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 $key => $child) {
 104                                                 /* Remove empty text nodes at the start or at the end of the children list */
 105                                                 if ($key > 0 && $key < $node->childNodes->length - 1 || $child->nodeName != '#text' || trim($child->nodeValue)) {
 106                                                         $newNode = $child->cloneNode(true);
 107                                                         $node->parentNode->insertBefore($newNode, $node);
 108                                                 }
 109                                         }
 110                                 }
 111
 112                                 $node->parentNode->insertBefore($EndCode, $node);
 113                                 $node->parentNode->removeChild($node);
 114                         }
 115                 }
 116
 117                 return $replace;
 118         }
 119
 120         /**
 121          * Made by: ike@piratenpartei.de
 122          * Originally made for the syncom project: http://wiki.piratenpartei.de/Syncom
 123          *                    https://github.com/annando/Syncom
 124          *
 125          * @brief Converter for HTML to BBCode
 126          * @param string $message
 127          * @param string $basepath
 128          * @return string
 129          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 130          */
 131         public static function toBBCode($message, $basepath = '')
 132         {
 133                 $message = str_replace("\r", "", $message);
 134
 135                 // Removing code blocks before the whitespace removal processing below
 136                 $codeblocks = [];
 137                 $message = preg_replace_callback(
 138                         '#<pre><code(?: class="language-([^"]*)")?>(.*)</code></pre>#iUs',
 139                         function ($matches) use (&$codeblocks) {
 140                                 $return = '[codeblock-' . count($codeblocks) . ']';
 141
 142                                 $prefix = '[code]';
 143                                 if ($matches[1] != '') {
 144                                         $prefix = '[code=' . $matches[1] . ']';
 145                                 }
 146
 147                                 $codeblocks[] = $prefix . PHP_EOL . trim($matches[2]) . PHP_EOL . '[/code]';
 148                                 return $return;
 149                         },
 150                         $message
 151                 );
 152
 153                 $message = str_replace(
 154                         [
 155                                 "<li><p>",
 156                                 "</p></li>",
 157                         ],
 158                         [
 159                                 "<li>",
 160                                 "</li>",
 161                         ],
 162                         $message
 163                 );
 164
 165                 // remove namespaces
 166                 $message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
 167                 $message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
 168
 169                 $doc = new DOMDocument();
 170                 $doc->preserveWhiteSpace = false;
 171
 172                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 173
 174                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 175
 176                 XML::deleteNode($doc, 'style');
 177                 XML::deleteNode($doc, 'head');
 178                 XML::deleteNode($doc, 'title');
 179                 XML::deleteNode($doc, 'meta');
 180                 XML::deleteNode($doc, 'xml');
 181                 XML::deleteNode($doc, 'removeme');
 182
 183                 $xpath = new DomXPath($doc);
 184                 $list = $xpath->query("//pre");
 185                 foreach ($list as $node) {
 186                         // Ensure to escape unescaped & - they will otherwise raise a warning
 187                         $safe_value = preg_replace('/&(?!\w+;)/', '&amp;', $node->nodeValue);
 188                         $node->nodeValue = str_replace("\n", "\r", $safe_value);
 189                 }
 190
 191                 $message = $doc->saveHTML();
 192                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br />", " ", ""], $message);
 193                 $message = preg_replace('= [\s]*=i', " ", $message);
 194
 195                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 196
 197                 self::tagToBBCode($doc, 'html', [], "", "");
 198                 self::tagToBBCode($doc, 'body', [], "", "");
 199
 200                 // Outlook-Quote - Variant 1
 201                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal', 'style' => 'margin-left:35.4pt'], '[quote]', '[/quote]');
 202
 203                 // Outlook-Quote - Variant 2
 204                 self::tagToBBCode(
 205                         $doc,
 206                         'div',
 207                         ['style' => 'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'],
 208                         '[quote]',
 209                         '[/quote]'
 210                 );
 211
 212                 // MyBB-Stuff
 213                 self::tagToBBCode($doc, 'span', ['style' => 'text-decoration: underline;'], '[u]', '[/u]');
 214                 self::tagToBBCode($doc, 'span', ['style' => 'font-style: italic;'], '[i]', '[/i]');
 215                 self::tagToBBCode($doc, 'span', ['style' => 'font-weight: bold;'], '[b]', '[/b]');
 216
 217                 /* self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
 218                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
 219                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
 220                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
 221                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
 222                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
 223                   self::node2BBCode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
 224                  */
 225                 // Untested
 226                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
 227                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
 228                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
 229
 230                 self::tagToBBCode($doc, 'span', ['style' => '/.*color:\s*(.+?)[,;].*/'], '[color="$1"]', '[/color]');
 231
 232                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 233                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
 234                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
 235                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 236                 // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
 237                 // Test
 238                 //self::node2BBCode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
 239                 self::tagToBBCode($doc, 'span', ['class' => 'type-link'], '[class=type-link]', '[/class]');
 240                 self::tagToBBCode($doc, 'span', ['class' => 'type-video'], '[class=type-video]', '[/class]');
 241
 242                 self::tagToBBCode($doc, 'strong', [], '[b]', '[/b]');
 243                 self::tagToBBCode($doc, 'em', [], '[i]', '[/i]');
 244                 self::tagToBBCode($doc, 'b', [], '[b]', '[/b]');
 245                 self::tagToBBCode($doc, 'i', [], '[i]', '[/i]');
 246                 self::tagToBBCode($doc, 'u', [], '[u]', '[/u]');
 247                 self::tagToBBCode($doc, 's', [], '[s]', '[/s]');
 248                 self::tagToBBCode($doc, 'del', [], '[s]', '[/s]');
 249                 self::tagToBBCode($doc, 'strike', [], '[s]', '[/s]');
 250
 251                 self::tagToBBCode($doc, 'big', [], "[size=large]", "[/size]");
 252                 self::tagToBBCode($doc, 'small', [], "[size=small]", "[/size]");
 253
 254                 self::tagToBBCode($doc, 'blockquote', [], '[quote]', '[/quote]');
 255
 256                 self::tagToBBCode($doc, 'br', [], "\n", '');
 257
 258                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal'], "\n", "");
 259                 self::tagToBBCode($doc, 'div', ['class' => 'MsoNormal'], "\r", "");
 260
 261                 self::tagToBBCode($doc, 'span', [], "", "");
 262
 263                 self::tagToBBCode($doc, 'span', [], "", "");
 264                 self::tagToBBCode($doc, 'pre', [], "", "");
 265
 266                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 267                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 268
 269                 self::tagToBBCode($doc, 'ul', [], "[list]", "[/list]");
 270                 self::tagToBBCode($doc, 'ol', [], "[list=1]", "[/list]");
 271                 self::tagToBBCode($doc, 'li', [], "[*]", "");
 272
 273                 self::tagToBBCode($doc, 'hr', [], "[hr]", "");
 274
 275                 self::tagToBBCode($doc, 'table', [], "", "");
 276                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 277                 self::tagToBBCode($doc, 'td', [], "\t", "");
 278                 //self::node2BBCode($doc, 'table', array(), "[table]", "[/table]");
 279                 //self::node2BBCode($doc, 'th', array(), "[th]", "[/th]");
 280                 //self::node2BBCode($doc, 'tr', array(), "[tr]", "[/tr]");
 281                 //self::node2BBCode($doc, 'td', array(), "[td]", "[/td]");
 282                 //self::node2BBCode($doc, 'h1', array(), "\n\n[size=xx-large][b]", "[/b][/size]\n");
 283                 //self::node2BBCode($doc, 'h2', array(), "\n\n[size=x-large][b]", "[/b][/size]\n");
 284                 //self::node2BBCode($doc, 'h3', array(), "\n\n[size=large][b]", "[/b][/size]\n");
 285                 //self::node2BBCode($doc, 'h4', array(), "\n\n[size=medium][b]", "[/b][/size]\n");
 286                 //self::node2BBCode($doc, 'h5', array(), "\n\n[size=small][b]", "[/b][/size]\n");
 287                 //self::node2BBCode($doc, 'h6', array(), "\n\n[size=x-small][b]", "[/b][/size]\n");
 288
 289                 self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 290                 self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 291                 self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 292                 self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 293                 self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 294                 self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 295
 296                 self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 297                 self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 298
 299                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'alt' => '/(.+)/'], '[img=$1]$2', '[/img]');
 300                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]');
 301                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]');
 302
 303
 304                 self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]');
 305                 self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]');
 306                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[iframe]$1', '[/iframe]');
 307
 308                 self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 309                 self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 310
 311                 $message = $doc->saveHTML();
 312
 313                 // I'm removing something really disturbing
 314                 // Don't know exactly what it is
 315                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 316
 317                 $message = str_replace("&nbsp;", " ", $message);
 318
 319                 // removing multiple DIVs
 320                 $message = preg_replace('=\r *\r=i', "\n", $message);
 321                 $message = str_replace("\r", "\n", $message);
 322
 323                 Hook::callAll('html2bbcode', $message);
 324
 325                 $message = strip_tags($message);
 326
 327                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 328
 329                 $message = str_replace(["<"], ["&lt;"], $message);
 330
 331                 // remove quotes if they don't make sense
 332                 $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 333
 334                 $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 335                 $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 336
 337                 do {
 338                         $oldmessage = $message;
 339                         $message = str_replace("\n \n", "\n\n", $message);
 340                 } while ($oldmessage != $message);
 341
 342                 do {
 343                         $oldmessage = $message;
 344                         $message = str_replace("\n\n\n", "\n\n", $message);
 345                 } while ($oldmessage != $message);
 346
 347                 do {
 348                         $oldmessage = $message;
 349                         $message = str_replace(
 350                                 [
 351                                 "[/size]\n\n",
 352                                 "\n[hr]",
 353                                 "[hr]\n",
 354                                 "\n[list",
 355                                 "[/list]\n",
 356                                 "\n[/",
 357                                 "[list]\n",
 358                                 "[list=1]\n",
 359                                 "\n[*]"],
 360                                 [
 361                                 "[/size]\n",
 362                                 "[hr]",
 363                                 "[hr]",
 364                                 "[list",
 365                                 "[/list]",
 366                                 "[/",
 367                                 "[list]",
 368                                 "[list=1]",
 369                                 "[*]"],
 370                                 $message
 371                         );
 372                 } while ($message != $oldmessage);
 373
 374                 $message = str_replace(
 375                         ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 376                         ['[b]', '[/b]', '[i]', '[/i]'],
 377                         $message
 378                 );
 379
 380                 // Handling Yahoo style of mails
 381                 $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 382
 383                 // Restore code blocks
 384                 $message = preg_replace_callback(
 385                         '#\[codeblock-([0-9]+)\]#iU',
 386                         function ($matches) use ($codeblocks) {
 387                                 $return = '';
 388                                 if (isset($codeblocks[intval($matches[1])])) {
 389                                         $return = $codeblocks[$matches[1]];
 390                                 }
 391                                 return $return;
 392                         },
 393                         $message
 394                 );
 395
 396                 $message = trim($message);
 397
 398                 if ($basepath != '') {
 399                         $message = self::qualifyURLs($message, $basepath);
 400                 }
 401
 402                 return $message;
 403         }
 404
 405         /**
 406          * @brief Sub function to complete incomplete URL
 407          *
 408          * @param array  $matches  Result of preg_replace_callback
 409          * @param string $basepath Basepath that is used to complete the URL
 410          *
 411          * @return string The expanded URL
 412          */
 413         private static function qualifyURLsSub($matches, $basepath)
 414         {
 415                 $base = parse_url($basepath);
 416                 unset($base['query']);
 417                 unset($base['fragment']);
 418
 419                 $link = $matches[0];
 420                 $url = $matches[1];
 421
 422                 if (empty($url) || empty(parse_url($url))) {
 423                         return $matches[0];
 424                 }
 425
 426                 $parts = array_merge($base, parse_url($url));
 427                 $url2 = Network::unparseURL($parts);
 428
 429                 return str_replace($url, $url2, $link);
 430         }
 431
 432         /**
 433          * @brief Complete incomplete URLs in BBCode
 434          *
 435          * @param string $body     Body with URLs
 436          * @param string $basepath Base path that is used to complete the URL
 437          *
 438          * @return string Body with expanded URLs
 439          */
 440         private static function qualifyURLs($body, $basepath)
 441         {
 442                 $URLSearchString = "^\[\]";
 443
 444                 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
 445                         "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
 446                         "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 447                         "/\[img\](.*?)\[\/img\]/ism",
 448                         "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 449                         "/\[zmg\](.*?)\[\/zmg\]/ism",
 450                         "/\[video\](.*?)\[\/video\]/ism",
 451                         "/\[audio\](.*?)\[\/audio\]/ism",
 452                 ];
 453
 454                 foreach ($matches as $match) {
 455                         $body = preg_replace_callback(
 456                                 $match,
 457                                 function ($match) use ($basepath) {
 458                                         return self::qualifyURLsSub($match, $basepath);
 459                                 },
 460                                 $body
 461                         );
 462                 }
 463                 return $body;
 464         }
 465
 466         private static function breakLines($line, $level, $wraplength = 75)
 467         {
 468                 if ($wraplength == 0) {
 469                         $wraplength = 2000000;
 470                 }
 471
 472                 $wraplen = $wraplength - $level;
 473
 474                 $newlines = [];
 475
 476                 do {
 477                         $oldline = $line;
 478
 479                         $subline = substr($line, 0, $wraplen);
 480
 481                         $pos = strrpos($subline, ' ');
 482
 483                         if ($pos == 0) {
 484                                 $pos = strpos($line, ' ');
 485                         }
 486
 487                         if (($pos > 0) && strlen($line) > $wraplen) {
 488                                 $newline = trim(substr($line, 0, $pos));
 489                                 if ($level > 0) {
 490                                         $newline = str_repeat(">", $level) . ' ' . $newline;
 491                                 }
 492
 493                                 $newlines[] = $newline . " ";
 494                                 $line = substr($line, $pos + 1);
 495                         }
 496                 } while ((strlen($line) > $wraplen) && !($oldline == $line));
 497
 498                 if ($level > 0) {
 499                         $line = str_repeat(">", $level) . ' ' . $line;
 500                 }
 501
 502                 $newlines[] = $line;
 503
 504                 return implode($newlines, "\n");
 505         }
 506
 507         private static function quoteLevel($message, $wraplength = 75)
 508         {
 509                 $lines = explode("\n", $message);
 510
 511                 $newlines = [];
 512                 $level = 0;
 513                 foreach ($lines as $line) {
 514                         $line = trim($line);
 515                         $startquote = false;
 516                         while (strpos("*" . $line, '[quote]') > 0) {
 517                                 $level++;
 518                                 $pos = strpos($line, '[quote]');
 519                                 $line = substr($line, 0, $pos) . substr($line, $pos + 7);
 520                                 $startquote = true;
 521                         }
 522
 523                         $currlevel = $level;
 524
 525                         while (strpos("*" . $line, '[/quote]') > 0) {
 526                                 $level--;
 527                                 if ($level < 0) {
 528                                         $level = 0;
 529                                 }
 530
 531                                 $pos = strpos($line, '[/quote]');
 532                                 $line = substr($line, 0, $pos) . substr($line, $pos + 8);
 533                         }
 534
 535                         if (!$startquote || ($line != '')) {
 536                                 $newlines[] = self::breakLines($line, $currlevel, $wraplength);
 537                         }
 538                 }
 539
 540                 return implode($newlines, "\n");
 541         }
 542
 543         private static function collectURLs($message)
 544         {
 545                 $pattern = '/<a.*?href="(.*?)".*?>(.*?)<\/a>/is';
 546                 preg_match_all($pattern, $message, $result, PREG_SET_ORDER);
 547
 548                 $urls = [];
 549                 foreach ($result as $treffer) {
 550                         $ignore = false;
 551
 552                         // A list of some links that should be ignored
 553                         $list = ["/user/", "/tag/", "/group/", "/profile/", "/search?search=", "/search?tag=", "mailto:", "/u/", "/node/",
 554                                 "//plus.google.com/", "//twitter.com/"];
 555                         foreach ($list as $listitem) {
 556                                 if (strpos($treffer[1], $listitem) !== false) {
 557                                         $ignore = true;
 558                                 }
 559                         }
 560
 561                         if ((strpos($treffer[1], "//twitter.com/") !== false) && (strpos($treffer[1], "/status/") !== false)) {
 562                                 $ignore = false;
 563                         }
 564
 565                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/posts") !== false)) {
 566                                 $ignore = false;
 567                         }
 568
 569                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/photos") !== false)) {
 570                                 $ignore = false;
 571                         }
 572
 573                         $ignore = $ignore || strpos($treffer[1], '#') === 0;
 574
 575                         if (!$ignore) {
 576                                 $urls[$treffer[1]] = $treffer[1];
 577                         }
 578                 }
 579
 580                 return $urls;
 581         }
 582
 583         /**
 584          * @param string $html
 585          * @param int    $wraplength Ensures individual lines aren't longer than this many characters. Doesn't break words.
 586          * @param bool   $compact    True: Completely strips image tags; False: Keeps image URLs
 587          * @return string
 588          */
 589         public static function toPlaintext(string $html, $wraplength = 75, $compact = false)
 590         {
 591                 $message = str_replace("\r", "", $html);
 592
 593                 $doc = new DOMDocument();
 594                 $doc->preserveWhiteSpace = false;
 595
 596                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 597
 598                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 599
 600                 $message = $doc->saveHTML();
 601                 // Remove eventual UTF-8 BOM
 602                 $message = str_replace("\xC3\x82\xC2\xA0", "", $message);
 603
 604                 // Collecting all links
 605                 $urls = self::collectURLs($message);
 606
 607                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 608
 609                 self::tagToBBCode($doc, 'html', [], '', '');
 610                 self::tagToBBCode($doc, 'body', [], '', '');
 611
 612                 if ($compact) {
 613                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 614                 } else {
 615                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 616                 }
 617
 618                 self::tagToBBCode($doc, 'br', [], "\n", '');
 619
 620                 self::tagToBBCode($doc, 'span', [], "", "");
 621                 self::tagToBBCode($doc, 'pre', [], "", "");
 622                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 623                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 624
 625                 self::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                 if (!$compact) {
 640                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
 641                 } else {
 642                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' ', ' ');
 643                 }
 644
 645                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], ' $1 ', '');
 646
 647                 $message = $doc->saveHTML();
 648
 649                 if (!$compact) {
 650                         $message = str_replace("[img]", "", $message);
 651                         $message = str_replace("[/img]", "", $message);
 652                 }
 653
 654                 // was ersetze ich da?
 655                 // Irgendein stoerrisches UTF-Zeug
 656                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 657
 658                 $message = str_replace("&nbsp;", " ", $message);
 659
 660                 // Aufeinanderfolgende DIVs
 661                 $message = preg_replace('=\r *\r=i', "\n", $message);
 662                 $message = str_replace("\r", "\n", $message);
 663
 664                 $message = strip_tags($message);
 665
 666                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 667
 668                 if (!$compact && ($message != '')) {
 669                         foreach ($urls as $id => $url) {
 670                                 if ($url != '' && strpos($message, $url) === false) {
 671                                         $message .= "\n" . $url . ' ';
 672                                 }
 673                         }
 674                 }
 675
 676                 $message = str_replace("\n«", "«\n", $message);
 677                 $message = str_replace("»\n", "\n»", $message);
 678
 679                 do {
 680                         $oldmessage = $message;
 681                         $message = str_replace("\n\n\n", "\n\n", $message);
 682                 } while ($oldmessage != $message);
 683
 684                 $message = self::quoteLevel(trim($message), $wraplength);
 685
 686                 return trim($message);
 687         }
 688
 689         /**
 690          * Converts provided HTML code to Markdown. The hardwrap parameter maximizes
 691          * compatibility with Diaspora in spite of the Markdown standards.
 692          *
 693          * @param string $html
 694          * @return string
 695          */
 696         public static function toMarkdown($html)
 697         {
 698                 $converter = new HtmlConverter(['hard_break' => true]);
 699                 $markdown = $converter->convert($html);
 700
 701                 return $markdown;
 702         }
 703
 704         /**
 705          * @brief Convert video HTML to BBCode tags
 706          *
 707          * @param string $s
 708          * @return string
 709          */
 710         public static function toBBCodeVideo($s)
 711         {
 712                 $s = preg_replace(
 713                         '#<object[^>]+>(.*?)https?://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+)(.*?)</object>#ism',
 714                         '[youtube]$2[/youtube]',
 715                         $s
 716                 );
 717
 718                 $s = preg_replace(
 719                         '#<iframe[^>](.*?)https?://www.youtube.com/embed/([A-Za-z0-9\-_=]+)(.*?)</iframe>#ism',
 720                         '[youtube]$2[/youtube]',
 721                         $s
 722                 );
 723
 724                 $s = preg_replace(
 725                         '#<iframe[^>](.*?)https?://player.vimeo.com/video/([0-9]+)(.*?)</iframe>#ism',
 726                         '[vimeo]$2[/vimeo]',
 727                         $s
 728                 );
 729
 730                 return $s;
 731         }
 732
 733         /**
 734          * transform link href and img src from relative to absolute
 735          *
 736          * @param string $text
 737          * @param string $base base url
 738          * @return string
 739          */
 740         public static function relToAbs($text, $base)
 741         {
 742                 if (empty($base)) {
 743                         return $text;
 744                 }
 745
 746                 $base = rtrim($base, '/');
 747
 748                 $base2 = $base . "/";
 749
 750                 // Replace links
 751                 $pattern = "/<a([^>]*) href=\"(?!http|https|\/)([^\"]*)\"/";
 752                 $replace = "<a\${1} href=\"" . $base2 . "\${2}\"";
 753                 $text = preg_replace($pattern, $replace, $text);
 754
 755                 $pattern = "/<a([^>]*) href=\"(?!http|https)([^\"]*)\"/";
 756                 $replace = "<a\${1} href=\"" . $base . "\${2}\"";
 757                 $text = preg_replace($pattern, $replace, $text);
 758
 759                 // Replace images
 760                 $pattern = "/<img([^>]*) src=\"(?!http|https|\/)([^\"]*)\"/";
 761                 $replace = "<img\${1} src=\"" . $base2 . "\${2}\"";
 762                 $text = preg_replace($pattern, $replace, $text);
 763
 764                 $pattern = "/<img([^>]*) src=\"(?!http|https)([^\"]*)\"/";
 765                 $replace = "<img\${1} src=\"" . $base . "\${2}\"";
 766                 $text = preg_replace($pattern, $replace, $text);
 767
 768
 769                 // Done
 770                 return $text;
 771         }
 772
 773         /**
 774          * return div element with class 'clear'
 775          * @return string
 776          * @deprecated
 777          */
 778         public static function clearDiv()
 779         {
 780                 return '<div class="clear"></div>';
 781         }
 782
 783         /**
 784          * Loader for infinite scrolling
 785          *
 786          * @return string html for loader
 787          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 788          */
 789         public static function scrollLoader()
 790         {
 791                 $tpl = Renderer::getMarkupTemplate("scroll_loader.tpl");
 792                 return Renderer::replaceMacros($tpl, [
 793                         'wait' => L10n::t('Loading more entries...'),
 794                         'end' => L10n::t('The end')
 795                 ]);
 796         }
 797
 798         /**
 799          * Get html for contact block.
 800          *
 801          * @deprecated since version 2019.03
 802          * @see ContactBlock::getHTML()
 803          * @return string
 804          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 805          * @throws \ImagickException
 806          */
 807         public static function contactBlock()
 808         {
 809                 $a = \get_app();
 810
 811                 return ContactBlock::getHTML($a->profile);
 812         }
 813
 814         /**
 815          * @brief Format contacts as picture links or as text links
 816          *
 817          * @param array   $contact  Array with contacts which contains an array with
 818          *                          int 'id' => The ID of the contact
 819          *                          int 'uid' => The user ID of the user who owns this data
 820          *                          string 'name' => The name of the contact
 821          *                          string 'url' => The url to the profile page of the contact
 822          *                          string 'addr' => The webbie of the contact (e.g.) username@friendica.com
 823          *                          string 'network' => The network to which the contact belongs to
 824          *                          string 'thumb' => The contact picture
 825          *                          string 'click' => js code which is performed when clicking on the contact
 826          * @param boolean $redirect If true try to use the redir url if it's possible
 827          * @param string  $class    CSS class for the
 828          * @param boolean $textmode If true display the contacts as text links
 829          *                          if false display the contacts as picture links
 830          * @return string Formatted html
 831          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 832          * @throws \ImagickException
 833          */
 834         public static function micropro($contact, $redirect = false, $class = '', $textmode = false)
 835         {
 836                 // Use the contact URL if no address is available
 837                 if (empty($contact['addr'])) {
 838                         $contact["addr"] = $contact["url"];
 839                 }
 840
 841                 $url = $contact['url'];
 842                 $sparkle = '';
 843                 $redir = false;
 844
 845                 if ($redirect) {
 846                         $url = Contact::magicLink($contact['url']);
 847                         if (strpos($url, 'redir/') === 0) {
 848                                 $sparkle = ' sparkle';
 849                         }
 850                 }
 851
 852                 // If there is some js available we don't need the url
 853                 if (!empty($contact['click'])) {
 854                         $url = '';
 855                 }
 856
 857                 return Renderer::replaceMacros(Renderer::getMarkupTemplate(($textmode)?'micropro_txt.tpl':'micropro_img.tpl'), [
 858                         '$click' => defaults($contact, 'click', ''),
 859                         '$class' => $class,
 860                         '$url' => $url,
 861                         '$photo' => ProxyUtils::proxifyUrl($contact['thumb'], false, ProxyUtils::SIZE_THUMB),
 862                         '$name' => $contact['name'],
 863                         'title' => $contact['name'] . ' [' . $contact['addr'] . ']',
 864                         '$parkle' => $sparkle,
 865                         '$redir' => $redir
 866                 ]);
 867         }
 868
 869         /**
 870          * Search box.
 871          *
 872          * @param string $s     Search query.
 873          * @param string $id    HTML id
 874          * @param string $url   Search url.
 875          * @param bool   $aside Display the search widgit aside.
 876          *
 877          * @return string Formatted HTML.
 878          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 879          */
 880         public static function search($s, $id = 'search-box', $url = 'search', $aside = true)
 881         {
 882                 $mode = 'text';
 883
 884                 if (strpos($s, '#') === 0) {
 885                         $mode = 'tag';
 886                 }
 887                 $save_label = $mode === 'text' ? L10n::t('Save') : L10n::t('Follow');
 888
 889                 $values = [
 890                                 '$s' => $s,
 891                                 '$id' => $id,
 892                                 '$action_url' => $url,
 893                                 '$search_label' => L10n::t('Search'),
 894                                 '$save_label' => $save_label,
 895                                 '$savedsearch' => 'savedsearch',
 896                                 '$search_hint' => L10n::t('@name, !forum, #tags, content'),
 897                                 '$mode' => $mode
 898                         ];
 899
 900                 if (!$aside) {
 901                         $values['$searchoption'] = [
 902                                                 L10n::t("Full Text"),
 903                                                 L10n::t("Tags"),
 904                                                 L10n::t("Contacts")];
 905
 906                         if (Config::get('system', 'poco_local_search')) {
 907                                 $values['$searchoption'][] = L10n::t("Forums");
 908                         }
 909                 }
 910
 911                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 912         }
 913
 914         /**
 915          * Replace naked text hyperlink with HTML formatted hyperlink
 916          *
 917          * @param string $s
 918          * @return string
 919          */
 920         public static function toLink($s)
 921         {
 922                 $s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank">$1</a>', $s);
 923                 $s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
 924                 return $s;
 925         }
 926
 927         /**
 928          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
 929          *
 930          * Reasons are expected to have been translated already.
 931          *
 932          * @param string $html
 933          * @param array  $reasons
 934          * @return string
 935          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 936          */
 937         public static function applyContentFilter($html, array $reasons)
 938         {
 939                 if (count($reasons)) {
 940                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
 941                         $html = Renderer::replaceMacros($tpl, [
 942                                 '$reasons'   => $reasons,
 943                                 '$rnd'       => Strings::getRandomHex(8),
 944                                 '$openclose' => L10n::t('Click to open/close'),
 945                                 '$html'      => $html
 946                         ]);
 947                 }
 948
 949                 return $html;
 950         }
 951
 952         /**
 953          * replace html amp entity with amp char
 954          * @param string $s
 955          * @return string
 956          */
 957         public static function unamp($s)
 958         {
 959                 return str_replace('&amp;', '&', $s);
 960         }
 961 }