $urls;
 580         }
 581
 582         /**
 583          * @param string $html
 584          * @param int    $wraplength Ensures individual lines aren't longer than this many characters. Doesn't break words.
 585          * @param bool   $compact    True: Completely strips image tags; False: Keeps image URLs
 586          * @return string
 587          */
 588         public static function toPlaintext(string $html, $wraplength = 75, $compact = false)
 589         {
 590                 DI::profiler()->startRecording('rendering');
 591                 $message = str_replace("\r", "", $html);
 592
 593                 $doc = new DOMDocument();
 594                 $doc->preserveWhiteSpace = false;
 595
 596                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 597
 598                 if (empty($message)) {
 599                         DI::profiler()->stopRecording();
 600                         return '';
 601                 }
 602
 603                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 604
 605                 $message = $doc->saveHTML();
 606                 // Remove eventual UTF-8 BOM
 607                 $message = str_replace("\xC3\x82\xC2\xA0", "", $message);
 608
 609                 // Collecting all links
 610                 $urls = self::collectURLs($message);
 611
 612                 if (empty($message)) {
 613                         DI::profiler()->stopRecording();
 614                         return '';
 615                 }
 616
 617                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 618
 619                 self::tagToBBCode($doc, 'html', [], '', '');
 620                 self::tagToBBCode($doc, 'body', [], '', '');
 621
 622                 if ($compact) {
 623                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 624                 } else {
 625                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 626                 }
 627
 628                 self::tagToBBCode($doc, 'br', [], "\n", '');
 629
 630                 self::tagToBBCode($doc, 'span', [], "", "");
 631                 self::tagToBBCode($doc, 'pre', [], "", "");
 632                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 633                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 634
 635                 self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 636
 637                 self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
 638
 639                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 640                 self::tagToBBCode($doc, 'td', [], "\t", "");
 641
 642                 self::tagToBBCode($doc, 'h1', [], "\n\n*", "*\n");
 643                 self::tagToBBCode($doc, 'h2', [], "\n\n*", "*\n");
 644                 self::tagToBBCode($doc, 'h3', [], "\n\n*", "*\n");
 645                 self::tagToBBCode($doc, 'h4', [], "\n\n*", "*\n");
 646                 self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
 647                 self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 648
 649                 if (!$compact) {
 650                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
 651                 } else {
 652                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' ', ' ');
 653                 }
 654
 655                 // Backward compatibility, [iframe] support has been removed in version 2020.12
 656                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], ' $1 ', '');
 657
 658                 $message = $doc->saveHTML();
 659
 660                 if (!$compact) {
 661                         $message = str_replace("[img]", "", $message);
 662                         $message = str_replace("[/img]", "", $message);
 663                 }
 664
 665                 // was ersetze ich da?
 666                 // Irgendein stoerrisches UTF-Zeug
 667                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 668
 669                 $message = str_replace("&nbsp;", " ", $message);
 670
 671                 // Aufeinanderfolgende DIVs
 672                 $message = preg_replace('=\r *\r=i', "\n", $message);
 673                 $message = str_replace("\r", "\n", $message);
 674
 675                 $message = strip_tags($message);
 676
 677                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 678
 679                 if (!$compact && ($message != '')) {
 680                         foreach ($urls as $id => $url) {
 681                                 if ($url != '' && strpos($message, $url) === false) {
 682                                         $message .= "\n" . $url . ' ';
 683                                 }
 684                         }
 685                 }
 686
 687                 $message = str_replace("\n«", "«\n", $message);
 688                 $message = str_replace("»\n", "\n»", $message);
 689
 690                 do {
 691                         $oldmessage = $message;
 692                         $message = str_replace("\n\n\n", "\n\n", $message);
 693                 } while ($oldmessage != $message);
 694
 695                 $message = self::quoteLevel(trim($message), $wraplength);
 696
 697                 DI::profiler()->stopRecording();
 698                 return trim($message);
 699         }
 700
 701         /**
 702          * Converts provided HTML code to Markdown. The hardwrap parameter maximizes
 703          * compatibility with Diaspora in spite of the Markdown standards.
 704          *
 705          * @param string $html
 706          * @return string
 707          */
 708         public static function toMarkdown($html)
 709         {
 710                 DI::profiler()->startRecording('rendering');
 711                 $converter = new HtmlConverter(['hard_break' => true]);
 712                 $markdown = $converter->convert($html);
 713
 714                 DI::profiler()->stopRecording();
 715                 return $markdown;
 716         }
 717
 718         /**
 719          * Convert video HTML to BBCode tags
 720          *
 721          * @param string $s
 722          * @return string
 723          */
 724         public static function toBBCodeVideo($s)
 725         {
 726                 $s = preg_replace(
 727                         '#<object[^>]+>(.*?)https?://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+)(.*?)</object>#ism',
 728                         '[youtube]$2[/youtube]',
 729                         $s
 730                 );
 731
 732                 $s = preg_replace(
 733                         '#<iframe[^>](.*?)https?://www.youtube.com/embed/([A-Za-z0-9\-_=]+)(.*?)</iframe>#ism',
 734                         '[youtube]$2[/youtube]',
 735                         $s
 736                 );
 737
 738                 $s = preg_replace(
 739                         '#<iframe[^>](.*?)https?://player.vimeo.com/video/([0-9]+)(.*?)</iframe>#ism',
 740                         '[vimeo]$2[/vimeo]',
 741                         $s
 742                 );
 743
 744                 return $s;
 745         }
 746
 747         /**
 748          * transform link href and img src from relative to absolute
 749          *
 750          * @param string $text
 751          * @param string $base base url
 752          * @return string
 753          */
 754         public static function relToAbs($text, $base)
 755         {
 756                 if (empty($base)) {
 757                         return $text;
 758                 }
 759
 760                 $base = rtrim($base, '/');
 761
 762                 $base2 = $base . "/";
 763
 764                 // Replace links
 765                 $pattern = "/<a([^>]*) href=\"(?!http|https|\/)([^\"]*)\"/";
 766                 $replace = "<a\${1} href=\"" . $base2 . "\${2}\"";
 767                 $text = preg_replace($pattern, $replace, $text);
 768
 769                 $pattern = "/<a([^>]*) href=\"(?!http|https)([^\"]*)\"/";
 770                 $replace = "<a\${1} href=\"" . $base . "\${2}\"";
 771                 $text = preg_replace($pattern, $replace, $text);
 772
 773                 // Replace images
 774                 $pattern = "/<img([^>]*) src=\"(?!http|https|\/)([^\"]*)\"/";
 775                 $replace = "<img\${1} src=\"" . $base2 . "\${2}\"";
 776                 $text = preg_replace($pattern, $replace, $text);
 777
 778                 $pattern = "/<img([^>]*) src=\"(?!http|https)([^\"]*)\"/";
 779                 $replace = "<img\${1} src=\"" . $base . "\${2}\"";
 780                 $text = preg_replace($pattern, $replace, $text);
 781
 782
 783                 // Done
 784                 return $text;
 785         }
 786
 787         /**
 788          * Loader for infinite scrolling
 789          *
 790          * @return string html for loader
 791          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 792          */
 793         public static function scrollLoader()
 794         {
 795                 $tpl = Renderer::getMarkupTemplate("scroll_loader.tpl");
 796                 return Renderer::replaceMacros($tpl, [
 797                         'wait' => DI::l10n()->t('Loading more entries...'),
 798                         'end' => DI::l10n()->t('The end')
 799                 ]);
 800         }
 801
 802         /**
 803          * Format contacts as picture links or as text links
 804          *
 805          * @param array   $contact  Array with contacts which contains an array with
 806          *                          int 'id' => The ID of the contact
 807          *                          int 'uid' => The user ID of the user who owns this data
 808          *                          string 'name' => The name of the contact
 809          *                          string 'url' => The url to the profile page of the contact
 810          *                          string 'addr' => The webbie of the contact (e.g.) username@friendica.com
 811          *                          string 'network' => The network to which the contact belongs to
 812          *                          string 'thumb' => The contact picture
 813          *                          string 'click' => js code which is performed when clicking on the contact
 814          * @param boolean $redirect If true try to use the redir url if it's possible
 815          * @param string  $class    CSS class for the
 816          * @param boolean $textmode If true display the contacts as text links
 817          *                          if false display the contacts as picture links
 818          * @return string Formatted html
 819          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 820          * @throws \ImagickException
 821          */
 822         public static function micropro($contact, $redirect = false, $class = '', $textmode = false)
 823         {
 824                 // Use the contact URL if no address is available
 825                 if (empty($contact['addr'])) {
 826                         $contact["addr"] = $contact["url"];
 827                 }
 828
 829                 $url = $contact['url'];
 830                 $sparkle = '';
 831                 $redir = false;
 832
 833                 if ($redirect) {
 834                         $url = Contact::magicLinkByContact($contact);
 835                         if (strpos($url, 'redir/') === 0) {
 836                                 $sparkle = ' sparkle';
 837                         }
 838                 }
 839
 840                 // If there is some js available we don't need the url
 841                 if (!empty($contact['click'])) {
 842                         $url = '';
 843                 }
 844
 845                 return Renderer::replaceMacros(Renderer::getMarkupTemplate($textmode ? 'micropro_txt.tpl' : 'micropro_img.tpl'), [
 846                         '$click' => $contact['click'] ?? '',
 847                         '$class' => $class,
 848                         '$url' => $url,
 849                         '$photo' => Contact::getThumb($contact),
 850                         '$name' => $contact['name'],
 851                         'title' => $contact['name'] . ' [' . $contact['addr'] . ']',
 852                         '$parkle' => $sparkle,
 853                         '$redir' => $redir
 854                 ]);
 855         }
 856
 857         /**
 858          * Search box.
 859          *
 860          * @param string $s     Search query.
 861          * @param string $id    HTML id
 862          * @param string $url   Search url.
 863          * @param bool   $aside Display the search widgit aside.
 864          *
 865          * @return string Formatted HTML.
 866          * @throws \Exception
 867          */
 868         public static function search($s, $id = 'search-box', $aside = true)
 869         {
 870                 $mode = 'text';
 871
 872                 if (strpos($s, '#') === 0) {
 873                         $mode = 'tag';
 874                 }
 875                 $save_label = $mode === 'text' ? DI::l10n()->t('Save') : DI::l10n()->t('Follow');
 876
 877                 $values = [
 878                         '$s'            => $s,
 879                         '$q'            => urlencode($s),
 880                         '$id'           => $id,
 881                         '$search_label' => DI::l10n()->t('Search'),
 882                         '$save_label'   => $save_label,
 883                         '$search_hint'  => DI::l10n()->t('@name, !forum, #tags, content'),
 884                         '$mode'         => $mode,
 885                         '$return_url'   => urlencode(Search::getSearchPath($s)),
 886                 ];
 887
 888                 if (!$aside) {
 889                         $values['$search_options'] = [
 890                                 'fulltext' => DI::l10n()->t('Full Text'),
 891                                 'tags'     => DI::l10n()->t('Tags'),
 892                                 'contacts' => DI::l10n()->t('Contacts')
 893                         ];
 894
 895                         if (DI::config()->get('system', 'poco_local_search')) {
 896                                 $values['$searchoption']['forums'] = DI::l10n()->t('Forums');
 897                         }
 898                 }
 899
 900                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 901         }
 902
 903         /**
 904          * Replace naked text hyperlink with HTML formatted hyperlink
 905          *
 906          * @param string $s
 907          * @return string
 908          */
 909         public static function toLink($s)
 910         {
 911                 $s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank" rel="noopener noreferrer">$1</a>', $s);
 912                 $s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
 913                 return $s;
 914         }
 915
 916         /**
 917          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
 918          *
 919          * Reasons are expected to have been translated already.
 920          *
 921          * @param string $html
 922          * @param array  $reasons
 923          * @return string
 924          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 925          */
 926         public static function applyContentFilter($html, array $reasons)
 927         {
 928                 if (count($reasons)) {
 929                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
 930                         $html = Renderer::replaceMacros($tpl, [
 931                                 '$reasons'   => $reasons,
 932                                 '$rnd'       => Strings::getRandomHex(8),
 933                                 '$openclose' => DI::l10n()->t('Click to open/close'),
 934                                 '$html'      => $html
 935                         ]);
 936                 }
 937
 938                 return $html;
 939         }
 940
 941         /**
 942          * replace html amp entity with amp char
 943          * @param string $s
 944          * @return string
 945          */
 946         public static function unamp($s)
 947         {
 948                 return str_replace('&amp;', '&', $s);
 949         }
 950
 951         /**
 952          * Clean an HTML text for potentially harmful code
 953          *
 954          * @param string $text
 955          * @param array  $allowedIframeDomains List of allowed iframe source domains without the scheme
 956          * @return string
 957          */
 958         public static function purify(string $text, array $allowedIframeDomains = []): string
 959         {
 960                 // Allows cid: URL scheme
 961                 \HTMLPurifier_URISchemeRegistry::instance()->register('cid', new HTMLPurifier_URIScheme_cid());
 962
 963                 $config = \HTMLPurifier_HTML5Config::createDefault();
 964                 $config->set('HTML.Doctype', 'HTML5');
 965
 966                 // Used to remove iframe with src attribute filtered out
 967                 $config->set('AutoFormat.RemoveEmpty', true);
 968
 969                 $config->set('HTML.SafeIframe', true);
 970
 971                 array_walk($allowedIframeDomains, function (&$domain) {
 972                         // Allow the domain and all its eventual sub-domains
 973                         $domain = '(?:(?!-)[A-Za-z0-9-]{1,63}(?<!-)\.)*' . preg_quote(trim($domain, '/'), '%');
 974                 });
 975
 976                 $config->set('URI.SafeIframeRegexp',
 977                         '%^https://(?:
 978                                 ' . implode('|', $allowedIframeDomains) . '
 979                         )
 980                         (?:/|$) # Prevents bogus domains like youtube.com.fake.tld
 981                         %xi'
 982                 );
 983
 984                 $config->set('Attr.AllowedRel', [
 985                         'noreferrer' => true,
 986                         'noopener' => true,
 987                         'tag' => true,
 988                 ]);
 989                 $config->set('Attr.AllowedFrameTargets', [
 990                         '_blank' => true,
 991                 ]);
 992
 993                 $config->set('AutoFormat.RemoveEmpty.Predicate', [
 994                         'colgroup' => [],        // |
 995                         'th'       => [],        // |
 996                         'td'       => [],        // |
 997                         'iframe'   => ['src'],   // ↳ Default HTMLPurify values
 998                         'i'        => ['class'], // Allows forkawesome icons
 999                 ]);
1000
1001                 // Uncomment to debug HTMLPurifier behavior
1002                 //$config->set('Core.CollectErrors', true);
1003                 //$config->set('Core.MaintainLineNumbers', true);
1004
1005                 $HTMLPurifier = new \HTMLPurifier($config);
1006
1007                 $text = $HTMLPurifier->purify($text);
1008
1009                 /** @var \HTMLPurifier_ErrorCollector $errorCollector */
1010                 // Uncomment to debug HTML Purifier behavior
1011                 //$errorCollector = $HTMLPurifier->context->get('ErrorCollector');
1012                 //var_dump($errorCollector->getRaw());
1013
1014                 return $text;
1015         }
1016 }