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\Model\Contact;
  16 use Friendica\Util\Network;
  17 use Friendica\Util\Proxy as ProxyUtils;
  18 use Friendica\Util\Strings;
  19 use Friendica\Util\XML;
  20 use League\HTMLToMarkdown\HtmlConverter;
  21
  22 class HTML
  23 {
  24         public static function sanitizeCSS($input)
  25         {
  26                 $cleaned = "";
  27
  28                 $input = strtolower($input);
  29
  30                 for ($i = 0; $i < strlen($input); $i++) {
  31                         $char = substr($input, $i, 1);
  32
  33                         if (($char >= "a") && ($char <= "z")) {
  34                                 $cleaned .= $char;
  35                         }
  36
  37                         if (!(strpos(" #;:0123456789-_.%", $char) === false)) {
  38                                 $cleaned .= $char;
  39                         }
  40                 }
  41
  42                 return $cleaned;
  43         }
  44
  45         private static function tagToBBCode(DOMDocument $doc, $tag, $attributes, $startbb, $endbb)
  46         {
  47                 do {
  48                         $done = self::tagToBBCodeSub($doc, $tag, $attributes, $startbb, $endbb);
  49                 } while ($done);
  50         }
  51
  52         private static function tagToBBCodeSub(DOMDocument $doc, $tag, $attributes, $startbb, $endbb)
  53         {
  54                 $savestart = str_replace('$', '\x01', $startbb);
  55                 $replace = false;
  56
  57                 $xpath = new DOMXPath($doc);
  58
  59                 $list = $xpath->query("//" . $tag);
  60                 foreach ($list as $node) {
  61                         $attr = [];
  62                         if ($node->attributes->length) {
  63                                 foreach ($node->attributes as $attribute) {
  64                                         $attr[$attribute->name] = $attribute->value;
  65                                 }
  66                         }
  67
  68                         $replace = true;
  69
  70                         $startbb = $savestart;
  71
  72                         $i = 0;
  73
  74                         foreach ($attributes as $attribute => $value) {
  75                                 $startbb = str_replace('\x01' . ++$i, '$1', $startbb);
  76                                 if (strpos('*' . $startbb, '$1') > 0) {
  77                                         if ($replace && (@$attr[$attribute] != '')) {
  78                                                 $startbb = preg_replace($value, $startbb, $attr[$attribute], -1, $count);
  79
  80                                                 // If nothing could be changed
  81                                                 if ($count == 0) {
  82                                                         $replace = false;
  83                                                 }
  84                                         } else {
  85                                                 $replace = false;
  86                                         }
  87                                 } else {
  88                                         if (@$attr[$attribute] != $value) {
  89                                                 $replace = false;
  90                                         }
  91                                 }
  92                         }
  93
  94                         if ($replace) {
  95                                 $StartCode = $doc->createTextNode($startbb);
  96                                 $EndCode = $doc->createTextNode($endbb);
  97
  98                                 $node->parentNode->insertBefore($StartCode, $node);
  99
 100                                 if ($node->hasChildNodes()) {
 101                                         foreach ($node->childNodes as $child) {
 102                                                 $newNode = $child->cloneNode(true);
 103                                                 $node->parentNode->insertBefore($newNode, $node);
 104                                         }
 105                                 }
 106
 107                                 $node->parentNode->insertBefore($EndCode, $node);
 108                                 $node->parentNode->removeChild($node);
 109                         }
 110                 }
 111
 112                 return $replace;
 113         }
 114
 115         /**
 116          * Made by: ike@piratenpartei.de
 117          * Originally made for the syncom project: http://wiki.piratenpartei.de/Syncom
 118          *                    https://github.com/annando/Syncom
 119          *
 120          * @brief Converter for HTML to BBCode
 121          * @param string $message
 122          * @param string $basepath
 123          * @return string
 124          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 125          */
 126         public static function toBBCode($message, $basepath = '')
 127         {
 128                 $message = str_replace("\r", "", $message);
 129
 130                 // Removing code blocks before the whitespace removal processing below
 131                 $codeblocks = [];
 132                 $message = preg_replace_callback(
 133                         '#<pre><code(?: class="language-([^"]*)")?>(.*)</code></pre>#iUs',
 134                         function ($matches) use (&$codeblocks) {
 135                                 $return = '[codeblock-' . count($codeblocks) . ']';
 136
 137                                 $prefix = '[code]';
 138                                 if ($matches[1] != '') {
 139                                         $prefix = '[code=' . $matches[1] . ']';
 140                                 }
 141
 142                                 $codeblocks[] = $prefix . PHP_EOL . trim($matches[2]) . PHP_EOL . '[/code]';
 143                                 return $return;
 144                         },
 145                         $message
 146                 );
 147
 148                 $message = str_replace(
 149                         [
 150                                 "<li><p>",
 151                                 "</p></li>",
 152                         ],
 153                         [
 154                                 "<li>",
 155                                 "</li>",
 156                         ],
 157                         $message
 158                 );
 159
 160                 // remove namespaces
 161                 $message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
 162                 $message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
 163
 164                 $doc = new DOMDocument();
 165                 $doc->preserveWhiteSpace = false;
 166
 167                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 168
 169                 @$doc->loadHTML($message);
 170
 171                 XML::deleteNode($doc, 'style');
 172                 XML::deleteNode($doc, 'head');
 173                 XML::deleteNode($doc, 'title');
 174                 XML::deleteNode($doc, 'meta');
 175                 XML::deleteNode($doc, 'xml');
 176                 XML::deleteNode($doc, 'removeme');
 177
 178                 $xpath = new DomXPath($doc);
 179                 $list = $xpath->query("//pre");
 180                 foreach ($list as $node) {
 181                         // Ensure to escape unescaped & - they will otherwise raise a warning
 182                         $safe_value = preg_replace('/&(?!\w+;)/', '&amp;', $node->nodeValue);
 183                         $node->nodeValue = str_replace("\n", "\r", $safe_value);
 184                 }
 185
 186                 $message = $doc->saveHTML();
 187                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br />", " ", ""], $message);
 188                 $message = preg_replace('= [\s]*=i', " ", $message);
 189                 @$doc->loadHTML($message);
 190
 191                 self::tagToBBCode($doc, 'html', [], "", "");
 192                 self::tagToBBCode($doc, 'body', [], "", "");
 193
 194                 // Outlook-Quote - Variant 1
 195                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal', 'style' => 'margin-left:35.4pt'], '[quote]', '[/quote]');
 196
 197                 // Outlook-Quote - Variant 2
 198                 self::tagToBBCode(
 199                         $doc,
 200                         'div',
 201                         ['style' => 'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'],
 202                         '[quote]',
 203                         '[/quote]'
 204                 );
 205
 206                 // MyBB-Stuff
 207                 self::tagToBBCode($doc, 'span', ['style' => 'text-decoration: underline;'], '[u]', '[/u]');
 208                 self::tagToBBCode($doc, 'span', ['style' => 'font-style: italic;'], '[i]', '[/i]');
 209                 self::tagToBBCode($doc, 'span', ['style' => 'font-weight: bold;'], '[b]', '[/b]');
 210
 211                 /* self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
 212                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
 213                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
 214                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
 215                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
 216                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
 217                   self::node2BBCode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
 218                  */
 219                 // Untested
 220                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
 221                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
 222                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
 223
 224                 self::tagToBBCode($doc, 'span', ['style' => '/.*color:\s*(.+?)[,;].*/'], '[color="$1"]', '[/color]');
 225
 226                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 227                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
 228                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
 229                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 230                 // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
 231                 // Test
 232                 //self::node2BBCode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
 233                 self::tagToBBCode($doc, 'span', ['class' => 'type-link'], '[class=type-link]', '[/class]');
 234                 self::tagToBBCode($doc, 'span', ['class' => 'type-video'], '[class=type-video]', '[/class]');
 235
 236                 self::tagToBBCode($doc, 'strong', [], '[b]', '[/b]');
 237                 self::tagToBBCode($doc, 'em', [], '[i]', '[/i]');
 238                 self::tagToBBCode($doc, 'b', [], '[b]', '[/b]');
 239                 self::tagToBBCode($doc, 'i', [], '[i]', '[/i]');
 240                 self::tagToBBCode($doc, 'u', [], '[u]', '[/u]');
 241                 self::tagToBBCode($doc, 's', [], '[s]', '[/s]');
 242                 self::tagToBBCode($doc, 'del', [], '[s]', '[/s]');
 243                 self::tagToBBCode($doc, 'strike', [], '[s]', '[/s]');
 244
 245                 self::tagToBBCode($doc, 'big', [], "[size=large]", "[/size]");
 246                 self::tagToBBCode($doc, 'small', [], "[size=small]", "[/size]");
 247
 248                 self::tagToBBCode($doc, 'blockquote', [], '[quote]', '[/quote]');
 249
 250                 self::tagToBBCode($doc, 'br', [], "\n", '');
 251
 252                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal'], "\n", "");
 253                 self::tagToBBCode($doc, 'div', ['class' => 'MsoNormal'], "\r", "");
 254
 255                 self::tagToBBCode($doc, 'span', [], "", "");
 256
 257                 self::tagToBBCode($doc, 'span', [], "", "");
 258                 self::tagToBBCode($doc, 'pre', [], "", "");
 259
 260                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 261                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 262
 263                 self::tagToBBCode($doc, 'ul', [], "[list]", "[/list]");
 264                 self::tagToBBCode($doc, 'ol', [], "[list=1]", "[/list]");
 265                 self::tagToBBCode($doc, 'li', [], "[*]", "");
 266
 267                 self::tagToBBCode($doc, 'hr', [], "[hr]", "");
 268
 269                 self::tagToBBCode($doc, 'table', [], "", "");
 270                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 271                 self::tagToBBCode($doc, 'td', [], "\t", "");
 272                 //self::node2BBCode($doc, 'table', array(), "[table]", "[/table]");
 273                 //self::node2BBCode($doc, 'th', array(), "[th]", "[/th]");
 274                 //self::node2BBCode($doc, 'tr', array(), "[tr]", "[/tr]");
 275                 //self::node2BBCode($doc, 'td', array(), "[td]", "[/td]");
 276                 //self::node2BBCode($doc, 'h1', array(), "\n\n[size=xx-large][b]", "[/b][/size]\n");
 277                 //self::node2BBCode($doc, 'h2', array(), "\n\n[size=x-large][b]", "[/b][/size]\n");
 278                 //self::node2BBCode($doc, 'h3', array(), "\n\n[size=large][b]", "[/b][/size]\n");
 279                 //self::node2BBCode($doc, 'h4', array(), "\n\n[size=medium][b]", "[/b][/size]\n");
 280                 //self::node2BBCode($doc, 'h5', array(), "\n\n[size=small][b]", "[/b][/size]\n");
 281                 //self::node2BBCode($doc, 'h6', array(), "\n\n[size=x-small][b]", "[/b][/size]\n");
 282
 283                 self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 284                 self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 285                 self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 286                 self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 287                 self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 288                 self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 289
 290                 self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 291                 self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 292
 293                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]');
 294                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]');
 295
 296
 297                 self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]');
 298                 self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]');
 299                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[iframe]$1', '[/iframe]');
 300
 301                 self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 302                 self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 303
 304                 $message = $doc->saveHTML();
 305
 306                 // I'm removing something really disturbing
 307                 // Don't know exactly what it is
 308                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 309
 310                 $message = str_replace("&nbsp;", " ", $message);
 311
 312                 // removing multiple DIVs
 313                 $message = preg_replace('=\r *\r=i', "\n", $message);
 314                 $message = str_replace("\r", "\n", $message);
 315
 316                 Hook::callAll('html2bbcode', $message);
 317
 318                 $message = strip_tags($message);
 319
 320                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 321
 322                 $message = str_replace(["<"], ["&lt;"], $message);
 323
 324                 // remove quotes if they don't make sense
 325                 $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 326
 327                 $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 328                 $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 329
 330                 do {
 331                         $oldmessage = $message;
 332                         $message = str_replace("\n \n", "\n\n", $message);
 333                 } while ($oldmessage != $message);
 334
 335                 do {
 336                         $oldmessage = $message;
 337                         $message = str_replace("\n\n\n", "\n\n", $message);
 338                 } while ($oldmessage != $message);
 339
 340                 do {
 341                         $oldmessage = $message;
 342                         $message = str_replace(
 343                                 [
 344                                 "[/size]\n\n",
 345                                 "\n[hr]",
 346                                 "[hr]\n",
 347                                 "\n[list",
 348                                 "[/list]\n",
 349                                 "\n[/",
 350                                 "[list]\n",
 351                                 "[list=1]\n",
 352                                 "\n[*]"],
 353                                 [
 354                                 "[/size]\n",
 355                                 "[hr]",
 356                                 "[hr]",
 357                                 "[list",
 358                                 "[/list]",
 359                                 "[/",
 360                                 "[list]",
 361                                 "[list=1]",
 362                                 "[*]"],
 363                                 $message
 364                         );
 365                 } while ($message != $oldmessage);
 366
 367                 $message = str_replace(
 368                         ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 369                         ['[b]', '[/b]', '[i]', '[/i]'],
 370                         $message
 371                 );
 372
 373                 // Handling Yahoo style of mails
 374                 $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 375
 376                 // Restore code blocks
 377                 $message = preg_replace_callback(
 378                         '#\[codeblock-([0-9]+)\]#iU',
 379                         function ($matches) use ($codeblocks) {
 380                                 $return = '';
 381                                 if (isset($codeblocks[intval($matches[1])])) {
 382                                         $return = $codeblocks[$matches[1]];
 383                                 }
 384                                 return $return;
 385                         },
 386                         $message
 387                 );
 388
 389                 $message = trim($message);
 390
 391                 if ($basepath != '') {
 392                         $message = self::qualifyURLs($message, $basepath);
 393                 }
 394
 395                 return $message;
 396         }
 397
 398         /**
 399          * @brief Sub function to complete incomplete URL
 400          *
 401          * @param array  $matches  Result of preg_replace_callback
 402          * @param string $basepath Basepath that is used to complete the URL
 403          *
 404          * @return string The expanded URL
 405          */
 406         private static function qualifyURLsSub($matches, $basepath)
 407         {
 408                 $base = parse_url($basepath);
 409                 unset($base['query']);
 410                 unset($base['fragment']);
 411
 412                 $link = $matches[0];
 413                 $url = $matches[1];
 414
 415                 $parts = array_merge($base, parse_url($url));
 416                 $url2 = Network::unparseURL($parts);
 417
 418                 return str_replace($url, $url2, $link);
 419         }
 420
 421         /**
 422          * @brief Complete incomplete URLs in BBCode
 423          *
 424          * @param string $body     Body with URLs
 425          * @param string $basepath Base path that is used to complete the URL
 426          *
 427          * @return string Body with expanded URLs
 428          */
 429         private static function qualifyURLs($body, $basepath)
 430         {
 431                 $URLSearchString = "^\[\]";
 432
 433                 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
 434                         "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
 435                         "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 436                         "/\[img\](.*?)\[\/img\]/ism",
 437                         "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 438                         "/\[zmg\](.*?)\[\/zmg\]/ism",
 439                         "/\[video\](.*?)\[\/video\]/ism",
 440                         "/\[audio\](.*?)\[\/audio\]/ism",
 441                 ];
 442
 443                 foreach ($matches as $match) {
 444                         $body = preg_replace_callback(
 445                                 $match,
 446                                 function ($match) use ($basepath) {
 447                                         return self::qualifyURLsSub($match, $basepath);
 448                                 },
 449                                 $body
 450                         );
 451                 }
 452                 return $body;
 453         }
 454
 455         private static function breakLines($line, $level, $wraplength = 75)
 456         {
 457                 if ($wraplength == 0) {
 458                         $wraplength = 2000000;
 459                 }
 460
 461                 $wraplen = $wraplength - $level;
 462
 463                 $newlines = [];
 464
 465                 do {
 466                         $oldline = $line;
 467
 468                         $subline = substr($line, 0, $wraplen);
 469
 470                         $pos = strrpos($subline, ' ');
 471
 472                         if ($pos == 0) {
 473                                 $pos = strpos($line, ' ');
 474                         }
 475
 476                         if (($pos > 0) && strlen($line) > $wraplen) {
 477                                 $newline = trim(substr($line, 0, $pos));
 478                                 if ($level > 0) {
 479                                         $newline = str_repeat(">", $level) . ' ' . $newline;
 480                                 }
 481
 482                                 $newlines[] = $newline . " ";
 483                                 $line = substr($line, $pos + 1);
 484                         }
 485                 } while ((strlen($line) > $wraplen) && !($oldline == $line));
 486
 487                 if ($level > 0) {
 488                         $line = str_repeat(">", $level) . ' ' . $line;
 489                 }
 490
 491                 $newlines[] = $line;
 492
 493                 return implode($newlines, "\n");
 494         }
 495
 496         private static function quoteLevel($message, $wraplength = 75)
 497         {
 498                 $lines = explode("\n", $message);
 499
 500                 $newlines = [];
 501                 $level = 0;
 502                 foreach ($lines as $line) {
 503                         $line = trim($line);
 504                         $startquote = false;
 505                         while (strpos("*" . $line, '[quote]') > 0) {
 506                                 $level++;
 507                                 $pos = strpos($line, '[quote]');
 508                                 $line = substr($line, 0, $pos) . substr($line, $pos + 7);
 509                                 $startquote = true;
 510                         }
 511
 512                         $currlevel = $level;
 513
 514                         while (strpos("*" . $line, '[/quote]') > 0) {
 515                                 $level--;
 516                                 if ($level < 0) {
 517                                         $level = 0;
 518                                 }
 519
 520                                 $pos = strpos($line, '[/quote]');
 521                                 $line = substr($line, 0, $pos) . substr($line, $pos + 8);
 522                         }
 523
 524                         if (!$startquote || ($line != '')) {
 525                                 $newlines[] = self::breakLines($line, $currlevel, $wraplength);
 526                         }
 527                 }
 528
 529                 return implode($newlines, "\n");
 530         }
 531
 532         private static function collectURLs($message)
 533         {
 534                 $pattern = '/<a.*?href="(.*?)".*?>(.*?)<\/a>/is';
 535                 preg_match_all($pattern, $message, $result, PREG_SET_ORDER);
 536
 537                 $urls = [];
 538                 foreach ($result as $treffer) {
 539                         $ignore = false;
 540
 541                         // A list of some links that should be ignored
 542                         $list = ["/user/", "/tag/", "/group/", "/profile/", "/search?search=", "/search?tag=", "mailto:", "/u/", "/node/",
 543                                 "//plus.google.com/", "//twitter.com/"];
 544                         foreach ($list as $listitem) {
 545                                 if (strpos($treffer[1], $listitem) !== false) {
 546                                         $ignore = true;
 547                                 }
 548                         }
 549
 550                         if ((strpos($treffer[1], "//twitter.com/") !== false) && (strpos($treffer[1], "/status/") !== false)) {
 551                                 $ignore = false;
 552                         }
 553
 554                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/posts") !== false)) {
 555                                 $ignore = false;
 556                         }
 557
 558                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/photos") !== false)) {
 559                                 $ignore = false;
 560                         }
 561
 562                         if (!$ignore) {
 563                                 $urls[$treffer[1]] = $treffer[1];
 564                         }
 565                 }
 566
 567                 return $urls;
 568         }
 569
 570         public static function toPlaintext($html, $wraplength = 75, $compact = false)
 571         {
 572                 $message = str_replace("\r", "", $html);
 573
 574                 $doc = new DOMDocument();
 575                 $doc->preserveWhiteSpace = false;
 576
 577                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 578
 579                 @$doc->loadHTML($message);
 580
 581                 $xpath = new DOMXPath($doc);
 582                 $list = $xpath->query("//pre");
 583                 foreach ($list as $node) {
 584                         $node->nodeValue = str_replace("\n", "\r", $node->nodeValue);
 585                 }
 586
 587                 $message = $doc->saveHTML();
 588                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br>", " ", ""], $message);
 589                 $message = preg_replace('= [\s]*=i', " ", $message);
 590
 591                 // Collecting all links
 592                 $urls = self::collectURLs($message);
 593
 594                 @$doc->loadHTML($message);
 595
 596                 self::tagToBBCode($doc, 'html', [], '', '');
 597                 self::tagToBBCode($doc, 'body', [], '', '');
 598
 599                 // MyBB-Auszeichnungen
 600                 /*
 601                   self::node2BBCode($doc, 'span', array('style'=>'text-decoration: underline;'), '_', '_');
 602                   self::node2BBCode($doc, 'span', array('style'=>'font-style: italic;'), '/', '/');
 603                   self::node2BBCode($doc, 'span', array('style'=>'font-weight: bold;'), '*', '*');
 604
 605                   self::node2BBCode($doc, 'strong', array(), '*', '*');
 606                   self::node2BBCode($doc, 'b', array(), '*', '*');
 607                   self::node2BBCode($doc, 'i', array(), '/', '/');
 608                   self::node2BBCode($doc, 'u', array(), '_', '_');
 609                  */
 610
 611                 if ($compact) {
 612                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 613                 } else {
 614                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 615                 }
 616
 617                 self::tagToBBCode($doc, 'br', [], "\n", '');
 618
 619                 self::tagToBBCode($doc, 'span', [], "", "");
 620                 self::tagToBBCode($doc, 'pre', [], "", "");
 621                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 622                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 623
 624                 //self::node2BBCode($doc, 'ul', array(), "\n[list]", "[/list]\n");
 625                 //self::node2BBCode($doc, 'ol', array(), "\n[list=1]", "[/list]\n");
 626                 self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 627
 628                 self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
 629
 630                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 631                 self::tagToBBCode($doc, 'td', [], "\t", "");
 632
 633                 self::tagToBBCode($doc, 'h1', [], "\n\n*", "*\n");
 634                 self::tagToBBCode($doc, 'h2', [], "\n\n*", "*\n");
 635                 self::tagToBBCode($doc, 'h3', [], "\n\n*", "*\n");
 636                 self::tagToBBCode($doc, 'h4', [], "\n\n*", "*\n");
 637                 self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
 638                 self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 639
 640                 // Problem: there is no reliable way to detect if it is a link to a tag or profile
 641                 //self::node2BBCode($doc, 'a', array('href'=>'/(.+)/'), ' $1 ', ' ', true);
 642                 //self::node2BBCode($doc, 'a', array('href'=>'/(.+)/', 'rel'=>'oembed'), ' $1 ', '', true);
 643                 //self::node2BBCode($doc, 'img', array('alt'=>'/(.+)/'), '$1', '');
 644                 //self::node2BBCode($doc, 'img', array('title'=>'/(.+)/'), '$1', '');
 645                 //self::node2BBCode($doc, 'img', array(), '', '');
 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          * @brief 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' => L10n::t('Loading more entries...'),
 801                         'end' => 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                 $a = \get_app();
 817
 818                 return ContactBlock::getHTML($a->profile);
 819         }
 820
 821         /**
 822          * @brief 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                         '$click' => defaults($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   $aside Display the search widgit aside.
 883          *
 884          * @return string Formatted HTML.
 885          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 886          */
 887         public static function search($s, $id = 'search-box', $url = 'search', $aside = true)
 888         {
 889                 $mode = 'text';
 890
 891                 if (strpos($s, '#') === 0) {
 892                         $mode = 'tag';
 893                 }
 894                 $save_label = $mode === 'text' ? L10n::t('Save') : L10n::t('Follow');
 895
 896                 $values = [
 897                                 '$s' => $s,
 898                                 '$id' => $id,
 899                                 '$action_url' => $url,
 900                                 '$search_label' => L10n::t('Search'),
 901                                 '$save_label' => $save_label,
 902                                 '$savedsearch' => 'savedsearch',
 903                                 '$search_hint' => L10n::t('@name, !forum, #tags, content'),
 904                                 '$mode' => $mode
 905                         ];
 906
 907                 if (!$aside) {
 908                         $values['$searchoption'] = [
 909                                                 L10n::t("Full Text"),
 910                                                 L10n::t("Tags"),
 911                                                 L10n::t("Contacts")];
 912
 913                         if (Config::get('system', 'poco_local_search')) {
 914                                 $values['$searchoption'][] = L10n::t("Forums");
 915                         }
 916                 }
 917
 918                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 919         }
 920
 921         /**
 922          * Replace naked text hyperlink with HTML formatted hyperlink
 923          *
 924          * @param string $s
 925          * @return string
 926          */
 927         public static function toLink($s)
 928         {
 929                 $s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank">$1</a>', $s);
 930                 $s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
 931                 return $s;
 932         }
 933
 934         /**
 935          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
 936          *
 937          * Reasons are expected to have been translated already.
 938          *
 939          * @param string $html
 940          * @param array  $reasons
 941          * @return string
 942          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 943          */
 944         public static function applyContentFilter($html, array $reasons)
 945         {
 946                 if (count($reasons)) {
 947                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
 948                         $html = Renderer::replaceMacros($tpl, [
 949                                 '$reasons'   => $reasons,
 950                                 '$rnd'       => Strings::getRandomHex(8),
 951                                 '$openclose' => L10n::t('Click to open/close'),
 952                                 '$html'      => $html
 953                         ]);
 954                 }
 955
 956                 return $html;
 957         }
 958
 959         /**
 960          * replace html amp entity with amp char
 961          * @param string $s
 962          * @return string
 963          */
 964         public static function unamp($s)
 965         {
 966                 return str_replace('&amp;', '&', $s);
 967         }
 968 }