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\L10n;
  13 use Friendica\Core\Config;
  14 use Friendica\Core\Renderer;
  15 use Friendica\DI;
  16 use Friendica\Model\Contact;
  17 use Friendica\Util\Network;
  18 use Friendica\Util\Proxy as ProxyUtils;
  19 use Friendica\Util\Strings;
  20 use Friendica\Util\XML;
  21 use League\HTMLToMarkdown\HtmlConverter;
  22
  23 class HTML
  24 {
  25         public static function sanitizeCSS($input)
  26         {
  27                 $cleaned = "";
  28
  29                 $input = strtolower($input);
  30
  31                 for ($i = 0; $i < strlen($input); $i++) {
  32                         $char = substr($input, $i, 1);
  33
  34                         if (($char >= "a") && ($char <= "z")) {
  35                                 $cleaned .= $char;
  36                         }
  37
  38                         if (!(strpos(" #;:0123456789-_.%", $char) === false)) {
  39                                 $cleaned .= $char;
  40                         }
  41                 }
  42
  43                 return $cleaned;
  44         }
  45
  46         /**
  47          * Search all instances of a specific HTML tag node in the provided DOM document and replaces them with BBCode text nodes.
  48          *
  49          * @see HTML::tagToBBCodeSub()
  50          */
  51         private static function tagToBBCode(DOMDocument $doc, string $tag, array $attributes, string $startbb, string $endbb, bool $ignoreChildren = false)
  52         {
  53                 do {
  54                         $done = self::tagToBBCodeSub($doc, $tag, $attributes, $startbb, $endbb, $ignoreChildren);
  55                 } while ($done);
  56         }
  57
  58         /**
  59          * Search the first specific HTML tag node in the provided DOM document and replaces it with BBCode text nodes.
  60          *
  61          * @param DOMDocument $doc
  62          * @param string      $tag            HTML tag name
  63          * @param array       $attributes     Array of attributes to match and optionally use the value from
  64          * @param string      $startbb        BBCode tag opening
  65          * @param string      $endbb          BBCode tag closing
  66          * @param bool        $ignoreChildren If set to false, the HTML tag children will be appended as text inside the BBCode tag
  67          *                                    Otherwise, they will be entirely ignored. Useful for simple BBCode that draw their
  68          *                                    inner value from an attribute value and disregard the tag children.
  69          * @return bool Whether a replacement was done
  70          */
  71         private static function tagToBBCodeSub(DOMDocument $doc, string $tag, array $attributes, string $startbb, string $endbb, bool $ignoreChildren = false)
  72         {
  73                 $savestart = str_replace('$', '\x01', $startbb);
  74                 $replace = false;
  75
  76                 $xpath = new DOMXPath($doc);
  77
  78                 /** @var \DOMNode[] $list */
  79                 $list = $xpath->query("//" . $tag);
  80                 foreach ($list as $node) {
  81                         $attr = [];
  82                         if ($node->attributes->length) {
  83                                 foreach ($node->attributes as $attribute) {
  84                                         $attr[$attribute->name] = $attribute->value;
  85                                 }
  86                         }
  87
  88                         $replace = true;
  89
  90                         $startbb = $savestart;
  91
  92                         $i = 0;
  93
  94                         foreach ($attributes as $attribute => $value) {
  95                                 $startbb = str_replace('\x01' . ++$i, '$1', $startbb);
  96                                 if (strpos('*' . $startbb, '$1') > 0) {
  97                                         if ($replace && (@$attr[$attribute] != '')) {
  98                                                 $startbb = preg_replace($value, $startbb, $attr[$attribute], -1, $count);
  99
 100                                                 // If nothing could be changed
 101                                                 if ($count == 0) {
 102                                                         $replace = false;
 103                                                 }
 104                                         } else {
 105                                                 $replace = false;
 106                                         }
 107                                 } else {
 108                                         if (@$attr[$attribute] != $value) {
 109                                                 $replace = false;
 110                                         }
 111                                 }
 112                         }
 113
 114                         if ($replace) {
 115                                 $StartCode = $doc->createTextNode($startbb);
 116                                 $EndCode = $doc->createTextNode($endbb);
 117
 118                                 $node->parentNode->insertBefore($StartCode, $node);
 119
 120                                 if (!$ignoreChildren && $node->hasChildNodes()) {
 121                                         /** @var \DOMNode $child */
 122                                         foreach ($node->childNodes as $key => $child) {
 123                                                 /* Remove empty text nodes at the start or at the end of the children list */
 124                                                 if ($key > 0 && $key < $node->childNodes->length - 1 || $child->nodeName != '#text' || trim($child->nodeValue)) {
 125                                                         $newNode = $child->cloneNode(true);
 126                                                         $node->parentNode->insertBefore($newNode, $node);
 127                                                 }
 128                                         }
 129                                 }
 130
 131                                 $node->parentNode->insertBefore($EndCode, $node);
 132                                 $node->parentNode->removeChild($node);
 133                         }
 134                 }
 135
 136                 return $replace;
 137         }
 138
 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          * @brief Converter for HTML to BBCode
 145          * @param string $message
 146          * @param string $basepath
 147          * @return string
 148          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 149          */
 150         public static function toBBCode($message, $basepath = '')
 151         {
 152                 $message = str_replace("\r", "", $message);
 153
 154                 // Removing code blocks before the whitespace removal processing below
 155                 $codeblocks = [];
 156                 $message = preg_replace_callback(
 157                         '#<pre><code(?: class="language-([^"]*)")?>(.*)</code></pre>#iUs',
 158                         function ($matches) use (&$codeblocks) {
 159                                 $return = '[codeblock-' . count($codeblocks) . ']';
 160
 161                                 $prefix = '[code]';
 162                                 if ($matches[1] != '') {
 163                                         $prefix = '[code=' . $matches[1] . ']';
 164                                 }
 165
 166                                 $codeblocks[] = $prefix . PHP_EOL . trim($matches[2]) . PHP_EOL . '[/code]';
 167                                 return $return;
 168                         },
 169                         $message
 170                 );
 171
 172                 $message = str_replace(
 173                         [
 174                                 "<li><p>",
 175                                 "</p></li>",
 176                         ],
 177                         [
 178                                 "<li>",
 179                                 "</li>",
 180                         ],
 181                         $message
 182                 );
 183
 184                 // remove namespaces
 185                 $message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
 186                 $message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
 187
 188                 $doc = new DOMDocument();
 189                 $doc->preserveWhiteSpace = false;
 190
 191                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 192
 193                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 194
 195                 XML::deleteNode($doc, 'style');
 196                 XML::deleteNode($doc, 'head');
 197                 XML::deleteNode($doc, 'title');
 198                 XML::deleteNode($doc, 'meta');
 199                 XML::deleteNode($doc, 'xml');
 200                 XML::deleteNode($doc, 'removeme');
 201
 202                 $xpath = new DomXPath($doc);
 203                 $list = $xpath->query("//pre");
 204                 foreach ($list as $node) {
 205                         // Ensure to escape unescaped & - they will otherwise raise a warning
 206                         $safe_value = preg_replace('/&(?!\w+;)/', '&amp;', $node->nodeValue);
 207                         $node->nodeValue = str_replace("\n", "\r", $safe_value);
 208                 }
 209
 210                 $message = $doc->saveHTML();
 211                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br />", " ", ""], $message);
 212                 $message = preg_replace('= [\s]*=i', " ", $message);
 213
 214                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 215
 216                 self::tagToBBCode($doc, 'html', [], "", "");
 217                 self::tagToBBCode($doc, 'body', [], "", "");
 218
 219                 // Outlook-Quote - Variant 1
 220                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal', 'style' => 'margin-left:35.4pt'], '[quote]', '[/quote]');
 221
 222                 // Outlook-Quote - Variant 2
 223                 self::tagToBBCode(
 224                         $doc,
 225                         'div',
 226                         ['style' => 'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'],
 227                         '[quote]',
 228                         '[/quote]'
 229                 );
 230
 231                 // MyBB-Stuff
 232                 self::tagToBBCode($doc, 'span', ['style' => 'text-decoration: underline;'], '[u]', '[/u]');
 233                 self::tagToBBCode($doc, 'span', ['style' => 'font-style: italic;'], '[i]', '[/i]');
 234                 self::tagToBBCode($doc, 'span', ['style' => 'font-weight: bold;'], '[b]', '[/b]');
 235
 236                 /* self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
 237                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
 238                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
 239                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
 240                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
 241                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
 242                   self::node2BBCode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
 243                  */
 244                 // Untested
 245                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
 246                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
 247                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
 248
 249                 self::tagToBBCode($doc, 'span', ['style' => '/.*color:\s*(.+?)[,;].*/'], '[color="$1"]', '[/color]');
 250
 251                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 252                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
 253                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
 254                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 255                 // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
 256                 // Test
 257                 //self::node2BBCode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
 258                 self::tagToBBCode($doc, 'span', ['class' => 'type-link'], '[class=type-link]', '[/class]');
 259                 self::tagToBBCode($doc, 'span', ['class' => 'type-video'], '[class=type-video]', '[/class]');
 260
 261                 self::tagToBBCode($doc, 'strong', [], '[b]', '[/b]');
 262                 self::tagToBBCode($doc, 'em', [], '[i]', '[/i]');
 263                 self::tagToBBCode($doc, 'b', [], '[b]', '[/b]');
 264                 self::tagToBBCode($doc, 'i', [], '[i]', '[/i]');
 265                 self::tagToBBCode($doc, 'u', [], '[u]', '[/u]');
 266                 self::tagToBBCode($doc, 's', [], '[s]', '[/s]');
 267                 self::tagToBBCode($doc, 'del', [], '[s]', '[/s]');
 268                 self::tagToBBCode($doc, 'strike', [], '[s]', '[/s]');
 269
 270                 self::tagToBBCode($doc, 'big', [], "[size=large]", "[/size]");
 271                 self::tagToBBCode($doc, 'small', [], "[size=small]", "[/size]");
 272
 273                 self::tagToBBCode($doc, 'blockquote', [], '[quote]', '[/quote]');
 274
 275                 self::tagToBBCode($doc, 'br', [], "\n", '');
 276
 277                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal'], "\n", "");
 278                 self::tagToBBCode($doc, 'div', ['class' => 'MsoNormal'], "\r", "");
 279
 280                 self::tagToBBCode($doc, 'span', [], "", "");
 281
 282                 self::tagToBBCode($doc, 'span', [], "", "");
 283                 self::tagToBBCode($doc, 'pre', [], "", "");
 284
 285                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 286                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 287
 288                 self::tagToBBCode($doc, 'ul', [], "[list]", "[/list]");
 289                 self::tagToBBCode($doc, 'ol', [], "[list=1]", "[/list]");
 290                 self::tagToBBCode($doc, 'li', [], "[*]", "");
 291
 292                 self::tagToBBCode($doc, 'hr', [], "[hr]", "");
 293
 294                 self::tagToBBCode($doc, 'table', [], "[table]", "[/table]");
 295                 self::tagToBBCode($doc, 'th', [], "[th]", "[/th]");
 296                 self::tagToBBCode($doc, 'tr', [], "[tr]", "[/tr]");
 297                 self::tagToBBCode($doc, 'td', [], "[td]", "[/td]");
 298
 299                 self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 300                 self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 301                 self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 302                 self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 303                 self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 304                 self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 305
 306                 self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 307                 self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 308
 309                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'alt' => '/(.+)/'], '[img=$1]$2', '[/img]', true);
 310                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]', true);
 311                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]', true);
 312
 313
 314                 self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]', true);
 315                 self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]', true);
 316                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[iframe]$1', '[/iframe]', true);
 317
 318                 self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 319                 self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 320
 321                 $message = $doc->saveHTML();
 322
 323                 // I'm removing something really disturbing
 324                 // Don't know exactly what it is
 325                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 326
 327                 $message = str_replace("&nbsp;", " ", $message);
 328
 329                 // removing multiple DIVs
 330                 $message = preg_replace('=\r *\r=i', "\n", $message);
 331                 $message = str_replace("\r", "\n", $message);
 332
 333                 Hook::callAll('html2bbcode', $message);
 334
 335                 $message = strip_tags($message);
 336
 337                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 338
 339                 // remove quotes if they don't make sense
 340                 $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 341
 342                 $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 343                 $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 344
 345                 do {
 346                         $oldmessage = $message;
 347                         $message = str_replace("\n \n", "\n\n", $message);
 348                 } while ($oldmessage != $message);
 349
 350                 do {
 351                         $oldmessage = $message;
 352                         $message = str_replace("\n\n\n", "\n\n", $message);
 353                 } while ($oldmessage != $message);
 354
 355                 do {
 356                         $oldmessage = $message;
 357                         $message = str_replace(
 358                                 [
 359                                 "[/size]\n\n",
 360                                 "\n[hr]",
 361                                 "[hr]\n",
 362                                 "\n[list",
 363                                 "[/list]\n",
 364                                 "\n[/",
 365                                 "[list]\n",
 366                                 "[list=1]\n",
 367                                 "\n[*]"],
 368                                 [
 369                                 "[/size]\n",
 370                                 "[hr]",
 371                                 "[hr]",
 372                                 "[list",
 373                                 "[/list]",
 374                                 "[/",
 375                                 "[list]",
 376                                 "[list=1]",
 377                                 "[*]"],
 378                                 $message
 379                         );
 380                 } while ($message != $oldmessage);
 381
 382                 $message = str_replace(
 383                         ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 384                         ['[b]', '[/b]', '[i]', '[/i]'],
 385                         $message
 386                 );
 387
 388                 // Handling Yahoo style of mails
 389                 $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 390
 391                 // Restore code blocks
 392                 $message = preg_replace_callback(
 393                         '#\[codeblock-([0-9]+)\]#iU',
 394                         function ($matches) use ($codeblocks) {
 395                                 $return = '';
 396                                 if (isset($codeblocks[intval($matches[1])])) {
 397                                         $return = $codeblocks[$matches[1]];
 398                                 }
 399                                 return $return;
 400                         },
 401                         $message
 402                 );
 403
 404                 $message = trim($message);
 405
 406                 if ($basepath != '') {
 407                         $message = self::qualifyURLs($message, $basepath);
 408                 }
 409
 410                 return $message;
 411         }
 412
 413         /**
 414          * @brief Sub function to complete incomplete URL
 415          *
 416          * @param array  $matches  Result of preg_replace_callback
 417          * @param string $basepath Basepath that is used to complete the URL
 418          *
 419          * @return string The expanded URL
 420          */
 421         private static function qualifyURLsSub($matches, $basepath)
 422         {
 423                 $base = parse_url($basepath);
 424                 unset($base['query']);
 425                 unset($base['fragment']);
 426
 427                 $link = $matches[0];
 428                 $url = $matches[1];
 429
 430                 if (empty($url) || empty(parse_url($url))) {
 431                         return $matches[0];
 432                 }
 433
 434                 $parts = array_merge($base, parse_url($url));
 435                 $url2 = Network::unparseURL($parts);
 436
 437                 return str_replace($url, $url2, $link);
 438         }
 439
 440         /**
 441          * @brief Complete incomplete URLs in BBCode
 442          *
 443          * @param string $body     Body with URLs
 444          * @param string $basepath Base path that is used to complete the URL
 445          *
 446          * @return string Body with expanded URLs
 447          */
 448         private static function qualifyURLs($body, $basepath)
 449         {
 450                 $URLSearchString = "^\[\]";
 451
 452                 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
 453                         "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
 454                         "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 455                         "/\[img\](.*?)\[\/img\]/ism",
 456                         "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 457                         "/\[zmg\](.*?)\[\/zmg\]/ism",
 458                         "/\[video\](.*?)\[\/video\]/ism",
 459                         "/\[audio\](.*?)\[\/audio\]/ism",
 460                 ];
 461
 462                 foreach ($matches as $match) {
 463                         $body = preg_replace_callback(
 464                                 $match,
 465                                 function ($match) use ($basepath) {
 466                                         return self::qualifyURLsSub($match, $basepath);
 467                                 },
 468                                 $body
 469                         );
 470                 }
 471                 return $body;
 472         }
 473
 474         private static function breakLines($line, $level, $wraplength = 75)
 475         {
 476                 if ($wraplength == 0) {
 477                         $wraplength = 2000000;
 478                 }
 479
 480                 $wraplen = $wraplength - $level;
 481
 482                 $newlines = [];
 483
 484                 do {
 485                         $oldline = $line;
 486
 487                         $subline = substr($line, 0, $wraplen);
 488
 489                         $pos = strrpos($subline, ' ');
 490
 491                         if ($pos == 0) {
 492                                 $pos = strpos($line, ' ');
 493                         }
 494
 495                         if (($pos > 0) && strlen($line) > $wraplen) {
 496                                 $newline = trim(substr($line, 0, $pos));
 497                                 if ($level > 0) {
 498                                         $newline = str_repeat(">", $level) . ' ' . $newline;
 499                                 }
 500
 501                                 $newlines[] = $newline . " ";
 502                                 $line = substr($line, $pos + 1);
 503                         }
 504                 } while ((strlen($line) > $wraplen) && !($oldline == $line));
 505
 506                 if ($level > 0) {
 507                         $line = str_repeat(">", $level) . ' ' . $line;
 508                 }
 509
 510                 $newlines[] = $line;
 511
 512                 return implode($newlines, "\n");
 513         }
 514
 515         private static function quoteLevel($message, $wraplength = 75)
 516         {
 517                 $lines = explode("\n", $message);
 518
 519                 $newlines = [];
 520                 $level = 0;
 521                 foreach ($lines as $line) {
 522                         $line = trim($line);
 523                         $startquote = false;
 524                         while (strpos("*" . $line, '[quote]') > 0) {
 525                                 $level++;
 526                                 $pos = strpos($line, '[quote]');
 527                                 $line = substr($line, 0, $pos) . substr($line, $pos + 7);
 528                                 $startquote = true;
 529                         }
 530
 531                         $currlevel = $level;
 532
 533                         while (strpos("*" . $line, '[/quote]') > 0) {
 534                                 $level--;
 535                                 if ($level < 0) {
 536                                         $level = 0;
 537                                 }
 538
 539                                 $pos = strpos($line, '[/quote]');
 540                                 $line = substr($line, 0, $pos) . substr($line, $pos + 8);
 541                         }
 542
 543                         if (!$startquote || ($line != '')) {
 544                                 $newlines[] = self::breakLines($line, $currlevel, $wraplength);
 545                         }
 546                 }
 547
 548                 return implode($newlines, "\n");
 549         }
 550
 551         private static function collectURLs($message)
 552         {
 553                 $pattern = '/<a.*?href="(.*?)".*?>(.*?)<\/a>/is';
 554                 preg_match_all($pattern, $message, $result, PREG_SET_ORDER);
 555
 556                 $urls = [];
 557                 foreach ($result as $treffer) {
 558                         $ignore = false;
 559
 560                         // A list of some links that should be ignored
 561                         $list = ["/user/", "/tag/", "/group/", "/profile/", "/search?search=", "/search?tag=", "mailto:", "/u/", "/node/",
 562                                 "//plus.google.com/", "//twitter.com/"];
 563                         foreach ($list as $listitem) {
 564                                 if (strpos($treffer[1], $listitem) !== false) {
 565                                         $ignore = true;
 566                                 }
 567                         }
 568
 569                         if ((strpos($treffer[1], "//twitter.com/") !== false) && (strpos($treffer[1], "/status/") !== false)) {
 570                                 $ignore = false;
 571                         }
 572
 573                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/posts") !== false)) {
 574                                 $ignore = false;
 575                         }
 576
 577                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/photos") !== false)) {
 578                                 $ignore = false;
 579                         }
 580
 581                         $ignore = $ignore || strpos($treffer[1], '#') === 0;
 582
 583                         if (!$ignore) {
 584                                 $urls[$treffer[1]] = $treffer[1];
 585                         }
 586                 }
 587
 588                 return $urls;
 589         }
 590
 591         /**
 592          * @param string $html
 593          * @param int    $wraplength Ensures individual lines aren't longer than this many characters. Doesn't break words.
 594          * @param bool   $compact    True: Completely strips image tags; False: Keeps image URLs
 595          * @return string
 596          */
 597         public static function toPlaintext(string $html, $wraplength = 75, $compact = false)
 598         {
 599                 $message = str_replace("\r", "", $html);
 600
 601                 $doc = new DOMDocument();
 602                 $doc->preserveWhiteSpace = false;
 603
 604                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 605
 606                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 607
 608                 $message = $doc->saveHTML();
 609                 // Remove eventual UTF-8 BOM
 610                 $message = str_replace("\xC3\x82\xC2\xA0", "", $message);
 611
 612                 // Collecting all links
 613                 $urls = self::collectURLs($message);
 614
 615                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 616
 617                 self::tagToBBCode($doc, 'html', [], '', '');
 618                 self::tagToBBCode($doc, 'body', [], '', '');
 619
 620                 if ($compact) {
 621                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 622                 } else {
 623                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 624                 }
 625
 626                 self::tagToBBCode($doc, 'br', [], "\n", '');
 627
 628                 self::tagToBBCode($doc, 'span', [], "", "");
 629                 self::tagToBBCode($doc, 'pre', [], "", "");
 630                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 631                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 632
 633                 self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 634
 635                 self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
 636
 637                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 638                 self::tagToBBCode($doc, 'td', [], "\t", "");
 639
 640                 self::tagToBBCode($doc, 'h1', [], "\n\n*", "*\n");
 641                 self::tagToBBCode($doc, 'h2', [], "\n\n*", "*\n");
 642                 self::tagToBBCode($doc, 'h3', [], "\n\n*", "*\n");
 643                 self::tagToBBCode($doc, 'h4', [], "\n\n*", "*\n");
 644                 self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
 645                 self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 646
 647                 if (!$compact) {
 648                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
 649                 } else {
 650                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' ', ' ');
 651                 }
 652
 653                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], ' $1 ', '');
 654
 655                 $message = $doc->saveHTML();
 656
 657                 if (!$compact) {
 658                         $message = str_replace("[img]", "", $message);
 659                         $message = str_replace("[/img]", "", $message);
 660                 }
 661
 662                 // was ersetze ich da?
 663                 // Irgendein stoerrisches UTF-Zeug
 664                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 665
 666                 $message = str_replace("&nbsp;", " ", $message);
 667
 668                 // Aufeinanderfolgende DIVs
 669                 $message = preg_replace('=\r *\r=i', "\n", $message);
 670                 $message = str_replace("\r", "\n", $message);
 671
 672                 $message = strip_tags($message);
 673
 674                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 675
 676                 if (!$compact && ($message != '')) {
 677                         foreach ($urls as $id => $url) {
 678                                 if ($url != '' && strpos($message, $url) === false) {
 679                                         $message .= "\n" . $url . ' ';
 680                                 }
 681                         }
 682                 }
 683
 684                 $message = str_replace("\n«", "«\n", $message);
 685                 $message = str_replace("»\n", "\n»", $message);
 686
 687                 do {
 688                         $oldmessage = $message;
 689                         $message = str_replace("\n\n\n", "\n\n", $message);
 690                 } while ($oldmessage != $message);
 691
 692                 $message = self::quoteLevel(trim($message), $wraplength);
 693
 694                 return trim($message);
 695         }
 696
 697         /**
 698          * Converts provided HTML code to Markdown. The hardwrap parameter maximizes
 699          * compatibility with Diaspora in spite of the Markdown standards.
 700          *
 701          * @param string $html
 702          * @return string
 703          */
 704         public static function toMarkdown($html)
 705         {
 706                 $converter = new HtmlConverter(['hard_break' => true]);
 707                 $markdown = $converter->convert($html);
 708
 709                 return $markdown;
 710         }
 711
 712         /**
 713          * @brief Convert video HTML to BBCode tags
 714          *
 715          * @param string $s
 716          * @return string
 717          */
 718         public static function toBBCodeVideo($s)
 719         {
 720                 $s = preg_replace(
 721                         '#<object[^>]+>(.*?)https?://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+)(.*?)</object>#ism',
 722                         '[youtube]$2[/youtube]',
 723                         $s
 724                 );
 725
 726                 $s = preg_replace(
 727                         '#<iframe[^>](.*?)https?://www.youtube.com/embed/([A-Za-z0-9\-_=]+)(.*?)</iframe>#ism',
 728                         '[youtube]$2[/youtube]',
 729                         $s
 730                 );
 731
 732                 $s = preg_replace(
 733                         '#<iframe[^>](.*?)https?://player.vimeo.com/video/([0-9]+)(.*?)</iframe>#ism',
 734                         '[vimeo]$2[/vimeo]',
 735                         $s
 736                 );
 737
 738                 return $s;
 739         }
 740
 741         /**
 742          * transform link href and img src from relative to absolute
 743          *
 744          * @param string $text
 745          * @param string $base base url
 746          * @return string
 747          */
 748         public static function relToAbs($text, $base)
 749         {
 750                 if (empty($base)) {
 751                         return $text;
 752                 }
 753
 754                 $base = rtrim($base, '/');
 755
 756                 $base2 = $base . "/";
 757
 758                 // Replace links
 759                 $pattern = "/<a([^>]*) href=\"(?!http|https|\/)([^\"]*)\"/";
 760                 $replace = "<a\${1} href=\"" . $base2 . "\${2}\"";
 761                 $text = preg_replace($pattern, $replace, $text);
 762
 763                 $pattern = "/<a([^>]*) href=\"(?!http|https)([^\"]*)\"/";
 764                 $replace = "<a\${1} href=\"" . $base . "\${2}\"";
 765                 $text = preg_replace($pattern, $replace, $text);
 766
 767                 // Replace images
 768                 $pattern = "/<img([^>]*) src=\"(?!http|https|\/)([^\"]*)\"/";
 769                 $replace = "<img\${1} src=\"" . $base2 . "\${2}\"";
 770                 $text = preg_replace($pattern, $replace, $text);
 771
 772                 $pattern = "/<img([^>]*) src=\"(?!http|https)([^\"]*)\"/";
 773                 $replace = "<img\${1} src=\"" . $base . "\${2}\"";
 774                 $text = preg_replace($pattern, $replace, $text);
 775
 776
 777                 // Done
 778                 return $text;
 779         }
 780
 781         /**
 782          * return div element with class 'clear'
 783          * @return string
 784          * @deprecated
 785          */
 786         public static function clearDiv()
 787         {
 788                 return '<div class="clear"></div>';
 789         }
 790
 791         /**
 792          * Loader for infinite scrolling
 793          *
 794          * @return string html for loader
 795          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 796          */
 797         public static function scrollLoader()
 798         {
 799                 $tpl = Renderer::getMarkupTemplate("scroll_loader.tpl");
 800                 return Renderer::replaceMacros($tpl, [
 801                         'wait' => L10n::t('Loading more entries...'),
 802                         'end' => L10n::t('The end')
 803                 ]);
 804         }
 805
 806         /**
 807          * Get html for contact block.
 808          *
 809          * @deprecated since version 2019.03
 810          * @see ContactBlock::getHTML()
 811          * @return string
 812          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 813          * @throws \ImagickException
 814          */
 815         public static function contactBlock()
 816         {
 817                 $a = DI::app();
 818
 819                 return ContactBlock::getHTML($a->profile);
 820         }
 821
 822         /**
 823          * @brief Format contacts as picture links or as text links
 824          *
 825          * @param array   $contact  Array with contacts which contains an array with
 826          *                          int 'id' => The ID of the contact
 827          *                          int 'uid' => The user ID of the user who owns this data
 828          *                          string 'name' => The name of the contact
 829          *                          string 'url' => The url to the profile page of the contact
 830          *                          string 'addr' => The webbie of the contact (e.g.) username@friendica.com
 831          *                          string 'network' => The network to which the contact belongs to
 832          *                          string 'thumb' => The contact picture
 833          *                          string 'click' => js code which is performed when clicking on the contact
 834          * @param boolean $redirect If true try to use the redir url if it's possible
 835          * @param string  $class    CSS class for the
 836          * @param boolean $textmode If true display the contacts as text links
 837          *                          if false display the contacts as picture links
 838          * @return string Formatted html
 839          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 840          * @throws \ImagickException
 841          */
 842         public static function micropro($contact, $redirect = false, $class = '', $textmode = false)
 843         {
 844                 // Use the contact URL if no address is available
 845                 if (empty($contact['addr'])) {
 846                         $contact["addr"] = $contact["url"];
 847                 }
 848
 849                 $url = $contact['url'];
 850                 $sparkle = '';
 851                 $redir = false;
 852
 853                 if ($redirect) {
 854                         $url = Contact::magicLink($contact['url']);
 855                         if (strpos($url, 'redir/') === 0) {
 856                                 $sparkle = ' sparkle';
 857                         }
 858                 }
 859
 860                 // If there is some js available we don't need the url
 861                 if (!empty($contact['click'])) {
 862                         $url = '';
 863                 }
 864
 865                 return Renderer::replaceMacros(Renderer::getMarkupTemplate($textmode ? 'micropro_txt.tpl' : 'micropro_img.tpl'), [
 866                         '$click' => $contact['click'] ?? '',
 867                         '$class' => $class,
 868                         '$url' => $url,
 869                         '$photo' => ProxyUtils::proxifyUrl($contact['thumb'], false, ProxyUtils::SIZE_THUMB),
 870                         '$name' => $contact['name'],
 871                         'title' => $contact['name'] . ' [' . $contact['addr'] . ']',
 872                         '$parkle' => $sparkle,
 873                         '$redir' => $redir
 874                 ]);
 875         }
 876
 877         /**
 878          * Search box.
 879          *
 880          * @param string $s     Search query.
 881          * @param string $id    HTML id
 882          * @param string $url   Search url.
 883          * @param bool   $aside Display the search widgit aside.
 884          *
 885          * @return string Formatted HTML.
 886          * @throws \Exception
 887          */
 888         public static function search($s, $id = 'search-box', $aside = true)
 889         {
 890                 $mode = 'text';
 891
 892                 if (strpos($s, '#') === 0) {
 893                         $mode = 'tag';
 894                 }
 895                 $save_label = $mode === 'text' ? L10n::t('Save') : L10n::t('Follow');
 896
 897                 $values = [
 898                         '$s'            => $s,
 899                         '$q'            => urlencode($s),
 900                         '$id'           => $id,
 901                         '$search_label' => L10n::t('Search'),
 902                         '$save_label'   => $save_label,
 903                         '$search_hint'  => L10n::t('@name, !forum, #tags, content'),
 904                         '$mode'         => $mode,
 905                         '$return_url'   => urlencode('search?q=' . urlencode($s)),
 906                 ];
 907
 908                 if (!$aside) {
 909                         $values['$search_options'] = [
 910                                 'fulltext' => L10n::t('Full Text'),
 911                                 'tags'     => L10n::t('Tags'),
 912                                 'contacts' => L10n::t('Contacts')
 913                         ];
 914
 915                         if (Config::get('system', 'poco_local_search')) {
 916                                 $values['$searchoption']['forums'] = L10n::t('Forums');
 917                         }
 918                 }
 919
 920                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 921         }
 922
 923         /**
 924          * Replace naked text hyperlink with HTML formatted hyperlink
 925          *
 926          * @param string $s
 927          * @return string
 928          */
 929         public static function toLink($s)
 930         {
 931                 $s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank">$1</a>', $s);
 932                 $s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
 933                 return $s;
 934         }
 935
 936         /**
 937          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
 938          *
 939          * Reasons are expected to have been translated already.
 940          *
 941          * @param string $html
 942          * @param array  $reasons
 943          * @return string
 944          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 945          */
 946         public static function applyContentFilter($html, array $reasons)
 947         {
 948                 if (count($reasons)) {
 949                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
 950                         $html = Renderer::replaceMacros($tpl, [
 951                                 '$reasons'   => $reasons,
 952                                 '$rnd'       => Strings::getRandomHex(8),
 953                                 '$openclose' => L10n::t('Click to open/close'),
 954                                 '$html'      => $html
 955                         ]);
 956                 }
 957
 958                 return $html;
 959         }
 960
 961         /**
 962          * replace html amp entity with amp char
 963          * @param string $s
 964          * @return string
 965          */
 966         public static function unamp($s)
 967         {
 968                 return str_replace('&amp;', '&', $s);
 969         }
 970 }