src/Content/Text/HTML.php

   1 <?php
   2 /**
   3  * @file src/Content/Text/HTML.php
   4  */
   5
   6 namespace Friendica\Content\Text;
   7
   8 use DOMDocument;
   9 use DOMXPath;
  10 use Friendica\Content\Widget\ContactBlock;
  11 use Friendica\Core\Hook;
  12 use Friendica\Core\L10n;
  13 use Friendica\Core\Config;
  14 use Friendica\Core\Renderer;
  15 use Friendica\Model\Contact;
  16 use Friendica\Util\Network;
  17 use Friendica\Util\Proxy as ProxyUtils;
  18 use Friendica\Util\Strings;
  19 use Friendica\Util\XML;
  20 use League\HTMLToMarkdown\HtmlConverter;
  21
  22 class HTML
  23 {
  24         public static function sanitizeCSS($input)
  25         {
  26                 $cleaned = "";
  27
  28                 $input = strtolower($input);
  29
  30                 for ($i = 0; $i < strlen($input); $i++) {
  31                         $char = substr($input, $i, 1);
  32
  33                         if (($char >= "a") && ($char <= "z")) {
  34                                 $cleaned .= $char;
  35                         }
  36
  37                         if (!(strpos(" #;:0123456789-_.%", $char) === false)) {
  38                                 $cleaned .= $char;
  39                         }
  40                 }
  41
  42                 return $cleaned;
  43         }
  44
  45         private static function tagToBBCode(DOMDocument $doc, $tag, $attributes, $startbb, $endbb)
  46         {
  47                 do {
  48                         $done = self::tagToBBCodeSub($doc, $tag, $attributes, $startbb, $endbb);
  49                 } while ($done);
  50         }
  51
  52         private static function tagToBBCodeSub(DOMDocument $doc, $tag, $attributes, $startbb, $endbb)
  53         {
  54                 $savestart = str_replace('$', '\x01', $startbb);
  55                 $replace = false;
  56
  57                 $xpath = new DOMXPath($doc);
  58
  59                 /** @var \DOMNode[] $list */
  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                                         /** @var \DOMNode $child */
 103                                         foreach ($node->childNodes as $child) {
 104                                                 if (trim($child->nodeValue)) {
 105                                                         $newNode = $child->cloneNode(true);
 106                                                         $node->parentNode->insertBefore($newNode, $node);
 107                                                 }
 108                                         }
 109                                 }
 110
 111                                 $node->parentNode->insertBefore($EndCode, $node);
 112                                 $node->parentNode->removeChild($node);
 113                         }
 114                 }
 115
 116                 return $replace;
 117         }
 118
 119         /**
 120          * Made by: ike@piratenpartei.de
 121          * Originally made for the syncom project: http://wiki.piratenpartei.de/Syncom
 122          *                    https://github.com/annando/Syncom
 123          *
 124          * @brief Converter for HTML to BBCode
 125          * @param string $message
 126          * @param string $basepath
 127          * @return string
 128          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 129          */
 130         public static function toBBCode($message, $basepath = '')
 131         {
 132                 $message = str_replace("\r", "", $message);
 133
 134                 // Removing code blocks before the whitespace removal processing below
 135                 $codeblocks = [];
 136                 $message = preg_replace_callback(
 137                         '#<pre><code(?: class="language-([^"]*)")?>(.*)</code></pre>#iUs',
 138                         function ($matches) use (&$codeblocks) {
 139                                 $return = '[codeblock-' . count($codeblocks) . ']';
 140
 141                                 $prefix = '[code]';
 142                                 if ($matches[1] != '') {
 143                                         $prefix = '[code=' . $matches[1] . ']';
 144                                 }
 145
 146                                 $codeblocks[] = $prefix . PHP_EOL . trim($matches[2]) . PHP_EOL . '[/code]';
 147                                 return $return;
 148                         },
 149                         $message
 150                 );
 151
 152                 $message = str_replace(
 153                         [
 154                                 "<li><p>",
 155                                 "</p></li>",
 156                         ],
 157                         [
 158                                 "<li>",
 159                                 "</li>",
 160                         ],
 161                         $message
 162                 );
 163
 164                 // remove namespaces
 165                 $message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
 166                 $message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
 167
 168                 $doc = new DOMDocument();
 169                 $doc->preserveWhiteSpace = false;
 170
 171                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 172
 173                 @$doc->loadHTML($message);
 174
 175                 XML::deleteNode($doc, 'style');
 176                 XML::deleteNode($doc, 'head');
 177                 XML::deleteNode($doc, 'title');
 178                 XML::deleteNode($doc, 'meta');
 179                 XML::deleteNode($doc, 'xml');
 180                 XML::deleteNode($doc, 'removeme');
 181
 182                 $xpath = new DomXPath($doc);
 183                 $list = $xpath->query("//pre");
 184                 foreach ($list as $node) {
 185                         // Ensure to escape unescaped & - they will otherwise raise a warning
 186                         $safe_value = preg_replace('/&(?!\w+;)/', '&amp;', $node->nodeValue);
 187                         $node->nodeValue = str_replace("\n", "\r", $safe_value);
 188                 }
 189
 190                 $message = $doc->saveHTML();
 191                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br />", " ", ""], $message);
 192                 $message = preg_replace('= [\s]*=i', " ", $message);
 193                 @$doc->loadHTML($message);
 194
 195                 self::tagToBBCode($doc, 'html', [], "", "");
 196                 self::tagToBBCode($doc, 'body', [], "", "");
 197
 198                 // Outlook-Quote - Variant 1
 199                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal', 'style' => 'margin-left:35.4pt'], '[quote]', '[/quote]');
 200
 201                 // Outlook-Quote - Variant 2
 202                 self::tagToBBCode(
 203                         $doc,
 204                         'div',
 205                         ['style' => 'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'],
 206                         '[quote]',
 207                         '[/quote]'
 208                 );
 209
 210                 // MyBB-Stuff
 211                 self::tagToBBCode($doc, 'span', ['style' => 'text-decoration: underline;'], '[u]', '[/u]');
 212                 self::tagToBBCode($doc, 'span', ['style' => 'font-style: italic;'], '[i]', '[/i]');
 213                 self::tagToBBCode($doc, 'span', ['style' => 'font-weight: bold;'], '[b]', '[/b]');
 214
 215                 /* self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
 216                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
 217                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
 218                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
 219                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
 220                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
 221                   self::node2BBCode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
 222                  */
 223                 // Untested
 224                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
 225                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
 226                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
 227
 228                 self::tagToBBCode($doc, 'span', ['style' => '/.*color:\s*(.+?)[,;].*/'], '[color="$1"]', '[/color]');
 229
 230                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 231                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
 232                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
 233                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 234                 // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
 235                 // Test
 236                 //self::node2BBCode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
 237                 self::tagToBBCode($doc, 'span', ['class' => 'type-link'], '[class=type-link]', '[/class]');
 238                 self::tagToBBCode($doc, 'span', ['class' => 'type-video'], '[class=type-video]', '[/class]');
 239
 240                 self::tagToBBCode($doc, 'strong', [], '[b]', '[/b]');
 241                 self::tagToBBCode($doc, 'em', [], '[i]', '[/i]');
 242                 self::tagToBBCode($doc, 'b', [], '[b]', '[/b]');
 243                 self::tagToBBCode($doc, 'i', [], '[i]', '[/i]');
 244                 self::tagToBBCode($doc, 'u', [], '[u]', '[/u]');
 245                 self::tagToBBCode($doc, 's', [], '[s]', '[/s]');
 246                 self::tagToBBCode($doc, 'del', [], '[s]', '[/s]');
 247                 self::tagToBBCode($doc, 'strike', [], '[s]', '[/s]');
 248
 249                 self::tagToBBCode($doc, 'big', [], "[size=large]", "[/size]");
 250                 self::tagToBBCode($doc, 'small', [], "[size=small]", "[/size]");
 251
 252                 self::tagToBBCode($doc, 'blockquote', [], '[quote]', '[/quote]');
 253
 254                 self::tagToBBCode($doc, 'br', [], "\n", '');
 255
 256                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal'], "\n", "");
 257                 self::tagToBBCode($doc, 'div', ['class' => 'MsoNormal'], "\r", "");
 258
 259                 self::tagToBBCode($doc, 'span', [], "", "");
 260
 261                 self::tagToBBCode($doc, 'span', [], "", "");
 262                 self::tagToBBCode($doc, 'pre', [], "", "");
 263
 264                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 265                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 266
 267                 self::tagToBBCode($doc, 'ul', [], "[list]", "[/list]");
 268                 self::tagToBBCode($doc, 'ol', [], "[list=1]", "[/list]");
 269                 self::tagToBBCode($doc, 'li', [], "[*]", "");
 270
 271                 self::tagToBBCode($doc, 'hr', [], "[hr]", "");
 272
 273                 self::tagToBBCode($doc, 'table', [], "", "");
 274                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 275                 self::tagToBBCode($doc, 'td', [], "\t", "");
 276                 //self::node2BBCode($doc, 'table', array(), "[table]", "[/table]");
 277                 //self::node2BBCode($doc, 'th', array(), "[th]", "[/th]");
 278                 //self::node2BBCode($doc, 'tr', array(), "[tr]", "[/tr]");
 279                 //self::node2BBCode($doc, 'td', array(), "[td]", "[/td]");
 280                 //self::node2BBCode($doc, 'h1', array(), "\n\n[size=xx-large][b]", "[/b][/size]\n");
 281                 //self::node2BBCode($doc, 'h2', array(), "\n\n[size=x-large][b]", "[/b][/size]\n");
 282                 //self::node2BBCode($doc, 'h3', array(), "\n\n[size=large][b]", "[/b][/size]\n");
 283                 //self::node2BBCode($doc, 'h4', array(), "\n\n[size=medium][b]", "[/b][/size]\n");
 284                 //self::node2BBCode($doc, 'h5', array(), "\n\n[size=small][b]", "[/b][/size]\n");
 285                 //self::node2BBCode($doc, 'h6', array(), "\n\n[size=x-small][b]", "[/b][/size]\n");
 286
 287                 self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 288                 self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 289                 self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 290                 self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 291                 self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 292                 self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 293
 294                 self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 295                 self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 296
 297                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'alt' => '/(.+)/'], '[img=$1]$2', '[/img]');
 298                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]');
 299                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]');
 300
 301
 302                 self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]');
 303                 self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]');
 304                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[iframe]$1', '[/iframe]');
 305
 306                 self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 307                 self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 308
 309                 $message = $doc->saveHTML();
 310
 311                 // I'm removing something really disturbing
 312                 // Don't know exactly what it is
 313                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 314
 315                 $message = str_replace("&nbsp;", " ", $message);
 316
 317                 // removing multiple DIVs
 318                 $message = preg_replace('=\r *\r=i', "\n", $message);
 319                 $message = str_replace("\r", "\n", $message);
 320
 321                 Hook::callAll('html2bbcode', $message);
 322
 323                 $message = strip_tags($message);
 324
 325                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 326
 327                 $message = str_replace(["<"], ["&lt;"], $message);
 328
 329                 // remove quotes if they don't make sense
 330                 $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 331
 332                 $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 333                 $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 334
 335                 do {
 336                         $oldmessage = $message;
 337                         $message = str_replace("\n \n", "\n\n", $message);
 338                 } while ($oldmessage != $message);
 339
 340                 do {
 341                         $oldmessage = $message;
 342                         $message = str_replace("\n\n\n", "\n\n", $message);
 343                 } while ($oldmessage != $message);
 344
 345                 do {
 346                         $oldmessage = $message;
 347                         $message = str_replace(
 348                                 [
 349                                 "[/size]\n\n",
 350                                 "\n[hr]",
 351                                 "[hr]\n",
 352                                 "\n[list",
 353                                 "[/list]\n",
 354                                 "\n[/",
 355                                 "[list]\n",
 356                                 "[list=1]\n",
 357                                 "\n[*]"],
 358                                 [
 359                                 "[/size]\n",
 360                                 "[hr]",
 361                                 "[hr]",
 362                                 "[list",
 363                                 "[/list]",
 364                                 "[/",
 365                                 "[list]",
 366                                 "[list=1]",
 367                                 "[*]"],
 368                                 $message
 369                         );
 370                 } while ($message != $oldmessage);
 371
 372                 $message = str_replace(
 373                         ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 374                         ['[b]', '[/b]', '[i]', '[/i]'],
 375                         $message
 376                 );
 377
 378                 // Handling Yahoo style of mails
 379                 $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 380
 381                 // Restore code blocks
 382                 $message = preg_replace_callback(
 383                         '#\[codeblock-([0-9]+)\]#iU',
 384                         function ($matches) use ($codeblocks) {
 385                                 $return = '';
 386                                 if (isset($codeblocks[intval($matches[1])])) {
 387                                         $return = $codeblocks[$matches[1]];
 388                                 }
 389                                 return $return;
 390                         },
 391                         $message
 392                 );
 393
 394                 $message = trim($message);
 395
 396                 if ($basepath != '') {
 397                         $message = self::qualifyURLs($message, $basepath);
 398                 }
 399
 400                 return $message;
 401         }
 402
 403         /**
 404          * @brief Sub function to complete incomplete URL
 405          *
 406          * @param array  $matches  Result of preg_replace_callback
 407          * @param string $basepath Basepath that is used to complete the URL
 408          *
 409          * @return string The expanded URL
 410          */
 411         private static function qualifyURLsSub($matches, $basepath)
 412         {
 413                 $base = parse_url($basepath);
 414                 unset($base['query']);
 415                 unset($base['fragment']);
 416
 417                 $link = $matches[0];
 418                 $url = $matches[1];
 419
 420                 $parts = array_merge($base, parse_url($url));
 421                 $url2 = Network::unparseURL($parts);
 422
 423                 return str_replace($url, $url2, $link);
 424         }
 425
 426         /**
 427          * @brief Complete incomplete URLs in BBCode
 428          *
 429          * @param string $body     Body with URLs
 430          * @param string $basepath Base path that is used to complete the URL
 431          *
 432          * @return string Body with expanded URLs
 433          */
 434         private static function qualifyURLs($body, $basepath)
 435         {
 436                 $URLSearchString = "^\[\]";
 437
 438                 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
 439                         "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
 440                         "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 441                         "/\[img\](.*?)\[\/img\]/ism",
 442                         "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 443                         "/\[zmg\](.*?)\[\/zmg\]/ism",
 444                         "/\[video\](.*?)\[\/video\]/ism",
 445                         "/\[audio\](.*?)\[\/audio\]/ism",
 446                 ];
 447
 448                 foreach ($matches as $match) {
 449                         $body = preg_replace_callback(
 450                                 $match,
 451                                 function ($match) use ($basepath) {
 452                                         return self::qualifyURLsSub($match, $basepath);
 453                                 },
 454                                 $body
 455                         );
 456                 }
 457                 return $body;
 458         }
 459
 460         private static function breakLines($line, $level, $wraplength = 75)
 461         {
 462                 if ($wraplength == 0) {
 463                         $wraplength = 2000000;
 464                 }
 465
 466                 $wraplen = $wraplength - $level;
 467
 468                 $newlines = [];
 469
 470                 do {
 471                         $oldline = $line;
 472
 473                         $subline = substr($line, 0, $wraplen);
 474
 475                         $pos = strrpos($subline, ' ');
 476
 477                         if ($pos == 0) {
 478                                 $pos = strpos($line, ' ');
 479                         }
 480
 481                         if (($pos > 0) && strlen($line) > $wraplen) {
 482                                 $newline = trim(substr($line, 0, $pos));
 483                                 if ($level > 0) {
 484                                         $newline = str_repeat(">", $level) . ' ' . $newline;
 485                                 }
 486
 487                                 $newlines[] = $newline . " ";
 488                                 $line = substr($line, $pos + 1);
 489                         }
 490                 } while ((strlen($line) > $wraplen) && !($oldline == $line));
 491
 492                 if ($level > 0) {
 493                         $line = str_repeat(">", $level) . ' ' . $line;
 494                 }
 495
 496                 $newlines[] = $line;
 497
 498                 return implode($newlines, "\n");
 499         }
 500
 501         private static function quoteLevel($message, $wraplength = 75)
 502         {
 503                 $lines = explode("\n", $message);
 504
 505                 $newlines = [];
 506                 $level = 0;
 507                 foreach ($lines as $line) {
 508                         $line = trim($line);
 509                         $startquote = false;
 510                         while (strpos("*" . $line, '[quote]') > 0) {
 511                                 $level++;
 512                                 $pos = strpos($line, '[quote]');
 513                                 $line = substr($line, 0, $pos) . substr($line, $pos + 7);
 514                                 $startquote = true;
 515                         }
 516
 517                         $currlevel = $level;
 518
 519                         while (strpos("*" . $line, '[/quote]') > 0) {
 520                                 $level--;
 521                                 if ($level < 0) {
 522                                         $level = 0;
 523                                 }
 524
 525                                 $pos = strpos($line, '[/quote]');
 526                                 $line = substr($line, 0, $pos) . substr($line, $pos + 8);
 527                         }
 528
 529                         if (!$startquote || ($line != '')) {
 530                                 $newlines[] = self::breakLines($line, $currlevel, $wraplength);
 531                         }
 532                 }
 533
 534                 return implode($newlines, "\n");
 535         }
 536
 537         private static function collectURLs($message)
 538         {
 539                 $pattern = '/<a.*?href="(.*?)".*?>(.*?)<\/a>/is';
 540                 preg_match_all($pattern, $message, $result, PREG_SET_ORDER);
 541
 542                 $urls = [];
 543                 foreach ($result as $treffer) {
 544                         $ignore = false;
 545
 546                         // A list of some links that should be ignored
 547                         $list = ["/user/", "/tag/", "/group/", "/profile/", "/search?search=", "/search?tag=", "mailto:", "/u/", "/node/",
 548                                 "//plus.google.com/", "//twitter.com/"];
 549                         foreach ($list as $listitem) {
 550                                 if (strpos($treffer[1], $listitem) !== false) {
 551                                         $ignore = true;
 552                                 }
 553                         }
 554
 555                         if ((strpos($treffer[1], "//twitter.com/") !== false) && (strpos($treffer[1], "/status/") !== false)) {
 556                                 $ignore = false;
 557                         }
 558
 559                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/posts") !== false)) {
 560                                 $ignore = false;
 561                         }
 562
 563                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/photos") !== false)) {
 564                                 $ignore = false;
 565                         }
 566
 567                         $ignore = $ignore || strpos($treffer[1], '#') === 0;
 568
 569                         if (!$ignore) {
 570                                 $urls[$treffer[1]] = $treffer[1];
 571                         }
 572                 }
 573
 574                 return $urls;
 575         }
 576
 577         /**
 578          * @param string $html
 579          * @param int    $wraplength Ensures individual lines aren't longer than this many characters. Doesn't break words.
 580          * @param bool   $compact    True: Completely strips image tags; False: Keeps image URLs
 581          * @return string
 582          */
 583         public static function toPlaintext(string $html, $wraplength = 75, $compact = false)
 584         {
 585                 $message = str_replace("\r", "", $html);
 586
 587                 $doc = new DOMDocument();
 588                 $doc->preserveWhiteSpace = false;
 589
 590                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 591
 592                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD | LIBXML_NOBLANKS);
 593
 594                 $message = $doc->saveHTML();
 595                 // Remove eventual UTF-8 BOM
 596                 $message = str_replace("\xC3\x82\xC2\xA0", "", $message);
 597
 598                 // Collecting all links
 599                 $urls = self::collectURLs($message);
 600
 601                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD | LIBXML_NOBLANKS);
 602
 603                 self::tagToBBCode($doc, 'html', [], '', '');
 604                 self::tagToBBCode($doc, 'body', [], '', '');
 605
 606                 if ($compact) {
 607                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 608                 } else {
 609                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 610                 }
 611
 612                 self::tagToBBCode($doc, 'br', [], "\n", '');
 613
 614                 self::tagToBBCode($doc, 'span', [], "", "");
 615                 self::tagToBBCode($doc, 'pre', [], "", "");
 616                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 617                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 618
 619                 self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 620
 621                 self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
 622
 623                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 624                 self::tagToBBCode($doc, 'td', [], "\t", "");
 625
 626                 self::tagToBBCode($doc, 'h1', [], "\n\n*", "*\n");
 627                 self::tagToBBCode($doc, 'h2', [], "\n\n*", "*\n");
 628                 self::tagToBBCode($doc, 'h3', [], "\n\n*", "*\n");
 629                 self::tagToBBCode($doc, 'h4', [], "\n\n*", "*\n");
 630                 self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
 631                 self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 632
 633                 if (!$compact) {
 634                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
 635                 } else {
 636                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' ', ' ');
 637                 }
 638
 639                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], ' $1 ', '');
 640
 641                 $message = $doc->saveHTML();
 642
 643                 if (!$compact) {
 644                         $message = str_replace("[img]", "", $message);
 645                         $message = str_replace("[/img]", "", $message);
 646                 }
 647
 648                 // was ersetze ich da?
 649                 // Irgendein stoerrisches UTF-Zeug
 650                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 651
 652                 $message = str_replace("&nbsp;", " ", $message);
 653
 654                 // Aufeinanderfolgende DIVs
 655                 $message = preg_replace('=\r *\r=i', "\n", $message);
 656                 $message = str_replace("\r", "\n", $message);
 657
 658                 $message = strip_tags($message);
 659
 660                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 661
 662                 if (!$compact && ($message != '')) {
 663                         foreach ($urls as $id => $url) {
 664                                 if ($url != '' && strpos($message, $url) === false) {
 665                                         $message .= "\n" . $url . ' ';
 666                                 }
 667                         }
 668                 }
 669
 670                 $message = str_replace("\n«", "«\n", $message);
 671                 $message = str_replace("»\n", "\n»", $message);
 672
 673                 do {
 674                         $oldmessage = $message;
 675                         $message = str_replace("\n\n\n", "\n\n", $message);
 676                 } while ($oldmessage != $message);
 677
 678                 $message = self::quoteLevel(trim($message), $wraplength);
 679
 680                 return trim($message);
 681         }
 682
 683         /**
 684          * Converts provided HTML code to Markdown. The hardwrap parameter maximizes
 685          * compatibility with Diaspora in spite of the Markdown standards.
 686          *
 687          * @param string $html
 688          * @return string
 689          */
 690         public static function toMarkdown($html)
 691         {
 692                 $converter = new HtmlConverter(['hard_break' => true]);
 693                 $markdown = $converter->convert($html);
 694
 695                 return $markdown;
 696         }
 697
 698         /**
 699          * @brief Convert video HTML to BBCode tags
 700          *
 701          * @param string $s
 702          * @return string
 703          */
 704         public static function toBBCodeVideo($s)
 705         {
 706                 $s = preg_replace(
 707                         '#<object[^>]+>(.*?)https?://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+)(.*?)</object>#ism',
 708                         '[youtube]$2[/youtube]',
 709                         $s
 710                 );
 711
 712                 $s = preg_replace(
 713                         '#<iframe[^>](.*?)https?://www.youtube.com/embed/([A-Za-z0-9\-_=]+)(.*?)</iframe>#ism',
 714                         '[youtube]$2[/youtube]',
 715                         $s
 716                 );
 717
 718                 $s = preg_replace(
 719                         '#<iframe[^>](.*?)https?://player.vimeo.com/video/([0-9]+)(.*?)</iframe>#ism',
 720                         '[vimeo]$2[/vimeo]',
 721                         $s
 722                 );
 723
 724                 return $s;
 725         }
 726
 727         /**
 728          * transform link href and img src from relative to absolute
 729          *
 730          * @param string $text
 731          * @param string $base base url
 732          * @return string
 733          */
 734         public static function relToAbs($text, $base)
 735         {
 736                 if (empty($base)) {
 737                         return $text;
 738                 }
 739
 740                 $base = rtrim($base, '/');
 741
 742                 $base2 = $base . "/";
 743
 744                 // Replace links
 745                 $pattern = "/<a([^>]*) href=\"(?!http|https|\/)([^\"]*)\"/";
 746                 $replace = "<a\${1} href=\"" . $base2 . "\${2}\"";
 747                 $text = preg_replace($pattern, $replace, $text);
 748
 749                 $pattern = "/<a([^>]*) href=\"(?!http|https)([^\"]*)\"/";
 750                 $replace = "<a\${1} href=\"" . $base . "\${2}\"";
 751                 $text = preg_replace($pattern, $replace, $text);
 752
 753                 // Replace images
 754                 $pattern = "/<img([^>]*) src=\"(?!http|https|\/)([^\"]*)\"/";
 755                 $replace = "<img\${1} src=\"" . $base2 . "\${2}\"";
 756                 $text = preg_replace($pattern, $replace, $text);
 757
 758                 $pattern = "/<img([^>]*) src=\"(?!http|https)([^\"]*)\"/";
 759                 $replace = "<img\${1} src=\"" . $base . "\${2}\"";
 760                 $text = preg_replace($pattern, $replace, $text);
 761
 762
 763                 // Done
 764                 return $text;
 765         }
 766
 767         /**
 768          * return div element with class 'clear'
 769          * @return string
 770          * @deprecated
 771          */
 772         public static function clearDiv()
 773         {
 774                 return '<div class="clear"></div>';
 775         }
 776
 777         /**
 778          * Loader for infinite scrolling
 779          *
 780          * @return string html for loader
 781          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 782          */
 783         public static function scrollLoader()
 784         {
 785                 $tpl = Renderer::getMarkupTemplate("scroll_loader.tpl");
 786                 return Renderer::replaceMacros($tpl, [
 787                         'wait' => L10n::t('Loading more entries...'),
 788                         'end' => L10n::t('The end')
 789                 ]);
 790         }
 791
 792         /**
 793          * Get html for contact block.
 794          *
 795          * @deprecated since version 2019.03
 796          * @see ContactBlock::getHTML()
 797          * @return string
 798          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 799          * @throws \ImagickException
 800          */
 801         public static function contactBlock()
 802         {
 803                 $a = \get_app();
 804
 805                 return ContactBlock::getHTML($a->profile);
 806         }
 807
 808         /**
 809          * @brief Format contacts as picture links or as text links
 810          *
 811          * @param array   $contact  Array with contacts which contains an array with
 812          *                          int 'id' => The ID of the contact
 813          *                          int 'uid' => The user ID of the user who owns this data
 814          *                          string 'name' => The name of the contact
 815          *                          string 'url' => The url to the profile page of the contact
 816          *                          string 'addr' => The webbie of the contact (e.g.) username@friendica.com
 817          *                          string 'network' => The network to which the contact belongs to
 818          *                          string 'thumb' => The contact picture
 819          *                          string 'click' => js code which is performed when clicking on the contact
 820          * @param boolean $redirect If true try to use the redir url if it's possible
 821          * @param string  $class    CSS class for the
 822          * @param boolean $textmode If true display the contacts as text links
 823          *                          if false display the contacts as picture links
 824          * @return string Formatted html
 825          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 826          * @throws \ImagickException
 827          */
 828         public static function micropro($contact, $redirect = false, $class = '', $textmode = false)
 829         {
 830                 // Use the contact URL if no address is available
 831                 if (empty($contact['addr'])) {
 832                         $contact["addr"] = $contact["url"];
 833                 }
 834
 835                 $url = $contact['url'];
 836                 $sparkle = '';
 837                 $redir = false;
 838
 839                 if ($redirect) {
 840                         $url = Contact::magicLink($contact['url']);
 841                         if (strpos($url, 'redir/') === 0) {
 842                                 $sparkle = ' sparkle';
 843                         }
 844                 }
 845
 846                 // If there is some js available we don't need the url
 847                 if (!empty($contact['click'])) {
 848                         $url = '';
 849                 }
 850
 851                 return Renderer::replaceMacros(Renderer::getMarkupTemplate(($textmode)?'micropro_txt.tpl':'micropro_img.tpl'), [
 852                         '$click' => defaults($contact, 'click', ''),
 853                         '$class' => $class,
 854                         '$url' => $url,
 855                         '$photo' => ProxyUtils::proxifyUrl($contact['thumb'], false, ProxyUtils::SIZE_THUMB),
 856                         '$name' => $contact['name'],
 857                         'title' => $contact['name'] . ' [' . $contact['addr'] . ']',
 858                         '$parkle' => $sparkle,
 859                         '$redir' => $redir
 860                 ]);
 861         }
 862
 863         /**
 864          * Search box.
 865          *
 866          * @param string $s     Search query.
 867          * @param string $id    HTML id
 868          * @param string $url   Search url.
 869          * @param bool   $aside Display the search widgit aside.
 870          *
 871          * @return string Formatted HTML.
 872          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 873          */
 874         public static function search($s, $id = 'search-box', $url = 'search', $aside = true)
 875         {
 876                 $mode = 'text';
 877
 878                 if (strpos($s, '#') === 0) {
 879                         $mode = 'tag';
 880                 }
 881                 $save_label = $mode === 'text' ? L10n::t('Save') : L10n::t('Follow');
 882
 883                 $values = [
 884                                 '$s' => $s,
 885                                 '$id' => $id,
 886                                 '$action_url' => $url,
 887                                 '$search_label' => L10n::t('Search'),
 888                                 '$save_label' => $save_label,
 889                                 '$savedsearch' => 'savedsearch',
 890                                 '$search_hint' => L10n::t('@name, !forum, #tags, content'),
 891                                 '$mode' => $mode
 892                         ];
 893
 894                 if (!$aside) {
 895                         $values['$searchoption'] = [
 896                                                 L10n::t("Full Text"),
 897                                                 L10n::t("Tags"),
 898                                                 L10n::t("Contacts")];
 899
 900                         if (Config::get('system', 'poco_local_search')) {
 901                                 $values['$searchoption'][] = L10n::t("Forums");
 902                         }
 903                 }
 904
 905                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 906         }
 907
 908         /**
 909          * Replace naked text hyperlink with HTML formatted hyperlink
 910          *
 911          * @param string $s
 912          * @return string
 913          */
 914         public static function toLink($s)
 915         {
 916                 $s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank">$1</a>', $s);
 917                 $s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
 918                 return $s;
 919         }
 920
 921         /**
 922          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
 923          *
 924          * Reasons are expected to have been translated already.
 925          *
 926          * @param string $html
 927          * @param array  $reasons
 928          * @return string
 929          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 930          */
 931         public static function applyContentFilter($html, array $reasons)
 932         {
 933                 if (count($reasons)) {
 934                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
 935                         $html = Renderer::replaceMacros($tpl, [
 936                                 '$reasons'   => $reasons,
 937                                 '$rnd'       => Strings::getRandomHex(8),
 938                                 '$openclose' => L10n::t('Click to open/close'),
 939                                 '$html'      => $html
 940                         ]);
 941                 }
 942
 943                 return $html;
 944         }
 945
 946         /**
 947          * replace html amp entity with amp char
 948          * @param string $s
 949          * @return string
 950          */
 951         public static function unamp($s)
 952         {
 953                 return str_replace('&amp;', '&', $s);
 954         }
 955 }