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\PConfig;
  15 use Friendica\Core\Protocol;
  16 use Friendica\Core\Renderer;
  17 use Friendica\Database\DBA;
  18 use Friendica\Model\Contact;
  19 use Friendica\Util\Network;
  20 use Friendica\Util\Proxy as ProxyUtils;
  21 use Friendica\Util\Strings;
  22 use Friendica\Util\XML;
  23 use League\HTMLToMarkdown\HtmlConverter;
  24
  25 class HTML
  26 {
  27         public static function sanitizeCSS($input)
  28         {
  29                 $cleaned = "";
  30
  31                 $input = strtolower($input);
  32
  33                 for ($i = 0; $i < strlen($input); $i++) {
  34                         $char = substr($input, $i, 1);
  35
  36                         if (($char >= "a") && ($char <= "z")) {
  37                                 $cleaned .= $char;
  38                         }
  39
  40                         if (!(strpos(" #;:0123456789-_.%", $char) === false)) {
  41                                 $cleaned .= $char;
  42                         }
  43                 }
  44
  45                 return $cleaned;
  46         }
  47
  48         private static function tagToBBCode(DOMDocument $doc, $tag, $attributes, $startbb, $endbb)
  49         {
  50                 do {
  51                         $done = self::tagToBBCodeSub($doc, $tag, $attributes, $startbb, $endbb);
  52                 } while ($done);
  53         }
  54
  55         private static function tagToBBCodeSub(DOMDocument $doc, $tag, $attributes, $startbb, $endbb)
  56         {
  57                 $savestart = str_replace('$', '\x01', $startbb);
  58                 $replace = false;
  59
  60                 $xpath = new DOMXPath($doc);
  61
  62                 $list = $xpath->query("//" . $tag);
  63                 foreach ($list as $node) {
  64                         $attr = [];
  65                         if ($node->attributes->length) {
  66                                 foreach ($node->attributes as $attribute) {
  67                                         $attr[$attribute->name] = $attribute->value;
  68                                 }
  69                         }
  70
  71                         $replace = true;
  72
  73                         $startbb = $savestart;
  74
  75                         $i = 0;
  76
  77                         foreach ($attributes as $attribute => $value) {
  78                                 $startbb = str_replace('\x01' . ++$i, '$1', $startbb);
  79                                 if (strpos('*' . $startbb, '$1') > 0) {
  80                                         if ($replace && (@$attr[$attribute] != '')) {
  81                                                 $startbb = preg_replace($value, $startbb, $attr[$attribute], -1, $count);
  82
  83                                                 // If nothing could be changed
  84                                                 if ($count == 0) {
  85                                                         $replace = false;
  86                                                 }
  87                                         } else {
  88                                                 $replace = false;
  89                                         }
  90                                 } else {
  91                                         if (@$attr[$attribute] != $value) {
  92                                                 $replace = false;
  93                                         }
  94                                 }
  95                         }
  96
  97                         if ($replace) {
  98                                 $StartCode = $doc->createTextNode($startbb);
  99                                 $EndCode = $doc->createTextNode($endbb);
 100
 101                                 $node->parentNode->insertBefore($StartCode, $node);
 102
 103                                 if ($node->hasChildNodes()) {
 104                                         foreach ($node->childNodes as $child) {
 105                                                 $newNode = $child->cloneNode(true);
 106                                                 $node->parentNode->insertBefore($newNode, $node);
 107                                         }
 108                                 }
 109
 110                                 $node->parentNode->insertBefore($EndCode, $node);
 111                                 $node->parentNode->removeChild($node);
 112                         }
 113                 }
 114
 115                 return $replace;
 116         }
 117
 118         /**
 119          * Made by: ike@piratenpartei.de
 120          * Originally made for the syncom project: http://wiki.piratenpartei.de/Syncom
 121          *                    https://github.com/annando/Syncom
 122          *
 123          * @brief Converter for HTML to BBCode
 124          * @param string $message
 125          * @param string $basepath
 126          * @return string
 127          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 128          */
 129         public static function toBBCode($message, $basepath = '')
 130         {
 131                 $message = str_replace("\r", "", $message);
 132
 133                 // Removing code blocks before the whitespace removal processing below
 134                 $codeblocks = [];
 135                 $message = preg_replace_callback(
 136                         '#<pre><code(?: class="language-([^"]*)")?>(.*)</code></pre>#iUs',
 137                         function ($matches) use (&$codeblocks) {
 138                                 $return = '[codeblock-' . count($codeblocks) . ']';
 139
 140                                 $prefix = '[code]';
 141                                 if ($matches[1] != '') {
 142                                         $prefix = '[code=' . $matches[1] . ']';
 143                                 }
 144
 145                                 $codeblocks[] = $prefix . PHP_EOL . trim($matches[2]) . PHP_EOL . '[/code]';
 146                                 return $return;
 147                         },
 148                         $message
 149                 );
 150
 151                 $message = str_replace(
 152                         [
 153                                 "<li><p>",
 154                                 "</p></li>",
 155                         ],
 156                         [
 157                                 "<li>",
 158                                 "</li>",
 159                         ],
 160                         $message
 161                 );
 162
 163                 // remove namespaces
 164                 $message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
 165                 $message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
 166
 167                 $doc = new DOMDocument();
 168                 $doc->preserveWhiteSpace = false;
 169
 170                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 171
 172                 @$doc->loadHTML($message);
 173
 174                 XML::deleteNode($doc, 'style');
 175                 XML::deleteNode($doc, 'head');
 176                 XML::deleteNode($doc, 'title');
 177                 XML::deleteNode($doc, 'meta');
 178                 XML::deleteNode($doc, 'xml');
 179                 XML::deleteNode($doc, 'removeme');
 180
 181                 $xpath = new DomXPath($doc);
 182                 $list = $xpath->query("//pre");
 183                 foreach ($list as $node) {
 184                         // Ensure to escape unescaped & - they will otherwise raise a warning
 185                         $safe_value = preg_replace('/&(?!\w+;)/', '&amp;', $node->nodeValue);
 186                         $node->nodeValue = str_replace("\n", "\r", $safe_value);
 187                 }
 188
 189                 $message = $doc->saveHTML();
 190                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br />", " ", ""], $message);
 191                 $message = preg_replace('= [\s]*=i', " ", $message);
 192                 @$doc->loadHTML($message);
 193
 194                 self::tagToBBCode($doc, 'html', [], "", "");
 195                 self::tagToBBCode($doc, 'body', [], "", "");
 196
 197                 // Outlook-Quote - Variant 1
 198                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal', 'style' => 'margin-left:35.4pt'], '[quote]', '[/quote]');
 199
 200                 // Outlook-Quote - Variant 2
 201                 self::tagToBBCode(
 202                         $doc,
 203                         'div',
 204                         ['style' => 'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'],
 205                         '[quote]',
 206                         '[/quote]'
 207                 );
 208
 209                 // MyBB-Stuff
 210                 self::tagToBBCode($doc, 'span', ['style' => 'text-decoration: underline;'], '[u]', '[/u]');
 211                 self::tagToBBCode($doc, 'span', ['style' => 'font-style: italic;'], '[i]', '[/i]');
 212                 self::tagToBBCode($doc, 'span', ['style' => 'font-weight: bold;'], '[b]', '[/b]');
 213
 214                 /* self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
 215                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
 216                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
 217                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
 218                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
 219                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
 220                   self::node2BBCode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
 221                  */
 222                 // Untested
 223                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
 224                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
 225                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
 226
 227                 self::tagToBBCode($doc, 'span', ['style' => '/.*color:\s*(.+?)[,;].*/'], '[color="$1"]', '[/color]');
 228
 229                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 230                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
 231                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
 232                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 233                 // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
 234                 // Test
 235                 //self::node2BBCode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
 236                 self::tagToBBCode($doc, 'span', ['class' => 'type-link'], '[class=type-link]', '[/class]');
 237                 self::tagToBBCode($doc, 'span', ['class' => 'type-video'], '[class=type-video]', '[/class]');
 238
 239                 self::tagToBBCode($doc, 'strong', [], '[b]', '[/b]');
 240                 self::tagToBBCode($doc, 'em', [], '[i]', '[/i]');
 241                 self::tagToBBCode($doc, 'b', [], '[b]', '[/b]');
 242                 self::tagToBBCode($doc, 'i', [], '[i]', '[/i]');
 243                 self::tagToBBCode($doc, 'u', [], '[u]', '[/u]');
 244                 self::tagToBBCode($doc, 's', [], '[s]', '[/s]');
 245                 self::tagToBBCode($doc, 'del', [], '[s]', '[/s]');
 246                 self::tagToBBCode($doc, 'strike', [], '[s]', '[/s]');
 247
 248                 self::tagToBBCode($doc, 'big', [], "[size=large]", "[/size]");
 249                 self::tagToBBCode($doc, 'small', [], "[size=small]", "[/size]");
 250
 251                 self::tagToBBCode($doc, 'blockquote', [], '[quote]', '[/quote]');
 252
 253                 self::tagToBBCode($doc, 'br', [], "\n", '');
 254
 255                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal'], "\n", "");
 256                 self::tagToBBCode($doc, 'div', ['class' => 'MsoNormal'], "\r", "");
 257
 258                 self::tagToBBCode($doc, 'span', [], "", "");
 259
 260                 self::tagToBBCode($doc, 'span', [], "", "");
 261                 self::tagToBBCode($doc, 'pre', [], "", "");
 262
 263                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 264                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 265
 266                 self::tagToBBCode($doc, 'ul', [], "[list]", "[/list]");
 267                 self::tagToBBCode($doc, 'ol', [], "[list=1]", "[/list]");
 268                 self::tagToBBCode($doc, 'li', [], "[*]", "");
 269
 270                 self::tagToBBCode($doc, 'hr', [], "[hr]", "");
 271
 272                 self::tagToBBCode($doc, 'table', [], "", "");
 273                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 274                 self::tagToBBCode($doc, 'td', [], "\t", "");
 275                 //self::node2BBCode($doc, 'table', array(), "[table]", "[/table]");
 276                 //self::node2BBCode($doc, 'th', array(), "[th]", "[/th]");
 277                 //self::node2BBCode($doc, 'tr', array(), "[tr]", "[/tr]");
 278                 //self::node2BBCode($doc, 'td', array(), "[td]", "[/td]");
 279                 //self::node2BBCode($doc, 'h1', array(), "\n\n[size=xx-large][b]", "[/b][/size]\n");
 280                 //self::node2BBCode($doc, 'h2', array(), "\n\n[size=x-large][b]", "[/b][/size]\n");
 281                 //self::node2BBCode($doc, 'h3', array(), "\n\n[size=large][b]", "[/b][/size]\n");
 282                 //self::node2BBCode($doc, 'h4', array(), "\n\n[size=medium][b]", "[/b][/size]\n");
 283                 //self::node2BBCode($doc, 'h5', array(), "\n\n[size=small][b]", "[/b][/size]\n");
 284                 //self::node2BBCode($doc, 'h6', array(), "\n\n[size=x-small][b]", "[/b][/size]\n");
 285
 286                 self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 287                 self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 288                 self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 289                 self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 290                 self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 291                 self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 292
 293                 self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 294                 self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 295
 296                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]');
 297                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]');
 298
 299
 300                 self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]');
 301                 self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]');
 302                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[iframe]$1', '[/iframe]');
 303
 304                 self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 305                 self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 306
 307                 $message = $doc->saveHTML();
 308
 309                 // I'm removing something really disturbing
 310                 // Don't know exactly what it is
 311                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 312
 313                 $message = str_replace("&nbsp;", " ", $message);
 314
 315                 // removing multiple DIVs
 316                 $message = preg_replace('=\r *\r=i', "\n", $message);
 317                 $message = str_replace("\r", "\n", $message);
 318
 319                 Hook::callAll('html2bbcode', $message);
 320
 321                 $message = strip_tags($message);
 322
 323                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 324
 325                 $message = str_replace(["<"], ["&lt;"], $message);
 326
 327                 // remove quotes if they don't make sense
 328                 $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 329
 330                 $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 331                 $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 332
 333                 do {
 334                         $oldmessage = $message;
 335                         $message = str_replace("\n \n", "\n\n", $message);
 336                 } while ($oldmessage != $message);
 337
 338                 do {
 339                         $oldmessage = $message;
 340                         $message = str_replace("\n\n\n", "\n\n", $message);
 341                 } while ($oldmessage != $message);
 342
 343                 do {
 344                         $oldmessage = $message;
 345                         $message = str_replace(
 346                                 [
 347                                 "[/size]\n\n",
 348                                 "\n[hr]",
 349                                 "[hr]\n",
 350                                 "\n[list",
 351                                 "[/list]\n",
 352                                 "\n[/",
 353                                 "[list]\n",
 354                                 "[list=1]\n",
 355                                 "\n[*]"],
 356                                 [
 357                                 "[/size]\n",
 358                                 "[hr]",
 359                                 "[hr]",
 360                                 "[list",
 361                                 "[/list]",
 362                                 "[/",
 363                                 "[list]",
 364                                 "[list=1]",
 365                                 "[*]"],
 366                                 $message
 367                         );
 368                 } while ($message != $oldmessage);
 369
 370                 $message = str_replace(
 371                         ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 372                         ['[b]', '[/b]', '[i]', '[/i]'],
 373                         $message
 374                 );
 375
 376                 // Handling Yahoo style of mails
 377                 $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 378
 379                 // Restore code blocks
 380                 $message = preg_replace_callback(
 381                         '#\[codeblock-([0-9]+)\]#iU',
 382                         function ($matches) use ($codeblocks) {
 383                                 $return = '';
 384                                 if (isset($codeblocks[intval($matches[1])])) {
 385                                         $return = $codeblocks[$matches[1]];
 386                                 }
 387                                 return $return;
 388                         },
 389                         $message
 390                 );
 391
 392                 $message = trim($message);
 393
 394                 if ($basepath != '') {
 395                         $message = self::qualifyURLs($message, $basepath);
 396                 }
 397
 398                 return $message;
 399         }
 400
 401         /**
 402          * @brief Sub function to complete incomplete URL
 403          *
 404          * @param array  $matches  Result of preg_replace_callback
 405          * @param string $basepath Basepath that is used to complete the URL
 406          *
 407          * @return string The expanded URL
 408          */
 409         private static function qualifyURLsSub($matches, $basepath)
 410         {
 411                 $base = parse_url($basepath);
 412                 unset($base['query']);
 413                 unset($base['fragment']);
 414
 415                 $link = $matches[0];
 416                 $url = $matches[1];
 417
 418                 $parts = array_merge($base, parse_url($url));
 419                 $url2 = Network::unparseURL($parts);
 420
 421                 return str_replace($url, $url2, $link);
 422         }
 423
 424         /**
 425          * @brief Complete incomplete URLs in BBCode
 426          *
 427          * @param string $body     Body with URLs
 428          * @param string $basepath Base path that is used to complete the URL
 429          *
 430          * @return string Body with expanded URLs
 431          */
 432         private static function qualifyURLs($body, $basepath)
 433         {
 434                 $URLSearchString = "^\[\]";
 435
 436                 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
 437                         "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
 438                         "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 439                         "/\[img\](.*?)\[\/img\]/ism",
 440                         "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 441                         "/\[zmg\](.*?)\[\/zmg\]/ism",
 442                         "/\[video\](.*?)\[\/video\]/ism",
 443                         "/\[audio\](.*?)\[\/audio\]/ism",
 444                 ];
 445
 446                 foreach ($matches as $match) {
 447                         $body = preg_replace_callback(
 448                                 $match,
 449                                 function ($match) use ($basepath) {
 450                                         return self::qualifyURLsSub($match, $basepath);
 451                                 },
 452                                 $body
 453                         );
 454                 }
 455                 return $body;
 456         }
 457
 458         private static function breakLines($line, $level, $wraplength = 75)
 459         {
 460                 if ($wraplength == 0) {
 461                         $wraplength = 2000000;
 462                 }
 463
 464                 $wraplen = $wraplength - $level;
 465
 466                 $newlines = [];
 467
 468                 do {
 469                         $oldline = $line;
 470
 471                         $subline = substr($line, 0, $wraplen);
 472
 473                         $pos = strrpos($subline, ' ');
 474
 475                         if ($pos == 0) {
 476                                 $pos = strpos($line, ' ');
 477                         }
 478
 479                         if (($pos > 0) && strlen($line) > $wraplen) {
 480                                 $newline = trim(substr($line, 0, $pos));
 481                                 if ($level > 0) {
 482                                         $newline = str_repeat(">", $level) . ' ' . $newline;
 483                                 }
 484
 485                                 $newlines[] = $newline . " ";
 486                                 $line = substr($line, $pos + 1);
 487                         }
 488                 } while ((strlen($line) > $wraplen) && !($oldline == $line));
 489
 490                 if ($level > 0) {
 491                         $line = str_repeat(">", $level) . ' ' . $line;
 492                 }
 493
 494                 $newlines[] = $line;
 495
 496                 return implode($newlines, "\n");
 497         }
 498
 499         private static function quoteLevel($message, $wraplength = 75)
 500         {
 501                 $lines = explode("\n", $message);
 502
 503                 $newlines = [];
 504                 $level = 0;
 505                 foreach ($lines as $line) {
 506                         $line = trim($line);
 507                         $startquote = false;
 508                         while (strpos("*" . $line, '[quote]') > 0) {
 509                                 $level++;
 510                                 $pos = strpos($line, '[quote]');
 511                                 $line = substr($line, 0, $pos) . substr($line, $pos + 7);
 512                                 $startquote = true;
 513                         }
 514
 515                         $currlevel = $level;
 516
 517                         while (strpos("*" . $line, '[/quote]') > 0) {
 518                                 $level--;
 519                                 if ($level < 0) {
 520                                         $level = 0;
 521                                 }
 522
 523                                 $pos = strpos($line, '[/quote]');
 524                                 $line = substr($line, 0, $pos) . substr($line, $pos + 8);
 525                         }
 526
 527                         if (!$startquote || ($line != '')) {
 528                                 $newlines[] = self::breakLines($line, $currlevel, $wraplength);
 529                         }
 530                 }
 531
 532                 return implode($newlines, "\n");
 533         }
 534
 535         private static function collectURLs($message)
 536         {
 537                 $pattern = '/<a.*?href="(.*?)".*?>(.*?)<\/a>/is';
 538                 preg_match_all($pattern, $message, $result, PREG_SET_ORDER);
 539
 540                 $urls = [];
 541                 foreach ($result as $treffer) {
 542                         $ignore = false;
 543
 544                         // A list of some links that should be ignored
 545                         $list = ["/user/", "/tag/", "/group/", "/profile/", "/search?search=", "/search?tag=", "mailto:", "/u/", "/node/",
 546                                 "//plus.google.com/", "//twitter.com/"];
 547                         foreach ($list as $listitem) {
 548                                 if (strpos($treffer[1], $listitem) !== false) {
 549                                         $ignore = true;
 550                                 }
 551                         }
 552
 553                         if ((strpos($treffer[1], "//twitter.com/") !== false) && (strpos($treffer[1], "/status/") !== false)) {
 554                                 $ignore = false;
 555                         }
 556
 557                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/posts") !== false)) {
 558                                 $ignore = false;
 559                         }
 560
 561                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/photos") !== false)) {
 562                                 $ignore = false;
 563                         }
 564
 565                         if (!$ignore) {
 566                                 $urls[$treffer[1]] = $treffer[1];
 567                         }
 568                 }
 569
 570                 return $urls;
 571         }
 572
 573         public static function toPlaintext($html, $wraplength = 75, $compact = false)
 574         {
 575                 $message = str_replace("\r", "", $html);
 576
 577                 $doc = new DOMDocument();
 578                 $doc->preserveWhiteSpace = false;
 579
 580                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 581
 582                 @$doc->loadHTML($message);
 583
 584                 $xpath = new DOMXPath($doc);
 585                 $list = $xpath->query("//pre");
 586                 foreach ($list as $node) {
 587                         $node->nodeValue = str_replace("\n", "\r", $node->nodeValue);
 588                 }
 589
 590                 $message = $doc->saveHTML();
 591                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br>", " ", ""], $message);
 592                 $message = preg_replace('= [\s]*=i', " ", $message);
 593
 594                 // Collecting all links
 595                 $urls = self::collectURLs($message);
 596
 597                 @$doc->loadHTML($message);
 598
 599                 self::tagToBBCode($doc, 'html', [], '', '');
 600                 self::tagToBBCode($doc, 'body', [], '', '');
 601
 602                 // MyBB-Auszeichnungen
 603                 /*
 604                   self::node2BBCode($doc, 'span', array('style'=>'text-decoration: underline;'), '_', '_');
 605                   self::node2BBCode($doc, 'span', array('style'=>'font-style: italic;'), '/', '/');
 606                   self::node2BBCode($doc, 'span', array('style'=>'font-weight: bold;'), '*', '*');
 607
 608                   self::node2BBCode($doc, 'strong', array(), '*', '*');
 609                   self::node2BBCode($doc, 'b', array(), '*', '*');
 610                   self::node2BBCode($doc, 'i', array(), '/', '/');
 611                   self::node2BBCode($doc, 'u', array(), '_', '_');
 612                  */
 613
 614                 if ($compact) {
 615                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 616                 } else {
 617                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 618                 }
 619
 620                 self::tagToBBCode($doc, 'br', [], "\n", '');
 621
 622                 self::tagToBBCode($doc, 'span', [], "", "");
 623                 self::tagToBBCode($doc, 'pre', [], "", "");
 624                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 625                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 626
 627                 //self::node2BBCode($doc, 'ul', array(), "\n[list]", "[/list]\n");
 628                 //self::node2BBCode($doc, 'ol', array(), "\n[list=1]", "[/list]\n");
 629                 self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 630
 631                 self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
 632
 633                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 634                 self::tagToBBCode($doc, 'td', [], "\t", "");
 635
 636                 self::tagToBBCode($doc, 'h1', [], "\n\n*", "*\n");
 637                 self::tagToBBCode($doc, 'h2', [], "\n\n*", "*\n");
 638                 self::tagToBBCode($doc, 'h3', [], "\n\n*", "*\n");
 639                 self::tagToBBCode($doc, 'h4', [], "\n\n*", "*\n");
 640                 self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
 641                 self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 642
 643                 // Problem: there is no reliable way to detect if it is a link to a tag or profile
 644                 //self::node2BBCode($doc, 'a', array('href'=>'/(.+)/'), ' $1 ', ' ', true);
 645                 //self::node2BBCode($doc, 'a', array('href'=>'/(.+)/', 'rel'=>'oembed'), ' $1 ', '', true);
 646                 //self::node2BBCode($doc, 'img', array('alt'=>'/(.+)/'), '$1', '');
 647                 //self::node2BBCode($doc, 'img', array('title'=>'/(.+)/'), '$1', '');
 648                 //self::node2BBCode($doc, 'img', array(), '', '');
 649                 if (!$compact) {
 650                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
 651                 } else {
 652                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' ', ' ');
 653                 }
 654
 655                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], ' $1 ', '');
 656
 657                 $message = $doc->saveHTML();
 658
 659                 if (!$compact) {
 660                         $message = str_replace("[img]", "", $message);
 661                         $message = str_replace("[/img]", "", $message);
 662                 }
 663
 664                 // was ersetze ich da?
 665                 // Irgendein stoerrisches UTF-Zeug
 666                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 667
 668                 $message = str_replace("&nbsp;", " ", $message);
 669
 670                 // Aufeinanderfolgende DIVs
 671                 $message = preg_replace('=\r *\r=i', "\n", $message);
 672                 $message = str_replace("\r", "\n", $message);
 673
 674                 $message = strip_tags($message);
 675
 676                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 677
 678                 if (!$compact && ($message != '')) {
 679                         foreach ($urls as $id => $url) {
 680                                 if ($url != '' && strpos($message, $url) === false) {
 681                                         $message .= "\n" . $url . ' ';
 682                                 }
 683                         }
 684                 }
 685
 686                 $message = str_replace("\n«", "«\n", $message);
 687                 $message = str_replace("»\n", "\n»", $message);
 688
 689                 do {
 690                         $oldmessage = $message;
 691                         $message = str_replace("\n\n\n", "\n\n", $message);
 692                 } while ($oldmessage != $message);
 693
 694                 $message = self::quoteLevel(trim($message), $wraplength);
 695
 696                 return trim($message);
 697         }
 698
 699         /**
 700          * Converts provided HTML code to Markdown. The hardwrap parameter maximizes
 701          * compatibility with Diaspora in spite of the Markdown standards.
 702          *
 703          * @param string $html
 704          * @return string
 705          */
 706         public static function toMarkdown($html)
 707         {
 708                 $converter = new HtmlConverter(['hard_break' => true]);
 709                 $markdown = $converter->convert($html);
 710
 711                 return $markdown;
 712         }
 713
 714         /**
 715          * @brief Convert video HTML to BBCode tags
 716          *
 717          * @param string $s
 718          * @return string
 719          */
 720         public static function toBBCodeVideo($s)
 721         {
 722                 $s = preg_replace(
 723                         '#<object[^>]+>(.*?)https?://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+)(.*?)</object>#ism',
 724                         '[youtube]$2[/youtube]',
 725                         $s
 726                 );
 727
 728                 $s = preg_replace(
 729                         '#<iframe[^>](.*?)https?://www.youtube.com/embed/([A-Za-z0-9\-_=]+)(.*?)</iframe>#ism',
 730                         '[youtube]$2[/youtube]',
 731                         $s
 732                 );
 733
 734                 $s = preg_replace(
 735                         '#<iframe[^>](.*?)https?://player.vimeo.com/video/([0-9]+)(.*?)</iframe>#ism',
 736                         '[vimeo]$2[/vimeo]',
 737                         $s
 738                 );
 739
 740                 return $s;
 741         }
 742
 743         /**
 744          * transform link href and img src from relative to absolute
 745          *
 746          * @param string $text
 747          * @param string $base base url
 748          * @return string
 749          */
 750         public static function relToAbs($text, $base)
 751         {
 752                 if (empty($base)) {
 753                         return $text;
 754                 }
 755
 756                 $base = rtrim($base, '/');
 757
 758                 $base2 = $base . "/";
 759
 760                 // Replace links
 761                 $pattern = "/<a([^>]*) href=\"(?!http|https|\/)([^\"]*)\"/";
 762                 $replace = "<a\${1} href=\"" . $base2 . "\${2}\"";
 763                 $text = preg_replace($pattern, $replace, $text);
 764
 765                 $pattern = "/<a([^>]*) href=\"(?!http|https)([^\"]*)\"/";
 766                 $replace = "<a\${1} href=\"" . $base . "\${2}\"";
 767                 $text = preg_replace($pattern, $replace, $text);
 768
 769                 // Replace images
 770                 $pattern = "/<img([^>]*) src=\"(?!http|https|\/)([^\"]*)\"/";
 771                 $replace = "<img\${1} src=\"" . $base2 . "\${2}\"";
 772                 $text = preg_replace($pattern, $replace, $text);
 773
 774                 $pattern = "/<img([^>]*) src=\"(?!http|https)([^\"]*)\"/";
 775                 $replace = "<img\${1} src=\"" . $base . "\${2}\"";
 776                 $text = preg_replace($pattern, $replace, $text);
 777
 778
 779                 // Done
 780                 return $text;
 781         }
 782
 783         /**
 784          * return div element with class 'clear'
 785          * @return string
 786          * @deprecated
 787          */
 788         public static function clearDiv()
 789         {
 790                 return '<div class="clear"></div>';
 791         }
 792
 793         /**
 794          * Loader for infinite scrolling
 795          *
 796          * @return string html for loader
 797          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 798          */
 799         public static function scrollLoader()
 800         {
 801                 $tpl = Renderer::getMarkupTemplate("scroll_loader.tpl");
 802                 return Renderer::replaceMacros($tpl, [
 803                         'wait' => L10n::t('Loading more entries...'),
 804                         'end' => L10n::t('The end')
 805                 ]);
 806         }
 807
 808         /**
 809          * Get html for contact block.
 810          *
 811          * @deprecated since version 2019.03
 812          * @see ContactBlock::getHTML()
 813          * @return string
 814          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 815          * @throws \ImagickException
 816          */
 817         public static function contactBlock()
 818         {
 819                 $a = \get_app();
 820
 821                 return ContactBlock::getHTML($a->profile);
 822         }
 823
 824         /**
 825          * @brief Format contacts as picture links or as text links
 826          *
 827          * @param array   $contact  Array with contacts which contains an array with
 828          *                          int 'id' => The ID of the contact
 829          *                          int 'uid' => The user ID of the user who owns this data
 830          *                          string 'name' => The name of the contact
 831          *                          string 'url' => The url to the profile page of the contact
 832          *                          string 'addr' => The webbie of the contact (e.g.) username@friendica.com
 833          *                          string 'network' => The network to which the contact belongs to
 834          *                          string 'thumb' => The contact picture
 835          *                          string 'click' => js code which is performed when clicking on the contact
 836          * @param boolean $redirect If true try to use the redir url if it's possible
 837          * @param string  $class    CSS class for the
 838          * @param boolean $textmode If true display the contacts as text links
 839          *                          if false display the contacts as picture links
 840          * @return string Formatted html
 841          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 842          * @throws \ImagickException
 843          */
 844         public static function micropro($contact, $redirect = false, $class = '', $textmode = false)
 845         {
 846                 // Use the contact URL if no address is available
 847                 if (empty($contact['addr'])) {
 848                         $contact["addr"] = $contact["url"];
 849                 }
 850
 851                 $url = $contact['url'];
 852                 $sparkle = '';
 853                 $redir = false;
 854
 855                 if ($redirect) {
 856                         $url = Contact::magicLink($contact['url']);
 857                         if (strpos($url, 'redir/') === 0) {
 858                                 $sparkle = ' sparkle';
 859                         }
 860                 }
 861
 862                 // If there is some js available we don't need the url
 863                 if (!empty($contact['click'])) {
 864                         $url = '';
 865                 }
 866
 867                 return Renderer::replaceMacros(Renderer::getMarkupTemplate(($textmode)?'micropro_txt.tpl':'micropro_img.tpl'), [
 868                         '$click' => defaults($contact, 'click', ''),
 869                         '$class' => $class,
 870                         '$url' => $url,
 871                         '$photo' => ProxyUtils::proxifyUrl($contact['thumb'], false, ProxyUtils::SIZE_THUMB),
 872                         '$name' => $contact['name'],
 873                         'title' => $contact['name'] . ' [' . $contact['addr'] . ']',
 874                         '$parkle' => $sparkle,
 875                         '$redir' => $redir
 876                 ]);
 877         }
 878
 879         /**
 880          * Search box.
 881          *
 882          * @param string $s     Search query.
 883          * @param string $id    HTML id
 884          * @param string $url   Search url.
 885          * @param bool   $aside Display the search widgit aside.
 886          *
 887          * @return string Formatted HTML.
 888          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 889          */
 890         public static function search($s, $id = 'search-box', $url = 'search', $aside = true)
 891         {
 892                 $mode = 'text';
 893
 894                 if (strpos($s, '#') === 0) {
 895                         $mode = 'tag';
 896                 }
 897                 $save_label = $mode === 'text' ? L10n::t('Save') : L10n::t('Follow');
 898
 899                 $values = [
 900                                 '$s' => $s,
 901                                 '$id' => $id,
 902                                 '$action_url' => $url,
 903                                 '$search_label' => L10n::t('Search'),
 904                                 '$save_label' => $save_label,
 905                                 '$savedsearch' => 'savedsearch',
 906                                 '$search_hint' => L10n::t('@name, !forum, #tags, content'),
 907                                 '$mode' => $mode
 908                         ];
 909
 910                 if (!$aside) {
 911                         $values['$searchoption'] = [
 912                                                 L10n::t("Full Text"),
 913                                                 L10n::t("Tags"),
 914                                                 L10n::t("Contacts")];
 915
 916                         if (Config::get('system', 'poco_local_search')) {
 917                                 $values['$searchoption'][] = L10n::t("Forums");
 918                         }
 919                 }
 920
 921                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 922         }
 923
 924         /**
 925          * Replace naked text hyperlink with HTML formatted hyperlink
 926          *
 927          * @param string $s
 928          * @return string
 929          */
 930         public static function toLink($s)
 931         {
 932                 $s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank">$1</a>', $s);
 933                 $s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
 934                 return $s;
 935         }
 936
 937         /**
 938          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
 939          *
 940          * Reasons are expected to have been translated already.
 941          *
 942          * @param string $html
 943          * @param array  $reasons
 944          * @return string
 945          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 946          */
 947         public static function applyContentFilter($html, array $reasons)
 948         {
 949                 if (count($reasons)) {
 950                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
 951                         $html = Renderer::replaceMacros($tpl, [
 952                                 '$reasons'   => $reasons,
 953                                 '$rnd'       => Strings::getRandomHex(8),
 954                                 '$openclose' => L10n::t('Click to open/close'),
 955                                 '$html'      => $html
 956                         ]);
 957                 }
 958
 959                 return $html;
 960         }
 961
 962         /**
 963          * replace html amp entity with amp char
 964          * @param string $s
 965          * @return string
 966          */
 967         public static function unamp($s)
 968         {
 969                 return str_replace('&amp;', '&', $s);
 970         }
 971 }