src/Content/Text/HTML.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2020, Friendica
   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\DI;
  30 use Friendica\Model\Contact;
  31 use Friendica\Util\Network;
  32 use Friendica\Util\Proxy as ProxyUtils;
  33 use Friendica\Util\Strings;
  34 use Friendica\Util\XML;
  35 use League\HTMLToMarkdown\HtmlConverter;
  36
  37 class HTML
  38 {
  39         public static function sanitizeCSS($input)
  40         {
  41                 $cleaned = "";
  42
  43                 $input = strtolower($input);
  44
  45                 for ($i = 0; $i < strlen($input); $i++) {
  46                         $char = substr($input, $i, 1);
  47
  48                         if (($char >= "a") && ($char <= "z")) {
  49                                 $cleaned .= $char;
  50                         }
  51
  52                         if (!(strpos(" #;:0123456789-_.%", $char) === false)) {
  53                                 $cleaned .= $char;
  54                         }
  55                 }
  56
  57                 return $cleaned;
  58         }
  59
  60         /**
  61          * Search all instances of a specific HTML tag node in the provided DOM document and replaces them with BBCode text nodes.
  62          *
  63          * @see HTML::tagToBBCodeSub()
  64          */
  65         private static function tagToBBCode(DOMDocument $doc, string $tag, array $attributes, string $startbb, string $endbb, bool $ignoreChildren = false)
  66         {
  67                 do {
  68                         $done = self::tagToBBCodeSub($doc, $tag, $attributes, $startbb, $endbb, $ignoreChildren);
  69                 } while ($done);
  70         }
  71
  72         /**
  73          * Search the first specific HTML tag node in the provided DOM document and replaces it with BBCode text nodes.
  74          *
  75          * @param DOMDocument $doc
  76          * @param string      $tag            HTML tag name
  77          * @param array       $attributes     Array of attributes to match and optionally use the value from
  78          * @param string      $startbb        BBCode tag opening
  79          * @param string      $endbb          BBCode tag closing
  80          * @param bool        $ignoreChildren If set to false, the HTML tag children will be appended as text inside the BBCode tag
  81          *                                    Otherwise, they will be entirely ignored. Useful for simple BBCode that draw their
  82          *                                    inner value from an attribute value and disregard the tag children.
  83          * @return bool Whether a replacement was done
  84          */
  85         private static function tagToBBCodeSub(DOMDocument $doc, string $tag, array $attributes, string $startbb, string $endbb, bool $ignoreChildren = false)
  86         {
  87                 $savestart = str_replace('$', '\x01', $startbb);
  88                 $replace = false;
  89
  90                 $xpath = new DOMXPath($doc);
  91
  92                 /** @var \DOMNode[] $list */
  93                 $list = $xpath->query("//" . $tag);
  94                 foreach ($list as $node) {
  95                         $attr = [];
  96                         if ($node->attributes->length) {
  97                                 foreach ($node->attributes as $attribute) {
  98                                         $attr[$attribute->name] = $attribute->value;
  99                                 }
 100                         }
 101
 102                         $replace = true;
 103
 104                         $startbb = $savestart;
 105
 106                         $i = 0;
 107
 108                         foreach ($attributes as $attribute => $value) {
 109                                 $startbb = str_replace('\x01' . ++$i, '$1', $startbb);
 110                                 if (strpos('*' . $startbb, '$1') > 0) {
 111                                         if ($replace && (@$attr[$attribute] != '')) {
 112                                                 $startbb = preg_replace($value, $startbb, $attr[$attribute], -1, $count);
 113
 114                                                 // If nothing could be changed
 115                                                 if ($count == 0) {
 116                                                         $replace = false;
 117                                                 }
 118                                         } else {
 119                                                 $replace = false;
 120                                         }
 121                                 } else {
 122                                         if (@$attr[$attribute] != $value) {
 123                                                 $replace = false;
 124                                         }
 125                                 }
 126                         }
 127
 128                         if ($replace) {
 129                                 $StartCode = $doc->createTextNode($startbb);
 130                                 $EndCode = $doc->createTextNode($endbb);
 131
 132                                 $node->parentNode->insertBefore($StartCode, $node);
 133
 134                                 if (!$ignoreChildren && $node->hasChildNodes()) {
 135                                         /** @var \DOMNode $child */
 136                                         foreach ($node->childNodes as $key => $child) {
 137                                                 /* Remove empty text nodes at the start or at the end of the children list */
 138                                                 if ($key > 0 && $key < $node->childNodes->length - 1 || $child->nodeName != '#text' || trim($child->nodeValue)) {
 139                                                         $newNode = $child->cloneNode(true);
 140                                                         $node->parentNode->insertBefore($newNode, $node);
 141                                                 }
 142                                         }
 143                                 }
 144
 145                                 $node->parentNode->insertBefore($EndCode, $node);
 146                                 $node->parentNode->removeChild($node);
 147                         }
 148                 }
 149
 150                 return $replace;
 151         }
 152
 153         /**
 154          * Converter for HTML to BBCode
 155          *
 156          * Made by: ike@piratenpartei.de
 157          * Originally made for the syncom project: http://wiki.piratenpartei.de/Syncom
 158          *                    https://github.com/annando/Syncom
 159          *
 160          * @param string $message
 161          * @param string $basepath
 162          * @return string
 163          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 164          */
 165         public static function toBBCode($message, $basepath = '')
 166         {
 167                 $message = str_replace("\r", "", $message);
 168
 169                 // Removing code blocks before the whitespace removal processing below
 170                 $codeblocks = [];
 171                 $message = preg_replace_callback(
 172                         '#<pre><code(?: class="language-([^"]*)")?>(.*)</code></pre>#iUs',
 173                         function ($matches) use (&$codeblocks) {
 174                                 $return = '[codeblock-' . count($codeblocks) . ']';
 175
 176                                 $prefix = '[code]';
 177                                 if ($matches[1] != '') {
 178                                         $prefix = '[code=' . $matches[1] . ']';
 179                                 }
 180
 181                                 $codeblocks[] = $prefix . PHP_EOL . trim($matches[2]) . PHP_EOL . '[/code]';
 182                                 return $return;
 183                         },
 184                         $message
 185                 );
 186
 187                 $message = str_replace(
 188                         [
 189                                 "<li><p>",
 190                                 "</p></li>",
 191                         ],
 192                         [
 193                                 "<li>",
 194                                 "</li>",
 195                         ],
 196                         $message
 197                 );
 198
 199                 // remove namespaces
 200                 $message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
 201                 $message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
 202
 203                 $doc = new DOMDocument();
 204                 $doc->preserveWhiteSpace = false;
 205
 206                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 207
 208                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 209
 210                 XML::deleteNode($doc, 'style');
 211                 XML::deleteNode($doc, 'head');
 212                 XML::deleteNode($doc, 'title');
 213                 XML::deleteNode($doc, 'meta');
 214                 XML::deleteNode($doc, 'xml');
 215                 XML::deleteNode($doc, 'removeme');
 216
 217                 $xpath = new DomXPath($doc);
 218                 $list = $xpath->query("//pre");
 219                 foreach ($list as $node) {
 220                         // Ensure to escape unescaped & - they will otherwise raise a warning
 221                         $safe_value = preg_replace('/&(?!\w+;)/', '&amp;', $node->nodeValue);
 222                         $node->nodeValue = str_replace("\n", "\r", $safe_value);
 223                 }
 224
 225                 $message = $doc->saveHTML();
 226                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br />", " ", ""], $message);
 227                 $message = preg_replace('= [\s]*=i', " ", $message);
 228
 229                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 230
 231                 self::tagToBBCode($doc, 'html', [], "", "");
 232                 self::tagToBBCode($doc, 'body', [], "", "");
 233
 234                 // Outlook-Quote - Variant 1
 235                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal', 'style' => 'margin-left:35.4pt'], '[quote]', '[/quote]');
 236
 237                 // Outlook-Quote - Variant 2
 238                 self::tagToBBCode(
 239                         $doc,
 240                         'div',
 241                         ['style' => 'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'],
 242                         '[quote]',
 243                         '[/quote]'
 244                 );
 245
 246                 // MyBB-Stuff
 247                 self::tagToBBCode($doc, 'span', ['style' => 'text-decoration: underline;'], '[u]', '[/u]');
 248                 self::tagToBBCode($doc, 'span', ['style' => 'font-style: italic;'], '[i]', '[/i]');
 249                 self::tagToBBCode($doc, 'span', ['style' => 'font-weight: bold;'], '[b]', '[/b]');
 250
 251                 /* self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
 252                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
 253                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
 254                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
 255                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
 256                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
 257                   self::node2BBCode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
 258                  */
 259                 // Untested
 260                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
 261                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
 262                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
 263
 264                 self::tagToBBCode($doc, 'span', ['style' => '/.*color:\s*(.+?)[,;].*/'], '[color="$1"]', '[/color]');
 265
 266                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 267                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
 268                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
 269                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 270                 // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
 271                 // Test
 272                 //self::node2BBCode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
 273                 self::tagToBBCode($doc, 'span', ['class' => 'type-link'], '[class=type-link]', '[/class]');
 274                 self::tagToBBCode($doc, 'span', ['class' => 'type-video'], '[class=type-video]', '[/class]');
 275
 276                 self::tagToBBCode($doc, 'strong', [], '[b]', '[/b]');
 277                 self::tagToBBCode($doc, 'em', [], '[i]', '[/i]');
 278                 self::tagToBBCode($doc, 'b', [], '[b]', '[/b]');
 279                 self::tagToBBCode($doc, 'i', [], '[i]', '[/i]');
 280                 self::tagToBBCode($doc, 'u', [], '[u]', '[/u]');
 281                 self::tagToBBCode($doc, 's', [], '[s]', '[/s]');
 282                 self::tagToBBCode($doc, 'del', [], '[s]', '[/s]');
 283                 self::tagToBBCode($doc, 'strike', [], '[s]', '[/s]');
 284
 285                 self::tagToBBCode($doc, 'big', [], "[size=large]", "[/size]");
 286                 self::tagToBBCode($doc, 'small', [], "[size=small]", "[/size]");
 287
 288                 self::tagToBBCode($doc, 'blockquote', [], '[quote]', '[/quote]');
 289
 290                 self::tagToBBCode($doc, 'br', [], "\n", '');
 291
 292                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal'], "\n", "");
 293                 self::tagToBBCode($doc, 'div', ['class' => 'MsoNormal'], "\r", "");
 294
 295                 self::tagToBBCode($doc, 'span', [], "", "");
 296
 297                 self::tagToBBCode($doc, 'span', [], "", "");
 298                 self::tagToBBCode($doc, 'pre', [], "", "");
 299
 300                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 301                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 302
 303                 self::tagToBBCode($doc, 'ul', [], "[list]", "[/list]");
 304                 self::tagToBBCode($doc, 'ol', [], "[list=1]", "[/list]");
 305                 self::tagToBBCode($doc, 'li', [], "[*]", "");
 306
 307                 self::tagToBBCode($doc, 'hr', [], "[hr]", "");
 308
 309                 self::tagToBBCode($doc, 'table', [], "[table]", "[/table]");
 310                 self::tagToBBCode($doc, 'th', [], "[th]", "[/th]");
 311                 self::tagToBBCode($doc, 'tr', [], "[tr]", "[/tr]");
 312                 self::tagToBBCode($doc, 'td', [], "[td]", "[/td]");
 313
 314                 self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 315                 self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 316                 self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 317                 self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 318                 self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 319                 self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 320
 321                 self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 322                 self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 323
 324                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'alt' => '/(.+)/'], '[img=$1]$2', '[/img]', true);
 325                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]', true);
 326                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]', true);
 327
 328
 329                 self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]', true);
 330                 self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]', true);
 331                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[iframe]$1', '[/iframe]', true);
 332
 333                 self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 334                 self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 335
 336                 $message = $doc->saveHTML();
 337
 338                 // I'm removing something really disturbing
 339                 // Don't know exactly what it is
 340                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 341
 342                 $message = str_replace("&nbsp;", " ", $message);
 343
 344                 // removing multiple DIVs
 345                 $message = preg_replace('=\r *\r=i', "\n", $message);
 346                 $message = str_replace("\r", "\n", $message);
 347
 348                 Hook::callAll('html2bbcode', $message);
 349
 350                 $message = strip_tags($message);
 351
 352                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 353
 354                 // remove quotes if they don't make sense
 355                 $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 356
 357                 $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 358                 $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 359
 360                 do {
 361                         $oldmessage = $message;
 362                         $message = str_replace("\n \n", "\n\n", $message);
 363                 } while ($oldmessage != $message);
 364
 365                 do {
 366                         $oldmessage = $message;
 367                         $message = str_replace("\n\n\n", "\n\n", $message);
 368                 } while ($oldmessage != $message);
 369
 370                 do {
 371                         $oldmessage = $message;
 372                         $message = str_replace(
 373                                 [
 374                                 "[/size]\n\n",
 375                                 "\n[hr]",
 376                                 "[hr]\n",
 377                                 "\n[list",
 378                                 "[/list]\n",
 379                                 "\n[/",
 380                                 "[list]\n",
 381                                 "[list=1]\n",
 382                                 "\n[*]"],
 383                                 [
 384                                 "[/size]\n",
 385                                 "[hr]",
 386                                 "[hr]",
 387                                 "[list",
 388                                 "[/list]",
 389                                 "[/",
 390                                 "[list]",
 391                                 "[list=1]",
 392                                 "[*]"],
 393                                 $message
 394                         );
 395                 } while ($message != $oldmessage);
 396
 397                 $message = str_replace(
 398                         ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 399                         ['[b]', '[/b]', '[i]', '[/i]'],
 400                         $message
 401                 );
 402
 403                 // Handling Yahoo style of mails
 404                 $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 405
 406                 // Restore code blocks
 407                 $message = preg_replace_callback(
 408                         '#\[codeblock-([0-9]+)\]#iU',
 409                         function ($matches) use ($codeblocks) {
 410                                 $return = '';
 411                                 if (isset($codeblocks[intval($matches[1])])) {
 412                                         $return = $codeblocks[$matches[1]];
 413                                 }
 414                                 return $return;
 415                         },
 416                         $message
 417                 );
 418
 419                 $message = trim($message);
 420
 421                 if ($basepath != '') {
 422                         $message = self::qualifyURLs($message, $basepath);
 423                 }
 424
 425                 return $message;
 426         }
 427
 428         /**
 429          * Sub function to complete incomplete URL
 430          *
 431          * @param array  $matches  Result of preg_replace_callback
 432          * @param string $basepath Basepath that is used to complete the URL
 433          *
 434          * @return string The expanded URL
 435          */
 436         private static function qualifyURLsSub($matches, $basepath)
 437         {
 438                 $base = parse_url($basepath);
 439                 unset($base['query']);
 440                 unset($base['fragment']);
 441
 442                 $link = $matches[0];
 443                 $url = $matches[1];
 444
 445                 if (empty($url) || empty(parse_url($url))) {
 446                         return $matches[0];
 447                 }
 448
 449                 $parts = array_merge($base, parse_url($url));
 450                 $url2 = Network::unparseURL($parts);
 451
 452                 return str_replace($url, $url2, $link);
 453         }
 454
 455         /**
 456          * Complete incomplete URLs in BBCode
 457          *
 458          * @param string $body     Body with URLs
 459          * @param string $basepath Base path that is used to complete the URL
 460          *
 461          * @return string Body with expanded URLs
 462          */
 463         private static function qualifyURLs($body, $basepath)
 464         {
 465                 $URLSearchString = "^\[\]";
 466
 467                 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
 468                         "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
 469                         "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 470                         "/\[img\](.*?)\[\/img\]/ism",
 471                         "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 472                         "/\[zmg\](.*?)\[\/zmg\]/ism",
 473                         "/\[video\](.*?)\[\/video\]/ism",
 474                         "/\[audio\](.*?)\[\/audio\]/ism",
 475                 ];
 476
 477                 foreach ($matches as $match) {
 478                         $body = preg_replace_callback(
 479                                 $match,
 480                                 function ($match) use ($basepath) {
 481                                         return self::qualifyURLsSub($match, $basepath);
 482                                 },
 483                                 $body
 484                         );
 485                 }
 486                 return $body;
 487         }
 488
 489         private static function breakLines($line, $level, $wraplength = 75)
 490         {
 491                 if ($wraplength == 0) {
 492                         $wraplength = 2000000;
 493                 }
 494
 495                 $wraplen = $wraplength - $level;
 496
 497                 $newlines = [];
 498
 499                 do {
 500                         $oldline = $line;
 501
 502                         $subline = substr($line, 0, $wraplen);
 503
 504                         $pos = strrpos($subline, ' ');
 505
 506                         if ($pos == 0) {
 507                                 $pos = strpos($line, ' ');
 508                         }
 509
 510                         if (($pos > 0) && strlen($line) > $wraplen) {
 511                                 $newline = trim(substr($line, 0, $pos));
 512                                 if ($level > 0) {
 513                                         $newline = str_repeat(">", $level) . ' ' . $newline;
 514                                 }
 515
 516                                 $newlines[] = $newline . " ";
 517                                 $line = substr($line, $pos + 1);
 518                         }
 519                 } while ((strlen($line) > $wraplen) && !($oldline == $line));
 520
 521                 if ($level > 0) {
 522                         $line = str_repeat(">", $level) . ' ' . $line;
 523                 }
 524
 525                 $newlines[] = $line;
 526
 527                 return implode("\n", $newlines);
 528         }
 529
 530         private static function quoteLevel($message, $wraplength = 75)
 531         {
 532                 $lines = explode("\n", $message);
 533
 534                 $newlines = [];
 535                 $level = 0;
 536                 foreach ($lines as $line) {
 537                         $line = trim($line);
 538                         $startquote = false;
 539                         while (strpos("*" . $line, '[quote]') > 0) {
 540                                 $level++;
 541                                 $pos = strpos($line, '[quote]');
 542                                 $line = substr($line, 0, $pos) . substr($line, $pos + 7);
 543                                 $startquote = true;
 544                         }
 545
 546                         $currlevel = $level;
 547
 548                         while (strpos("*" . $line, '[/quote]') > 0) {
 549                                 $level--;
 550                                 if ($level < 0) {
 551                                         $level = 0;
 552                                 }
 553
 554                                 $pos = strpos($line, '[/quote]');
 555                                 $line = substr($line, 0, $pos) . substr($line, $pos + 8);
 556                         }
 557
 558                         if (!$startquote || ($line != '')) {
 559                                 $newlines[] = self::breakLines($line, $currlevel, $wraplength);
 560                         }
 561                 }
 562
 563                 return implode("\n", $newlines);
 564         }
 565
 566         private static function collectURLs($message)
 567         {
 568                 $pattern = '/<a.*?href="(.*?)".*?>(.*?)<\/a>/is';
 569                 preg_match_all($pattern, $message, $result, PREG_SET_ORDER);
 570
 571                 $urls = [];
 572                 foreach ($result as $treffer) {
 573                         $ignore = false;
 574
 575                         // A list of some links that should be ignored
 576                         $list = ["/user/", "/tag/", "/group/", "/profile/", "/search?search=", "/search?tag=", "mailto:", "/u/", "/node/",
 577                                 "//plus.google.com/", "//twitter.com/"];
 578                         foreach ($list as $listitem) {
 579                                 if (strpos($treffer[1], $listitem) !== false) {
 580                                         $ignore = true;
 581                                 }
 582                         }
 583
 584                         if ((strpos($treffer[1], "//twitter.com/") !== false) && (strpos($treffer[1], "/status/") !== false)) {
 585                                 $ignore = false;
 586                         }
 587
 588                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/posts") !== false)) {
 589                                 $ignore = false;
 590                         }
 591
 592                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/photos") !== false)) {
 593                                 $ignore = false;
 594                         }
 595
 596                         $ignore = $ignore || strpos($treffer[1], '#') === 0;
 597
 598                         if (!$ignore) {
 599                                 $urls[$treffer[1]] = $treffer[1];
 600                         }
 601                 }
 602
 603                 return $urls;
 604         }
 605
 606         /**
 607          * @param string $html
 608          * @param int    $wraplength Ensures individual lines aren't longer than this many characters. Doesn't break words.
 609          * @param bool   $compact    True: Completely strips image tags; False: Keeps image URLs
 610          * @return string
 611          */
 612         public static function toPlaintext(string $html, $wraplength = 75, $compact = false)
 613         {
 614                 $message = str_replace("\r", "", $html);
 615
 616                 $doc = new DOMDocument();
 617                 $doc->preserveWhiteSpace = false;
 618
 619                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 620
 621                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 622
 623                 $message = $doc->saveHTML();
 624                 // Remove eventual UTF-8 BOM
 625                 $message = str_replace("\xC3\x82\xC2\xA0", "", $message);
 626
 627                 // Collecting all links
 628                 $urls = self::collectURLs($message);
 629
 630                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 631
 632                 self::tagToBBCode($doc, 'html', [], '', '');
 633                 self::tagToBBCode($doc, 'body', [], '', '');
 634
 635                 if ($compact) {
 636                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 637                 } else {
 638                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 639                 }
 640
 641                 self::tagToBBCode($doc, 'br', [], "\n", '');
 642
 643                 self::tagToBBCode($doc, 'span', [], "", "");
 644                 self::tagToBBCode($doc, 'pre', [], "", "");
 645                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 646                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 647
 648                 self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 649
 650                 self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
 651
 652                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 653                 self::tagToBBCode($doc, 'td', [], "\t", "");
 654
 655                 self::tagToBBCode($doc, 'h1', [], "\n\n*", "*\n");
 656                 self::tagToBBCode($doc, 'h2', [], "\n\n*", "*\n");
 657                 self::tagToBBCode($doc, 'h3', [], "\n\n*", "*\n");
 658                 self::tagToBBCode($doc, 'h4', [], "\n\n*", "*\n");
 659                 self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
 660                 self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 661
 662                 if (!$compact) {
 663                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
 664                 } else {
 665                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' ', ' ');
 666                 }
 667
 668                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], ' $1 ', '');
 669
 670                 $message = $doc->saveHTML();
 671
 672                 if (!$compact) {
 673                         $message = str_replace("[img]", "", $message);
 674                         $message = str_replace("[/img]", "", $message);
 675                 }
 676
 677                 // was ersetze ich da?
 678                 // Irgendein stoerrisches UTF-Zeug
 679                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 680
 681                 $message = str_replace("&nbsp;", " ", $message);
 682
 683                 // Aufeinanderfolgende DIVs
 684                 $message = preg_replace('=\r *\r=i', "\n", $message);
 685                 $message = str_replace("\r", "\n", $message);
 686
 687                 $message = strip_tags($message);
 688
 689                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 690
 691                 if (!$compact && ($message != '')) {
 692                         foreach ($urls as $id => $url) {
 693                                 if ($url != '' && strpos($message, $url) === false) {
 694                                         $message .= "\n" . $url . ' ';
 695                                 }
 696                         }
 697                 }
 698
 699                 $message = str_replace("\n«", "«\n", $message);
 700                 $message = str_replace("»\n", "\n»", $message);
 701
 702                 do {
 703                         $oldmessage = $message;
 704                         $message = str_replace("\n\n\n", "\n\n", $message);
 705                 } while ($oldmessage != $message);
 706
 707                 $message = self::quoteLevel(trim($message), $wraplength);
 708
 709                 return trim($message);
 710         }
 711
 712         /**
 713          * Converts provided HTML code to Markdown. The hardwrap parameter maximizes
 714          * compatibility with Diaspora in spite of the Markdown standards.
 715          *
 716          * @param string $html
 717          * @return string
 718          */
 719         public static function toMarkdown($html)
 720         {
 721                 $converter = new HtmlConverter(['hard_break' => true]);
 722                 $markdown = $converter->convert($html);
 723
 724                 return $markdown;
 725         }
 726
 727         /**
 728          * Convert video HTML to BBCode tags
 729          *
 730          * @param string $s
 731          * @return string
 732          */
 733         public static function toBBCodeVideo($s)
 734         {
 735                 $s = preg_replace(
 736                         '#<object[^>]+>(.*?)https?://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+)(.*?)</object>#ism',
 737                         '[youtube]$2[/youtube]',
 738                         $s
 739                 );
 740
 741                 $s = preg_replace(
 742                         '#<iframe[^>](.*?)https?://www.youtube.com/embed/([A-Za-z0-9\-_=]+)(.*?)</iframe>#ism',
 743                         '[youtube]$2[/youtube]',
 744                         $s
 745                 );
 746
 747                 $s = preg_replace(
 748                         '#<iframe[^>](.*?)https?://player.vimeo.com/video/([0-9]+)(.*?)</iframe>#ism',
 749                         '[vimeo]$2[/vimeo]',
 750                         $s
 751                 );
 752
 753                 return $s;
 754         }
 755
 756         /**
 757          * transform link href and img src from relative to absolute
 758          *
 759          * @param string $text
 760          * @param string $base base url
 761          * @return string
 762          */
 763         public static function relToAbs($text, $base)
 764         {
 765                 if (empty($base)) {
 766                         return $text;
 767                 }
 768
 769                 $base = rtrim($base, '/');
 770
 771                 $base2 = $base . "/";
 772
 773                 // Replace links
 774                 $pattern = "/<a([^>]*) href=\"(?!http|https|\/)([^\"]*)\"/";
 775                 $replace = "<a\${1} href=\"" . $base2 . "\${2}\"";
 776                 $text = preg_replace($pattern, $replace, $text);
 777
 778                 $pattern = "/<a([^>]*) href=\"(?!http|https)([^\"]*)\"/";
 779                 $replace = "<a\${1} href=\"" . $base . "\${2}\"";
 780                 $text = preg_replace($pattern, $replace, $text);
 781
 782                 // Replace images
 783                 $pattern = "/<img([^>]*) src=\"(?!http|https|\/)([^\"]*)\"/";
 784                 $replace = "<img\${1} src=\"" . $base2 . "\${2}\"";
 785                 $text = preg_replace($pattern, $replace, $text);
 786
 787                 $pattern = "/<img([^>]*) src=\"(?!http|https)([^\"]*)\"/";
 788                 $replace = "<img\${1} src=\"" . $base . "\${2}\"";
 789                 $text = preg_replace($pattern, $replace, $text);
 790
 791
 792                 // Done
 793                 return $text;
 794         }
 795
 796         /**
 797          * return div element with class 'clear'
 798          * @return string
 799          * @deprecated
 800          */
 801         public static function clearDiv()
 802         {
 803                 return '<div class="clear"></div>';
 804         }
 805
 806         /**
 807          * Loader for infinite scrolling
 808          *
 809          * @return string html for loader
 810          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 811          */
 812         public static function scrollLoader()
 813         {
 814                 $tpl = Renderer::getMarkupTemplate("scroll_loader.tpl");
 815                 return Renderer::replaceMacros($tpl, [
 816                         'wait' => DI::l10n()->t('Loading more entries...'),
 817                         'end' => DI::l10n()->t('The end')
 818                 ]);
 819         }
 820
 821         /**
 822          * Get html for contact block.
 823          *
 824          * @deprecated since version 2019.03
 825          * @see ContactBlock::getHTML()
 826          * @return string
 827          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 828          * @throws \ImagickException
 829          */
 830         public static function contactBlock()
 831         {
 832                 $a = DI::app();
 833
 834                 return ContactBlock::getHTML($a->profile);
 835         }
 836
 837         /**
 838          * Format contacts as picture links or as text links
 839          *
 840          * @param array   $contact  Array with contacts which contains an array with
 841          *                          int 'id' => The ID of the contact
 842          *                          int 'uid' => The user ID of the user who owns this data
 843          *                          string 'name' => The name of the contact
 844          *                          string 'url' => The url to the profile page of the contact
 845          *                          string 'addr' => The webbie of the contact (e.g.) username@friendica.com
 846          *                          string 'network' => The network to which the contact belongs to
 847          *                          string 'thumb' => The contact picture
 848          *                          string 'click' => js code which is performed when clicking on the contact
 849          * @param boolean $redirect If true try to use the redir url if it's possible
 850          * @param string  $class    CSS class for the
 851          * @param boolean $textmode If true display the contacts as text links
 852          *                          if false display the contacts as picture links
 853          * @return string Formatted html
 854          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 855          * @throws \ImagickException
 856          */
 857         public static function micropro($contact, $redirect = false, $class = '', $textmode = false)
 858         {
 859                 // Use the contact URL if no address is available
 860                 if (empty($contact['addr'])) {
 861                         $contact["addr"] = $contact["url"];
 862                 }
 863
 864                 $url = $contact['url'];
 865                 $sparkle = '';
 866                 $redir = false;
 867
 868                 if ($redirect) {
 869                         $url = Contact::magicLink($contact['url']);
 870                         if (strpos($url, 'redir/') === 0) {
 871                                 $sparkle = ' sparkle';
 872                         }
 873                 }
 874
 875                 // If there is some js available we don't need the url
 876                 if (!empty($contact['click'])) {
 877                         $url = '';
 878                 }
 879
 880                 return Renderer::replaceMacros(Renderer::getMarkupTemplate($textmode ? 'micropro_txt.tpl' : 'micropro_img.tpl'), [
 881                         '$click' => $contact['click'] ?? '',
 882                         '$class' => $class,
 883                         '$url' => $url,
 884                         '$photo' => ProxyUtils::proxifyUrl($contact['thumb'], false, ProxyUtils::SIZE_THUMB),
 885                         '$name' => $contact['name'],
 886                         'title' => $contact['name'] . ' [' . $contact['addr'] . ']',
 887                         '$parkle' => $sparkle,
 888                         '$redir' => $redir
 889                 ]);
 890         }
 891
 892         /**
 893          * Search box.
 894          *
 895          * @param string $s     Search query.
 896          * @param string $id    HTML id
 897          * @param string $url   Search url.
 898          * @param bool   $aside Display the search widgit aside.
 899          *
 900          * @return string Formatted HTML.
 901          * @throws \Exception
 902          */
 903         public static function search($s, $id = 'search-box', $aside = true)
 904         {
 905                 $mode = 'text';
 906
 907                 if (strpos($s, '#') === 0) {
 908                         $mode = 'tag';
 909                 }
 910                 $save_label = $mode === 'text' ? DI::l10n()->t('Save') : DI::l10n()->t('Follow');
 911
 912                 $values = [
 913                         '$s'            => $s,
 914                         '$q'            => urlencode($s),
 915                         '$id'           => $id,
 916                         '$search_label' => DI::l10n()->t('Search'),
 917                         '$save_label'   => $save_label,
 918                         '$search_hint'  => DI::l10n()->t('@name, !forum, #tags, content'),
 919                         '$mode'         => $mode,
 920                         '$return_url'   => urlencode('search?q=' . urlencode($s)),
 921                 ];
 922
 923                 if (!$aside) {
 924                         $values['$search_options'] = [
 925                                 'fulltext' => DI::l10n()->t('Full Text'),
 926                                 'tags'     => DI::l10n()->t('Tags'),
 927                                 'contacts' => DI::l10n()->t('Contacts')
 928                         ];
 929
 930                         if (DI::config()->get('system', 'poco_local_search')) {
 931                                 $values['$searchoption']['forums'] = DI::l10n()->t('Forums');
 932                         }
 933                 }
 934
 935                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 936         }
 937
 938         /**
 939          * Replace naked text hyperlink with HTML formatted hyperlink
 940          *
 941          * @param string $s
 942          * @return string
 943          */
 944         public static function toLink($s)
 945         {
 946                 $s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank" rel="noopener noreferrer">$1</a>', $s);
 947                 $s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
 948                 return $s;
 949         }
 950
 951         /**
 952          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
 953          *
 954          * Reasons are expected to have been translated already.
 955          *
 956          * @param string $html
 957          * @param array  $reasons
 958          * @return string
 959          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 960          */
 961         public static function applyContentFilter($html, array $reasons)
 962         {
 963                 if (count($reasons)) {
 964                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
 965                         $html = Renderer::replaceMacros($tpl, [
 966                                 '$reasons'   => $reasons,
 967                                 '$rnd'       => Strings::getRandomHex(8),
 968                                 '$openclose' => DI::l10n()->t('Click to open/close'),
 969                                 '$html'      => $html
 970                         ]);
 971                 }
 972
 973                 return $html;
 974         }
 975
 976         /**
 977          * replace html amp entity with amp char
 978          * @param string $s
 979          * @return string
 980          */
 981         public static function unamp($s)
 982         {
 983                 return str_replace('&amp;', '&', $s);
 984         }
 985 }