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\Core\Hook;
  11 use Friendica\Core\L10n;
  12 use Friendica\Core\Config;
  13 use Friendica\Core\PConfig;
  14 use Friendica\Core\Protocol;
  15 use Friendica\Core\Renderer;
  16 use Friendica\Database\DBA;
  17 use Friendica\Model\Contact;
  18 use Friendica\Util\Network;
  19 use Friendica\Util\Proxy as ProxyUtils;
  20 use Friendica\Util\Strings;
  21 use Friendica\Util\XML;
  22 use League\HTMLToMarkdown\HtmlConverter;
  23
  24 class HTML
  25 {
  26         public static function sanitizeCSS($input)
  27         {
  28                 $cleaned = "";
  29
  30                 $input = strtolower($input);
  31
  32                 for ($i = 0; $i < strlen($input); $i++) {
  33                         $char = substr($input, $i, 1);
  34
  35                         if (($char >= "a") && ($char <= "z")) {
  36                                 $cleaned .= $char;
  37                         }
  38
  39                         if (!(strpos(" #;:0123456789-_.%", $char) === false)) {
  40                                 $cleaned .= $char;
  41                         }
  42                 }
  43
  44                 return $cleaned;
  45         }
  46
  47         private static function tagToBBCode(DOMDocument $doc, $tag, $attributes, $startbb, $endbb)
  48         {
  49                 do {
  50                         $done = self::tagToBBCodeSub($doc, $tag, $attributes, $startbb, $endbb);
  51                 } while ($done);
  52         }
  53
  54         private static function tagToBBCodeSub(DOMDocument $doc, $tag, $attributes, $startbb, $endbb)
  55         {
  56                 $savestart = str_replace('$', '\x01', $startbb);
  57                 $replace = false;
  58
  59                 $xpath = new DOMXPath($doc);
  60
  61                 $list = $xpath->query("//" . $tag);
  62                 foreach ($list as $node) {
  63                         $attr = [];
  64                         if ($node->attributes->length) {
  65                                 foreach ($node->attributes as $attribute) {
  66                                         $attr[$attribute->name] = $attribute->value;
  67                                 }
  68                         }
  69
  70                         $replace = true;
  71
  72                         $startbb = $savestart;
  73
  74                         $i = 0;
  75
  76                         foreach ($attributes as $attribute => $value) {
  77                                 $startbb = str_replace('\x01' . ++$i, '$1', $startbb);
  78                                 if (strpos('*' . $startbb, '$1') > 0) {
  79                                         if ($replace && (@$attr[$attribute] != '')) {
  80                                                 $startbb = preg_replace($value, $startbb, $attr[$attribute], -1, $count);
  81
  82                                                 // If nothing could be changed
  83                                                 if ($count == 0) {
  84                                                         $replace = false;
  85                                                 }
  86                                         } else {
  87                                                 $replace = false;
  88                                         }
  89                                 } else {
  90                                         if (@$attr[$attribute] != $value) {
  91                                                 $replace = false;
  92                                         }
  93                                 }
  94                         }
  95
  96                         if ($replace) {
  97                                 $StartCode = $doc->createTextNode($startbb);
  98                                 $EndCode = $doc->createTextNode($endbb);
  99
 100                                 $node->parentNode->insertBefore($StartCode, $node);
 101
 102                                 if ($node->hasChildNodes()) {
 103                                         foreach ($node->childNodes as $child) {
 104                                                 $newNode = $child->cloneNode(true);
 105                                                 $node->parentNode->insertBefore($newNode, $node);
 106                                         }
 107                                 }
 108
 109                                 $node->parentNode->insertBefore($EndCode, $node);
 110                                 $node->parentNode->removeChild($node);
 111                         }
 112                 }
 113
 114                 return $replace;
 115         }
 116
 117         /**
 118          * Made by: ike@piratenpartei.de
 119          * Originally made for the syncom project: http://wiki.piratenpartei.de/Syncom
 120          *                                      https://github.com/annando/Syncom
 121          *
 122          * @brief Converter for HTML to BBCode
 123          * @param string $message
 124          * @param string $basepath
 125          * @return string
 126          */
 127         public static function toBBCode($message, $basepath = '')
 128         {
 129                 $message = str_replace("\r", "", $message);
 130
 131                 // Removing code blocks before the whitespace removal processing below
 132                 $codeblocks = [];
 133                 $message = preg_replace_callback(
 134                         '#<pre><code(?: class="language-([^"]*)")?>(.*)</code></pre>#iUs',
 135                         function ($matches) use (&$codeblocks) {
 136                                 $return = '[codeblock-' . count($codeblocks) . ']';
 137
 138                                 $prefix = '[code]';
 139                                 if ($matches[1] != '') {
 140                                         $prefix = '[code=' . $matches[1] . ']';
 141                                 }
 142
 143                                 $codeblocks[] = $prefix . PHP_EOL . trim($matches[2]) . PHP_EOL . '[/code]';
 144                                 return $return;
 145                         },
 146                         $message
 147                 );
 148
 149                 $message = str_replace(
 150                         [
 151                                 "<li><p>",
 152                                 "</p></li>",
 153                         ],
 154                         [
 155                                 "<li>",
 156                                 "</li>",
 157                         ],
 158                         $message
 159                 );
 160
 161                 // remove namespaces
 162                 $message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
 163                 $message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
 164
 165                 $doc = new DOMDocument();
 166                 $doc->preserveWhiteSpace = false;
 167
 168                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 169
 170                 @$doc->loadHTML($message);
 171
 172                 XML::deleteNode($doc, 'style');
 173                 XML::deleteNode($doc, 'head');
 174                 XML::deleteNode($doc, 'title');
 175                 XML::deleteNode($doc, 'meta');
 176                 XML::deleteNode($doc, 'xml');
 177                 XML::deleteNode($doc, 'removeme');
 178
 179                 $xpath = new DomXPath($doc);
 180                 $list = $xpath->query("//pre");
 181                 foreach ($list as $node) {
 182                         // Ensure to escape unescaped & - they will otherwise raise a warning
 183                         $safe_value = preg_replace('/&(?!\w+;)/', '&amp;', $node->nodeValue);
 184                         $node->nodeValue = str_replace("\n", "\r", $safe_value);
 185                 }
 186
 187                 $message = $doc->saveHTML();
 188                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br />", " ", ""], $message);
 189                 $message = preg_replace('= [\s]*=i', " ", $message);
 190                 @$doc->loadHTML($message);
 191
 192                 self::tagToBBCode($doc, 'html', [], "", "");
 193                 self::tagToBBCode($doc, 'body', [], "", "");
 194
 195                 // Outlook-Quote - Variant 1
 196                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal', 'style' => 'margin-left:35.4pt'], '[quote]', '[/quote]');
 197
 198                 // Outlook-Quote - Variant 2
 199                 self::tagToBBCode(
 200                         $doc,
 201                         'div',
 202                         ['style' => 'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'],
 203                         '[quote]',
 204                         '[/quote]'
 205                 );
 206
 207                 // MyBB-Stuff
 208                 self::tagToBBCode($doc, 'span', ['style' => 'text-decoration: underline;'], '[u]', '[/u]');
 209                 self::tagToBBCode($doc, 'span', ['style' => 'font-style: italic;'], '[i]', '[/i]');
 210                 self::tagToBBCode($doc, 'span', ['style' => 'font-weight: bold;'], '[b]', '[/b]');
 211
 212                 /* self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
 213                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
 214                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
 215                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
 216                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
 217                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
 218                   self::node2BBCode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
 219                  */
 220                 // Untested
 221                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
 222                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
 223                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
 224
 225                 self::tagToBBCode($doc, 'span', ['style' => '/.*color:\s*(.+?)[,;].*/'], '[color="$1"]', '[/color]');
 226
 227                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 228                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
 229                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
 230                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 231                 // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
 232                 // Test
 233                 //self::node2BBCode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
 234                 self::tagToBBCode($doc, 'span', ['class' => 'type-link'], '[class=type-link]', '[/class]');
 235                 self::tagToBBCode($doc, 'span', ['class' => 'type-video'], '[class=type-video]', '[/class]');
 236
 237                 self::tagToBBCode($doc, 'strong', [], '[b]', '[/b]');
 238                 self::tagToBBCode($doc, 'em', [], '[i]', '[/i]');
 239                 self::tagToBBCode($doc, 'b', [], '[b]', '[/b]');
 240                 self::tagToBBCode($doc, 'i', [], '[i]', '[/i]');
 241                 self::tagToBBCode($doc, 'u', [], '[u]', '[/u]');
 242                 self::tagToBBCode($doc, 's', [], '[s]', '[/s]');
 243                 self::tagToBBCode($doc, 'del', [], '[s]', '[/s]');
 244                 self::tagToBBCode($doc, 'strike', [], '[s]', '[/s]');
 245
 246                 self::tagToBBCode($doc, 'big', [], "[size=large]", "[/size]");
 247                 self::tagToBBCode($doc, 'small', [], "[size=small]", "[/size]");
 248
 249                 self::tagToBBCode($doc, 'blockquote', [], '[quote]', '[/quote]');
 250
 251                 self::tagToBBCode($doc, 'br', [], "\n", '');
 252
 253                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal'], "\n", "");
 254                 self::tagToBBCode($doc, 'div', ['class' => 'MsoNormal'], "\r", "");
 255
 256                 self::tagToBBCode($doc, 'span', [], "", "");
 257
 258                 self::tagToBBCode($doc, 'span', [], "", "");
 259                 self::tagToBBCode($doc, 'pre', [], "", "");
 260
 261                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 262                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 263
 264                 self::tagToBBCode($doc, 'ul', [], "[list]", "[/list]");
 265                 self::tagToBBCode($doc, 'ol', [], "[list=1]", "[/list]");
 266                 self::tagToBBCode($doc, 'li', [], "[*]", "");
 267
 268                 self::tagToBBCode($doc, 'hr', [], "[hr]", "");
 269
 270                 self::tagToBBCode($doc, 'table', [], "", "");
 271                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 272                 self::tagToBBCode($doc, 'td', [], "\t", "");
 273                 //self::node2BBCode($doc, 'table', array(), "[table]", "[/table]");
 274                 //self::node2BBCode($doc, 'th', array(), "[th]", "[/th]");
 275                 //self::node2BBCode($doc, 'tr', array(), "[tr]", "[/tr]");
 276                 //self::node2BBCode($doc, 'td', array(), "[td]", "[/td]");
 277                 //self::node2BBCode($doc, 'h1', array(), "\n\n[size=xx-large][b]", "[/b][/size]\n");
 278                 //self::node2BBCode($doc, 'h2', array(), "\n\n[size=x-large][b]", "[/b][/size]\n");
 279                 //self::node2BBCode($doc, 'h3', array(), "\n\n[size=large][b]", "[/b][/size]\n");
 280                 //self::node2BBCode($doc, 'h4', array(), "\n\n[size=medium][b]", "[/b][/size]\n");
 281                 //self::node2BBCode($doc, 'h5', array(), "\n\n[size=small][b]", "[/b][/size]\n");
 282                 //self::node2BBCode($doc, 'h6', array(), "\n\n[size=x-small][b]", "[/b][/size]\n");
 283
 284                 self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 285                 self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 286                 self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 287                 self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 288                 self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 289                 self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 290
 291                 self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 292                 self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 293
 294                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]');
 295                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]');
 296
 297
 298                 self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]');
 299                 self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]');
 300                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[iframe]$1', '[/iframe]');
 301
 302                 self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 303                 self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 304
 305                 $message = $doc->saveHTML();
 306
 307                 // I'm removing something really disturbing
 308                 // Don't know exactly what it is
 309                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 310
 311                 $message = str_replace("&nbsp;", " ", $message);
 312
 313                 // removing multiple DIVs
 314                 $message = preg_replace('=\r *\r=i', "\n", $message);
 315                 $message = str_replace("\r", "\n", $message);
 316
 317                 Hook::callAll('html2bbcode', $message);
 318
 319                 $message = strip_tags($message);
 320
 321                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 322
 323                 $message = str_replace(["<"], ["&lt;"], $message);
 324
 325                 // remove quotes if they don't make sense
 326                 $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 327
 328                 $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 329                 $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 330
 331                 do {
 332                         $oldmessage = $message;
 333                         $message = str_replace("\n \n", "\n\n", $message);
 334                 } while ($oldmessage != $message);
 335
 336                 do {
 337                         $oldmessage = $message;
 338                         $message = str_replace("\n\n\n", "\n\n", $message);
 339                 } while ($oldmessage != $message);
 340
 341                 do {
 342                         $oldmessage = $message;
 343                         $message = str_replace(
 344                                 [
 345                                 "[/size]\n\n",
 346                                 "\n[hr]",
 347                                 "[hr]\n",
 348                                 "\n[list",
 349                                 "[/list]\n",
 350                                 "\n[/",
 351                                 "[list]\n",
 352                                 "[list=1]\n",
 353                                 "\n[*]"],
 354                                 [
 355                                 "[/size]\n",
 356                                 "[hr]",
 357                                 "[hr]",
 358                                 "[list",
 359                                 "[/list]",
 360                                 "[/",
 361                                 "[list]",
 362                                 "[list=1]",
 363                                 "[*]"],
 364                                 $message
 365                         );
 366                 } while ($message != $oldmessage);
 367
 368                 $message = str_replace(
 369                         ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 370                         ['[b]', '[/b]', '[i]', '[/i]'],
 371                         $message
 372                 );
 373
 374                 // Handling Yahoo style of mails
 375                 $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 376
 377                 // Restore code blocks
 378                 $message = preg_replace_callback(
 379                         '#\[codeblock-([0-9]+)\]#iU',
 380                         function ($matches) use ($codeblocks) {
 381                                 $return = '';
 382                                 if (isset($codeblocks[intval($matches[1])])) {
 383                                         $return = $codeblocks[$matches[1]];
 384                                 }
 385                                 return $return;
 386                         },
 387                         $message
 388                 );
 389
 390                 $message = trim($message);
 391
 392                 if ($basepath != '') {
 393                         $message = self::qualifyURLs($message, $basepath);
 394                 }
 395
 396                 return $message;
 397         }
 398
 399         /**
 400          * @brief Sub function to complete incomplete URL
 401          *
 402          * @param array  $matches  Result of preg_replace_callback
 403          * @param string $basepath Basepath that is used to complete the URL
 404          *
 405          * @return string The expanded URL
 406          */
 407         private static function qualifyURLsSub($matches, $basepath)
 408         {
 409                 $base = parse_url($basepath);
 410                 unset($base['query']);
 411                 unset($base['fragment']);
 412
 413                 $link = $matches[0];
 414                 $url = $matches[1];
 415
 416                 $parts = array_merge($base, parse_url($url));
 417                 $url2 = Network::unparseURL($parts);
 418
 419                 return str_replace($url, $url2, $link);
 420         }
 421
 422         /**
 423          * @brief Complete incomplete URLs in BBCode
 424          *
 425          * @param string $body     Body with URLs
 426          * @param string $basepath Base path that is used to complete the URL
 427          *
 428          * @return string Body with expanded URLs
 429          */
 430         private static function qualifyURLs($body, $basepath)
 431         {
 432                 $URLSearchString = "^\[\]";
 433
 434                 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
 435                         "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
 436                         "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 437                         "/\[img\](.*?)\[\/img\]/ism",
 438                         "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 439                         "/\[zmg\](.*?)\[\/zmg\]/ism",
 440                         "/\[video\](.*?)\[\/video\]/ism",
 441                         "/\[audio\](.*?)\[\/audio\]/ism",
 442                 ];
 443
 444                 foreach ($matches as $match) {
 445                         $body = preg_replace_callback(
 446                                 $match,
 447                                 function ($match) use ($basepath) {
 448                                         return self::qualifyURLsSub($match, $basepath);
 449                                 },
 450                                 $body
 451                         );
 452                 }
 453                 return $body;
 454         }
 455
 456         private static function breakLines($line, $level, $wraplength = 75)
 457         {
 458                 if ($wraplength == 0) {
 459                         $wraplength = 2000000;
 460                 }
 461
 462                 $wraplen = $wraplength - $level;
 463
 464                 $newlines = [];
 465
 466                 do {
 467                         $oldline = $line;
 468
 469                         $subline = substr($line, 0, $wraplen);
 470
 471                         $pos = strrpos($subline, ' ');
 472
 473                         if ($pos == 0) {
 474                                 $pos = strpos($line, ' ');
 475                         }
 476
 477                         if (($pos > 0) && strlen($line) > $wraplen) {
 478                                 $newline = trim(substr($line, 0, $pos));
 479                                 if ($level > 0) {
 480                                         $newline = str_repeat(">", $level) . ' ' . $newline;
 481                                 }
 482
 483                                 $newlines[] = $newline . " ";
 484                                 $line = substr($line, $pos + 1);
 485                         }
 486                 } while ((strlen($line) > $wraplen) && !($oldline == $line));
 487
 488                 if ($level > 0) {
 489                         $line = str_repeat(">", $level) . ' ' . $line;
 490                 }
 491
 492                 $newlines[] = $line;
 493
 494                 return implode($newlines, "\n");
 495         }
 496
 497         private static function quoteLevel($message, $wraplength = 75)
 498         {
 499                 $lines = explode("\n", $message);
 500
 501                 $newlines = [];
 502                 $level = 0;
 503                 foreach ($lines as $line) {
 504                         $line = trim($line);
 505                         $startquote = false;
 506                         while (strpos("*" . $line, '[quote]') > 0) {
 507                                 $level++;
 508                                 $pos = strpos($line, '[quote]');
 509                                 $line = substr($line, 0, $pos) . substr($line, $pos + 7);
 510                                 $startquote = true;
 511                         }
 512
 513                         $currlevel = $level;
 514
 515                         while (strpos("*" . $line, '[/quote]') > 0) {
 516                                 $level--;
 517                                 if ($level < 0) {
 518                                         $level = 0;
 519                                 }
 520
 521                                 $pos = strpos($line, '[/quote]');
 522                                 $line = substr($line, 0, $pos) . substr($line, $pos + 8);
 523                         }
 524
 525                         if (!$startquote || ($line != '')) {
 526                                 $newlines[] = self::breakLines($line, $currlevel, $wraplength);
 527                         }
 528                 }
 529
 530                 return implode($newlines, "\n");
 531         }
 532
 533         private static function collectURLs($message)
 534         {
 535                 $pattern = '/<a.*?href="(.*?)".*?>(.*?)<\/a>/is';
 536                 preg_match_all($pattern, $message, $result, PREG_SET_ORDER);
 537
 538                 $urls = [];
 539                 foreach ($result as $treffer) {
 540                         $ignore = false;
 541
 542                         // A list of some links that should be ignored
 543                         $list = ["/user/", "/tag/", "/group/", "/profile/", "/search?search=", "/search?tag=", "mailto:", "/u/", "/node/",
 544                                 "//plus.google.com/", "//twitter.com/"];
 545                         foreach ($list as $listitem) {
 546                                 if (strpos($treffer[1], $listitem) !== false) {
 547                                         $ignore = true;
 548                                 }
 549                         }
 550
 551                         if ((strpos($treffer[1], "//twitter.com/") !== false) && (strpos($treffer[1], "/status/") !== false)) {
 552                                 $ignore = false;
 553                         }
 554
 555                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/posts") !== false)) {
 556                                 $ignore = false;
 557                         }
 558
 559                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/photos") !== false)) {
 560                                 $ignore = false;
 561                         }
 562
 563                         if (!$ignore) {
 564                                 $urls[$treffer[1]] = $treffer[1];
 565                         }
 566                 }
 567
 568                 return $urls;
 569         }
 570
 571         public static function toPlaintext($html, $wraplength = 75, $compact = false)
 572         {
 573                 $message = str_replace("\r", "", $html);
 574
 575                 $doc = new DOMDocument();
 576                 $doc->preserveWhiteSpace = false;
 577
 578                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 579
 580                 @$doc->loadHTML($message);
 581
 582                 $xpath = new DOMXPath($doc);
 583                 $list = $xpath->query("//pre");
 584                 foreach ($list as $node) {
 585                         $node->nodeValue = str_replace("\n", "\r", $node->nodeValue);
 586                 }
 587
 588                 $message = $doc->saveHTML();
 589                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br>", " ", ""], $message);
 590                 $message = preg_replace('= [\s]*=i', " ", $message);
 591
 592                 // Collecting all links
 593                 $urls = self::collectURLs($message);
 594
 595                 @$doc->loadHTML($message);
 596
 597                 self::tagToBBCode($doc, 'html', [], '', '');
 598                 self::tagToBBCode($doc, 'body', [], '', '');
 599
 600                 // MyBB-Auszeichnungen
 601                 /*
 602                   self::node2BBCode($doc, 'span', array('style'=>'text-decoration: underline;'), '_', '_');
 603                   self::node2BBCode($doc, 'span', array('style'=>'font-style: italic;'), '/', '/');
 604                   self::node2BBCode($doc, 'span', array('style'=>'font-weight: bold;'), '*', '*');
 605
 606                   self::node2BBCode($doc, 'strong', array(), '*', '*');
 607                   self::node2BBCode($doc, 'b', array(), '*', '*');
 608                   self::node2BBCode($doc, 'i', array(), '/', '/');
 609                   self::node2BBCode($doc, 'u', array(), '_', '_');
 610                  */
 611
 612                 if ($compact) {
 613                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 614                 } else {
 615                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 616                 }
 617
 618                 self::tagToBBCode($doc, 'br', [], "\n", '');
 619
 620                 self::tagToBBCode($doc, 'span', [], "", "");
 621                 self::tagToBBCode($doc, 'pre', [], "", "");
 622                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 623                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 624
 625                 //self::node2BBCode($doc, 'ul', array(), "\n[list]", "[/list]\n");
 626                 //self::node2BBCode($doc, 'ol', array(), "\n[list=1]", "[/list]\n");
 627                 self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 628
 629                 self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
 630
 631                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 632                 self::tagToBBCode($doc, 'td', [], "\t", "");
 633
 634                 self::tagToBBCode($doc, 'h1', [], "\n\n*", "*\n");
 635                 self::tagToBBCode($doc, 'h2', [], "\n\n*", "*\n");
 636                 self::tagToBBCode($doc, 'h3', [], "\n\n*", "*\n");
 637                 self::tagToBBCode($doc, 'h4', [], "\n\n*", "*\n");
 638                 self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
 639                 self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 640
 641                 // Problem: there is no reliable way to detect if it is a link to a tag or profile
 642                 //self::node2BBCode($doc, 'a', array('href'=>'/(.+)/'), ' $1 ', ' ', true);
 643                 //self::node2BBCode($doc, 'a', array('href'=>'/(.+)/', 'rel'=>'oembed'), ' $1 ', '', true);
 644                 //self::node2BBCode($doc, 'img', array('alt'=>'/(.+)/'), '$1', '');
 645                 //self::node2BBCode($doc, 'img', array('title'=>'/(.+)/'), '$1', '');
 646                 //self::node2BBCode($doc, 'img', array(), '', '');
 647                 if (!$compact) {
 648                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
 649                 } else {
 650                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' ', ' ');
 651                 }
 652
 653                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], ' $1 ', '');
 654
 655                 $message = $doc->saveHTML();
 656
 657                 if (!$compact) {
 658                         $message = str_replace("[img]", "", $message);
 659                         $message = str_replace("[/img]", "", $message);
 660                 }
 661
 662                 // was ersetze ich da?
 663                 // Irgendein stoerrisches UTF-Zeug
 664                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 665
 666                 $message = str_replace("&nbsp;", " ", $message);
 667
 668                 // Aufeinanderfolgende DIVs
 669                 $message = preg_replace('=\r *\r=i', "\n", $message);
 670                 $message = str_replace("\r", "\n", $message);
 671
 672                 $message = strip_tags($message);
 673
 674                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 675
 676                 if (!$compact && ($message != '')) {
 677                         foreach ($urls as $id => $url) {
 678                                 if ($url != '' && strpos($message, $url) === false) {
 679                                         $message .= "\n" . $url . ' ';
 680                                 }
 681                         }
 682                 }
 683
 684                 $message = str_replace("\n«", "«\n", $message);
 685                 $message = str_replace("»\n", "\n»", $message);
 686
 687                 do {
 688                         $oldmessage = $message;
 689                         $message = str_replace("\n\n\n", "\n\n", $message);
 690                 } while ($oldmessage != $message);
 691
 692                 $message = self::quoteLevel(trim($message), $wraplength);
 693
 694                 return trim($message);
 695         }
 696
 697         /**
 698          * Converts provided HTML code to Markdown. The hardwrap parameter maximizes
 699          * compatibility with Diaspora in spite of the Markdown standards.
 700          *
 701          * @param string $html
 702          * @return string
 703          */
 704         public static function toMarkdown($html)
 705         {
 706                 $converter = new HtmlConverter(['hard_break' => true]);
 707                 $markdown = $converter->convert($html);
 708
 709                 return $markdown;
 710         }
 711
 712         /**
 713          * @brief Convert video HTML to BBCode tags
 714          *
 715          * @param string $s
 716          */
 717         public static function toBBCodeVideo($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          * @return string html for loader
 793          */
 794         public static function scrollLoader()
 795         {
 796                 $tpl = Renderer::getMarkupTemplate("scroll_loader.tpl");
 797                 return Renderer::replaceMacros($tpl, [
 798                         'wait' => L10n::t('Loading more entries...'),
 799                         'end' => L10n::t('The end')
 800                 ]);
 801         }
 802
 803         /**
 804          * Get html for contact block.
 805          *
 806          * @template contact_block.tpl
 807          * @hook contact_block_end (contacts=>array, output=>string)
 808          * @return string
 809          */
 810         public static function contactBlock()
 811         {
 812                 $o = '';
 813                 $a = \get_app();
 814
 815                 $shown = PConfig::get($a->profile['uid'], 'system', 'display_friend_count', 24);
 816                 if ($shown == 0) {
 817                         return;
 818                 }
 819
 820                 if (!is_array($a->profile) || $a->profile['hide-friends']) {
 821                         return $o;
 822                 }
 823
 824                 $r = q("SELECT COUNT(*) AS `total` FROM `contact`
 825                                 WHERE `uid` = %d AND NOT `self` AND NOT `blocked`
 826                                         AND NOT `pending` AND NOT `hidden` AND NOT `archive`
 827                                         AND `network` IN ('%s', '%s', '%s')",
 828                         intval($a->profile['uid']),
 829                         DBA::escape(Protocol::DFRN),
 830                         DBA::escape(Protocol::OSTATUS),
 831                         DBA::escape(Protocol::DIASPORA)
 832                 );
 833
 834                 if (DBA::isResult($r)) {
 835                         $total = intval($r[0]['total']);
 836                 }
 837
 838                 if (!$total) {
 839                         $contacts = L10n::t('No contacts');
 840                         $micropro = null;
 841                 } else {
 842                         // Splitting the query in two parts makes it much faster
 843                         $r = q("SELECT `id` FROM `contact`
 844                                         WHERE `uid` = %d AND NOT `self` AND NOT `blocked`
 845                                                 AND NOT `pending` AND NOT `hidden` AND NOT `archive`
 846                                                 AND `network` IN ('%s', '%s', '%s')
 847                                         ORDER BY RAND() LIMIT %d",
 848                                 intval($a->profile['uid']),
 849                                 DBA::escape(Protocol::DFRN),
 850                                 DBA::escape(Protocol::OSTATUS),
 851                                 DBA::escape(Protocol::DIASPORA),
 852                                 intval($shown)
 853                         );
 854
 855                         if (DBA::isResult($r)) {
 856                                 $contacts = [];
 857                                 foreach ($r as $contact) {
 858                                         $contacts[] = $contact["id"];
 859                                 }
 860
 861                                 $r = q("SELECT `id`, `uid`, `addr`, `url`, `name`, `thumb`, `network` FROM `contact` WHERE `id` IN (%s)",
 862                                         DBA::escape(implode(",", $contacts))
 863                                 );
 864
 865                                 if (DBA::isResult($r)) {
 866                                         $contacts = L10n::tt('%d Contact', '%d Contacts', $total);
 867                                         $micropro = [];
 868                                         foreach ($r as $rr) {
 869                                                 $micropro[] = self::micropro($rr, true, 'mpfriend');
 870                                         }
 871                                 }
 872                         }
 873                 }
 874
 875                 $tpl = Renderer::getMarkupTemplate('contact_block.tpl');
 876                 $o = Renderer::replaceMacros($tpl, [
 877                         '$contacts' => $contacts,
 878                         '$nickname' => $a->profile['nickname'],
 879                         '$viewcontacts' => L10n::t('View Contacts'),
 880                         '$micropro' => $micropro,
 881                 ]);
 882
 883                 $arr = ['contacts' => $r, 'output' => $o];
 884
 885                 Hook::callAll('contact_block_end', $arr);
 886
 887                 return $o;
 888         }
 889
 890         /**
 891          * @brief Format contacts as picture links or as text links
 892          *
 893          * @param array $contact Array with contacts which contains an array with
 894          *      int 'id' => The ID of the contact
 895         *       int 'uid' => The user ID of the user who owns this data
 896         *       string 'name' => The name of the contact
 897         *       string 'url' => The url to the profile page of the contact
 898         *       string 'addr' => The webbie of the contact (e.g.) username@friendica.com
 899         *       string 'network' => The network to which the contact belongs to
 900         *       string 'thumb' => The contact picture
 901         *       string 'click' => js code which is performed when clicking on the contact
 902         * @param boolean $redirect If true try to use the redir url if it's possible
 903         * @param string $class CSS class for the
 904         * @param boolean $textmode If true display the contacts as text links
 905         *       if false display the contacts as picture links
 906
 907         * @return string Formatted html
 908         */
 909         public static function micropro($contact, $redirect = false, $class = '', $textmode = false)
 910         {
 911                 // Use the contact URL if no address is available
 912                 if (empty($contact['addr'])) {
 913                         $contact["addr"] = $contact["url"];
 914                 }
 915
 916                 $url = $contact['url'];
 917                 $sparkle = '';
 918                 $redir = false;
 919
 920                 if ($redirect) {
 921                         $url = Contact::magicLink($contact['url']);
 922                         if (strpos($url, 'redir/') === 0) {
 923                                 $sparkle = ' sparkle';
 924                         }
 925                 }
 926
 927                 // If there is some js available we don't need the url
 928                 if (!empty($contact['click'])) {
 929                         $url = '';
 930                 }
 931
 932                 return Renderer::replaceMacros(Renderer::getMarkupTemplate(($textmode)?'micropro_txt.tpl':'micropro_img.tpl'), [
 933                         '$click' => defaults($contact, 'click', ''),
 934                         '$class' => $class,
 935                         '$url' => $url,
 936                         '$photo' => ProxyUtils::proxifyUrl($contact['thumb'], false, ProxyUtils::SIZE_THUMB),
 937                         '$name' => $contact['name'],
 938                         'title' => $contact['name'] . ' [' . $contact['addr'] . ']',
 939                         '$parkle' => $sparkle,
 940                         '$redir' => $redir
 941                 ]);
 942         }
 943
 944         /**
 945          * Search box.
 946          *
 947          * @param string $s     Search query.
 948          * @param string $id    HTML id
 949          * @param string $url   Search url.
 950          * @param bool   $save  Show save search button.
 951          * @param bool   $aside Display the search widgit aside.
 952          *
 953          * @return string Formatted HTML.
 954          */
 955         public static function search($s, $id = 'search-box', $url = 'search', $aside = true)
 956         {
 957                 $mode = 'text';
 958
 959                 if (strpos($s, '#') === 0) {
 960                         $mode = 'tag';
 961                 }
 962                 $save_label = $mode === 'text' ? L10n::t('Save') : L10n::t('Follow');
 963
 964                 $values = [
 965                                 '$s' => $s,
 966                                 '$id' => $id,
 967                                 '$action_url' => $url,
 968                                 '$search_label' => L10n::t('Search'),
 969                                 '$save_label' => $save_label,
 970                                 '$savedsearch' => 'savedsearch',
 971                                 '$search_hint' => L10n::t('@name, !forum, #tags, content'),
 972                                 '$mode' => $mode
 973                         ];
 974
 975                 if (!$aside) {
 976                         $values['$searchoption'] = [
 977                                                 L10n::t("Full Text"),
 978                                                 L10n::t("Tags"),
 979                                                 L10n::t("Contacts")];
 980
 981                         if (Config::get('system', 'poco_local_search')) {
 982                                 $values['$searchoption'][] = L10n::t("Forums");
 983                         }
 984                 }
 985
 986                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 987         }
 988
 989         /**
 990          * Replace naked text hyperlink with HTML formatted hyperlink
 991          *
 992          * @param string $s
 993          */
 994         public static function toLink($s)
 995         {
 996                 $s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank">$1</a>', $s);
 997                 $s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
 998                 return $s;
 999         }
1000
1001         /**
1002          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
1003          *
1004          * Reasons are expected to have been translated already.
1005          *
1006          * @param string $html
1007          * @param array  $reasons
1008          * @return string
1009          */
1010         public static function applyContentFilter($html, array $reasons)
1011         {
1012                 if (count($reasons)) {
1013                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
1014                         $html = Renderer::replaceMacros($tpl, [
1015                                 '$reasons'   => $reasons,
1016                                 '$rnd'       => Strings::getRandomHex(8),
1017                                 '$openclose' => L10n::t('Click to open/close'),
1018                                 '$html'      => $html
1019                         ]);
1020                 }
1021
1022                 return $html;
1023         }
1024
1025         /**
1026          * replace html amp entity with amp char
1027          * @param string $s
1028          * @return string
1029          */
1030         public static function unamp($s)
1031         {
1032                 return str_replace('&amp;', '&', $s);
1033         }
1034 }