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\Addon;
  11 use Friendica\Core\L10n;
  12 use Friendica\Core\Config;
  13 use Friendica\Core\PConfig;
  14 use Friendica\Core\Renderer;
  15 use Friendica\Database\DBA;
  16 use Friendica\Model\Contact;
  17 use Friendica\Util\Network;
  18 use Friendica\Util\Proxy as ProxyUtils;
  19 use Friendica\Util\XML;
  20 use League\HTMLToMarkdown\HtmlConverter;
  21
  22 class HTML
  23 {
  24         public static function sanitizeCSS($input)
  25         {
  26                 $cleaned = "";
  27
  28                 $input = strtolower($input);
  29
  30                 for ($i = 0; $i < strlen($input); $i++) {
  31                         $char = substr($input, $i, 1);
  32
  33                         if (($char >= "a") && ($char <= "z")) {
  34                                 $cleaned .= $char;
  35                         }
  36
  37                         if (!(strpos(" #;:0123456789-_.%", $char) === false)) {
  38                                 $cleaned .= $char;
  39                         }
  40                 }
  41
  42                 return $cleaned;
  43         }
  44
  45         private static function tagToBBCode(DOMDocument $doc, $tag, $attributes, $startbb, $endbb)
  46         {
  47                 do {
  48                         $done = self::tagToBBCodeSub($doc, $tag, $attributes, $startbb, $endbb);
  49                 } while ($done);
  50         }
  51
  52         private static function tagToBBCodeSub(DOMDocument $doc, $tag, $attributes, $startbb, $endbb)
  53         {
  54                 $savestart = str_replace('$', '\x01', $startbb);
  55                 $replace = false;
  56
  57                 $xpath = new DOMXPath($doc);
  58
  59                 $list = $xpath->query("//" . $tag);
  60                 foreach ($list as $node) {
  61                         $attr = [];
  62                         if ($node->attributes->length) {
  63                                 foreach ($node->attributes as $attribute) {
  64                                         $attr[$attribute->name] = $attribute->value;
  65                                 }
  66                         }
  67
  68                         $replace = true;
  69
  70                         $startbb = $savestart;
  71
  72                         $i = 0;
  73
  74                         foreach ($attributes as $attribute => $value) {
  75                                 $startbb = str_replace('\x01' . ++$i, '$1', $startbb);
  76                                 if (strpos('*' . $startbb, '$1') > 0) {
  77                                         if ($replace && (@$attr[$attribute] != '')) {
  78                                                 $startbb = preg_replace($value, $startbb, $attr[$attribute], -1, $count);
  79
  80                                                 // If nothing could be changed
  81                                                 if ($count == 0) {
  82                                                         $replace = false;
  83                                                 }
  84                                         } else {
  85                                                 $replace = false;
  86                                         }
  87                                 } else {
  88                                         if (@$attr[$attribute] != $value) {
  89                                                 $replace = false;
  90                                         }
  91                                 }
  92                         }
  93
  94                         if ($replace) {
  95                                 $StartCode = $doc->createTextNode($startbb);
  96                                 $EndCode = $doc->createTextNode($endbb);
  97
  98                                 $node->parentNode->insertBefore($StartCode, $node);
  99
 100                                 if ($node->hasChildNodes()) {
 101                                         foreach ($node->childNodes as $child) {
 102                                                 $newNode = $child->cloneNode(true);
 103                                                 $node->parentNode->insertBefore($newNode, $node);
 104                                         }
 105                                 }
 106
 107                                 $node->parentNode->insertBefore($EndCode, $node);
 108                                 $node->parentNode->removeChild($node);
 109                         }
 110                 }
 111
 112                 return $replace;
 113         }
 114
 115         /**
 116          * Made by: ike@piratenpartei.de
 117          * Originally made for the syncom project: http://wiki.piratenpartei.de/Syncom
 118          *                                      https://github.com/annando/Syncom
 119          *
 120          * @brief Converter for HTML to BBCode
 121          * @param string $message
 122          * @param string $basepath
 123          * @return string
 124          */
 125         public static function toBBCode($message, $basepath = '')
 126         {
 127                 $message = str_replace("\r", "", $message);
 128
 129                 // Removing code blocks before the whitespace removal processing below
 130                 $codeblocks = [];
 131                 $message = preg_replace_callback(
 132                         '#<pre><code(?: class="language-([^"]*)")?>(.*)</code></pre>#iUs',
 133                         function ($matches) use (&$codeblocks) {
 134                                 $return = '[codeblock-' . count($codeblocks) . ']';
 135
 136                                 $prefix = '[code]';
 137                                 if ($matches[1] != '') {
 138                                         $prefix = '[code=' . $matches[1] . ']';
 139                                 }
 140
 141                                 $codeblocks[] = $prefix . PHP_EOL . trim($matches[2]) . PHP_EOL . '[/code]';
 142                                 return $return;
 143                         },
 144                         $message
 145                 );
 146
 147                 $message = str_replace(
 148                         [
 149                                 "<li><p>",
 150                                 "</p></li>",
 151                         ],
 152                         [
 153                                 "<li>",
 154                                 "</li>",
 155                         ],
 156                         message
 157                 );
 158
 159                 // remove namespaces
 160                 $message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
 161                 $message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
 162
 163                 $doc = new DOMDocument();
 164                 $doc->preserveWhiteSpace = false;
 165
 166                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 167
 168                 @$doc->loadHTML($message);
 169
 170                 XML::deleteNode($doc, 'style');
 171                 XML::deleteNode($doc, 'head');
 172                 XML::deleteNode($doc, 'title');
 173                 XML::deleteNode($doc, 'meta');
 174                 XML::deleteNode($doc, 'xml');
 175                 XML::deleteNode($doc, 'removeme');
 176
 177                 $xpath = new DomXPath($doc);
 178                 $list = $xpath->query("//pre");
 179                 foreach ($list as $node) {
 180                         $node->nodeValue = str_replace("\n", "\r", $node->nodeValue);
 181                 }
 182
 183                 $message = $doc->saveHTML();
 184                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br />", " ", ""], $message);
 185                 $message = preg_replace('= [\s]*=i', " ", $message);
 186                 @$doc->loadHTML($message);
 187
 188                 self::tagToBBCode($doc, 'html', [], "", "");
 189                 self::tagToBBCode($doc, 'body', [], "", "");
 190
 191                 // Outlook-Quote - Variant 1
 192                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal', 'style' => 'margin-left:35.4pt'], '[quote]', '[/quote]');
 193
 194                 // Outlook-Quote - Variant 2
 195                 self::tagToBBCode(
 196                         $doc,
 197                         'div',
 198                         ['style' => 'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'],
 199                         '[quote]',
 200                         '[/quote]'
 201                 );
 202
 203                 // MyBB-Stuff
 204                 self::tagToBBCode($doc, 'span', ['style' => 'text-decoration: underline;'], '[u]', '[/u]');
 205                 self::tagToBBCode($doc, 'span', ['style' => 'font-style: italic;'], '[i]', '[/i]');
 206                 self::tagToBBCode($doc, 'span', ['style' => 'font-weight: bold;'], '[b]', '[/b]');
 207
 208                 /* self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
 209                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
 210                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
 211                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
 212                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
 213                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
 214                   self::node2BBCode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
 215                  */
 216                 // Untested
 217                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
 218                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
 219                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
 220
 221                 self::tagToBBCode($doc, 'span', ['style' => '/.*color:\s*(.+?)[,;].*/'], '[color="$1"]', '[/color]');
 222
 223                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 224                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
 225                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
 226                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 227                 // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
 228                 // Test
 229                 //self::node2BBCode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
 230                 self::tagToBBCode($doc, 'span', ['class' => 'type-link'], '[class=type-link]', '[/class]');
 231                 self::tagToBBCode($doc, 'span', ['class' => 'type-video'], '[class=type-video]', '[/class]');
 232
 233                 self::tagToBBCode($doc, 'strong', [], '[b]', '[/b]');
 234                 self::tagToBBCode($doc, 'em', [], '[i]', '[/i]');
 235                 self::tagToBBCode($doc, 'b', [], '[b]', '[/b]');
 236                 self::tagToBBCode($doc, 'i', [], '[i]', '[/i]');
 237                 self::tagToBBCode($doc, 'u', [], '[u]', '[/u]');
 238                 self::tagToBBCode($doc, 's', [], '[s]', '[/s]');
 239                 self::tagToBBCode($doc, 'del', [], '[s]', '[/s]');
 240                 self::tagToBBCode($doc, 'strike', [], '[s]', '[/s]');
 241
 242                 self::tagToBBCode($doc, 'big', [], "[size=large]", "[/size]");
 243                 self::tagToBBCode($doc, 'small', [], "[size=small]", "[/size]");
 244
 245                 self::tagToBBCode($doc, 'blockquote', [], '[quote]', '[/quote]');
 246
 247                 self::tagToBBCode($doc, 'br', [], "\n", '');
 248
 249                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal'], "\n", "");
 250                 self::tagToBBCode($doc, 'div', ['class' => 'MsoNormal'], "\r", "");
 251
 252                 self::tagToBBCode($doc, 'span', [], "", "");
 253
 254                 self::tagToBBCode($doc, 'span', [], "", "");
 255                 self::tagToBBCode($doc, 'pre', [], "", "");
 256
 257                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 258                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 259
 260                 self::tagToBBCode($doc, 'ul', [], "[list]", "[/list]");
 261                 self::tagToBBCode($doc, 'ol', [], "[list=1]", "[/list]");
 262                 self::tagToBBCode($doc, 'li', [], "[*]", "");
 263
 264                 self::tagToBBCode($doc, 'hr', [], "[hr]", "");
 265
 266                 self::tagToBBCode($doc, 'table', [], "", "");
 267                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 268                 self::tagToBBCode($doc, 'td', [], "\t", "");
 269                 //self::node2BBCode($doc, 'table', array(), "[table]", "[/table]");
 270                 //self::node2BBCode($doc, 'th', array(), "[th]", "[/th]");
 271                 //self::node2BBCode($doc, 'tr', array(), "[tr]", "[/tr]");
 272                 //self::node2BBCode($doc, 'td', array(), "[td]", "[/td]");
 273                 //self::node2BBCode($doc, 'h1', array(), "\n\n[size=xx-large][b]", "[/b][/size]\n");
 274                 //self::node2BBCode($doc, 'h2', array(), "\n\n[size=x-large][b]", "[/b][/size]\n");
 275                 //self::node2BBCode($doc, 'h3', array(), "\n\n[size=large][b]", "[/b][/size]\n");
 276                 //self::node2BBCode($doc, 'h4', array(), "\n\n[size=medium][b]", "[/b][/size]\n");
 277                 //self::node2BBCode($doc, 'h5', array(), "\n\n[size=small][b]", "[/b][/size]\n");
 278                 //self::node2BBCode($doc, 'h6', array(), "\n\n[size=x-small][b]", "[/b][/size]\n");
 279
 280                 self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 281                 self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 282                 self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 283                 self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 284                 self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 285                 self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 286
 287                 self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 288                 self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 289
 290                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]');
 291                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]');
 292
 293
 294                 self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]');
 295                 self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]');
 296                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[iframe]$1', '[/iframe]');
 297
 298                 self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 299                 self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 300
 301                 $message = $doc->saveHTML();
 302
 303                 // I'm removing something really disturbing
 304                 // Don't know exactly what it is
 305                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 306
 307                 $message = str_replace("&nbsp;", " ", $message);
 308
 309                 // removing multiple DIVs
 310                 $message = preg_replace('=\r *\r=i', "\n", $message);
 311                 $message = str_replace("\r", "\n", $message);
 312
 313                 Addon::callHooks('html2bbcode', $message);
 314
 315                 $message = strip_tags($message);
 316
 317                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 318
 319                 $message = str_replace(["<"], ["&lt;"], $message);
 320
 321                 // remove quotes if they don't make sense
 322                 $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 323
 324                 $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 325                 $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 326
 327                 do {
 328                         $oldmessage = $message;
 329                         $message = str_replace("\n \n", "\n\n", $message);
 330                 } while ($oldmessage != $message);
 331
 332                 do {
 333                         $oldmessage = $message;
 334                         $message = str_replace("\n\n\n", "\n\n", $message);
 335                 } while ($oldmessage != $message);
 336
 337                 do {
 338                         $oldmessage = $message;
 339                         $message = str_replace(
 340                                 [
 341                                 "[/size]\n\n",
 342                                 "\n[hr]",
 343                                 "[hr]\n",
 344                                 "\n[list",
 345                                 "[/list]\n",
 346                                 "\n[/",
 347                                 "[list]\n",
 348                                 "[list=1]\n",
 349                                 "\n[*]"],
 350                                 [
 351                                 "[/size]\n",
 352                                 "[hr]",
 353                                 "[hr]",
 354                                 "[list",
 355                                 "[/list]",
 356                                 "[/",
 357                                 "[list]",
 358                                 "[list=1]",
 359                                 "[*]"],
 360                                 $message
 361                         );
 362                 } while ($message != $oldmessage);
 363
 364                 $message = str_replace(
 365                         ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 366                         ['[b]', '[/b]', '[i]', '[/i]'],
 367                         $message
 368                 );
 369
 370                 // Handling Yahoo style of mails
 371                 $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 372
 373                 // Restore code blocks
 374                 $message = preg_replace_callback(
 375                         '#\[codeblock-([0-9]+)\]#iU',
 376                         function ($matches) use ($codeblocks) {
 377                                 $return = '';
 378                                 if (isset($codeblocks[intval($matches[1])])) {
 379                                         $return = $codeblocks[$matches[1]];
 380                                 }
 381                                 return $return;
 382                         },
 383                         $message
 384                 );
 385
 386                 $message = trim($message);
 387
 388                 if ($basepath != '') {
 389                         $message = self::qualifyURLs($message, $basepath);
 390                 }
 391
 392                 return $message;
 393         }
 394
 395         /**
 396          * @brief Sub function to complete incomplete URL
 397          *
 398          * @param array  $matches  Result of preg_replace_callback
 399          * @param string $basepath Basepath that is used to complete the URL
 400          *
 401          * @return string The expanded URL
 402          */
 403         private static function qualifyURLsSub($matches, $basepath)
 404         {
 405                 $base = parse_url($basepath);
 406                 unset($base['query']);
 407                 unset($base['fragment']);
 408
 409                 $link = $matches[0];
 410                 $url = $matches[1];
 411
 412                 $parts = array_merge($base, parse_url($url));
 413                 $url2 = Network::unparseURL($parts);
 414
 415                 return str_replace($url, $url2, $link);
 416         }
 417
 418         /**
 419          * @brief Complete incomplete URLs in BBCode
 420          *
 421          * @param string $body     Body with URLs
 422          * @param string $basepath Base path that is used to complete the URL
 423          *
 424          * @return string Body with expanded URLs
 425          */
 426         private static function qualifyURLs($body, $basepath)
 427         {
 428                 $URLSearchString = "^\[\]";
 429
 430                 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
 431                         "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
 432                         "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 433                         "/\[img\](.*?)\[\/img\]/ism",
 434                         "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 435                         "/\[zmg\](.*?)\[\/zmg\]/ism",
 436                         "/\[video\](.*?)\[\/video\]/ism",
 437                         "/\[audio\](.*?)\[\/audio\]/ism",
 438                 ];
 439
 440                 foreach ($matches as $match) {
 441                         $body = preg_replace_callback(
 442                                 $match,
 443                                 function ($match) use ($basepath) {
 444                                         return self::qualifyURLsSub($match, $basepath);
 445                                 },
 446                                 $body
 447                         );
 448                 }
 449                 return $body;
 450         }
 451
 452         private static function breakLines($line, $level, $wraplength = 75)
 453         {
 454                 if ($wraplength == 0) {
 455                         $wraplength = 2000000;
 456                 }
 457
 458                 $wraplen = $wraplength - $level;
 459
 460                 $newlines = [];
 461
 462                 do {
 463                         $oldline = $line;
 464
 465                         $subline = substr($line, 0, $wraplen);
 466
 467                         $pos = strrpos($subline, ' ');
 468
 469                         if ($pos == 0) {
 470                                 $pos = strpos($line, ' ');
 471                         }
 472
 473                         if (($pos > 0) && strlen($line) > $wraplen) {
 474                                 $newline = trim(substr($line, 0, $pos));
 475                                 if ($level > 0) {
 476                                         $newline = str_repeat(">", $level) . ' ' . $newline;
 477                                 }
 478
 479                                 $newlines[] = $newline . " ";
 480                                 $line = substr($line, $pos + 1);
 481                         }
 482                 } while ((strlen($line) > $wraplen) && !($oldline == $line));
 483
 484                 if ($level > 0) {
 485                         $line = str_repeat(">", $level) . ' ' . $line;
 486                 }
 487
 488                 $newlines[] = $line;
 489
 490                 return implode($newlines, "\n");
 491         }
 492
 493         private static function quoteLevel($message, $wraplength = 75)
 494         {
 495                 $lines = explode("\n", $message);
 496
 497                 $newlines = [];
 498                 $level = 0;
 499                 foreach ($lines as $line) {
 500                         $line = trim($line);
 501                         $startquote = false;
 502                         while (strpos("*" . $line, '[quote]') > 0) {
 503                                 $level++;
 504                                 $pos = strpos($line, '[quote]');
 505                                 $line = substr($line, 0, $pos) . substr($line, $pos + 7);
 506                                 $startquote = true;
 507                         }
 508
 509                         $currlevel = $level;
 510
 511                         while (strpos("*" . $line, '[/quote]') > 0) {
 512                                 $level--;
 513                                 if ($level < 0) {
 514                                         $level = 0;
 515                                 }
 516
 517                                 $pos = strpos($line, '[/quote]');
 518                                 $line = substr($line, 0, $pos) . substr($line, $pos + 8);
 519                         }
 520
 521                         if (!$startquote || ($line != '')) {
 522                                 $newlines[] = self::breakLines($line, $currlevel, $wraplength);
 523                         }
 524                 }
 525
 526                 return implode($newlines, "\n");
 527         }
 528
 529         private static function collectURLs($message)
 530         {
 531                 $pattern = '/<a.*?href="(.*?)".*?>(.*?)<\/a>/is';
 532                 preg_match_all($pattern, $message, $result, PREG_SET_ORDER);
 533
 534                 $urls = [];
 535                 foreach ($result as $treffer) {
 536                         $ignore = false;
 537
 538                         // A list of some links that should be ignored
 539                         $list = ["/user/", "/tag/", "/group/", "/profile/", "/search?search=", "/search?tag=", "mailto:", "/u/", "/node/",
 540                                 "//plus.google.com/", "//twitter.com/"];
 541                         foreach ($list as $listitem) {
 542                                 if (strpos($treffer[1], $listitem) !== false) {
 543                                         $ignore = true;
 544                                 }
 545                         }
 546
 547                         if ((strpos($treffer[1], "//twitter.com/") !== false) && (strpos($treffer[1], "/status/") !== false)) {
 548                                 $ignore = false;
 549                         }
 550
 551                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/posts") !== false)) {
 552                                 $ignore = false;
 553                         }
 554
 555                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/photos") !== false)) {
 556                                 $ignore = false;
 557                         }
 558
 559                         if (!$ignore) {
 560                                 $urls[$treffer[1]] = $treffer[1];
 561                         }
 562                 }
 563
 564                 return $urls;
 565         }
 566
 567         public static function toPlaintext($html, $wraplength = 75, $compact = false)
 568         {
 569                 $message = str_replace("\r", "", $html);
 570
 571                 $doc = new DOMDocument();
 572                 $doc->preserveWhiteSpace = false;
 573
 574                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 575
 576                 @$doc->loadHTML($message);
 577
 578                 $xpath = new DOMXPath($doc);
 579                 $list = $xpath->query("//pre");
 580                 foreach ($list as $node) {
 581                         $node->nodeValue = str_replace("\n", "\r", $node->nodeValue);
 582                 }
 583
 584                 $message = $doc->saveHTML();
 585                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br>", " ", ""], $message);
 586                 $message = preg_replace('= [\s]*=i', " ", $message);
 587
 588                 // Collecting all links
 589                 $urls = self::collectURLs($message);
 590
 591                 @$doc->loadHTML($message);
 592
 593                 self::tagToBBCode($doc, 'html', [], '', '');
 594                 self::tagToBBCode($doc, 'body', [], '', '');
 595
 596                 // MyBB-Auszeichnungen
 597                 /*
 598                   self::node2BBCode($doc, 'span', array('style'=>'text-decoration: underline;'), '_', '_');
 599                   self::node2BBCode($doc, 'span', array('style'=>'font-style: italic;'), '/', '/');
 600                   self::node2BBCode($doc, 'span', array('style'=>'font-weight: bold;'), '*', '*');
 601
 602                   self::node2BBCode($doc, 'strong', array(), '*', '*');
 603                   self::node2BBCode($doc, 'b', array(), '*', '*');
 604                   self::node2BBCode($doc, 'i', array(), '/', '/');
 605                   self::node2BBCode($doc, 'u', array(), '_', '_');
 606                  */
 607
 608                 if ($compact) {
 609                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 610                 } else {
 611                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 612                 }
 613
 614                 self::tagToBBCode($doc, 'br', [], "\n", '');
 615
 616                 self::tagToBBCode($doc, 'span', [], "", "");
 617                 self::tagToBBCode($doc, 'pre', [], "", "");
 618                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 619                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 620
 621                 //self::node2BBCode($doc, 'ul', array(), "\n[list]", "[/list]\n");
 622                 //self::node2BBCode($doc, 'ol', array(), "\n[list=1]", "[/list]\n");
 623                 self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 624
 625                 self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
 626
 627                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 628                 self::tagToBBCode($doc, 'td', [], "\t", "");
 629
 630                 self::tagToBBCode($doc, 'h1', [], "\n\n*", "*\n");
 631                 self::tagToBBCode($doc, 'h2', [], "\n\n*", "*\n");
 632                 self::tagToBBCode($doc, 'h3', [], "\n\n*", "*\n");
 633                 self::tagToBBCode($doc, 'h4', [], "\n\n*", "*\n");
 634                 self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
 635                 self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 636
 637                 // Problem: there is no reliable way to detect if it is a link to a tag or profile
 638                 //self::node2BBCode($doc, 'a', array('href'=>'/(.+)/'), ' $1 ', ' ', true);
 639                 //self::node2BBCode($doc, 'a', array('href'=>'/(.+)/', 'rel'=>'oembed'), ' $1 ', '', true);
 640                 //self::node2BBCode($doc, 'img', array('alt'=>'/(.+)/'), '$1', '');
 641                 //self::node2BBCode($doc, 'img', array('title'=>'/(.+)/'), '$1', '');
 642                 //self::node2BBCode($doc, 'img', array(), '', '');
 643                 if (!$compact) {
 644                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
 645                 } else {
 646                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' ', ' ');
 647                 }
 648
 649                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], ' $1 ', '');
 650
 651                 $message = $doc->saveHTML();
 652
 653                 if (!$compact) {
 654                         $message = str_replace("[img]", "", $message);
 655                         $message = str_replace("[/img]", "", $message);
 656                 }
 657
 658                 // was ersetze ich da?
 659                 // Irgendein stoerrisches UTF-Zeug
 660                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 661
 662                 $message = str_replace("&nbsp;", " ", $message);
 663
 664                 // Aufeinanderfolgende DIVs
 665                 $message = preg_replace('=\r *\r=i', "\n", $message);
 666                 $message = str_replace("\r", "\n", $message);
 667
 668                 $message = strip_tags($message);
 669
 670                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 671
 672                 if (!$compact && ($message != '')) {
 673                         foreach ($urls as $id => $url) {
 674                                 if ($url != '' && strpos($message, $url) === false) {
 675                                         $message .= "\n" . $url . ' ';
 676                                 }
 677                         }
 678                 }
 679
 680                 $message = str_replace("\n«", "«\n", $message);
 681                 $message = str_replace("»\n", "\n»", $message);
 682
 683                 do {
 684                         $oldmessage = $message;
 685                         $message = str_replace("\n\n\n", "\n\n", $message);
 686                 } while ($oldmessage != $message);
 687
 688                 $message = self::quoteLevel(trim($message), $wraplength);
 689
 690                 return trim($message);
 691         }
 692
 693         /**
 694          * Converts provided HTML code to Markdown. The hardwrap parameter maximizes
 695          * compatibility with Diaspora in spite of the Markdown standards.
 696          *
 697          * @param string $html
 698          * @return string
 699          */
 700         public static function toMarkdown($html)
 701         {
 702                 $converter = new HtmlConverter(['hard_break' => true]);
 703                 $markdown = $converter->convert($html);
 704
 705                 return $markdown;
 706         }
 707
 708         /**
 709          * @brief Convert video HTML to BBCode tags
 710          *
 711          * @param string $s
 712          */
 713         public static function htmlToBBVideo($s)
 714         {
 715                 $s = preg_replace(
 716                         '#<object[^>]+>(.*?)https?://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+)(.*?)</object>#ism',
 717                         '[youtube]$2[/youtube]',
 718                         $s
 719                 );
 720
 721                 $s = preg_replace(
 722                         '#<iframe[^>](.*?)https?://www.youtube.com/embed/([A-Za-z0-9\-_=]+)(.*?)</iframe>#ism',
 723                         '[youtube]$2[/youtube]',
 724                         $s
 725                 );
 726
 727                 $s = preg_replace(
 728                         '#<iframe[^>](.*?)https?://player.vimeo.com/video/([0-9]+)(.*?)</iframe>#ism',
 729                         '[vimeo]$2[/vimeo]',
 730                         $s
 731                 );
 732
 733                 return $s;
 734         }
 735
 736         /**
 737          * transform link href and img src from relative to absolute
 738          *
 739          * @param string $text
 740          * @param string $base base url
 741          * @return string
 742          */
 743         public static function relToAbs($text, $base)
 744         {
 745                 if (empty($base)) {
 746                         return $text;
 747                 }
 748
 749                 $base = rtrim($base, '/');
 750
 751                 $base2 = $base . "/";
 752
 753                 // Replace links
 754                 $pattern = "/<a([^>]*) href=\"(?!http|https|\/)([^\"]*)\"/";
 755                 $replace = "<a\${1} href=\"" . $base2 . "\${2}\"";
 756                 $text = preg_replace($pattern, $replace, $text);
 757
 758                 $pattern = "/<a([^>]*) href=\"(?!http|https)([^\"]*)\"/";
 759                 $replace = "<a\${1} href=\"" . $base . "\${2}\"";
 760                 $text = preg_replace($pattern, $replace, $text);
 761
 762                 // Replace images
 763                 $pattern = "/<img([^>]*) src=\"(?!http|https|\/)([^\"]*)\"/";
 764                 $replace = "<img\${1} src=\"" . $base2 . "\${2}\"";
 765                 $text = preg_replace($pattern, $replace, $text);
 766
 767                 $pattern = "/<img([^>]*) src=\"(?!http|https)([^\"]*)\"/";
 768                 $replace = "<img\${1} src=\"" . $base . "\${2}\"";
 769                 $text = preg_replace($pattern, $replace, $text);
 770
 771
 772                 // Done
 773                 return $text;
 774         }
 775
 776         /**
 777          * return div element with class 'clear'
 778          * @return string
 779          * @deprecated
 780          */
 781         public static function clearDiv()
 782         {
 783                 return '<div class="clear"></div>';
 784         }
 785
 786         /**
 787          * Loader for infinite scrolling
 788          * @return string html for loader
 789          */
 790         public static function scrollLoader()
 791         {
 792                 $tpl = Renderer::getMarkupTemplate("scroll_loader.tpl");
 793                 return Renderer::replaceMacros($tpl, [
 794                         'wait' => L10n::t('Loading more entries...'),
 795                         'end' => L10n::t('The end')
 796                 ]);
 797         }
 798
 799         /**
 800          * Get html for contact block.
 801          *
 802          * @template contact_block.tpl
 803          * @hook contact_block_end (contacts=>array, output=>string)
 804          * @return string
 805          */
 806         public static function contactBlock()
 807         {
 808                 $o = '';
 809                 $a = get_app();
 810
 811                 $shown = PConfig::get($a->profile['uid'], 'system', 'display_friend_count', 24);
 812                 if ($shown == 0) {
 813                         return;
 814                 }
 815
 816                 if (!is_array($a->profile) || $a->profile['hide-friends']) {
 817                         return $o;
 818                 }
 819
 820                 $r = q("SELECT COUNT(*) AS `total` FROM `contact`
 821                                 WHERE `uid` = %d AND NOT `self` AND NOT `blocked`
 822                                         AND NOT `pending` AND NOT `hidden` AND NOT `archive`
 823                                         AND `network` IN ('%s', '%s', '%s')",
 824                         intval($a->profile['uid']),
 825                         DBA::escape(Protocol::DFRN),
 826                         DBA::escape(Protocol::OSTATUS),
 827                         DBA::escape(Protocol::DIASPORA)
 828                 );
 829
 830                 if (DBA::isResult($r)) {
 831                         $total = intval($r[0]['total']);
 832                 }
 833
 834                 if (!$total) {
 835                         $contacts = L10n::t('No contacts');
 836                         $micropro = null;
 837                 } else {
 838                         // Splitting the query in two parts makes it much faster
 839                         $r = q("SELECT `id` FROM `contact`
 840                                         WHERE `uid` = %d AND NOT `self` AND NOT `blocked`
 841                                                 AND NOT `pending` AND NOT `hidden` AND NOT `archive`
 842                                                 AND `network` IN ('%s', '%s', '%s')
 843                                         ORDER BY RAND() LIMIT %d",
 844                                 intval($a->profile['uid']),
 845                                 DBA::escape(Protocol::DFRN),
 846                                 DBA::escape(Protocol::OSTATUS),
 847                                 DBA::escape(Protocol::DIASPORA),
 848                                 intval($shown)
 849                         );
 850
 851                         if (DBA::isResult($r)) {
 852                                 $contacts = [];
 853                                 foreach ($r as $contact) {
 854                                         $contacts[] = $contact["id"];
 855                                 }
 856
 857                                 $r = q("SELECT `id`, `uid`, `addr`, `url`, `name`, `thumb`, `network` FROM `contact` WHERE `id` IN (%s)",
 858                                         DBA::escape(implode(",", $contacts))
 859                                 );
 860
 861                                 if (DBA::isResult($r)) {
 862                                         $contacts = L10n::tt('%d Contact', '%d Contacts', $total);
 863                                         $micropro = [];
 864                                         foreach ($r as $rr) {
 865                                                 $micropro[] = micropro($rr, true, 'mpfriend');
 866                                         }
 867                                 }
 868                         }
 869                 }
 870
 871                 $tpl = Renderer::getMarkupTemplate('contact_block.tpl');
 872                 $o = Renderer::replaceMacros($tpl, [
 873                         '$contacts' => $contacts,
 874                         '$nickname' => $a->profile['nickname'],
 875                         '$viewcontacts' => L10n::t('View Contacts'),
 876                         '$micropro' => $micropro,
 877                 ]);
 878
 879                 $arr = ['contacts' => $r, 'output' => $o];
 880
 881                 Addon::callHooks('contact_block_end', $arr);
 882
 883                 return $o;
 884         }
 885
 886         /**
 887          * @brief Format contacts as picture links or as texxt links
 888          *
 889          * @param array $contact Array with contacts which contains an array with
 890          *      int 'id' => The ID of the contact
 891         *       int 'uid' => The user ID of the user who owns this data
 892         *       string 'name' => The name of the contact
 893         *       string 'url' => The url to the profile page of the contact
 894         *       string 'addr' => The webbie of the contact (e.g.) username@friendica.com
 895         *       string 'network' => The network to which the contact belongs to
 896         *       string 'thumb' => The contact picture
 897         *       string 'click' => js code which is performed when clicking on the contact
 898         * @param boolean $redirect If true try to use the redir url if it's possible
 899         * @param string $class CSS class for the
 900         * @param boolean $textmode If true display the contacts as text links
 901         *       if false display the contacts as picture links
 902
 903         * @return string Formatted html
 904         */
 905         public static function micropro($contact, $redirect = false, $class = '', $textmode = false)
 906         {
 907                 // Use the contact URL if no address is available
 908                 if (!x($contact, "addr")) {
 909                         $contact["addr"] = $contact["url"];
 910                 }
 911
 912                 $url = $contact['url'];
 913                 $sparkle = '';
 914                 $redir = false;
 915
 916                 if ($redirect) {
 917                         $url = Contact::magicLink($contact['url']);
 918                         if (strpos($url, 'redir/') === 0) {
 919                                 $sparkle = ' sparkle';
 920                         }
 921                 }
 922
 923                 // If there is some js available we don't need the url
 924                 if (x($contact, 'click')) {
 925                         $url = '';
 926                 }
 927
 928                 return Renderer::replaceMacros(Renderer::getMarkupTemplate(($textmode)?'micropro_txt.tpl':'micropro_img.tpl'), [
 929                         '$click' => defaults($contact, 'click', ''),
 930                         '$class' => $class,
 931                         '$url' => $url,
 932                         '$photo' => ProxyUtils::proxifyUrl($contact['thumb'], false, ProxyUtils::SIZE_THUMB),
 933                         '$name' => $contact['name'],
 934                         'title' => $contact['name'] . ' [' . $contact['addr'] . ']',
 935                         '$parkle' => $sparkle,
 936                         '$redir' => $redir
 937                 ]);
 938         }
 939
 940         /**
 941          * Search box.
 942          *
 943          * @param string $s     Search query.
 944          * @param string $id    HTML id
 945          * @param string $url   Search url.
 946          * @param bool   $save  Show save search button.
 947          * @param bool   $aside Display the search widgit aside.
 948          *
 949          * @return string Formatted HTML.
 950          */
 951         public static function search($s, $id = 'search-box', $url = 'search', $save = false, $aside = true)
 952         {
 953                 $mode = 'text';
 954
 955                 if (strpos($s, '#') === 0) {
 956                         $mode = 'tag';
 957                 }
 958                 $save_label = $mode === 'text' ? L10n::t('Save') : L10n::t('Follow');
 959
 960                 $values = [
 961                                 '$s' => htmlspecialchars($s),
 962                                 '$id' => $id,
 963                                 '$action_url' => $url,
 964                                 '$search_label' => L10n::t('Search'),
 965                                 '$save_label' => $save_label,
 966                                 '$savedsearch' => local_user() && Feature::isEnabled(local_user(), 'savedsearch'),
 967                                 '$search_hint' => L10n::t('@name, !forum, #tags, content'),
 968                                 '$mode' => $mode
 969                         ];
 970
 971                 if (!$aside) {
 972                         $values['$searchoption'] = [
 973                                                 L10n::t("Full Text"),
 974                                                 L10n::t("Tags"),
 975                                                 L10n::t("Contacts")];
 976
 977                         if (Config::get('system', 'poco_local_search')) {
 978                                 $values['$searchoption'][] = L10n::t("Forums");
 979                         }
 980                 }
 981
 982                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 983         }
 984
 985         /**
 986          * Replace naked text hyperlink with HTML formatted hyperlink
 987          *
 988          * @param string $s
 989          */
 990         public static function toLink($s)
 991         {
 992                 $s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank">$1</a>', $s);
 993                 $s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
 994                 return $s;
 995         }
 996
 997         /**
 998          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
 999          *
1000          * Reasons are expected to have been translated already.
1001          *
1002          * @param string $html
1003          * @param array  $reasons
1004          * @return string
1005          */
1006         public static function applyContentFilter($html, array $reasons)
1007         {
1008                 if (count($reasons)) {
1009                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
1010                         $html = Renderer::replaceMacros($tpl, [
1011                                 '$reasons'   => $reasons,
1012                                 '$rnd'       => random_string(8),
1013                                 '$openclose' => L10n::t('Click to open/close'),
1014                                 '$html'      => $html
1015                         ]);
1016                 }
1017
1018                 return $html;
1019         }
1020
1021         /**
1022          * replace html amp entity with amp char
1023          * @param string $s
1024          * @return string
1025          */
1026         public static function unamp($s)
1027         {
1028                 return str_replace('&amp;', '&', $s);
1029         }
1030 }