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\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\XML;
  21 use League\HTMLToMarkdown\HtmlConverter;
  22
  23 class HTML
  24 {
  25         public static function sanitizeCSS($input)
  26         {
  27                 $cleaned = "";
  28
  29                 $input = strtolower($input);
  30
  31                 for ($i = 0; $i < strlen($input); $i++) {
  32                         $char = substr($input, $i, 1);
  33
  34                         if (($char >= "a") && ($char <= "z")) {
  35                                 $cleaned .= $char;
  36                         }
  37
  38                         if (!(strpos(" #;:0123456789-_.%", $char) === false)) {
  39                                 $cleaned .= $char;
  40                         }
  41                 }
  42
  43                 return $cleaned;
  44         }
  45
  46         private static function tagToBBCode(DOMDocument $doc, $tag, $attributes, $startbb, $endbb)
  47         {
  48                 do {
  49                         $done = self::tagToBBCodeSub($doc, $tag, $attributes, $startbb, $endbb);
  50                 } while ($done);
  51         }
  52
  53         private static function tagToBBCodeSub(DOMDocument $doc, $tag, $attributes, $startbb, $endbb)
  54         {
  55                 $savestart = str_replace('$', '\x01', $startbb);
  56                 $replace = false;
  57
  58                 $xpath = new DOMXPath($doc);
  59
  60                 $list = $xpath->query("//" . $tag);
  61                 foreach ($list as $node) {
  62                         $attr = [];
  63                         if ($node->attributes->length) {
  64                                 foreach ($node->attributes as $attribute) {
  65                                         $attr[$attribute->name] = $attribute->value;
  66                                 }
  67                         }
  68
  69                         $replace = true;
  70
  71                         $startbb = $savestart;
  72
  73                         $i = 0;
  74
  75                         foreach ($attributes as $attribute => $value) {
  76                                 $startbb = str_replace('\x01' . ++$i, '$1', $startbb);
  77                                 if (strpos('*' . $startbb, '$1') > 0) {
  78                                         if ($replace && (@$attr[$attribute] != '')) {
  79                                                 $startbb = preg_replace($value, $startbb, $attr[$attribute], -1, $count);
  80
  81                                                 // If nothing could be changed
  82                                                 if ($count == 0) {
  83                                                         $replace = false;
  84                                                 }
  85                                         } else {
  86                                                 $replace = false;
  87                                         }
  88                                 } else {
  89                                         if (@$attr[$attribute] != $value) {
  90                                                 $replace = false;
  91                                         }
  92                                 }
  93                         }
  94
  95                         if ($replace) {
  96                                 $StartCode = $doc->createTextNode($startbb);
  97                                 $EndCode = $doc->createTextNode($endbb);
  98
  99                                 $node->parentNode->insertBefore($StartCode, $node);
 100
 101                                 if ($node->hasChildNodes()) {
 102                                         foreach ($node->childNodes as $child) {
 103                                                 $newNode = $child->cloneNode(true);
 104                                                 $node->parentNode->insertBefore($newNode, $node);
 105                                         }
 106                                 }
 107
 108                                 $node->parentNode->insertBefore($EndCode, $node);
 109                                 $node->parentNode->removeChild($node);
 110                         }
 111                 }
 112
 113                 return $replace;
 114         }
 115
 116         /**
 117          * Made by: ike@piratenpartei.de
 118          * Originally made for the syncom project: http://wiki.piratenpartei.de/Syncom
 119          *                                      https://github.com/annando/Syncom
 120          *
 121          * @brief Converter for HTML to BBCode
 122          * @param string $message
 123          * @param string $basepath
 124          * @return string
 125          */
 126         public static function toBBCode($message, $basepath = '')
 127         {
 128                 $message = str_replace("\r", "", $message);
 129
 130                 // Removing code blocks before the whitespace removal processing below
 131                 $codeblocks = [];
 132                 $message = preg_replace_callback(
 133                         '#<pre><code(?: class="language-([^"]*)")?>(.*)</code></pre>#iUs',
 134                         function ($matches) use (&$codeblocks) {
 135                                 $return = '[codeblock-' . count($codeblocks) . ']';
 136
 137                                 $prefix = '[code]';
 138                                 if ($matches[1] != '') {
 139                                         $prefix = '[code=' . $matches[1] . ']';
 140                                 }
 141
 142                                 $codeblocks[] = $prefix . PHP_EOL . trim($matches[2]) . PHP_EOL . '[/code]';
 143                                 return $return;
 144                         },
 145                         $message
 146                 );
 147
 148                 $message = str_replace(
 149                         [
 150                                 "<li><p>",
 151                                 "</p></li>",
 152                         ],
 153                         [
 154                                 "<li>",
 155                                 "</li>",
 156                         ],
 157                         $message
 158                 );
 159
 160                 // remove namespaces
 161                 $message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
 162                 $message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
 163
 164                 $doc = new DOMDocument();
 165                 $doc->preserveWhiteSpace = false;
 166
 167                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 168
 169                 @$doc->loadHTML($message);
 170
 171                 XML::deleteNode($doc, 'style');
 172                 XML::deleteNode($doc, 'head');
 173                 XML::deleteNode($doc, 'title');
 174                 XML::deleteNode($doc, 'meta');
 175                 XML::deleteNode($doc, 'xml');
 176                 XML::deleteNode($doc, 'removeme');
 177
 178                 $xpath = new DomXPath($doc);
 179                 $list = $xpath->query("//pre");
 180                 foreach ($list as $node) {
 181                         $node->nodeValue = str_replace("\n", "\r", $node->nodeValue);
 182                 }
 183
 184                 $message = $doc->saveHTML();
 185                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br />", " ", ""], $message);
 186                 $message = preg_replace('= [\s]*=i', " ", $message);
 187                 @$doc->loadHTML($message);
 188
 189                 self::tagToBBCode($doc, 'html', [], "", "");
 190                 self::tagToBBCode($doc, 'body', [], "", "");
 191
 192                 // Outlook-Quote - Variant 1
 193                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal', 'style' => 'margin-left:35.4pt'], '[quote]', '[/quote]');
 194
 195                 // Outlook-Quote - Variant 2
 196                 self::tagToBBCode(
 197                         $doc,
 198                         'div',
 199                         ['style' => 'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'],
 200                         '[quote]',
 201                         '[/quote]'
 202                 );
 203
 204                 // MyBB-Stuff
 205                 self::tagToBBCode($doc, 'span', ['style' => 'text-decoration: underline;'], '[u]', '[/u]');
 206                 self::tagToBBCode($doc, 'span', ['style' => 'font-style: italic;'], '[i]', '[/i]');
 207                 self::tagToBBCode($doc, 'span', ['style' => 'font-weight: bold;'], '[b]', '[/b]');
 208
 209                 /* self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
 210                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
 211                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
 212                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
 213                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
 214                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
 215                   self::node2BBCode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
 216                  */
 217                 // Untested
 218                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
 219                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
 220                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
 221
 222                 self::tagToBBCode($doc, 'span', ['style' => '/.*color:\s*(.+?)[,;].*/'], '[color="$1"]', '[/color]');
 223
 224                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 225                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
 226                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
 227                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 228                 // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
 229                 // Test
 230                 //self::node2BBCode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
 231                 self::tagToBBCode($doc, 'span', ['class' => 'type-link'], '[class=type-link]', '[/class]');
 232                 self::tagToBBCode($doc, 'span', ['class' => 'type-video'], '[class=type-video]', '[/class]');
 233
 234                 self::tagToBBCode($doc, 'strong', [], '[b]', '[/b]');
 235                 self::tagToBBCode($doc, 'em', [], '[i]', '[/i]');
 236                 self::tagToBBCode($doc, 'b', [], '[b]', '[/b]');
 237                 self::tagToBBCode($doc, 'i', [], '[i]', '[/i]');
 238                 self::tagToBBCode($doc, 'u', [], '[u]', '[/u]');
 239                 self::tagToBBCode($doc, 's', [], '[s]', '[/s]');
 240                 self::tagToBBCode($doc, 'del', [], '[s]', '[/s]');
 241                 self::tagToBBCode($doc, 'strike', [], '[s]', '[/s]');
 242
 243                 self::tagToBBCode($doc, 'big', [], "[size=large]", "[/size]");
 244                 self::tagToBBCode($doc, 'small', [], "[size=small]", "[/size]");
 245
 246                 self::tagToBBCode($doc, 'blockquote', [], '[quote]', '[/quote]');
 247
 248                 self::tagToBBCode($doc, 'br', [], "\n", '');
 249
 250                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal'], "\n", "");
 251                 self::tagToBBCode($doc, 'div', ['class' => 'MsoNormal'], "\r", "");
 252
 253                 self::tagToBBCode($doc, 'span', [], "", "");
 254
 255                 self::tagToBBCode($doc, 'span', [], "", "");
 256                 self::tagToBBCode($doc, 'pre', [], "", "");
 257
 258                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 259                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 260
 261                 self::tagToBBCode($doc, 'ul', [], "[list]", "[/list]");
 262                 self::tagToBBCode($doc, 'ol', [], "[list=1]", "[/list]");
 263                 self::tagToBBCode($doc, 'li', [], "[*]", "");
 264
 265                 self::tagToBBCode($doc, 'hr', [], "[hr]", "");
 266
 267                 self::tagToBBCode($doc, 'table', [], "", "");
 268                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 269                 self::tagToBBCode($doc, 'td', [], "\t", "");
 270                 //self::node2BBCode($doc, 'table', array(), "[table]", "[/table]");
 271                 //self::node2BBCode($doc, 'th', array(), "[th]", "[/th]");
 272                 //self::node2BBCode($doc, 'tr', array(), "[tr]", "[/tr]");
 273                 //self::node2BBCode($doc, 'td', array(), "[td]", "[/td]");
 274                 //self::node2BBCode($doc, 'h1', array(), "\n\n[size=xx-large][b]", "[/b][/size]\n");
 275                 //self::node2BBCode($doc, 'h2', array(), "\n\n[size=x-large][b]", "[/b][/size]\n");
 276                 //self::node2BBCode($doc, 'h3', array(), "\n\n[size=large][b]", "[/b][/size]\n");
 277                 //self::node2BBCode($doc, 'h4', array(), "\n\n[size=medium][b]", "[/b][/size]\n");
 278                 //self::node2BBCode($doc, 'h5', array(), "\n\n[size=small][b]", "[/b][/size]\n");
 279                 //self::node2BBCode($doc, 'h6', array(), "\n\n[size=x-small][b]", "[/b][/size]\n");
 280
 281                 self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 282                 self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 283                 self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 284                 self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 285                 self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 286                 self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 287
 288                 self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 289                 self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 290
 291                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]');
 292                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]');
 293
 294
 295                 self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]');
 296                 self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]');
 297                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[iframe]$1', '[/iframe]');
 298
 299                 self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 300                 self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 301
 302                 $message = $doc->saveHTML();
 303
 304                 // I'm removing something really disturbing
 305                 // Don't know exactly what it is
 306                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 307
 308                 $message = str_replace("&nbsp;", " ", $message);
 309
 310                 // removing multiple DIVs
 311                 $message = preg_replace('=\r *\r=i', "\n", $message);
 312                 $message = str_replace("\r", "\n", $message);
 313
 314                 Addon::callHooks('html2bbcode', $message);
 315
 316                 $message = strip_tags($message);
 317
 318                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 319
 320                 $message = str_replace(["<"], ["&lt;"], $message);
 321
 322                 // remove quotes if they don't make sense
 323                 $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 324
 325                 $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 326                 $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 327
 328                 do {
 329                         $oldmessage = $message;
 330                         $message = str_replace("\n \n", "\n\n", $message);
 331                 } while ($oldmessage != $message);
 332
 333                 do {
 334                         $oldmessage = $message;
 335                         $message = str_replace("\n\n\n", "\n\n", $message);
 336                 } while ($oldmessage != $message);
 337
 338                 do {
 339                         $oldmessage = $message;
 340                         $message = str_replace(
 341                                 [
 342                                 "[/size]\n\n",
 343                                 "\n[hr]",
 344                                 "[hr]\n",
 345                                 "\n[list",
 346                                 "[/list]\n",
 347                                 "\n[/",
 348                                 "[list]\n",
 349                                 "[list=1]\n",
 350                                 "\n[*]"],
 351                                 [
 352                                 "[/size]\n",
 353                                 "[hr]",
 354                                 "[hr]",
 355                                 "[list",
 356                                 "[/list]",
 357                                 "[/",
 358                                 "[list]",
 359                                 "[list=1]",
 360                                 "[*]"],
 361                                 $message
 362                         );
 363                 } while ($message != $oldmessage);
 364
 365                 $message = str_replace(
 366                         ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 367                         ['[b]', '[/b]', '[i]', '[/i]'],
 368                         $message
 369                 );
 370
 371                 // Handling Yahoo style of mails
 372                 $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 373
 374                 // Restore code blocks
 375                 $message = preg_replace_callback(
 376                         '#\[codeblock-([0-9]+)\]#iU',
 377                         function ($matches) use ($codeblocks) {
 378                                 $return = '';
 379                                 if (isset($codeblocks[intval($matches[1])])) {
 380                                         $return = $codeblocks[$matches[1]];
 381                                 }
 382                                 return $return;
 383                         },
 384                         $message
 385                 );
 386
 387                 $message = trim($message);
 388
 389                 if ($basepath != '') {
 390                         $message = self::qualifyURLs($message, $basepath);
 391                 }
 392
 393                 return $message;
 394         }
 395
 396         /**
 397          * @brief Sub function to complete incomplete URL
 398          *
 399          * @param array  $matches  Result of preg_replace_callback
 400          * @param string $basepath Basepath that is used to complete the URL
 401          *
 402          * @return string The expanded URL
 403          */
 404         private static function qualifyURLsSub($matches, $basepath)
 405         {
 406                 $base = parse_url($basepath);
 407                 unset($base['query']);
 408                 unset($base['fragment']);
 409
 410                 $link = $matches[0];
 411                 $url = $matches[1];
 412
 413                 $parts = array_merge($base, parse_url($url));
 414                 $url2 = Network::unparseURL($parts);
 415
 416                 return str_replace($url, $url2, $link);
 417         }
 418
 419         /**
 420          * @brief Complete incomplete URLs in BBCode
 421          *
 422          * @param string $body     Body with URLs
 423          * @param string $basepath Base path that is used to complete the URL
 424          *
 425          * @return string Body with expanded URLs
 426          */
 427         private static function qualifyURLs($body, $basepath)
 428         {
 429                 $URLSearchString = "^\[\]";
 430
 431                 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
 432                         "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
 433                         "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 434                         "/\[img\](.*?)\[\/img\]/ism",
 435                         "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 436                         "/\[zmg\](.*?)\[\/zmg\]/ism",
 437                         "/\[video\](.*?)\[\/video\]/ism",
 438                         "/\[audio\](.*?)\[\/audio\]/ism",
 439                 ];
 440
 441                 foreach ($matches as $match) {
 442                         $body = preg_replace_callback(
 443                                 $match,
 444                                 function ($match) use ($basepath) {
 445                                         return self::qualifyURLsSub($match, $basepath);
 446                                 },
 447                                 $body
 448                         );
 449                 }
 450                 return $body;
 451         }
 452
 453         private static function breakLines($line, $level, $wraplength = 75)
 454         {
 455                 if ($wraplength == 0) {
 456                         $wraplength = 2000000;
 457                 }
 458
 459                 $wraplen = $wraplength - $level;
 460
 461                 $newlines = [];
 462
 463                 do {
 464                         $oldline = $line;
 465
 466                         $subline = substr($line, 0, $wraplen);
 467
 468                         $pos = strrpos($subline, ' ');
 469
 470                         if ($pos == 0) {
 471                                 $pos = strpos($line, ' ');
 472                         }
 473
 474                         if (($pos > 0) && strlen($line) > $wraplen) {
 475                                 $newline = trim(substr($line, 0, $pos));
 476                                 if ($level > 0) {
 477                                         $newline = str_repeat(">", $level) . ' ' . $newline;
 478                                 }
 479
 480                                 $newlines[] = $newline . " ";
 481                                 $line = substr($line, $pos + 1);
 482                         }
 483                 } while ((strlen($line) > $wraplen) && !($oldline == $line));
 484
 485                 if ($level > 0) {
 486                         $line = str_repeat(">", $level) . ' ' . $line;
 487                 }
 488
 489                 $newlines[] = $line;
 490
 491                 return implode($newlines, "\n");
 492         }
 493
 494         private static function quoteLevel($message, $wraplength = 75)
 495         {
 496                 $lines = explode("\n", $message);
 497
 498                 $newlines = [];
 499                 $level = 0;
 500                 foreach ($lines as $line) {
 501                         $line = trim($line);
 502                         $startquote = false;
 503                         while (strpos("*" . $line, '[quote]') > 0) {
 504                                 $level++;
 505                                 $pos = strpos($line, '[quote]');
 506                                 $line = substr($line, 0, $pos) . substr($line, $pos + 7);
 507                                 $startquote = true;
 508                         }
 509
 510                         $currlevel = $level;
 511
 512                         while (strpos("*" . $line, '[/quote]') > 0) {
 513                                 $level--;
 514                                 if ($level < 0) {
 515                                         $level = 0;
 516                                 }
 517
 518                                 $pos = strpos($line, '[/quote]');
 519                                 $line = substr($line, 0, $pos) . substr($line, $pos + 8);
 520                         }
 521
 522                         if (!$startquote || ($line != '')) {
 523                                 $newlines[] = self::breakLines($line, $currlevel, $wraplength);
 524                         }
 525                 }
 526
 527                 return implode($newlines, "\n");
 528         }
 529
 530         private static function collectURLs($message)
 531         {
 532                 $pattern = '/<a.*?href="(.*?)".*?>(.*?)<\/a>/is';
 533                 preg_match_all($pattern, $message, $result, PREG_SET_ORDER);
 534
 535                 $urls = [];
 536                 foreach ($result as $treffer) {
 537                         $ignore = false;
 538
 539                         // A list of some links that should be ignored
 540                         $list = ["/user/", "/tag/", "/group/", "/profile/", "/search?search=", "/search?tag=", "mailto:", "/u/", "/node/",
 541                                 "//plus.google.com/", "//twitter.com/"];
 542                         foreach ($list as $listitem) {
 543                                 if (strpos($treffer[1], $listitem) !== false) {
 544                                         $ignore = true;
 545                                 }
 546                         }
 547
 548                         if ((strpos($treffer[1], "//twitter.com/") !== false) && (strpos($treffer[1], "/status/") !== false)) {
 549                                 $ignore = false;
 550                         }
 551
 552                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/posts") !== false)) {
 553                                 $ignore = false;
 554                         }
 555
 556                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/photos") !== false)) {
 557                                 $ignore = false;
 558                         }
 559
 560                         if (!$ignore) {
 561                                 $urls[$treffer[1]] = $treffer[1];
 562                         }
 563                 }
 564
 565                 return $urls;
 566         }
 567
 568         public static function toPlaintext($html, $wraplength = 75, $compact = false)
 569         {
 570                 $message = str_replace("\r", "", $html);
 571
 572                 $doc = new DOMDocument();
 573                 $doc->preserveWhiteSpace = false;
 574
 575                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 576
 577                 @$doc->loadHTML($message);
 578
 579                 $xpath = new DOMXPath($doc);
 580                 $list = $xpath->query("//pre");
 581                 foreach ($list as $node) {
 582                         $node->nodeValue = str_replace("\n", "\r", $node->nodeValue);
 583                 }
 584
 585                 $message = $doc->saveHTML();
 586                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br>", " ", ""], $message);
 587                 $message = preg_replace('= [\s]*=i', " ", $message);
 588
 589                 // Collecting all links
 590                 $urls = self::collectURLs($message);
 591
 592                 @$doc->loadHTML($message);
 593
 594                 self::tagToBBCode($doc, 'html', [], '', '');
 595                 self::tagToBBCode($doc, 'body', [], '', '');
 596
 597                 // MyBB-Auszeichnungen
 598                 /*
 599                   self::node2BBCode($doc, 'span', array('style'=>'text-decoration: underline;'), '_', '_');
 600                   self::node2BBCode($doc, 'span', array('style'=>'font-style: italic;'), '/', '/');
 601                   self::node2BBCode($doc, 'span', array('style'=>'font-weight: bold;'), '*', '*');
 602
 603                   self::node2BBCode($doc, 'strong', array(), '*', '*');
 604                   self::node2BBCode($doc, 'b', array(), '*', '*');
 605                   self::node2BBCode($doc, 'i', array(), '/', '/');
 606                   self::node2BBCode($doc, 'u', array(), '_', '_');
 607                  */
 608
 609                 if ($compact) {
 610                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 611                 } else {
 612                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 613                 }
 614
 615                 self::tagToBBCode($doc, 'br', [], "\n", '');
 616
 617                 self::tagToBBCode($doc, 'span', [], "", "");
 618                 self::tagToBBCode($doc, 'pre', [], "", "");
 619                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 620                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 621
 622                 //self::node2BBCode($doc, 'ul', array(), "\n[list]", "[/list]\n");
 623                 //self::node2BBCode($doc, 'ol', array(), "\n[list=1]", "[/list]\n");
 624                 self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 625
 626                 self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
 627
 628                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 629                 self::tagToBBCode($doc, 'td', [], "\t", "");
 630
 631                 self::tagToBBCode($doc, 'h1', [], "\n\n*", "*\n");
 632                 self::tagToBBCode($doc, 'h2', [], "\n\n*", "*\n");
 633                 self::tagToBBCode($doc, 'h3', [], "\n\n*", "*\n");
 634                 self::tagToBBCode($doc, 'h4', [], "\n\n*", "*\n");
 635                 self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
 636                 self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 637
 638                 // Problem: there is no reliable way to detect if it is a link to a tag or profile
 639                 //self::node2BBCode($doc, 'a', array('href'=>'/(.+)/'), ' $1 ', ' ', true);
 640                 //self::node2BBCode($doc, 'a', array('href'=>'/(.+)/', 'rel'=>'oembed'), ' $1 ', '', true);
 641                 //self::node2BBCode($doc, 'img', array('alt'=>'/(.+)/'), '$1', '');
 642                 //self::node2BBCode($doc, 'img', array('title'=>'/(.+)/'), '$1', '');
 643                 //self::node2BBCode($doc, 'img', array(), '', '');
 644                 if (!$compact) {
 645                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
 646                 } else {
 647                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' ', ' ');
 648                 }
 649
 650                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], ' $1 ', '');
 651
 652                 $message = $doc->saveHTML();
 653
 654                 if (!$compact) {
 655                         $message = str_replace("[img]", "", $message);
 656                         $message = str_replace("[/img]", "", $message);
 657                 }
 658
 659                 // was ersetze ich da?
 660                 // Irgendein stoerrisches UTF-Zeug
 661                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 662
 663                 $message = str_replace("&nbsp;", " ", $message);
 664
 665                 // Aufeinanderfolgende DIVs
 666                 $message = preg_replace('=\r *\r=i', "\n", $message);
 667                 $message = str_replace("\r", "\n", $message);
 668
 669                 $message = strip_tags($message);
 670
 671                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 672
 673                 if (!$compact && ($message != '')) {
 674                         foreach ($urls as $id => $url) {
 675                                 if ($url != '' && strpos($message, $url) === false) {
 676                                         $message .= "\n" . $url . ' ';
 677                                 }
 678                         }
 679                 }
 680
 681                 $message = str_replace("\n«", "«\n", $message);
 682                 $message = str_replace("»\n", "\n»", $message);
 683
 684                 do {
 685                         $oldmessage = $message;
 686                         $message = str_replace("\n\n\n", "\n\n", $message);
 687                 } while ($oldmessage != $message);
 688
 689                 $message = self::quoteLevel(trim($message), $wraplength);
 690
 691                 return trim($message);
 692         }
 693
 694         /**
 695          * Converts provided HTML code to Markdown. The hardwrap parameter maximizes
 696          * compatibility with Diaspora in spite of the Markdown standards.
 697          *
 698          * @param string $html
 699          * @return string
 700          */
 701         public static function toMarkdown($html)
 702         {
 703                 $converter = new HtmlConverter(['hard_break' => true]);
 704                 $markdown = $converter->convert($html);
 705
 706                 return $markdown;
 707         }
 708
 709         /**
 710          * @brief Convert video HTML to BBCode tags
 711          *
 712          * @param string $s
 713          */
 714         public static function htmlToBBVideo($s)
 715         {
 716                 $s = preg_replace(
 717                         '#<object[^>]+>(.*?)https?://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+)(.*?)</object>#ism',
 718                         '[youtube]$2[/youtube]',
 719                         $s
 720                 );
 721
 722                 $s = preg_replace(
 723                         '#<iframe[^>](.*?)https?://www.youtube.com/embed/([A-Za-z0-9\-_=]+)(.*?)</iframe>#ism',
 724                         '[youtube]$2[/youtube]',
 725                         $s
 726                 );
 727
 728                 $s = preg_replace(
 729                         '#<iframe[^>](.*?)https?://player.vimeo.com/video/([0-9]+)(.*?)</iframe>#ism',
 730                         '[vimeo]$2[/vimeo]',
 731                         $s
 732                 );
 733
 734                 return $s;
 735         }
 736
 737         /**
 738          * transform link href and img src from relative to absolute
 739          *
 740          * @param string $text
 741          * @param string $base base url
 742          * @return string
 743          */
 744         public static function relToAbs($text, $base)
 745         {
 746                 if (empty($base)) {
 747                         return $text;
 748                 }
 749
 750                 $base = rtrim($base, '/');
 751
 752                 $base2 = $base . "/";
 753
 754                 // Replace links
 755                 $pattern = "/<a([^>]*) href=\"(?!http|https|\/)([^\"]*)\"/";
 756                 $replace = "<a\${1} href=\"" . $base2 . "\${2}\"";
 757                 $text = preg_replace($pattern, $replace, $text);
 758
 759                 $pattern = "/<a([^>]*) href=\"(?!http|https)([^\"]*)\"/";
 760                 $replace = "<a\${1} href=\"" . $base . "\${2}\"";
 761                 $text = preg_replace($pattern, $replace, $text);
 762
 763                 // Replace images
 764                 $pattern = "/<img([^>]*) src=\"(?!http|https|\/)([^\"]*)\"/";
 765                 $replace = "<img\${1} src=\"" . $base2 . "\${2}\"";
 766                 $text = preg_replace($pattern, $replace, $text);
 767
 768                 $pattern = "/<img([^>]*) src=\"(?!http|https)([^\"]*)\"/";
 769                 $replace = "<img\${1} src=\"" . $base . "\${2}\"";
 770                 $text = preg_replace($pattern, $replace, $text);
 771
 772
 773                 // Done
 774                 return $text;
 775         }
 776
 777         /**
 778          * return div element with class 'clear'
 779          * @return string
 780          * @deprecated
 781          */
 782         public static function clearDiv()
 783         {
 784                 return '<div class="clear"></div>';
 785         }
 786
 787         /**
 788          * Loader for infinite scrolling
 789          * @return string html for loader
 790          */
 791         public static function scrollLoader()
 792         {
 793                 $tpl = Renderer::getMarkupTemplate("scroll_loader.tpl");
 794                 return Renderer::replaceMacros($tpl, [
 795                         'wait' => L10n::t('Loading more entries...'),
 796                         'end' => L10n::t('The end')
 797                 ]);
 798         }
 799
 800         /**
 801          * Get html for contact block.
 802          *
 803          * @template contact_block.tpl
 804          * @hook contact_block_end (contacts=>array, output=>string)
 805          * @return string
 806          */
 807         public static function contactBlock()
 808         {
 809                 $o = '';
 810                 $a = get_app();
 811
 812                 $shown = PConfig::get($a->profile['uid'], 'system', 'display_friend_count', 24);
 813                 if ($shown == 0) {
 814                         return;
 815                 }
 816
 817                 if (!is_array($a->profile) || $a->profile['hide-friends']) {
 818                         return $o;
 819                 }
 820
 821                 $r = q("SELECT COUNT(*) AS `total` FROM `contact`
 822                                 WHERE `uid` = %d AND NOT `self` AND NOT `blocked`
 823                                         AND NOT `pending` AND NOT `hidden` AND NOT `archive`
 824                                         AND `network` IN ('%s', '%s', '%s')",
 825                         intval($a->profile['uid']),
 826                         DBA::escape(Protocol::DFRN),
 827                         DBA::escape(Protocol::OSTATUS),
 828                         DBA::escape(Protocol::DIASPORA)
 829                 );
 830
 831                 if (DBA::isResult($r)) {
 832                         $total = intval($r[0]['total']);
 833                 }
 834
 835                 if (!$total) {
 836                         $contacts = L10n::t('No contacts');
 837                         $micropro = null;
 838                 } else {
 839                         // Splitting the query in two parts makes it much faster
 840                         $r = q("SELECT `id` FROM `contact`
 841                                         WHERE `uid` = %d AND NOT `self` AND NOT `blocked`
 842                                                 AND NOT `pending` AND NOT `hidden` AND NOT `archive`
 843                                                 AND `network` IN ('%s', '%s', '%s')
 844                                         ORDER BY RAND() LIMIT %d",
 845                                 intval($a->profile['uid']),
 846                                 DBA::escape(Protocol::DFRN),
 847                                 DBA::escape(Protocol::OSTATUS),
 848                                 DBA::escape(Protocol::DIASPORA),
 849                                 intval($shown)
 850                         );
 851
 852                         if (DBA::isResult($r)) {
 853                                 $contacts = [];
 854                                 foreach ($r as $contact) {
 855                                         $contacts[] = $contact["id"];
 856                                 }
 857
 858                                 $r = q("SELECT `id`, `uid`, `addr`, `url`, `name`, `thumb`, `network` FROM `contact` WHERE `id` IN (%s)",
 859                                         DBA::escape(implode(",", $contacts))
 860                                 );
 861
 862                                 if (DBA::isResult($r)) {
 863                                         $contacts = L10n::tt('%d Contact', '%d Contacts', $total);
 864                                         $micropro = [];
 865                                         foreach ($r as $rr) {
 866                                                 $micropro[] = self::micropro($rr, true, 'mpfriend');
 867                                         }
 868                                 }
 869                         }
 870                 }
 871
 872                 $tpl = Renderer::getMarkupTemplate('contact_block.tpl');
 873                 $o = Renderer::replaceMacros($tpl, [
 874                         '$contacts' => $contacts,
 875                         '$nickname' => $a->profile['nickname'],
 876                         '$viewcontacts' => L10n::t('View Contacts'),
 877                         '$micropro' => $micropro,
 878                 ]);
 879
 880                 $arr = ['contacts' => $r, 'output' => $o];
 881
 882                 Addon::callHooks('contact_block_end', $arr);
 883
 884                 return $o;
 885         }
 886
 887         /**
 888          * @brief Format contacts as picture links or as texxt links
 889          *
 890          * @param array $contact Array with contacts which contains an array with
 891          *      int 'id' => The ID of the contact
 892         *       int 'uid' => The user ID of the user who owns this data
 893         *       string 'name' => The name of the contact
 894         *       string 'url' => The url to the profile page of the contact
 895         *       string 'addr' => The webbie of the contact (e.g.) username@friendica.com
 896         *       string 'network' => The network to which the contact belongs to
 897         *       string 'thumb' => The contact picture
 898         *       string 'click' => js code which is performed when clicking on the contact
 899         * @param boolean $redirect If true try to use the redir url if it's possible
 900         * @param string $class CSS class for the
 901         * @param boolean $textmode If true display the contacts as text links
 902         *       if false display the contacts as picture links
 903
 904         * @return string Formatted html
 905         */
 906         public static function micropro($contact, $redirect = false, $class = '', $textmode = false)
 907         {
 908                 // Use the contact URL if no address is available
 909                 if (!x($contact, "addr")) {
 910                         $contact["addr"] = $contact["url"];
 911                 }
 912
 913                 $url = $contact['url'];
 914                 $sparkle = '';
 915                 $redir = false;
 916
 917                 if ($redirect) {
 918                         $url = Contact::magicLink($contact['url']);
 919                         if (strpos($url, 'redir/') === 0) {
 920                                 $sparkle = ' sparkle';
 921                         }
 922                 }
 923
 924                 // If there is some js available we don't need the url
 925                 if (x($contact, 'click')) {
 926                         $url = '';
 927                 }
 928
 929                 return Renderer::replaceMacros(Renderer::getMarkupTemplate(($textmode)?'micropro_txt.tpl':'micropro_img.tpl'), [
 930                         '$click' => defaults($contact, 'click', ''),
 931                         '$class' => $class,
 932                         '$url' => $url,
 933                         '$photo' => ProxyUtils::proxifyUrl($contact['thumb'], false, ProxyUtils::SIZE_THUMB),
 934                         '$name' => $contact['name'],
 935                         'title' => $contact['name'] . ' [' . $contact['addr'] . ']',
 936                         '$parkle' => $sparkle,
 937                         '$redir' => $redir
 938                 ]);
 939         }
 940
 941         /**
 942          * Search box.
 943          *
 944          * @param string $s     Search query.
 945          * @param string $id    HTML id
 946          * @param string $url   Search url.
 947          * @param bool   $save  Show save search button.
 948          * @param bool   $aside Display the search widgit aside.
 949          *
 950          * @return string Formatted HTML.
 951          */
 952         public static function search($s, $id = 'search-box', $url = 'search', $save = false, $aside = true)
 953         {
 954                 $mode = 'text';
 955
 956                 if (strpos($s, '#') === 0) {
 957                         $mode = 'tag';
 958                 }
 959                 $save_label = $mode === 'text' ? L10n::t('Save') : L10n::t('Follow');
 960
 961                 $values = [
 962                                 '$s' => htmlspecialchars($s),
 963                                 '$id' => $id,
 964                                 '$action_url' => $url,
 965                                 '$search_label' => L10n::t('Search'),
 966                                 '$save_label' => $save_label,
 967                                 '$savedsearch' => local_user() && Feature::isEnabled(local_user(), 'savedsearch'),
 968                                 '$search_hint' => L10n::t('@name, !forum, #tags, content'),
 969                                 '$mode' => $mode
 970                         ];
 971
 972                 if (!$aside) {
 973                         $values['$searchoption'] = [
 974                                                 L10n::t("Full Text"),
 975                                                 L10n::t("Tags"),
 976                                                 L10n::t("Contacts")];
 977
 978                         if (Config::get('system', 'poco_local_search')) {
 979                                 $values['$searchoption'][] = L10n::t("Forums");
 980                         }
 981                 }
 982
 983                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 984         }
 985
 986         /**
 987          * Replace naked text hyperlink with HTML formatted hyperlink
 988          *
 989          * @param string $s
 990          */
 991         public static function toLink($s)
 992         {
 993                 $s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank">$1</a>', $s);
 994                 $s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
 995                 return $s;
 996         }
 997
 998         /**
 999          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
1000          *
1001          * Reasons are expected to have been translated already.
1002          *
1003          * @param string $html
1004          * @param array  $reasons
1005          * @return string
1006          */
1007         public static function applyContentFilter($html, array $reasons)
1008         {
1009                 if (count($reasons)) {
1010                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
1011                         $html = Renderer::replaceMacros($tpl, [
1012                                 '$reasons'   => $reasons,
1013                                 '$rnd'       => random_string(8),
1014                                 '$openclose' => L10n::t('Click to open/close'),
1015                                 '$html'      => $html
1016                         ]);
1017                 }
1018
1019                 return $html;
1020         }
1021
1022         /**
1023          * replace html amp entity with amp char
1024          * @param string $s
1025          * @return string
1026          */
1027         public static function unamp($s)
1028         {
1029                 return str_replace('&amp;', '&', $s);
1030         }
1031 }