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 use Friendica\Content\Feature;
  23
  24 class HTML
  25 {
  26         public static function sanitizeCSS($input)
  27         {
  28                 $cleaned = "";
  29
  30                 $input = strtolower($input);
  31
  32                 for ($i = 0; $i < strlen($input); $i++) {
  33                         $char = substr($input, $i, 1);
  34
  35                         if (($char >= "a") && ($char <= "z")) {
  36                                 $cleaned .= $char;
  37                         }
  38
  39                         if (!(strpos(" #;:0123456789-_.%", $char) === false)) {
  40                                 $cleaned .= $char;
  41                         }
  42                 }
  43
  44                 return $cleaned;
  45         }
  46
  47         private static function tagToBBCode(DOMDocument $doc, $tag, $attributes, $startbb, $endbb)
  48         {
  49                 do {
  50                         $done = self::tagToBBCodeSub($doc, $tag, $attributes, $startbb, $endbb);
  51                 } while ($done);
  52         }
  53
  54         private static function tagToBBCodeSub(DOMDocument $doc, $tag, $attributes, $startbb, $endbb)
  55         {
  56                 $savestart = str_replace('$', '\x01', $startbb);
  57                 $replace = false;
  58
  59                 $xpath = new DOMXPath($doc);
  60
  61                 $list = $xpath->query("//" . $tag);
  62                 foreach ($list as $node) {
  63                         $attr = [];
  64                         if ($node->attributes->length) {
  65                                 foreach ($node->attributes as $attribute) {
  66                                         $attr[$attribute->name] = $attribute->value;
  67                                 }
  68                         }
  69
  70                         $replace = true;
  71
  72                         $startbb = $savestart;
  73
  74                         $i = 0;
  75
  76                         foreach ($attributes as $attribute => $value) {
  77                                 $startbb = str_replace('\x01' . ++$i, '$1', $startbb);
  78                                 if (strpos('*' . $startbb, '$1') > 0) {
  79                                         if ($replace && (@$attr[$attribute] != '')) {
  80                                                 $startbb = preg_replace($value, $startbb, $attr[$attribute], -1, $count);
  81
  82                                                 // If nothing could be changed
  83                                                 if ($count == 0) {
  84                                                         $replace = false;
  85                                                 }
  86                                         } else {
  87                                                 $replace = false;
  88                                         }
  89                                 } else {
  90                                         if (@$attr[$attribute] != $value) {
  91                                                 $replace = false;
  92                                         }
  93                                 }
  94                         }
  95
  96                         if ($replace) {
  97                                 $StartCode = $doc->createTextNode($startbb);
  98                                 $EndCode = $doc->createTextNode($endbb);
  99
 100                                 $node->parentNode->insertBefore($StartCode, $node);
 101
 102                                 if ($node->hasChildNodes()) {
 103                                         foreach ($node->childNodes as $child) {
 104                                                 $newNode = $child->cloneNode(true);
 105                                                 $node->parentNode->insertBefore($newNode, $node);
 106                                         }
 107                                 }
 108
 109                                 $node->parentNode->insertBefore($EndCode, $node);
 110                                 $node->parentNode->removeChild($node);
 111                         }
 112                 }
 113
 114                 return $replace;
 115         }
 116
 117         /**
 118          * Made by: ike@piratenpartei.de
 119          * Originally made for the syncom project: http://wiki.piratenpartei.de/Syncom
 120          *                                      https://github.com/annando/Syncom
 121          *
 122          * @brief Converter for HTML to BBCode
 123          * @param string $message
 124          * @param string $basepath
 125          * @return string
 126          */
 127         public static function toBBCode($message, $basepath = '')
 128         {
 129                 $message = str_replace("\r", "", $message);
 130
 131                 // Removing code blocks before the whitespace removal processing below
 132                 $codeblocks = [];
 133                 $message = preg_replace_callback(
 134                         '#<pre><code(?: class="language-([^"]*)")?>(.*)</code></pre>#iUs',
 135                         function ($matches) use (&$codeblocks) {
 136                                 $return = '[codeblock-' . count($codeblocks) . ']';
 137
 138                                 $prefix = '[code]';
 139                                 if ($matches[1] != '') {
 140                                         $prefix = '[code=' . $matches[1] . ']';
 141                                 }
 142
 143                                 $codeblocks[] = $prefix . PHP_EOL . trim($matches[2]) . PHP_EOL . '[/code]';
 144                                 return $return;
 145                         },
 146                         $message
 147                 );
 148
 149                 $message = str_replace(
 150                         [
 151                                 "<li><p>",
 152                                 "</p></li>",
 153                         ],
 154                         [
 155                                 "<li>",
 156                                 "</li>",
 157                         ],
 158                         $message
 159                 );
 160
 161                 // remove namespaces
 162                 $message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
 163                 $message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
 164
 165                 $doc = new DOMDocument();
 166                 $doc->preserveWhiteSpace = false;
 167
 168                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 169
 170                 @$doc->loadHTML($message);
 171
 172                 XML::deleteNode($doc, 'style');
 173                 XML::deleteNode($doc, 'head');
 174                 XML::deleteNode($doc, 'title');
 175                 XML::deleteNode($doc, 'meta');
 176                 XML::deleteNode($doc, 'xml');
 177                 XML::deleteNode($doc, 'removeme');
 178
 179                 $xpath = new DomXPath($doc);
 180                 $list = $xpath->query("//pre");
 181                 foreach ($list as $node) {
 182                         $node->nodeValue = str_replace("\n", "\r", $node->nodeValue);
 183                 }
 184
 185                 $message = $doc->saveHTML();
 186                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br />", " ", ""], $message);
 187                 $message = preg_replace('= [\s]*=i', " ", $message);
 188                 @$doc->loadHTML($message);
 189
 190                 self::tagToBBCode($doc, 'html', [], "", "");
 191                 self::tagToBBCode($doc, 'body', [], "", "");
 192
 193                 // Outlook-Quote - Variant 1
 194                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal', 'style' => 'margin-left:35.4pt'], '[quote]', '[/quote]');
 195
 196                 // Outlook-Quote - Variant 2
 197                 self::tagToBBCode(
 198                         $doc,
 199                         'div',
 200                         ['style' => 'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'],
 201                         '[quote]',
 202                         '[/quote]'
 203                 );
 204
 205                 // MyBB-Stuff
 206                 self::tagToBBCode($doc, 'span', ['style' => 'text-decoration: underline;'], '[u]', '[/u]');
 207                 self::tagToBBCode($doc, 'span', ['style' => 'font-style: italic;'], '[i]', '[/i]');
 208                 self::tagToBBCode($doc, 'span', ['style' => 'font-weight: bold;'], '[b]', '[/b]');
 209
 210                 /* self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
 211                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
 212                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
 213                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
 214                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
 215                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
 216                   self::node2BBCode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
 217                  */
 218                 // Untested
 219                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
 220                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
 221                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
 222
 223                 self::tagToBBCode($doc, 'span', ['style' => '/.*color:\s*(.+?)[,;].*/'], '[color="$1"]', '[/color]');
 224
 225                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 226                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
 227                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
 228                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 229                 // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
 230                 // Test
 231                 //self::node2BBCode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
 232                 self::tagToBBCode($doc, 'span', ['class' => 'type-link'], '[class=type-link]', '[/class]');
 233                 self::tagToBBCode($doc, 'span', ['class' => 'type-video'], '[class=type-video]', '[/class]');
 234
 235                 self::tagToBBCode($doc, 'strong', [], '[b]', '[/b]');
 236                 self::tagToBBCode($doc, 'em', [], '[i]', '[/i]');
 237                 self::tagToBBCode($doc, 'b', [], '[b]', '[/b]');
 238                 self::tagToBBCode($doc, 'i', [], '[i]', '[/i]');
 239                 self::tagToBBCode($doc, 'u', [], '[u]', '[/u]');
 240                 self::tagToBBCode($doc, 's', [], '[s]', '[/s]');
 241                 self::tagToBBCode($doc, 'del', [], '[s]', '[/s]');
 242                 self::tagToBBCode($doc, 'strike', [], '[s]', '[/s]');
 243
 244                 self::tagToBBCode($doc, 'big', [], "[size=large]", "[/size]");
 245                 self::tagToBBCode($doc, 'small', [], "[size=small]", "[/size]");
 246
 247                 self::tagToBBCode($doc, 'blockquote', [], '[quote]', '[/quote]');
 248
 249                 self::tagToBBCode($doc, 'br', [], "\n", '');
 250
 251                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal'], "\n", "");
 252                 self::tagToBBCode($doc, 'div', ['class' => 'MsoNormal'], "\r", "");
 253
 254                 self::tagToBBCode($doc, 'span', [], "", "");
 255
 256                 self::tagToBBCode($doc, 'span', [], "", "");
 257                 self::tagToBBCode($doc, 'pre', [], "", "");
 258
 259                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 260                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 261
 262                 self::tagToBBCode($doc, 'ul', [], "[list]", "[/list]");
 263                 self::tagToBBCode($doc, 'ol', [], "[list=1]", "[/list]");
 264                 self::tagToBBCode($doc, 'li', [], "[*]", "");
 265
 266                 self::tagToBBCode($doc, 'hr', [], "[hr]", "");
 267
 268                 self::tagToBBCode($doc, 'table', [], "", "");
 269                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 270                 self::tagToBBCode($doc, 'td', [], "\t", "");
 271                 //self::node2BBCode($doc, 'table', array(), "[table]", "[/table]");
 272                 //self::node2BBCode($doc, 'th', array(), "[th]", "[/th]");
 273                 //self::node2BBCode($doc, 'tr', array(), "[tr]", "[/tr]");
 274                 //self::node2BBCode($doc, 'td', array(), "[td]", "[/td]");
 275                 //self::node2BBCode($doc, 'h1', array(), "\n\n[size=xx-large][b]", "[/b][/size]\n");
 276                 //self::node2BBCode($doc, 'h2', array(), "\n\n[size=x-large][b]", "[/b][/size]\n");
 277                 //self::node2BBCode($doc, 'h3', array(), "\n\n[size=large][b]", "[/b][/size]\n");
 278                 //self::node2BBCode($doc, 'h4', array(), "\n\n[size=medium][b]", "[/b][/size]\n");
 279                 //self::node2BBCode($doc, 'h5', array(), "\n\n[size=small][b]", "[/b][/size]\n");
 280                 //self::node2BBCode($doc, 'h6', array(), "\n\n[size=x-small][b]", "[/b][/size]\n");
 281
 282                 self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 283                 self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 284                 self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 285                 self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 286                 self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 287                 self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 288
 289                 self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 290                 self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 291
 292                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]');
 293                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]');
 294
 295
 296                 self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]');
 297                 self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]');
 298                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[iframe]$1', '[/iframe]');
 299
 300                 self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 301                 self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 302
 303                 $message = $doc->saveHTML();
 304
 305                 // I'm removing something really disturbing
 306                 // Don't know exactly what it is
 307                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 308
 309                 $message = str_replace("&nbsp;", " ", $message);
 310
 311                 // removing multiple DIVs
 312                 $message = preg_replace('=\r *\r=i', "\n", $message);
 313                 $message = str_replace("\r", "\n", $message);
 314
 315                 Addon::callHooks('html2bbcode', $message);
 316
 317                 $message = strip_tags($message);
 318
 319                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 320
 321                 $message = str_replace(["<"], ["&lt;"], $message);
 322
 323                 // remove quotes if they don't make sense
 324                 $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 325
 326                 $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 327                 $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 328
 329                 do {
 330                         $oldmessage = $message;
 331                         $message = str_replace("\n \n", "\n\n", $message);
 332                 } while ($oldmessage != $message);
 333
 334                 do {
 335                         $oldmessage = $message;
 336                         $message = str_replace("\n\n\n", "\n\n", $message);
 337                 } while ($oldmessage != $message);
 338
 339                 do {
 340                         $oldmessage = $message;
 341                         $message = str_replace(
 342                                 [
 343                                 "[/size]\n\n",
 344                                 "\n[hr]",
 345                                 "[hr]\n",
 346                                 "\n[list",
 347                                 "[/list]\n",
 348                                 "\n[/",
 349                                 "[list]\n",
 350                                 "[list=1]\n",
 351                                 "\n[*]"],
 352                                 [
 353                                 "[/size]\n",
 354                                 "[hr]",
 355                                 "[hr]",
 356                                 "[list",
 357                                 "[/list]",
 358                                 "[/",
 359                                 "[list]",
 360                                 "[list=1]",
 361                                 "[*]"],
 362                                 $message
 363                         );
 364                 } while ($message != $oldmessage);
 365
 366                 $message = str_replace(
 367                         ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 368                         ['[b]', '[/b]', '[i]', '[/i]'],
 369                         $message
 370                 );
 371
 372                 // Handling Yahoo style of mails
 373                 $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 374
 375                 // Restore code blocks
 376                 $message = preg_replace_callback(
 377                         '#\[codeblock-([0-9]+)\]#iU',
 378                         function ($matches) use ($codeblocks) {
 379                                 $return = '';
 380                                 if (isset($codeblocks[intval($matches[1])])) {
 381                                         $return = $codeblocks[$matches[1]];
 382                                 }
 383                                 return $return;
 384                         },
 385                         $message
 386                 );
 387
 388                 $message = trim($message);
 389
 390                 if ($basepath != '') {
 391                         $message = self::qualifyURLs($message, $basepath);
 392                 }
 393
 394                 return $message;
 395         }
 396
 397         /**
 398          * @brief Sub function to complete incomplete URL
 399          *
 400          * @param array  $matches  Result of preg_replace_callback
 401          * @param string $basepath Basepath that is used to complete the URL
 402          *
 403          * @return string The expanded URL
 404          */
 405         private static function qualifyURLsSub($matches, $basepath)
 406         {
 407                 $base = parse_url($basepath);
 408                 unset($base['query']);
 409                 unset($base['fragment']);
 410
 411                 $link = $matches[0];
 412                 $url = $matches[1];
 413
 414                 $parts = array_merge($base, parse_url($url));
 415                 $url2 = Network::unparseURL($parts);
 416
 417                 return str_replace($url, $url2, $link);
 418         }
 419
 420         /**
 421          * @brief Complete incomplete URLs in BBCode
 422          *
 423          * @param string $body     Body with URLs
 424          * @param string $basepath Base path that is used to complete the URL
 425          *
 426          * @return string Body with expanded URLs
 427          */
 428         private static function qualifyURLs($body, $basepath)
 429         {
 430                 $URLSearchString = "^\[\]";
 431
 432                 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
 433                         "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
 434                         "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 435                         "/\[img\](.*?)\[\/img\]/ism",
 436                         "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 437                         "/\[zmg\](.*?)\[\/zmg\]/ism",
 438                         "/\[video\](.*?)\[\/video\]/ism",
 439                         "/\[audio\](.*?)\[\/audio\]/ism",
 440                 ];
 441
 442                 foreach ($matches as $match) {
 443                         $body = preg_replace_callback(
 444                                 $match,
 445                                 function ($match) use ($basepath) {
 446                                         return self::qualifyURLsSub($match, $basepath);
 447                                 },
 448                                 $body
 449                         );
 450                 }
 451                 return $body;
 452         }
 453
 454         private static function breakLines($line, $level, $wraplength = 75)
 455         {
 456                 if ($wraplength == 0) {
 457                         $wraplength = 2000000;
 458                 }
 459
 460                 $wraplen = $wraplength - $level;
 461
 462                 $newlines = [];
 463
 464                 do {
 465                         $oldline = $line;
 466
 467                         $subline = substr($line, 0, $wraplen);
 468
 469                         $pos = strrpos($subline, ' ');
 470
 471                         if ($pos == 0) {
 472                                 $pos = strpos($line, ' ');
 473                         }
 474
 475                         if (($pos > 0) && strlen($line) > $wraplen) {
 476                                 $newline = trim(substr($line, 0, $pos));
 477                                 if ($level > 0) {
 478                                         $newline = str_repeat(">", $level) . ' ' . $newline;
 479                                 }
 480
 481                                 $newlines[] = $newline . " ";
 482                                 $line = substr($line, $pos + 1);
 483                         }
 484                 } while ((strlen($line) > $wraplen) && !($oldline == $line));
 485
 486                 if ($level > 0) {
 487                         $line = str_repeat(">", $level) . ' ' . $line;
 488                 }
 489
 490                 $newlines[] = $line;
 491
 492                 return implode($newlines, "\n");
 493         }
 494
 495         private static function quoteLevel($message, $wraplength = 75)
 496         {
 497                 $lines = explode("\n", $message);
 498
 499                 $newlines = [];
 500                 $level = 0;
 501                 foreach ($lines as $line) {
 502                         $line = trim($line);
 503                         $startquote = false;
 504                         while (strpos("*" . $line, '[quote]') > 0) {
 505                                 $level++;
 506                                 $pos = strpos($line, '[quote]');
 507                                 $line = substr($line, 0, $pos) . substr($line, $pos + 7);
 508                                 $startquote = true;
 509                         }
 510
 511                         $currlevel = $level;
 512
 513                         while (strpos("*" . $line, '[/quote]') > 0) {
 514                                 $level--;
 515                                 if ($level < 0) {
 516                                         $level = 0;
 517                                 }
 518
 519                                 $pos = strpos($line, '[/quote]');
 520                                 $line = substr($line, 0, $pos) . substr($line, $pos + 8);
 521                         }
 522
 523                         if (!$startquote || ($line != '')) {
 524                                 $newlines[] = self::breakLines($line, $currlevel, $wraplength);
 525                         }
 526                 }
 527
 528                 return implode($newlines, "\n");
 529         }
 530
 531         private static function collectURLs($message)
 532         {
 533                 $pattern = '/<a.*?href="(.*?)".*?>(.*?)<\/a>/is';
 534                 preg_match_all($pattern, $message, $result, PREG_SET_ORDER);
 535
 536                 $urls = [];
 537                 foreach ($result as $treffer) {
 538                         $ignore = false;
 539
 540                         // A list of some links that should be ignored
 541                         $list = ["/user/", "/tag/", "/group/", "/profile/", "/search?search=", "/search?tag=", "mailto:", "/u/", "/node/",
 542                                 "//plus.google.com/", "//twitter.com/"];
 543                         foreach ($list as $listitem) {
 544                                 if (strpos($treffer[1], $listitem) !== false) {
 545                                         $ignore = true;
 546                                 }
 547                         }
 548
 549                         if ((strpos($treffer[1], "//twitter.com/") !== false) && (strpos($treffer[1], "/status/") !== false)) {
 550                                 $ignore = false;
 551                         }
 552
 553                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/posts") !== false)) {
 554                                 $ignore = false;
 555                         }
 556
 557                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/photos") !== false)) {
 558                                 $ignore = false;
 559                         }
 560
 561                         if (!$ignore) {
 562                                 $urls[$treffer[1]] = $treffer[1];
 563                         }
 564                 }
 565
 566                 return $urls;
 567         }
 568
 569         public static function toPlaintext($html, $wraplength = 75, $compact = false)
 570         {
 571                 $message = str_replace("\r", "", $html);
 572
 573                 $doc = new DOMDocument();
 574                 $doc->preserveWhiteSpace = false;
 575
 576                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 577
 578                 @$doc->loadHTML($message);
 579
 580                 $xpath = new DOMXPath($doc);
 581                 $list = $xpath->query("//pre");
 582                 foreach ($list as $node) {
 583                         $node->nodeValue = str_replace("\n", "\r", $node->nodeValue);
 584                 }
 585
 586                 $message = $doc->saveHTML();
 587                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br>", " ", ""], $message);
 588                 $message = preg_replace('= [\s]*=i', " ", $message);
 589
 590                 // Collecting all links
 591                 $urls = self::collectURLs($message);
 592
 593                 @$doc->loadHTML($message);
 594
 595                 self::tagToBBCode($doc, 'html', [], '', '');
 596                 self::tagToBBCode($doc, 'body', [], '', '');
 597
 598                 // MyBB-Auszeichnungen
 599                 /*
 600                   self::node2BBCode($doc, 'span', array('style'=>'text-decoration: underline;'), '_', '_');
 601                   self::node2BBCode($doc, 'span', array('style'=>'font-style: italic;'), '/', '/');
 602                   self::node2BBCode($doc, 'span', array('style'=>'font-weight: bold;'), '*', '*');
 603
 604                   self::node2BBCode($doc, 'strong', array(), '*', '*');
 605                   self::node2BBCode($doc, 'b', array(), '*', '*');
 606                   self::node2BBCode($doc, 'i', array(), '/', '/');
 607                   self::node2BBCode($doc, 'u', array(), '_', '_');
 608                  */
 609
 610                 if ($compact) {
 611                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 612                 } else {
 613                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 614                 }
 615
 616                 self::tagToBBCode($doc, 'br', [], "\n", '');
 617
 618                 self::tagToBBCode($doc, 'span', [], "", "");
 619                 self::tagToBBCode($doc, 'pre', [], "", "");
 620                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 621                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 622
 623                 //self::node2BBCode($doc, 'ul', array(), "\n[list]", "[/list]\n");
 624                 //self::node2BBCode($doc, 'ol', array(), "\n[list=1]", "[/list]\n");
 625                 self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 626
 627                 self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
 628
 629                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 630                 self::tagToBBCode($doc, 'td', [], "\t", "");
 631
 632                 self::tagToBBCode($doc, 'h1', [], "\n\n*", "*\n");
 633                 self::tagToBBCode($doc, 'h2', [], "\n\n*", "*\n");
 634                 self::tagToBBCode($doc, 'h3', [], "\n\n*", "*\n");
 635                 self::tagToBBCode($doc, 'h4', [], "\n\n*", "*\n");
 636                 self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
 637                 self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 638
 639                 // Problem: there is no reliable way to detect if it is a link to a tag or profile
 640                 //self::node2BBCode($doc, 'a', array('href'=>'/(.+)/'), ' $1 ', ' ', true);
 641                 //self::node2BBCode($doc, 'a', array('href'=>'/(.+)/', 'rel'=>'oembed'), ' $1 ', '', true);
 642                 //self::node2BBCode($doc, 'img', array('alt'=>'/(.+)/'), '$1', '');
 643                 //self::node2BBCode($doc, 'img', array('title'=>'/(.+)/'), '$1', '');
 644                 //self::node2BBCode($doc, 'img', array(), '', '');
 645                 if (!$compact) {
 646                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
 647                 } else {
 648                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' ', ' ');
 649                 }
 650
 651                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], ' $1 ', '');
 652
 653                 $message = $doc->saveHTML();
 654
 655                 if (!$compact) {
 656                         $message = str_replace("[img]", "", $message);
 657                         $message = str_replace("[/img]", "", $message);
 658                 }
 659
 660                 // was ersetze ich da?
 661                 // Irgendein stoerrisches UTF-Zeug
 662                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 663
 664                 $message = str_replace("&nbsp;", " ", $message);
 665
 666                 // Aufeinanderfolgende DIVs
 667                 $message = preg_replace('=\r *\r=i', "\n", $message);
 668                 $message = str_replace("\r", "\n", $message);
 669
 670                 $message = strip_tags($message);
 671
 672                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 673
 674                 if (!$compact && ($message != '')) {
 675                         foreach ($urls as $id => $url) {
 676                                 if ($url != '' && strpos($message, $url) === false) {
 677                                         $message .= "\n" . $url . ' ';
 678                                 }
 679                         }
 680                 }
 681
 682                 $message = str_replace("\n«", "«\n", $message);
 683                 $message = str_replace("»\n", "\n»", $message);
 684
 685                 do {
 686                         $oldmessage = $message;
 687                         $message = str_replace("\n\n\n", "\n\n", $message);
 688                 } while ($oldmessage != $message);
 689
 690                 $message = self::quoteLevel(trim($message), $wraplength);
 691
 692                 return trim($message);
 693         }
 694
 695         /**
 696          * Converts provided HTML code to Markdown. The hardwrap parameter maximizes
 697          * compatibility with Diaspora in spite of the Markdown standards.
 698          *
 699          * @param string $html
 700          * @return string
 701          */
 702         public static function toMarkdown($html)
 703         {
 704                 $converter = new HtmlConverter(['hard_break' => true]);
 705                 $markdown = $converter->convert($html);
 706
 707                 return $markdown;
 708         }
 709
 710         /**
 711          * @brief Convert video HTML to BBCode tags
 712          *
 713          * @param string $s
 714          */
 715         public static function toBBCodeVideo($s)
 716         {
 717                 $s = preg_replace(
 718                         '#<object[^>]+>(.*?)https?://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+)(.*?)</object>#ism',
 719                         '[youtube]$2[/youtube]',
 720                         $s
 721                 );
 722
 723                 $s = preg_replace(
 724                         '#<iframe[^>](.*?)https?://www.youtube.com/embed/([A-Za-z0-9\-_=]+)(.*?)</iframe>#ism',
 725                         '[youtube]$2[/youtube]',
 726                         $s
 727                 );
 728
 729                 $s = preg_replace(
 730                         '#<iframe[^>](.*?)https?://player.vimeo.com/video/([0-9]+)(.*?)</iframe>#ism',
 731                         '[vimeo]$2[/vimeo]',
 732                         $s
 733                 );
 734
 735                 return $s;
 736         }
 737
 738         /**
 739          * transform link href and img src from relative to absolute
 740          *
 741          * @param string $text
 742          * @param string $base base url
 743          * @return string
 744          */
 745         public static function relToAbs($text, $base)
 746         {
 747                 if (empty($base)) {
 748                         return $text;
 749                 }
 750
 751                 $base = rtrim($base, '/');
 752
 753                 $base2 = $base . "/";
 754
 755                 // Replace links
 756                 $pattern = "/<a([^>]*) href=\"(?!http|https|\/)([^\"]*)\"/";
 757                 $replace = "<a\${1} href=\"" . $base2 . "\${2}\"";
 758                 $text = preg_replace($pattern, $replace, $text);
 759
 760                 $pattern = "/<a([^>]*) href=\"(?!http|https)([^\"]*)\"/";
 761                 $replace = "<a\${1} href=\"" . $base . "\${2}\"";
 762                 $text = preg_replace($pattern, $replace, $text);
 763
 764                 // Replace images
 765                 $pattern = "/<img([^>]*) src=\"(?!http|https|\/)([^\"]*)\"/";
 766                 $replace = "<img\${1} src=\"" . $base2 . "\${2}\"";
 767                 $text = preg_replace($pattern, $replace, $text);
 768
 769                 $pattern = "/<img([^>]*) src=\"(?!http|https)([^\"]*)\"/";
 770                 $replace = "<img\${1} src=\"" . $base . "\${2}\"";
 771                 $text = preg_replace($pattern, $replace, $text);
 772
 773
 774                 // Done
 775                 return $text;
 776         }
 777
 778         /**
 779          * return div element with class 'clear'
 780          * @return string
 781          * @deprecated
 782          */
 783         public static function clearDiv()
 784         {
 785                 return '<div class="clear"></div>';
 786         }
 787
 788         /**
 789          * Loader for infinite scrolling
 790          * @return string html for loader
 791          */
 792         public static function scrollLoader()
 793         {
 794                 $tpl = Renderer::getMarkupTemplate("scroll_loader.tpl");
 795                 return Renderer::replaceMacros($tpl, [
 796                         'wait' => L10n::t('Loading more entries...'),
 797                         'end' => L10n::t('The end')
 798                 ]);
 799         }
 800
 801         /**
 802          * Get html for contact block.
 803          *
 804          * @template contact_block.tpl
 805          * @hook contact_block_end (contacts=>array, output=>string)
 806          * @return string
 807          */
 808         public static function contactBlock()
 809         {
 810                 $o = '';
 811                 $a = get_app();
 812
 813                 $shown = PConfig::get($a->profile['uid'], 'system', 'display_friend_count', 24);
 814                 if ($shown == 0) {
 815                         return;
 816                 }
 817
 818                 if (!is_array($a->profile) || $a->profile['hide-friends']) {
 819                         return $o;
 820                 }
 821
 822                 $r = q("SELECT COUNT(*) AS `total` FROM `contact`
 823                                 WHERE `uid` = %d AND NOT `self` AND NOT `blocked`
 824                                         AND NOT `pending` AND NOT `hidden` AND NOT `archive`
 825                                         AND `network` IN ('%s', '%s', '%s')",
 826                         intval($a->profile['uid']),
 827                         DBA::escape(Protocol::DFRN),
 828                         DBA::escape(Protocol::OSTATUS),
 829                         DBA::escape(Protocol::DIASPORA)
 830                 );
 831
 832                 if (DBA::isResult($r)) {
 833                         $total = intval($r[0]['total']);
 834                 }
 835
 836                 if (!$total) {
 837                         $contacts = L10n::t('No contacts');
 838                         $micropro = null;
 839                 } else {
 840                         // Splitting the query in two parts makes it much faster
 841                         $r = q("SELECT `id` FROM `contact`
 842                                         WHERE `uid` = %d AND NOT `self` AND NOT `blocked`
 843                                                 AND NOT `pending` AND NOT `hidden` AND NOT `archive`
 844                                                 AND `network` IN ('%s', '%s', '%s')
 845                                         ORDER BY RAND() LIMIT %d",
 846                                 intval($a->profile['uid']),
 847                                 DBA::escape(Protocol::DFRN),
 848                                 DBA::escape(Protocol::OSTATUS),
 849                                 DBA::escape(Protocol::DIASPORA),
 850                                 intval($shown)
 851                         );
 852
 853                         if (DBA::isResult($r)) {
 854                                 $contacts = [];
 855                                 foreach ($r as $contact) {
 856                                         $contacts[] = $contact["id"];
 857                                 }
 858
 859                                 $r = q("SELECT `id`, `uid`, `addr`, `url`, `name`, `thumb`, `network` FROM `contact` WHERE `id` IN (%s)",
 860                                         DBA::escape(implode(",", $contacts))
 861                                 );
 862
 863                                 if (DBA::isResult($r)) {
 864                                         $contacts = L10n::tt('%d Contact', '%d Contacts', $total);
 865                                         $micropro = [];
 866                                         foreach ($r as $rr) {
 867                                                 $micropro[] = self::micropro($rr, true, 'mpfriend');
 868                                         }
 869                                 }
 870                         }
 871                 }
 872
 873                 $tpl = Renderer::getMarkupTemplate('contact_block.tpl');
 874                 $o = Renderer::replaceMacros($tpl, [
 875                         '$contacts' => $contacts,
 876                         '$nickname' => $a->profile['nickname'],
 877                         '$viewcontacts' => L10n::t('View Contacts'),
 878                         '$micropro' => $micropro,
 879                 ]);
 880
 881                 $arr = ['contacts' => $r, 'output' => $o];
 882
 883                 Addon::callHooks('contact_block_end', $arr);
 884
 885                 return $o;
 886         }
 887
 888         /**
 889          * @brief Format contacts as picture links or as texxt links
 890          *
 891          * @param array $contact Array with contacts which contains an array with
 892          *      int 'id' => The ID of the contact
 893         *       int 'uid' => The user ID of the user who owns this data
 894         *       string 'name' => The name of the contact
 895         *       string 'url' => The url to the profile page of the contact
 896         *       string 'addr' => The webbie of the contact (e.g.) username@friendica.com
 897         *       string 'network' => The network to which the contact belongs to
 898         *       string 'thumb' => The contact picture
 899         *       string 'click' => js code which is performed when clicking on the contact
 900         * @param boolean $redirect If true try to use the redir url if it's possible
 901         * @param string $class CSS class for the
 902         * @param boolean $textmode If true display the contacts as text links
 903         *       if false display the contacts as picture links
 904
 905         * @return string Formatted html
 906         */
 907         public static function micropro($contact, $redirect = false, $class = '', $textmode = false)
 908         {
 909                 // Use the contact URL if no address is available
 910                 if (!x($contact, "addr")) {
 911                         $contact["addr"] = $contact["url"];
 912                 }
 913
 914                 $url = $contact['url'];
 915                 $sparkle = '';
 916                 $redir = false;
 917
 918                 if ($redirect) {
 919                         $url = Contact::magicLink($contact['url']);
 920                         if (strpos($url, 'redir/') === 0) {
 921                                 $sparkle = ' sparkle';
 922                         }
 923                 }
 924
 925                 // If there is some js available we don't need the url
 926                 if (x($contact, 'click')) {
 927                         $url = '';
 928                 }
 929
 930                 return Renderer::replaceMacros(Renderer::getMarkupTemplate(($textmode)?'micropro_txt.tpl':'micropro_img.tpl'), [
 931                         '$click' => defaults($contact, 'click', ''),
 932                         '$class' => $class,
 933                         '$url' => $url,
 934                         '$photo' => ProxyUtils::proxifyUrl($contact['thumb'], false, ProxyUtils::SIZE_THUMB),
 935                         '$name' => $contact['name'],
 936                         'title' => $contact['name'] . ' [' . $contact['addr'] . ']',
 937                         '$parkle' => $sparkle,
 938                         '$redir' => $redir
 939                 ]);
 940         }
 941
 942         /**
 943          * Search box.
 944          *
 945          * @param string $s     Search query.
 946          * @param string $id    HTML id
 947          * @param string $url   Search url.
 948          * @param bool   $save  Show save search button.
 949          * @param bool   $aside Display the search widgit aside.
 950          *
 951          * @return string Formatted HTML.
 952          */
 953         public static function search($s, $id = 'search-box', $url = 'search', $save = false, $aside = true)
 954         {
 955                 $mode = 'text';
 956
 957                 if (strpos($s, '#') === 0) {
 958                         $mode = 'tag';
 959                 }
 960                 $save_label = $mode === 'text' ? L10n::t('Save') : L10n::t('Follow');
 961
 962                 $values = [
 963                                 '$s' => htmlspecialchars($s),
 964                                 '$id' => $id,
 965                                 '$action_url' => $url,
 966                                 '$search_label' => L10n::t('Search'),
 967                                 '$save_label' => $save_label,
 968                                 '$savedsearch' => local_user() && Feature::isEnabled(local_user(), 'savedsearch'),
 969                                 '$search_hint' => L10n::t('@name, !forum, #tags, content'),
 970                                 '$mode' => $mode
 971                         ];
 972
 973                 if (!$aside) {
 974                         $values['$searchoption'] = [
 975                                                 L10n::t("Full Text"),
 976                                                 L10n::t("Tags"),
 977                                                 L10n::t("Contacts")];
 978
 979                         if (Config::get('system', 'poco_local_search')) {
 980                                 $values['$searchoption'][] = L10n::t("Forums");
 981                         }
 982                 }
 983
 984                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 985         }
 986
 987         /**
 988          * Replace naked text hyperlink with HTML formatted hyperlink
 989          *
 990          * @param string $s
 991          */
 992         public static function toLink($s)
 993         {
 994                 $s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank">$1</a>', $s);
 995                 $s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
 996                 return $s;
 997         }
 998
 999         /**
1000          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
1001          *
1002          * Reasons are expected to have been translated already.
1003          *
1004          * @param string $html
1005          * @param array  $reasons
1006          * @return string
1007          */
1008         public static function applyContentFilter($html, array $reasons)
1009         {
1010                 if (count($reasons)) {
1011                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
1012                         $html = Renderer::replaceMacros($tpl, [
1013                                 '$reasons'   => $reasons,
1014                                 '$rnd'       => random_string(8),
1015                                 '$openclose' => L10n::t('Click to open/close'),
1016                                 '$html'      => $html
1017                         ]);
1018                 }
1019
1020                 return $html;
1021         }
1022
1023         /**
1024          * replace html amp entity with amp char
1025          * @param string $s
1026          * @return string
1027          */
1028         public static function unamp($s)
1029         {
1030                 return str_replace('&amp;', '&', $s);
1031         }
1032 }