src/Content/Text/HTML.php

   1 <?php
   2 /**
   3  * @file src/Content/Text/HTML.php
   4  */
   5
   6 namespace Friendica\Content\Text;
   7
   8 use DOMDocument;
   9 use DOMXPath;
  10 use Friendica\Content\Widget\ContactBlock;
  11 use Friendica\Core\Hook;
  12 use Friendica\Core\Renderer;
  13 use Friendica\DI;
  14 use Friendica\Model\Contact;
  15 use Friendica\Util\Network;
  16 use Friendica\Util\Proxy as ProxyUtils;
  17 use Friendica\Util\Strings;
  18 use Friendica\Util\XML;
  19 use League\HTMLToMarkdown\HtmlConverter;
  20
  21 class HTML
  22 {
  23         public static function sanitizeCSS($input)
  24         {
  25                 $cleaned = "";
  26
  27                 $input = strtolower($input);
  28
  29                 for ($i = 0; $i < strlen($input); $i++) {
  30                         $char = substr($input, $i, 1);
  31
  32                         if (($char >= "a") && ($char <= "z")) {
  33                                 $cleaned .= $char;
  34                         }
  35
  36                         if (!(strpos(" #;:0123456789-_.%", $char) === false)) {
  37                                 $cleaned .= $char;
  38                         }
  39                 }
  40
  41                 return $cleaned;
  42         }
  43
  44         /**
  45          * Search all instances of a specific HTML tag node in the provided DOM document and replaces them with BBCode text nodes.
  46          *
  47          * @see HTML::tagToBBCodeSub()
  48          */
  49         private static function tagToBBCode(DOMDocument $doc, string $tag, array $attributes, string $startbb, string $endbb, bool $ignoreChildren = false)
  50         {
  51                 do {
  52                         $done = self::tagToBBCodeSub($doc, $tag, $attributes, $startbb, $endbb, $ignoreChildren);
  53                 } while ($done);
  54         }
  55
  56         /**
  57          * Search the first specific HTML tag node in the provided DOM document and replaces it with BBCode text nodes.
  58          *
  59          * @param DOMDocument $doc
  60          * @param string      $tag            HTML tag name
  61          * @param array       $attributes     Array of attributes to match and optionally use the value from
  62          * @param string      $startbb        BBCode tag opening
  63          * @param string      $endbb          BBCode tag closing
  64          * @param bool        $ignoreChildren If set to false, the HTML tag children will be appended as text inside the BBCode tag
  65          *                                    Otherwise, they will be entirely ignored. Useful for simple BBCode that draw their
  66          *                                    inner value from an attribute value and disregard the tag children.
  67          * @return bool Whether a replacement was done
  68          */
  69         private static function tagToBBCodeSub(DOMDocument $doc, string $tag, array $attributes, string $startbb, string $endbb, bool $ignoreChildren = false)
  70         {
  71                 $savestart = str_replace('$', '\x01', $startbb);
  72                 $replace = false;
  73
  74                 $xpath = new DOMXPath($doc);
  75
  76                 /** @var \DOMNode[] $list */
  77                 $list = $xpath->query("//" . $tag);
  78                 foreach ($list as $node) {
  79                         $attr = [];
  80                         if ($node->attributes->length) {
  81                                 foreach ($node->attributes as $attribute) {
  82                                         $attr[$attribute->name] = $attribute->value;
  83                                 }
  84                         }
  85
  86                         $replace = true;
  87
  88                         $startbb = $savestart;
  89
  90                         $i = 0;
  91
  92                         foreach ($attributes as $attribute => $value) {
  93                                 $startbb = str_replace('\x01' . ++$i, '$1', $startbb);
  94                                 if (strpos('*' . $startbb, '$1') > 0) {
  95                                         if ($replace && (@$attr[$attribute] != '')) {
  96                                                 $startbb = preg_replace($value, $startbb, $attr[$attribute], -1, $count);
  97
  98                                                 // If nothing could be changed
  99                                                 if ($count == 0) {
 100                                                         $replace = false;
 101                                                 }
 102                                         } else {
 103                                                 $replace = false;
 104                                         }
 105                                 } else {
 106                                         if (@$attr[$attribute] != $value) {
 107                                                 $replace = false;
 108                                         }
 109                                 }
 110                         }
 111
 112                         if ($replace) {
 113                                 $StartCode = $doc->createTextNode($startbb);
 114                                 $EndCode = $doc->createTextNode($endbb);
 115
 116                                 $node->parentNode->insertBefore($StartCode, $node);
 117
 118                                 if (!$ignoreChildren && $node->hasChildNodes()) {
 119                                         /** @var \DOMNode $child */
 120                                         foreach ($node->childNodes as $key => $child) {
 121                                                 /* Remove empty text nodes at the start or at the end of the children list */
 122                                                 if ($key > 0 && $key < $node->childNodes->length - 1 || $child->nodeName != '#text' || trim($child->nodeValue)) {
 123                                                         $newNode = $child->cloneNode(true);
 124                                                         $node->parentNode->insertBefore($newNode, $node);
 125                                                 }
 126                                         }
 127                                 }
 128
 129                                 $node->parentNode->insertBefore($EndCode, $node);
 130                                 $node->parentNode->removeChild($node);
 131                         }
 132                 }
 133
 134                 return $replace;
 135         }
 136
 137         /**
 138          * Converter for HTML to BBCode
 139          *
 140          * Made by: ike@piratenpartei.de
 141          * Originally made for the syncom project: http://wiki.piratenpartei.de/Syncom
 142          *                    https://github.com/annando/Syncom
 143          *
 144          * @param string $message
 145          * @param string $basepath
 146          * @return string
 147          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 148          */
 149         public static function toBBCode($message, $basepath = '')
 150         {
 151                 $message = str_replace("\r", "", $message);
 152
 153                 // Removing code blocks before the whitespace removal processing below
 154                 $codeblocks = [];
 155                 $message = preg_replace_callback(
 156                         '#<pre><code(?: class="language-([^"]*)")?>(.*)</code></pre>#iUs',
 157                         function ($matches) use (&$codeblocks) {
 158                                 $return = '[codeblock-' . count($codeblocks) . ']';
 159
 160                                 $prefix = '[code]';
 161                                 if ($matches[1] != '') {
 162                                         $prefix = '[code=' . $matches[1] . ']';
 163                                 }
 164
 165                                 $codeblocks[] = $prefix . PHP_EOL . trim($matches[2]) . PHP_EOL . '[/code]';
 166                                 return $return;
 167                         },
 168                         $message
 169                 );
 170
 171                 $message = str_replace(
 172                         [
 173                                 "<li><p>",
 174                                 "</p></li>",
 175                         ],
 176                         [
 177                                 "<li>",
 178                                 "</li>",
 179                         ],
 180                         $message
 181                 );
 182
 183                 // remove namespaces
 184                 $message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
 185                 $message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
 186
 187                 $doc = new DOMDocument();
 188                 $doc->preserveWhiteSpace = false;
 189
 190                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 191
 192                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 193
 194                 XML::deleteNode($doc, 'style');
 195                 XML::deleteNode($doc, 'head');
 196                 XML::deleteNode($doc, 'title');
 197                 XML::deleteNode($doc, 'meta');
 198                 XML::deleteNode($doc, 'xml');
 199                 XML::deleteNode($doc, 'removeme');
 200
 201                 $xpath = new DomXPath($doc);
 202                 $list = $xpath->query("//pre");
 203                 foreach ($list as $node) {
 204                         // Ensure to escape unescaped & - they will otherwise raise a warning
 205                         $safe_value = preg_replace('/&(?!\w+;)/', '&amp;', $node->nodeValue);
 206                         $node->nodeValue = str_replace("\n", "\r", $safe_value);
 207                 }
 208
 209                 $message = $doc->saveHTML();
 210                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br />", " ", ""], $message);
 211                 $message = preg_replace('= [\s]*=i', " ", $message);
 212
 213                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 214
 215                 self::tagToBBCode($doc, 'html', [], "", "");
 216                 self::tagToBBCode($doc, 'body', [], "", "");
 217
 218                 // Outlook-Quote - Variant 1
 219                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal', 'style' => 'margin-left:35.4pt'], '[quote]', '[/quote]');
 220
 221                 // Outlook-Quote - Variant 2
 222                 self::tagToBBCode(
 223                         $doc,
 224                         'div',
 225                         ['style' => 'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'],
 226                         '[quote]',
 227                         '[/quote]'
 228                 );
 229
 230                 // MyBB-Stuff
 231                 self::tagToBBCode($doc, 'span', ['style' => 'text-decoration: underline;'], '[u]', '[/u]');
 232                 self::tagToBBCode($doc, 'span', ['style' => 'font-style: italic;'], '[i]', '[/i]');
 233                 self::tagToBBCode($doc, 'span', ['style' => 'font-weight: bold;'], '[b]', '[/b]');
 234
 235                 /* self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
 236                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
 237                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
 238                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
 239                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
 240                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
 241                   self::node2BBCode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
 242                  */
 243                 // Untested
 244                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
 245                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
 246                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
 247
 248                 self::tagToBBCode($doc, 'span', ['style' => '/.*color:\s*(.+?)[,;].*/'], '[color="$1"]', '[/color]');
 249
 250                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 251                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
 252                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
 253                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 254                 // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
 255                 // Test
 256                 //self::node2BBCode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
 257                 self::tagToBBCode($doc, 'span', ['class' => 'type-link'], '[class=type-link]', '[/class]');
 258                 self::tagToBBCode($doc, 'span', ['class' => 'type-video'], '[class=type-video]', '[/class]');
 259
 260                 self::tagToBBCode($doc, 'strong', [], '[b]', '[/b]');
 261                 self::tagToBBCode($doc, 'em', [], '[i]', '[/i]');
 262                 self::tagToBBCode($doc, 'b', [], '[b]', '[/b]');
 263                 self::tagToBBCode($doc, 'i', [], '[i]', '[/i]');
 264                 self::tagToBBCode($doc, 'u', [], '[u]', '[/u]');
 265                 self::tagToBBCode($doc, 's', [], '[s]', '[/s]');
 266                 self::tagToBBCode($doc, 'del', [], '[s]', '[/s]');
 267                 self::tagToBBCode($doc, 'strike', [], '[s]', '[/s]');
 268
 269                 self::tagToBBCode($doc, 'big', [], "[size=large]", "[/size]");
 270                 self::tagToBBCode($doc, 'small', [], "[size=small]", "[/size]");
 271
 272                 self::tagToBBCode($doc, 'blockquote', [], '[quote]', '[/quote]');
 273
 274                 self::tagToBBCode($doc, 'br', [], "\n", '');
 275
 276                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal'], "\n", "");
 277                 self::tagToBBCode($doc, 'div', ['class' => 'MsoNormal'], "\r", "");
 278
 279                 self::tagToBBCode($doc, 'span', [], "", "");
 280
 281                 self::tagToBBCode($doc, 'span', [], "", "");
 282                 self::tagToBBCode($doc, 'pre', [], "", "");
 283
 284                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 285                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 286
 287                 self::tagToBBCode($doc, 'ul', [], "[list]", "[/list]");
 288                 self::tagToBBCode($doc, 'ol', [], "[list=1]", "[/list]");
 289                 self::tagToBBCode($doc, 'li', [], "[*]", "");
 290
 291                 self::tagToBBCode($doc, 'hr', [], "[hr]", "");
 292
 293                 self::tagToBBCode($doc, 'table', [], "[table]", "[/table]");
 294                 self::tagToBBCode($doc, 'th', [], "[th]", "[/th]");
 295                 self::tagToBBCode($doc, 'tr', [], "[tr]", "[/tr]");
 296                 self::tagToBBCode($doc, 'td', [], "[td]", "[/td]");
 297
 298                 self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 299                 self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 300                 self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 301                 self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 302                 self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 303                 self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 304
 305                 self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 306                 self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 307
 308                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'alt' => '/(.+)/'], '[img=$1]$2', '[/img]', true);
 309                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]', true);
 310                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]', true);
 311
 312
 313                 self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]', true);
 314                 self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]', true);
 315                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[iframe]$1', '[/iframe]', true);
 316
 317                 self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 318                 self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 319
 320                 $message = $doc->saveHTML();
 321
 322                 // I'm removing something really disturbing
 323                 // Don't know exactly what it is
 324                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 325
 326                 $message = str_replace("&nbsp;", " ", $message);
 327
 328                 // removing multiple DIVs
 329                 $message = preg_replace('=\r *\r=i', "\n", $message);
 330                 $message = str_replace("\r", "\n", $message);
 331
 332                 Hook::callAll('html2bbcode', $message);
 333
 334                 $message = strip_tags($message);
 335
 336                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 337
 338                 // remove quotes if they don't make sense
 339                 $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 340
 341                 $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 342                 $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 343
 344                 do {
 345                         $oldmessage = $message;
 346                         $message = str_replace("\n \n", "\n\n", $message);
 347                 } while ($oldmessage != $message);
 348
 349                 do {
 350                         $oldmessage = $message;
 351                         $message = str_replace("\n\n\n", "\n\n", $message);
 352                 } while ($oldmessage != $message);
 353
 354                 do {
 355                         $oldmessage = $message;
 356                         $message = str_replace(
 357                                 [
 358                                 "[/size]\n\n",
 359                                 "\n[hr]",
 360                                 "[hr]\n",
 361                                 "\n[list",
 362                                 "[/list]\n",
 363                                 "\n[/",
 364                                 "[list]\n",
 365                                 "[list=1]\n",
 366                                 "\n[*]"],
 367                                 [
 368                                 "[/size]\n",
 369                                 "[hr]",
 370                                 "[hr]",
 371                                 "[list",
 372                                 "[/list]",
 373                                 "[/",
 374                                 "[list]",
 375                                 "[list=1]",
 376                                 "[*]"],
 377                                 $message
 378                         );
 379                 } while ($message != $oldmessage);
 380
 381                 $message = str_replace(
 382                         ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 383                         ['[b]', '[/b]', '[i]', '[/i]'],
 384                         $message
 385                 );
 386
 387                 // Handling Yahoo style of mails
 388                 $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 389
 390                 // Restore code blocks
 391                 $message = preg_replace_callback(
 392                         '#\[codeblock-([0-9]+)\]#iU',
 393                         function ($matches) use ($codeblocks) {
 394                                 $return = '';
 395                                 if (isset($codeblocks[intval($matches[1])])) {
 396                                         $return = $codeblocks[$matches[1]];
 397                                 }
 398                                 return $return;
 399                         },
 400                         $message
 401                 );
 402
 403                 $message = trim($message);
 404
 405                 if ($basepath != '') {
 406                         $message = self::qualifyURLs($message, $basepath);
 407                 }
 408
 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 $basepath Basepath that is used to complete the URL
 417          *
 418          * @return string The expanded URL
 419          */
 420         private static function qualifyURLsSub($matches, $basepath)
 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 $basepath Base path that is used to complete the URL
 444          *
 445          * @return string Body with expanded URLs
 446          */
 447         private static function qualifyURLs($body, $basepath)
 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 $body;
 471         }
 472
 473         private static function breakLines($line, $level, $wraplength = 75)
 474         {
 475                 if ($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", $newlines);
 512         }
 513
 514         private static function quoteLevel($message, $wraplength = 75)
 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", $newlines);
 548         }
 549
 550         private static function collectURLs($message)
 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, $wraplength = 75, $compact = false)
 597         {
 598                 $message = str_replace("\r", "", $html);
 599
 600                 $doc = new DOMDocument();
 601                 $doc->preserveWhiteSpace = false;
 602
 603                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 604
 605                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 606
 607                 $message = $doc->saveHTML();
 608                 // Remove eventual UTF-8 BOM
 609                 $message = str_replace("\xC3\x82\xC2\xA0", "", $message);
 610
 611                 // Collecting all links
 612                 $urls = self::collectURLs($message);
 613
 614                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 615
 616                 self::tagToBBCode($doc, 'html', [], '', '');
 617                 self::tagToBBCode($doc, 'body', [], '', '');
 618
 619                 if ($compact) {
 620                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 621                 } else {
 622                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 623                 }
 624
 625                 self::tagToBBCode($doc, 'br', [], "\n", '');
 626
 627                 self::tagToBBCode($doc, 'span', [], "", "");
 628                 self::tagToBBCode($doc, 'pre', [], "", "");
 629                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 630                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 631
 632                 self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 633
 634                 self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
 635
 636                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 637                 self::tagToBBCode($doc, 'td', [], "\t", "");
 638
 639                 self::tagToBBCode($doc, 'h1', [], "\n\n*", "*\n");
 640                 self::tagToBBCode($doc, 'h2', [], "\n\n*", "*\n");
 641                 self::tagToBBCode($doc, 'h3', [], "\n\n*", "*\n");
 642                 self::tagToBBCode($doc, 'h4', [], "\n\n*", "*\n");
 643                 self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
 644                 self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 645
 646                 if (!$compact) {
 647                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
 648                 } else {
 649                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' ', ' ');
 650                 }
 651
 652                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], ' $1 ', '');
 653
 654                 $message = $doc->saveHTML();
 655
 656                 if (!$compact) {
 657                         $message = str_replace("[img]", "", $message);
 658                         $message = str_replace("[/img]", "", $message);
 659                 }
 660
 661                 // was ersetze ich da?
 662                 // Irgendein stoerrisches UTF-Zeug
 663                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 664
 665                 $message = str_replace("&nbsp;", " ", $message);
 666
 667                 // Aufeinanderfolgende DIVs
 668                 $message = preg_replace('=\r *\r=i', "\n", $message);
 669                 $message = str_replace("\r", "\n", $message);
 670
 671                 $message = strip_tags($message);
 672
 673                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 674
 675                 if (!$compact && ($message != '')) {
 676                         foreach ($urls as $id => $url) {
 677                                 if ($url != '' && strpos($message, $url) === false) {
 678                                         $message .= "\n" . $url . ' ';
 679                                 }
 680                         }
 681                 }
 682
 683                 $message = str_replace("\n«", "«\n", $message);
 684                 $message = str_replace("»\n", "\n»", $message);
 685
 686                 do {
 687                         $oldmessage = $message;
 688                         $message = str_replace("\n\n\n", "\n\n", $message);
 689                 } while ($oldmessage != $message);
 690
 691                 $message = self::quoteLevel(trim($message), $wraplength);
 692
 693                 return trim($message);
 694         }
 695
 696         /**
 697          * Converts provided HTML code to Markdown. The hardwrap parameter maximizes
 698          * compatibility with Diaspora in spite of the Markdown standards.
 699          *
 700          * @param string $html
 701          * @return string
 702          */
 703         public static function toMarkdown($html)
 704         {
 705                 $converter = new HtmlConverter(['hard_break' => true]);
 706                 $markdown = $converter->convert($html);
 707
 708                 return $markdown;
 709         }
 710
 711         /**
 712          * Convert video HTML to BBCode tags
 713          *
 714          * @param string $s
 715          * @return string
 716          */
 717         public static function toBBCodeVideo($s)
 718         {
 719                 $s = preg_replace(
 720                         '#<object[^>]+>(.*?)https?://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+)(.*?)</object>#ism',
 721                         '[youtube]$2[/youtube]',
 722                         $s
 723                 );
 724
 725                 $s = preg_replace(
 726                         '#<iframe[^>](.*?)https?://www.youtube.com/embed/([A-Za-z0-9\-_=]+)(.*?)</iframe>#ism',
 727                         '[youtube]$2[/youtube]',
 728                         $s
 729                 );
 730
 731                 $s = preg_replace(
 732                         '#<iframe[^>](.*?)https?://player.vimeo.com/video/([0-9]+)(.*?)</iframe>#ism',
 733                         '[vimeo]$2[/vimeo]',
 734                         $s
 735                 );
 736
 737                 return $s;
 738         }
 739
 740         /**
 741          * transform link href and img src from relative to absolute
 742          *
 743          * @param string $text
 744          * @param string $base base url
 745          * @return string
 746          */
 747         public static function relToAbs($text, $base)
 748         {
 749                 if (empty($base)) {
 750                         return $text;
 751                 }
 752
 753                 $base = rtrim($base, '/');
 754
 755                 $base2 = $base . "/";
 756
 757                 // Replace links
 758                 $pattern = "/<a([^>]*) href=\"(?!http|https|\/)([^\"]*)\"/";
 759                 $replace = "<a\${1} href=\"" . $base2 . "\${2}\"";
 760                 $text = preg_replace($pattern, $replace, $text);
 761
 762                 $pattern = "/<a([^>]*) href=\"(?!http|https)([^\"]*)\"/";
 763                 $replace = "<a\${1} href=\"" . $base . "\${2}\"";
 764                 $text = preg_replace($pattern, $replace, $text);
 765
 766                 // Replace images
 767                 $pattern = "/<img([^>]*) src=\"(?!http|https|\/)([^\"]*)\"/";
 768                 $replace = "<img\${1} src=\"" . $base2 . "\${2}\"";
 769                 $text = preg_replace($pattern, $replace, $text);
 770
 771                 $pattern = "/<img([^>]*) src=\"(?!http|https)([^\"]*)\"/";
 772                 $replace = "<img\${1} src=\"" . $base . "\${2}\"";
 773                 $text = preg_replace($pattern, $replace, $text);
 774
 775
 776                 // Done
 777                 return $text;
 778         }
 779
 780         /**
 781          * return div element with class 'clear'
 782          * @return string
 783          * @deprecated
 784          */
 785         public static function clearDiv()
 786         {
 787                 return '<div class="clear"></div>';
 788         }
 789
 790         /**
 791          * Loader for infinite scrolling
 792          *
 793          * @return string html for loader
 794          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 795          */
 796         public static function scrollLoader()
 797         {
 798                 $tpl = Renderer::getMarkupTemplate("scroll_loader.tpl");
 799                 return Renderer::replaceMacros($tpl, [
 800                         'wait' => DI::l10n()->t('Loading more entries...'),
 801                         'end' => DI::l10n()->t('The end')
 802                 ]);
 803         }
 804
 805         /**
 806          * Get html for contact block.
 807          *
 808          * @deprecated since version 2019.03
 809          * @see ContactBlock::getHTML()
 810          * @return string
 811          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 812          * @throws \ImagickException
 813          */
 814         public static function contactBlock()
 815         {
 816                 $a = DI::app();
 817
 818                 return ContactBlock::getHTML($a->profile);
 819         }
 820
 821         /**
 822          * Format contacts as picture links or as text links
 823          *
 824          * @param array   $contact  Array with contacts which contains an array with
 825          *                          int 'id' => The ID of the contact
 826          *                          int 'uid' => The user ID of the user who owns this data
 827          *                          string 'name' => The name of the contact
 828          *                          string 'url' => The url to the profile page of the contact
 829          *                          string 'addr' => The webbie of the contact (e.g.) username@friendica.com
 830          *                          string 'network' => The network to which the contact belongs to
 831          *                          string 'thumb' => The contact picture
 832          *                          string 'click' => js code which is performed when clicking on the contact
 833          * @param boolean $redirect If true try to use the redir url if it's possible
 834          * @param string  $class    CSS class for the
 835          * @param boolean $textmode If true display the contacts as text links
 836          *                          if false display the contacts as picture links
 837          * @return string Formatted html
 838          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 839          * @throws \ImagickException
 840          */
 841         public static function micropro($contact, $redirect = false, $class = '', $textmode = false)
 842         {
 843                 // Use the contact URL if no address is available
 844                 if (empty($contact['addr'])) {
 845                         $contact["addr"] = $contact["url"];
 846                 }
 847
 848                 $url = $contact['url'];
 849                 $sparkle = '';
 850                 $redir = false;
 851
 852                 if ($redirect) {
 853                         $url = Contact::magicLink($contact['url']);
 854                         if (strpos($url, 'redir/') === 0) {
 855                                 $sparkle = ' sparkle';
 856                         }
 857                 }
 858
 859                 // If there is some js available we don't need the url
 860                 if (!empty($contact['click'])) {
 861                         $url = '';
 862                 }
 863
 864                 return Renderer::replaceMacros(Renderer::getMarkupTemplate($textmode ? 'micropro_txt.tpl' : 'micropro_img.tpl'), [
 865                         '$click' => $contact['click'] ?? '',
 866                         '$class' => $class,
 867                         '$url' => $url,
 868                         '$photo' => ProxyUtils::proxifyUrl($contact['thumb'], false, ProxyUtils::SIZE_THUMB),
 869                         '$name' => $contact['name'],
 870                         'title' => $contact['name'] . ' [' . $contact['addr'] . ']',
 871                         '$parkle' => $sparkle,
 872                         '$redir' => $redir
 873                 ]);
 874         }
 875
 876         /**
 877          * Search box.
 878          *
 879          * @param string $s     Search query.
 880          * @param string $id    HTML id
 881          * @param string $url   Search url.
 882          * @param bool   $aside Display the search widgit aside.
 883          *
 884          * @return string Formatted HTML.
 885          * @throws \Exception
 886          */
 887         public static function search($s, $id = 'search-box', $aside = true)
 888         {
 889                 $mode = 'text';
 890
 891                 if (strpos($s, '#') === 0) {
 892                         $mode = 'tag';
 893                 }
 894                 $save_label = $mode === 'text' ? DI::l10n()->t('Save') : DI::l10n()->t('Follow');
 895
 896                 $values = [
 897                         '$s'            => $s,
 898                         '$q'            => urlencode($s),
 899                         '$id'           => $id,
 900                         '$search_label' => DI::l10n()->t('Search'),
 901                         '$save_label'   => $save_label,
 902                         '$search_hint'  => DI::l10n()->t('@name, !forum, #tags, content'),
 903                         '$mode'         => $mode,
 904                         '$return_url'   => urlencode('search?q=' . urlencode($s)),
 905                 ];
 906
 907                 if (!$aside) {
 908                         $values['$search_options'] = [
 909                                 'fulltext' => DI::l10n()->t('Full Text'),
 910                                 'tags'     => DI::l10n()->t('Tags'),
 911                                 'contacts' => DI::l10n()->t('Contacts')
 912                         ];
 913
 914                         if (DI::config()->get('system', 'poco_local_search')) {
 915                                 $values['$searchoption']['forums'] = DI::l10n()->t('Forums');
 916                         }
 917                 }
 918
 919                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 920         }
 921
 922         /**
 923          * Replace naked text hyperlink with HTML formatted hyperlink
 924          *
 925          * @param string $s
 926          * @return string
 927          */
 928         public static function toLink($s)
 929         {
 930                 $s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank">$1</a>', $s);
 931                 $s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
 932                 return $s;
 933         }
 934
 935         /**
 936          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
 937          *
 938          * Reasons are expected to have been translated already.
 939          *
 940          * @param string $html
 941          * @param array  $reasons
 942          * @return string
 943          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 944          */
 945         public static function applyContentFilter($html, array $reasons)
 946         {
 947                 if (count($reasons)) {
 948                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
 949                         $html = Renderer::replaceMacros($tpl, [
 950                                 '$reasons'   => $reasons,
 951                                 '$rnd'       => Strings::getRandomHex(8),
 952                                 '$openclose' => DI::l10n()->t('Click to open/close'),
 953                                 '$html'      => $html
 954                         ]);
 955                 }
 956
 957                 return $html;
 958         }
 959
 960         /**
 961          * replace html amp entity with amp char
 962          * @param string $s
 963          * @return string
 964          */
 965         public static function unamp($s)
 966         {
 967                 return str_replace('&amp;', '&', $s);
 968         }
 969 }