src/Content/Text/HTML.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2021, 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                 $message = str_replace("\r", "", $message);
 147
 148                 $message = Strings::performWithEscapedBlocks($message, '#<pre><code.*</code></pre>#iUs', function ($message) {
 149                         $message = str_replace(
 150                                 [
 151                                         "<li><p>",
 152                                         "</p></li>",
 153                                 ],
 154                                 [
 155                                         "<li>",
 156                                         "</li>",
 157                                 ],
 158                                 $message
 159                         );
 160
 161                         // remove namespaces
 162                         $message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
 163                         $message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
 164
 165                         $doc = new DOMDocument();
 166                         $doc->preserveWhiteSpace = false;
 167
 168                         $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 169
 170                         if (empty($message)) {
 171                                 return '';
 172                         }
 173
 174                         @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 175
 176                         XML::deleteNode($doc, 'style');
 177                         XML::deleteNode($doc, 'head');
 178                         XML::deleteNode($doc, 'title');
 179                         XML::deleteNode($doc, 'meta');
 180                         XML::deleteNode($doc, 'xml');
 181                         XML::deleteNode($doc, 'removeme');
 182
 183                         $xpath = new DomXPath($doc);
 184                         $list = $xpath->query("//pre");
 185                         foreach ($list as $node) {
 186                                 // Ensure to escape unescaped & - they will otherwise raise a warning
 187                                 $safe_value = preg_replace('/&(?!\w+;)/', '&amp;', $node->nodeValue);
 188                                 $node->nodeValue = str_replace("\n", "\r", $safe_value);
 189                         }
 190
 191                         $message = $doc->saveHTML();
 192                         $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br />", " ", ""], $message);
 193                         $message = preg_replace('= [\s]*=i', " ", $message);
 194
 195                         if (empty($message)) {
 196                                 return '';
 197                         }
 198
 199                         @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 200
 201                         self::tagToBBCode($doc, 'html', [], "", "");
 202                         self::tagToBBCode($doc, 'body', [], "", "");
 203
 204                         // Outlook-Quote - Variant 1
 205                         self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal', 'style' => 'margin-left:35.4pt'], '[quote]', '[/quote]');
 206
 207                         // Outlook-Quote - Variant 2
 208                         self::tagToBBCode(
 209                                 $doc,
 210                                 'div',
 211                                 ['style' => 'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'],
 212                                 '[quote]',
 213                                 '[/quote]'
 214                         );
 215
 216                         // MyBB-Stuff
 217                         self::tagToBBCode($doc, 'span', ['style' => 'text-decoration: underline;'], '[u]', '[/u]');
 218                         self::tagToBBCode($doc, 'span', ['style' => 'font-style: italic;'], '[i]', '[/i]');
 219                         self::tagToBBCode($doc, 'span', ['style' => 'font-weight: bold;'], '[b]', '[/b]');
 220
 221                         /* self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
 222                           self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
 223                           self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
 224                           self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
 225                           self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
 226                           self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
 227                           self::node2BBCode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
 228                          */
 229                         // Untested
 230                         //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
 231                         //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
 232                         //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
 233
 234                         self::tagToBBCode($doc, 'span', ['style' => '/.*color:\s*(.+?)[,;].*/'], '[color="$1"]', '[/color]');
 235
 236                         //self::node2BBCode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 237                         //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
 238                         //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
 239                         //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 240                         // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
 241                         // Test
 242                         //self::node2BBCode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
 243                         self::tagToBBCode($doc, 'span', ['class' => 'type-link'], '[class=type-link]', '[/class]');
 244                         self::tagToBBCode($doc, 'span', ['class' => 'type-video'], '[class=type-video]', '[/class]');
 245
 246                         self::tagToBBCode($doc, 'strong', [], '[b]', '[/b]');
 247                         self::tagToBBCode($doc, 'em', [], '[i]', '[/i]');
 248                         self::tagToBBCode($doc, 'b', [], '[b]', '[/b]');
 249                         self::tagToBBCode($doc, 'i', [], '[i]', '[/i]');
 250                         self::tagToBBCode($doc, 'u', [], '[u]', '[/u]');
 251                         self::tagToBBCode($doc, 's', [], '[s]', '[/s]');
 252                         self::tagToBBCode($doc, 'del', [], '[s]', '[/s]');
 253                         self::tagToBBCode($doc, 'strike', [], '[s]', '[/s]');
 254
 255                         self::tagToBBCode($doc, 'big', [], "[size=large]", "[/size]");
 256                         self::tagToBBCode($doc, 'small', [], "[size=small]", "[/size]");
 257
 258                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', '[/quote]');
 259
 260                         self::tagToBBCode($doc, 'br', [], "\n", '');
 261
 262                         self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal'], "\n", "");
 263                         self::tagToBBCode($doc, 'div', ['class' => 'MsoNormal'], "\r", "");
 264
 265                         self::tagToBBCode($doc, 'span', [], "", "");
 266
 267                         self::tagToBBCode($doc, 'span', [], "", "");
 268                         self::tagToBBCode($doc, 'pre', [], "", "");
 269
 270                         self::tagToBBCode($doc, 'div', [], "\r", "\r");
 271                         self::tagToBBCode($doc, 'p', [], "\n", "\n");
 272
 273                         self::tagToBBCode($doc, 'ul', [], "[list]", "[/list]");
 274                         self::tagToBBCode($doc, 'ol', [], "[list=1]", "[/list]");
 275                         self::tagToBBCode($doc, 'li', [], "[*]", "");
 276
 277                         self::tagToBBCode($doc, 'hr', [], "[hr]", "");
 278
 279                         self::tagToBBCode($doc, 'table', [], "[table]", "[/table]");
 280                         self::tagToBBCode($doc, 'th', [], "[th]", "[/th]");
 281                         self::tagToBBCode($doc, 'tr', [], "[tr]", "[/tr]");
 282                         self::tagToBBCode($doc, 'td', [], "[td]", "[/td]");
 283
 284                         self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 285                         self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 286                         self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 287                         self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 288                         self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 289                         self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 290
 291                         self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 292                         self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 293
 294                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'alt' => '/(.+)/'], '[img=$1]$2', '[/img]', true);
 295                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]', true);
 296                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]', true);
 297
 298
 299                         self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]', true);
 300                         self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]', true);
 301                         // Backward compatibility, [iframe] support has been removed in version 2020.12
 302                         self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[url]$1', '[/url]', true);
 303
 304                         self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 305                         self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 306
 307                         $message = $doc->saveHTML();
 308
 309                         // I'm removing something really disturbing
 310                         // Don't know exactly what it is
 311                         $message = str_replace(chr(194) . chr(160), ' ', $message);
 312
 313                         $message = str_replace("&nbsp;", " ", $message);
 314
 315                         // removing multiple DIVs
 316                         $message = preg_replace('=\r *\r=i', "\n", $message);
 317                         $message = str_replace("\r", "\n", $message);
 318
 319                         Hook::callAll('html2bbcode', $message);
 320
 321                         $message = strip_tags($message);
 322
 323                         $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 324
 325                         // remove quotes if they don't make sense
 326                         $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 327
 328                         $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 329                         $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 330
 331                         do {
 332                                 $oldmessage = $message;
 333                                 $message = str_replace("\n \n", "\n\n", $message);
 334                         } while ($oldmessage != $message);
 335
 336                         do {
 337                                 $oldmessage = $message;
 338                                 $message = str_replace("\n\n\n", "\n\n", $message);
 339                         } while ($oldmessage != $message);
 340
 341                         do {
 342                                 $oldmessage = $message;
 343                                 $message = str_replace(
 344                                         [
 345                                                 "[/size]\n\n",
 346                                                 "\n[hr]",
 347                                                 "[hr]\n",
 348                                                 "\n[list",
 349                                                 "[/list]\n",
 350                                                 "\n[/",
 351                                                 "[list]\n",
 352                                                 "[list=1]\n",
 353                                                 "\n[*]"],
 354                                         [
 355                                                 "[/size]\n",
 356                                                 "[hr]",
 357                                                 "[hr]",
 358                                                 "[list",
 359                                                 "[/list]",
 360                                                 "[/",
 361                                                 "[list]",
 362                                                 "[list=1]",
 363                                                 "[*]"],
 364                                         $message
 365                                 );
 366                         } while ($message != $oldmessage);
 367
 368                         $message = str_replace(
 369                                 ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 370                                 ['[b]', '[/b]', '[i]', '[/i]'],
 371                                 $message
 372                         );
 373
 374                         // Handling Yahoo style of mails
 375                         $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 376
 377                         return $message;
 378                 });
 379
 380                 $message = preg_replace_callback(
 381                         '#<pre><code(?: class="language-([^"]*)")?>(.*)</code></pre>#iUs',
 382                         function ($matches) {
 383                                 $prefix = '[code]';
 384                                 if ($matches[1] != '') {
 385                                         $prefix = '[code=' . $matches[1] . ']';
 386                                 }
 387
 388                                 return $prefix . PHP_EOL . trim($matches[2]) . PHP_EOL . '[/code]';
 389                         },
 390                         $message
 391                 );
 392
 393                 $message = trim($message);
 394
 395                 if ($basepath != '') {
 396                         $message = self::qualifyURLs($message, $basepath);
 397                 }
 398
 399                 return $message;
 400         }
 401
 402         /**
 403          * Sub function to complete incomplete URL
 404          *
 405          * @param array  $matches  Result of preg_replace_callback
 406          * @param string $basepath Basepath that is used to complete the URL
 407          *
 408          * @return string The expanded URL
 409          */
 410         private static function qualifyURLsSub($matches, $basepath)
 411         {
 412                 $base = parse_url($basepath);
 413                 unset($base['query']);
 414                 unset($base['fragment']);
 415
 416                 $link = $matches[0];
 417                 $url = $matches[1];
 418
 419                 if (empty($url) || empty(parse_url($url))) {
 420                         return $matches[0];
 421                 }
 422
 423                 $parts = array_merge($base, parse_url($url));
 424                 $url2 = Network::unparseURL($parts);
 425
 426                 return str_replace($url, $url2, $link);
 427         }
 428
 429         /**
 430          * Complete incomplete URLs in BBCode
 431          *
 432          * @param string $body     Body with URLs
 433          * @param string $basepath Base path that is used to complete the URL
 434          *
 435          * @return string Body with expanded URLs
 436          */
 437         private static function qualifyURLs($body, $basepath)
 438         {
 439                 $URLSearchString = "^\[\]";
 440
 441                 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
 442                         "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
 443                         "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 444                         "/\[img\](.*?)\[\/img\]/ism",
 445                         "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 446                         "/\[zmg\](.*?)\[\/zmg\]/ism",
 447                         "/\[video\](.*?)\[\/video\]/ism",
 448                         "/\[audio\](.*?)\[\/audio\]/ism",
 449                 ];
 450
 451                 foreach ($matches as $match) {
 452                         $body = preg_replace_callback(
 453                                 $match,
 454                                 function ($match) use ($basepath) {
 455                                         return self::qualifyURLsSub($match, $basepath);
 456                                 },
 457                                 $body
 458                         );
 459                 }
 460                 return $body;
 461         }
 462
 463         private static function breakLines($line, $level, $wraplength = 75)
 464         {
 465                 if ($wraplength == 0) {
 466                         $wraplength = 2000000;
 467                 }
 468
 469                 $wraplen = $wraplength - $level;
 470
 471                 $newlines = [];
 472
 473                 do {
 474                         $oldline = $line;
 475
 476                         $subline = substr($line, 0, $wraplen);
 477
 478                         $pos = strrpos($subline, ' ');
 479
 480                         if ($pos == 0) {
 481                                 $pos = strpos($line, ' ');
 482                         }
 483
 484                         if (($pos > 0) && strlen($line) > $wraplen) {
 485                                 $newline = trim(substr($line, 0, $pos));
 486                                 if ($level > 0) {
 487                                         $newline = str_repeat(">", $level) . ' ' . $newline;
 488                                 }
 489
 490                                 $newlines[] = $newline . " ";
 491                                 $line = substr($line, $pos + 1);
 492                         }
 493                 } while ((strlen($line) > $wraplen) && !($oldline == $line));
 494
 495                 if ($level > 0) {
 496                         $line = str_repeat(">", $level) . ' ' . $line;
 497                 }
 498
 499                 $newlines[] = $line;
 500
 501                 return implode("\n", $newlines);
 502         }
 503
 504         private static function quoteLevel($message, $wraplength = 75)
 505         {
 506                 $lines = explode("\n", $message);
 507
 508                 $newlines = [];
 509                 $level = 0;
 510                 foreach ($lines as $line) {
 511                         $line = trim($line);
 512                         $startquote = false;
 513                         while (strpos("*" . $line, '[quote]') > 0) {
 514                                 $level++;
 515                                 $pos = strpos($line, '[quote]');
 516                                 $line = substr($line, 0, $pos) . substr($line, $pos + 7);
 517                                 $startquote = true;
 518                         }
 519
 520                         $currlevel = $level;
 521
 522                         while (strpos("*" . $line, '[/quote]') > 0) {
 523                                 $level--;
 524                                 if ($level < 0) {
 525                                         $level = 0;
 526                                 }
 527
 528                                 $pos = strpos($line, '[/quote]');
 529                                 $line = substr($line, 0, $pos) . substr($line, $pos + 8);
 530                         }
 531
 532                         if (!$startquote || ($line != '')) {
 533                                 $newlines[] = self::breakLines($line, $currlevel, $wraplength);
 534                         }
 535                 }
 536
 537                 return implode("\n", $newlines);
 538         }
 539
 540         private static function collectURLs($message)
 541         {
 542                 $pattern = '/<a.*?href="(.*?)".*?>(.*?)<\/a>/is';
 543                 preg_match_all($pattern, $message, $result, PREG_SET_ORDER);
 544
 545                 $urls = [];
 546                 foreach ($result as $treffer) {
 547                         $ignore = false;
 548
 549                         // A list of some links that should be ignored
 550                         $list = ["/user/", "/tag/", "/group/", "/profile/", "/search?search=", "/search?tag=", "mailto:", "/u/", "/node/",
 551                                 "//plus.google.com/", "//twitter.com/"];
 552                         foreach ($list as $listitem) {
 553                                 if (strpos($treffer[1], $listitem) !== false) {
 554                                         $ignore = true;
 555                                 }
 556                         }
 557
 558                         if ((strpos($treffer[1], "//twitter.com/") !== false) && (strpos($treffer[1], "/status/") !== false)) {
 559                                 $ignore = false;
 560                         }
 561
 562                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/posts") !== false)) {
 563                                 $ignore = false;
 564                         }
 565
 566                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/photos") !== false)) {
 567                                 $ignore = false;
 568                         }
 569
 570                         $ignore = $ignore || strpos($treffer[1], '#') === 0;
 571
 572                         if (!$ignore) {
 573                                 $urls[$treffer[1]] = $treffer[1];
 574                         }
 575                 }
 576
 577                 return $urls;
 578         }
 579
 580         /**
 581          * @param string $html
 582          * @param int    $wraplength Ensures individual lines aren't longer than this many characters. Doesn't break words.
 583          * @param bool   $compact    True: Completely strips image tags; False: Keeps image URLs
 584          * @return string
 585          */
 586         public static function toPlaintext(string $html, $wraplength = 75, $compact = false)
 587         {
 588                 $message = str_replace("\r", "", $html);
 589
 590                 $doc = new DOMDocument();
 591                 $doc->preserveWhiteSpace = false;
 592
 593                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 594
 595                 if (empty($message)) {
 596                         return '';
 597                 }
 598
 599                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 600
 601                 $message = $doc->saveHTML();
 602                 // Remove eventual UTF-8 BOM
 603                 $message = str_replace("\xC3\x82\xC2\xA0", "", $message);
 604
 605                 // Collecting all links
 606                 $urls = self::collectURLs($message);
 607
 608                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 609
 610                 self::tagToBBCode($doc, 'html', [], '', '');
 611                 self::tagToBBCode($doc, 'body', [], '', '');
 612
 613                 if ($compact) {
 614                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 615                 } else {
 616                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 617                 }
 618
 619                 self::tagToBBCode($doc, 'br', [], "\n", '');
 620
 621                 self::tagToBBCode($doc, 'span', [], "", "");
 622                 self::tagToBBCode($doc, 'pre', [], "", "");
 623                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 624                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 625
 626                 self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 627
 628                 self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
 629
 630                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 631                 self::tagToBBCode($doc, 'td', [], "\t", "");
 632
 633                 self::tagToBBCode($doc, 'h1', [], "\n\n*", "*\n");
 634                 self::tagToBBCode($doc, 'h2', [], "\n\n*", "*\n");
 635                 self::tagToBBCode($doc, 'h3', [], "\n\n*", "*\n");
 636                 self::tagToBBCode($doc, 'h4', [], "\n\n*", "*\n");
 637                 self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
 638                 self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 639
 640                 if (!$compact) {
 641                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
 642                 } else {
 643                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' ', ' ');
 644                 }
 645
 646                 // Backward compatibility, [iframe] support has been removed in version 2020.12
 647                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], ' $1 ', '');
 648
 649                 $message = $doc->saveHTML();
 650
 651                 if (!$compact) {
 652                         $message = str_replace("[img]", "", $message);
 653                         $message = str_replace("[/img]", "", $message);
 654                 }
 655
 656                 // was ersetze ich da?
 657                 // Irgendein stoerrisches UTF-Zeug
 658                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 659
 660                 $message = str_replace("&nbsp;", " ", $message);
 661
 662                 // Aufeinanderfolgende DIVs
 663                 $message = preg_replace('=\r *\r=i', "\n", $message);
 664                 $message = str_replace("\r", "\n", $message);
 665
 666                 $message = strip_tags($message);
 667
 668                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 669
 670                 if (!$compact && ($message != '')) {
 671                         foreach ($urls as $id => $url) {
 672                                 if ($url != '' && strpos($message, $url) === false) {
 673                                         $message .= "\n" . $url . ' ';
 674                                 }
 675                         }
 676                 }
 677
 678                 $message = str_replace("\n«", "«\n", $message);
 679                 $message = str_replace("»\n", "\n»", $message);
 680
 681                 do {
 682                         $oldmessage = $message;
 683                         $message = str_replace("\n\n\n", "\n\n", $message);
 684                 } while ($oldmessage != $message);
 685
 686                 $message = self::quoteLevel(trim($message), $wraplength);
 687
 688                 return trim($message);
 689         }
 690
 691         /**
 692          * Converts provided HTML code to Markdown. The hardwrap parameter maximizes
 693          * compatibility with Diaspora in spite of the Markdown standards.
 694          *
 695          * @param string $html
 696          * @return string
 697          */
 698         public static function toMarkdown($html)
 699         {
 700                 $converter = new HtmlConverter(['hard_break' => true]);
 701                 $markdown = $converter->convert($html);
 702
 703                 return $markdown;
 704         }
 705
 706         /**
 707          * Convert video HTML to BBCode tags
 708          *
 709          * @param string $s
 710          * @return string
 711          */
 712         public static function toBBCodeVideo($s)
 713         {
 714                 $s = preg_replace(
 715                         '#<object[^>]+>(.*?)https?://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+)(.*?)</object>#ism',
 716                         '[youtube]$2[/youtube]',
 717                         $s
 718                 );
 719
 720                 $s = preg_replace(
 721                         '#<iframe[^>](.*?)https?://www.youtube.com/embed/([A-Za-z0-9\-_=]+)(.*?)</iframe>#ism',
 722                         '[youtube]$2[/youtube]',
 723                         $s
 724                 );
 725
 726                 $s = preg_replace(
 727                         '#<iframe[^>](.*?)https?://player.vimeo.com/video/([0-9]+)(.*?)</iframe>#ism',
 728                         '[vimeo]$2[/vimeo]',
 729                         $s
 730                 );
 731
 732                 return $s;
 733         }
 734
 735         /**
 736          * transform link href and img src from relative to absolute
 737          *
 738          * @param string $text
 739          * @param string $base base url
 740          * @return string
 741          */
 742         public static function relToAbs($text, $base)
 743         {
 744                 if (empty($base)) {
 745                         return $text;
 746                 }
 747
 748                 $base = rtrim($base, '/');
 749
 750                 $base2 = $base . "/";
 751
 752                 // Replace links
 753                 $pattern = "/<a([^>]*) href=\"(?!http|https|\/)([^\"]*)\"/";
 754                 $replace = "<a\${1} href=\"" . $base2 . "\${2}\"";
 755                 $text = preg_replace($pattern, $replace, $text);
 756
 757                 $pattern = "/<a([^>]*) href=\"(?!http|https)([^\"]*)\"/";
 758                 $replace = "<a\${1} href=\"" . $base . "\${2}\"";
 759                 $text = preg_replace($pattern, $replace, $text);
 760
 761                 // Replace images
 762                 $pattern = "/<img([^>]*) src=\"(?!http|https|\/)([^\"]*)\"/";
 763                 $replace = "<img\${1} src=\"" . $base2 . "\${2}\"";
 764                 $text = preg_replace($pattern, $replace, $text);
 765
 766                 $pattern = "/<img([^>]*) src=\"(?!http|https)([^\"]*)\"/";
 767                 $replace = "<img\${1} src=\"" . $base . "\${2}\"";
 768                 $text = preg_replace($pattern, $replace, $text);
 769
 770
 771                 // Done
 772                 return $text;
 773         }
 774
 775         /**
 776          * return div element with class 'clear'
 777          * @return string
 778          * @deprecated
 779          */
 780         public static function clearDiv()
 781         {
 782                 return '<div class="clear"></div>';
 783         }
 784
 785         /**
 786          * Loader for infinite scrolling
 787          *
 788          * @return string html for loader
 789          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 790          */
 791         public static function scrollLoader()
 792         {
 793                 $tpl = Renderer::getMarkupTemplate("scroll_loader.tpl");
 794                 return Renderer::replaceMacros($tpl, [
 795                         'wait' => DI::l10n()->t('Loading more entries...'),
 796                         'end' => DI::l10n()->t('The end')
 797                 ]);
 798         }
 799
 800         /**
 801          * Get html for contact block.
 802          *
 803          * @deprecated since version 2019.03
 804          * @see ContactBlock::getHTML()
 805          * @return string
 806          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 807          * @throws \ImagickException
 808          */
 809         public static function contactBlock()
 810         {
 811                 $a = DI::app();
 812
 813                 return ContactBlock::getHTML($a->profile);
 814         }
 815
 816         /**
 817          * Format contacts as picture links or as text links
 818          *
 819          * @param array   $contact  Array with contacts which contains an array with
 820          *                          int 'id' => The ID of the contact
 821          *                          int 'uid' => The user ID of the user who owns this data
 822          *                          string 'name' => The name of the contact
 823          *                          string 'url' => The url to the profile page of the contact
 824          *                          string 'addr' => The webbie of the contact (e.g.) username@friendica.com
 825          *                          string 'network' => The network to which the contact belongs to
 826          *                          string 'thumb' => The contact picture
 827          *                          string 'click' => js code which is performed when clicking on the contact
 828          * @param boolean $redirect If true try to use the redir url if it's possible
 829          * @param string  $class    CSS class for the
 830          * @param boolean $textmode If true display the contacts as text links
 831          *                          if false display the contacts as picture links
 832          * @return string Formatted html
 833          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 834          * @throws \ImagickException
 835          */
 836         public static function micropro($contact, $redirect = false, $class = '', $textmode = false)
 837         {
 838                 // Use the contact URL if no address is available
 839                 if (empty($contact['addr'])) {
 840                         $contact["addr"] = $contact["url"];
 841                 }
 842
 843                 $url = $contact['url'];
 844                 $sparkle = '';
 845                 $redir = false;
 846
 847                 if ($redirect) {
 848                         $url = Contact::magicLinkByContact($contact);
 849                         if (strpos($url, 'redir/') === 0) {
 850                                 $sparkle = ' sparkle';
 851                         }
 852                 }
 853
 854                 // If there is some js available we don't need the url
 855                 if (!empty($contact['click'])) {
 856                         $url = '';
 857                 }
 858
 859                 return Renderer::replaceMacros(Renderer::getMarkupTemplate($textmode ? 'micropro_txt.tpl' : 'micropro_img.tpl'), [
 860                         '$click' => $contact['click'] ?? '',
 861                         '$class' => $class,
 862                         '$url' => $url,
 863                         '$photo' => Contact::getThumb($contact),
 864                         '$name' => $contact['name'],
 865                         'title' => $contact['name'] . ' [' . $contact['addr'] . ']',
 866                         '$parkle' => $sparkle,
 867                         '$redir' => $redir
 868                 ]);
 869         }
 870
 871         /**
 872          * Search box.
 873          *
 874          * @param string $s     Search query.
 875          * @param string $id    HTML id
 876          * @param string $url   Search url.
 877          * @param bool   $aside Display the search widgit aside.
 878          *
 879          * @return string Formatted HTML.
 880          * @throws \Exception
 881          */
 882         public static function search($s, $id = 'search-box', $aside = true)
 883         {
 884                 $mode = 'text';
 885
 886                 if (strpos($s, '#') === 0) {
 887                         $mode = 'tag';
 888                 }
 889                 $save_label = $mode === 'text' ? DI::l10n()->t('Save') : DI::l10n()->t('Follow');
 890
 891                 $values = [
 892                         '$s'            => $s,
 893                         '$q'            => urlencode($s),
 894                         '$id'           => $id,
 895                         '$search_label' => DI::l10n()->t('Search'),
 896                         '$save_label'   => $save_label,
 897                         '$search_hint'  => DI::l10n()->t('@name, !forum, #tags, content'),
 898                         '$mode'         => $mode,
 899                         '$return_url'   => urlencode(Search::getSearchPath($s)),
 900                 ];
 901
 902                 if (!$aside) {
 903                         $values['$search_options'] = [
 904                                 'fulltext' => DI::l10n()->t('Full Text'),
 905                                 'tags'     => DI::l10n()->t('Tags'),
 906                                 'contacts' => DI::l10n()->t('Contacts')
 907                         ];
 908
 909                         if (DI::config()->get('system', 'poco_local_search')) {
 910                                 $values['$searchoption']['forums'] = DI::l10n()->t('Forums');
 911                         }
 912                 }
 913
 914                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 915         }
 916
 917         /**
 918          * Replace naked text hyperlink with HTML formatted hyperlink
 919          *
 920          * @param string $s
 921          * @return string
 922          */
 923         public static function toLink($s)
 924         {
 925                 $s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank" rel="noopener noreferrer">$1</a>', $s);
 926                 $s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
 927                 return $s;
 928         }
 929
 930         /**
 931          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
 932          *
 933          * Reasons are expected to have been translated already.
 934          *
 935          * @param string $html
 936          * @param array  $reasons
 937          * @return string
 938          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 939          */
 940         public static function applyContentFilter($html, array $reasons)
 941         {
 942                 if (count($reasons)) {
 943                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
 944                         $html = Renderer::replaceMacros($tpl, [
 945                                 '$reasons'   => $reasons,
 946                                 '$rnd'       => Strings::getRandomHex(8),
 947                                 '$openclose' => DI::l10n()->t('Click to open/close'),
 948                                 '$html'      => $html
 949                         ]);
 950                 }
 951
 952                 return $html;
 953         }
 954
 955         /**
 956          * replace html amp entity with amp char
 957          * @param string $s
 958          * @return string
 959          */
 960         public static function unamp($s)
 961         {
 962                 return str_replace('&amp;', '&', $s);
 963         }
 964
 965         /**
 966          * Clean an HTML text for potentially harmful code
 967          *
 968          * @param string $text
 969          * @param array  $allowedIframeDomains List of allowed iframe source domains without the scheme
 970          * @return string
 971          */
 972         public static function purify(string $text, array $allowedIframeDomains = []): string
 973         {
 974                 // Allows cid: URL scheme
 975                 \HTMLPurifier_URISchemeRegistry::instance()->register('cid', new HTMLPurifier_URIScheme_cid());
 976
 977                 $config = \HTMLPurifier_HTML5Config::createDefault();
 978                 $config->set('HTML.Doctype', 'HTML5');
 979
 980                 // Used to remove iframe with src attribute filtered out
 981                 $config->set('AutoFormat.RemoveEmpty', true);
 982
 983                 $config->set('HTML.SafeIframe', true);
 984
 985                 array_walk($allowedIframeDomains, function (&$domain) {
 986                         // Allow the domain and all its eventual sub-domains
 987                         $domain = '(?:(?!-)[A-Za-z0-9-]{1,63}(?<!-)\.)*' . preg_quote(trim($domain, '/'), '%');
 988                 });
 989
 990                 $config->set('URI.SafeIframeRegexp',
 991                         '%^https://(?:
 992                                 ' . implode('|', $allowedIframeDomains) . '
 993                         )
 994                         (?:/|$) # Prevents bogus domains like youtube.com.fake.tld
 995                         %xi'
 996                 );
 997
 998                 $config->set('Attr.AllowedRel', [
 999                         'noreferrer' => true,
1000                         'noopener' => true,
1001                 ]);
1002                 $config->set('Attr.AllowedFrameTargets', [
1003                         '_blank' => true,
1004                 ]);
1005
1006                 $config->set('AutoFormat.RemoveEmpty.Predicate', [
1007                         'colgroup' => [],        // |
1008                         'th'       => [],        // |
1009                         'td'       => [],        // |
1010                         'iframe'   => ['src'],   // ↳ Default HTMLPurify values
1011                         'i'        => ['class'], // Allows forkawesome icons
1012                 ]);
1013
1014                 // Uncomment to debug HTMLPurifier behavior
1015                 //$config->set('Core.CollectErrors', true);
1016                 //$config->set('Core.MaintainLineNumbers', true);
1017
1018                 $HTMLPurifier = new \HTMLPurifier($config);
1019
1020                 $text = $HTMLPurifier->purify($text);
1021
1022                 /** @var \HTMLPurifier_ErrorCollector $errorCollector */
1023                 // Uncomment to debug HTML Purifier behavior
1024                 //$errorCollector = $HTMLPurifier->context->get('ErrorCollector');
1025                 //var_dump($errorCollector->getRaw());
1026
1027                 return $text;
1028         }
1029 }