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