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         /**
  46          * Search all instances of a specific HTML tag node in the provided DOM document and replaces them with BBCode text nodes.
  47          *
  48          * @see HTML::tagToBBCodeSub()
  49          */
  50         private static function tagToBBCode(DOMDocument $doc, string $tag, array $attributes, string $startbb, string $endbb, bool $ignoreChildren = false)
  51         {
  52                 do {
  53                         $done = self::tagToBBCodeSub($doc, $tag, $attributes, $startbb, $endbb, $ignoreChildren);
  54                 } while ($done);
  55         }
  56
  57         /**
  58          * Search the first specific HTML tag node in the provided DOM document and replaces it with BBCode text nodes.
  59          *
  60          * @param DOMDocument $doc
  61          * @param string      $tag            HTML tag name
  62          * @param array       $attributes     Array of attributes to match and optionally use the value from
  63          * @param string      $startbb        BBCode tag opening
  64          * @param string      $endbb          BBCode tag closing
  65          * @param bool        $ignoreChildren If set to false, the HTML tag children will be appended as text inside the BBCode tag
  66          *                                    Otherwise, they will be entirely ignored. Useful for simple BBCode that draw their
  67          *                                    inner value from an attribute value and disregard the tag children.
  68          * @return bool Whether a replacement was done
  69          */
  70         private static function tagToBBCodeSub(DOMDocument $doc, string $tag, array $attributes, string $startbb, string $endbb, bool $ignoreChildren = false)
  71         {
  72                 $savestart = str_replace('$', '\x01', $startbb);
  73                 $replace = false;
  74
  75                 $xpath = new DOMXPath($doc);
  76
  77                 /** @var \DOMNode[] $list */
  78                 $list = $xpath->query("//" . $tag);
  79                 foreach ($list as $node) {
  80                         $attr = [];
  81                         if ($node->attributes->length) {
  82                                 foreach ($node->attributes as $attribute) {
  83                                         $attr[$attribute->name] = $attribute->value;
  84                                 }
  85                         }
  86
  87                         $replace = true;
  88
  89                         $startbb = $savestart;
  90
  91                         $i = 0;
  92
  93                         foreach ($attributes as $attribute => $value) {
  94                                 $startbb = str_replace('\x01' . ++$i, '$1', $startbb);
  95                                 if (strpos('*' . $startbb, '$1') > 0) {
  96                                         if ($replace && (@$attr[$attribute] != '')) {
  97                                                 $startbb = preg_replace($value, $startbb, $attr[$attribute], -1, $count);
  98
  99                                                 // If nothing could be changed
 100                                                 if ($count == 0) {
 101                                                         $replace = false;
 102                                                 }
 103                                         } else {
 104                                                 $replace = false;
 105                                         }
 106                                 } else {
 107                                         if (@$attr[$attribute] != $value) {
 108                                                 $replace = false;
 109                                         }
 110                                 }
 111                         }
 112
 113                         if ($replace) {
 114                                 $StartCode = $doc->createTextNode($startbb);
 115                                 $EndCode = $doc->createTextNode($endbb);
 116
 117                                 $node->parentNode->insertBefore($StartCode, $node);
 118
 119                                 if (!$ignoreChildren && $node->hasChildNodes()) {
 120                                         /** @var \DOMNode $child */
 121                                         foreach ($node->childNodes as $key => $child) {
 122                                                 /* Remove empty text nodes at the start or at the end of the children list */
 123                                                 if ($key > 0 && $key < $node->childNodes->length - 1 || $child->nodeName != '#text' || trim($child->nodeValue)) {
 124                                                         $newNode = $child->cloneNode(true);
 125                                                         $node->parentNode->insertBefore($newNode, $node);
 126                                                 }
 127                                         }
 128                                 }
 129
 130                                 $node->parentNode->insertBefore($EndCode, $node);
 131                                 $node->parentNode->removeChild($node);
 132                         }
 133                 }
 134
 135                 return $replace;
 136         }
 137
 138         /**
 139          * Made by: ike@piratenpartei.de
 140          * Originally made for the syncom project: http://wiki.piratenpartei.de/Syncom
 141          *                    https://github.com/annando/Syncom
 142          *
 143          * @brief Converter for HTML to BBCode
 144          * @param string $message
 145          * @param string $basepath
 146          * @return string
 147          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 148          */
 149         public static function toBBCode($message, $basepath = '')
 150         {
 151                 $message = str_replace("\r", "", $message);
 152
 153                 // Removing code blocks before the whitespace removal processing below
 154                 $codeblocks = [];
 155                 $message = preg_replace_callback(
 156                         '#<pre><code(?: class="language-([^"]*)")?>(.*)</code></pre>#iUs',
 157                         function ($matches) use (&$codeblocks) {
 158                                 $return = '[codeblock-' . count($codeblocks) . ']';
 159
 160                                 $prefix = '[code]';
 161                                 if ($matches[1] != '') {
 162                                         $prefix = '[code=' . $matches[1] . ']';
 163                                 }
 164
 165                                 $codeblocks[] = $prefix . PHP_EOL . trim($matches[2]) . PHP_EOL . '[/code]';
 166                                 return $return;
 167                         },
 168                         $message
 169                 );
 170
 171                 $message = str_replace(
 172                         [
 173                                 "<li><p>",
 174                                 "</p></li>",
 175                         ],
 176                         [
 177                                 "<li>",
 178                                 "</li>",
 179                         ],
 180                         $message
 181                 );
 182
 183                 // remove namespaces
 184                 $message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
 185                 $message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
 186
 187                 $doc = new DOMDocument();
 188                 $doc->preserveWhiteSpace = false;
 189
 190                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 191
 192                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 193
 194                 XML::deleteNode($doc, 'style');
 195                 XML::deleteNode($doc, 'head');
 196                 XML::deleteNode($doc, 'title');
 197                 XML::deleteNode($doc, 'meta');
 198                 XML::deleteNode($doc, 'xml');
 199                 XML::deleteNode($doc, 'removeme');
 200
 201                 $xpath = new DomXPath($doc);
 202                 $list = $xpath->query("//pre");
 203                 foreach ($list as $node) {
 204                         // Ensure to escape unescaped & - they will otherwise raise a warning
 205                         $safe_value = preg_replace('/&(?!\w+;)/', '&amp;', $node->nodeValue);
 206                         $node->nodeValue = str_replace("\n", "\r", $safe_value);
 207                 }
 208
 209                 $message = $doc->saveHTML();
 210                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br />", " ", ""], $message);
 211                 $message = preg_replace('= [\s]*=i', " ", $message);
 212
 213                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 214
 215                 self::tagToBBCode($doc, 'html', [], "", "");
 216                 self::tagToBBCode($doc, 'body', [], "", "");
 217
 218                 // Outlook-Quote - Variant 1
 219                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal', 'style' => 'margin-left:35.4pt'], '[quote]', '[/quote]');
 220
 221                 // Outlook-Quote - Variant 2
 222                 self::tagToBBCode(
 223                         $doc,
 224                         'div',
 225                         ['style' => 'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'],
 226                         '[quote]',
 227                         '[/quote]'
 228                 );
 229
 230                 // MyBB-Stuff
 231                 self::tagToBBCode($doc, 'span', ['style' => 'text-decoration: underline;'], '[u]', '[/u]');
 232                 self::tagToBBCode($doc, 'span', ['style' => 'font-style: italic;'], '[i]', '[/i]');
 233                 self::tagToBBCode($doc, 'span', ['style' => 'font-weight: bold;'], '[b]', '[/b]');
 234
 235                 /* self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
 236                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
 237                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
 238                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
 239                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
 240                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
 241                   self::node2BBCode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
 242                  */
 243                 // Untested
 244                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
 245                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
 246                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
 247
 248                 self::tagToBBCode($doc, 'span', ['style' => '/.*color:\s*(.+?)[,;].*/'], '[color="$1"]', '[/color]');
 249
 250                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 251                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
 252                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
 253                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 254                 // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
 255                 // Test
 256                 //self::node2BBCode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
 257                 self::tagToBBCode($doc, 'span', ['class' => 'type-link'], '[class=type-link]', '[/class]');
 258                 self::tagToBBCode($doc, 'span', ['class' => 'type-video'], '[class=type-video]', '[/class]');
 259
 260                 self::tagToBBCode($doc, 'strong', [], '[b]', '[/b]');
 261                 self::tagToBBCode($doc, 'em', [], '[i]', '[/i]');
 262                 self::tagToBBCode($doc, 'b', [], '[b]', '[/b]');
 263                 self::tagToBBCode($doc, 'i', [], '[i]', '[/i]');
 264                 self::tagToBBCode($doc, 'u', [], '[u]', '[/u]');
 265                 self::tagToBBCode($doc, 's', [], '[s]', '[/s]');
 266                 self::tagToBBCode($doc, 'del', [], '[s]', '[/s]');
 267                 self::tagToBBCode($doc, 'strike', [], '[s]', '[/s]');
 268
 269                 self::tagToBBCode($doc, 'big', [], "[size=large]", "[/size]");
 270                 self::tagToBBCode($doc, 'small', [], "[size=small]", "[/size]");
 271
 272                 self::tagToBBCode($doc, 'blockquote', [], '[quote]', '[/quote]');
 273
 274                 self::tagToBBCode($doc, 'br', [], "\n", '');
 275
 276                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal'], "\n", "");
 277                 self::tagToBBCode($doc, 'div', ['class' => 'MsoNormal'], "\r", "");
 278
 279                 self::tagToBBCode($doc, 'span', [], "", "");
 280
 281                 self::tagToBBCode($doc, 'span', [], "", "");
 282                 self::tagToBBCode($doc, 'pre', [], "", "");
 283
 284                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 285                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 286
 287                 self::tagToBBCode($doc, 'ul', [], "[list]", "[/list]");
 288                 self::tagToBBCode($doc, 'ol', [], "[list=1]", "[/list]");
 289                 self::tagToBBCode($doc, 'li', [], "[*]", "");
 290
 291                 self::tagToBBCode($doc, 'hr', [], "[hr]", "");
 292
 293                 self::tagToBBCode($doc, 'table', [], "[table]", "[/table]");
 294                 self::tagToBBCode($doc, 'th', [], "[th]", "[/th]");
 295                 self::tagToBBCode($doc, 'tr', [], "[tr]", "[/tr]");
 296                 self::tagToBBCode($doc, 'td', [], "[td]", "[/td]");
 297
 298                 self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 299                 self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 300                 self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 301                 self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 302                 self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 303                 self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 304
 305                 self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 306                 self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 307
 308                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'alt' => '/(.+)/'], '[img=$1]$2', '[/img]', true);
 309                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]', true);
 310                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]', true);
 311
 312
 313                 self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]', true);
 314                 self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]', true);
 315                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[iframe]$1', '[/iframe]', true);
 316
 317                 self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 318                 self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 319
 320                 $message = $doc->saveHTML();
 321
 322                 // I'm removing something really disturbing
 323                 // Don't know exactly what it is
 324                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 325
 326                 $message = str_replace("&nbsp;", " ", $message);
 327
 328                 // removing multiple DIVs
 329                 $message = preg_replace('=\r *\r=i', "\n", $message);
 330                 $message = str_replace("\r", "\n", $message);
 331
 332                 Hook::callAll('html2bbcode', $message);
 333
 334                 $message = strip_tags($message);
 335
 336                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 337
 338                 $message = str_replace(["<"], ["&lt;"], $message);
 339
 340                 // remove quotes if they don't make sense
 341                 $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 342
 343                 $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 344                 $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 345
 346                 do {
 347                         $oldmessage = $message;
 348                         $message = str_replace("\n \n", "\n\n", $message);
 349                 } while ($oldmessage != $message);
 350
 351                 do {
 352                         $oldmessage = $message;
 353                         $message = str_replace("\n\n\n", "\n\n", $message);
 354                 } while ($oldmessage != $message);
 355
 356                 do {
 357                         $oldmessage = $message;
 358                         $message = str_replace(
 359                                 [
 360                                 "[/size]\n\n",
 361                                 "\n[hr]",
 362                                 "[hr]\n",
 363                                 "\n[list",
 364                                 "[/list]\n",
 365                                 "\n[/",
 366                                 "[list]\n",
 367                                 "[list=1]\n",
 368                                 "\n[*]"],
 369                                 [
 370                                 "[/size]\n",
 371                                 "[hr]",
 372                                 "[hr]",
 373                                 "[list",
 374                                 "[/list]",
 375                                 "[/",
 376                                 "[list]",
 377                                 "[list=1]",
 378                                 "[*]"],
 379                                 $message
 380                         );
 381                 } while ($message != $oldmessage);
 382
 383                 $message = str_replace(
 384                         ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 385                         ['[b]', '[/b]', '[i]', '[/i]'],
 386                         $message
 387                 );
 388
 389                 // Handling Yahoo style of mails
 390                 $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 391
 392                 // Restore code blocks
 393                 $message = preg_replace_callback(
 394                         '#\[codeblock-([0-9]+)\]#iU',
 395                         function ($matches) use ($codeblocks) {
 396                                 $return = '';
 397                                 if (isset($codeblocks[intval($matches[1])])) {
 398                                         $return = $codeblocks[$matches[1]];
 399                                 }
 400                                 return $return;
 401                         },
 402                         $message
 403                 );
 404
 405                 $message = trim($message);
 406
 407                 if ($basepath != '') {
 408                         $message = self::qualifyURLs($message, $basepath);
 409                 }
 410
 411                 return $message;
 412         }
 413
 414         /**
 415          * @brief Sub function to complete incomplete URL
 416          *
 417          * @param array  $matches  Result of preg_replace_callback
 418          * @param string $basepath Basepath that is used to complete the URL
 419          *
 420          * @return string The expanded URL
 421          */
 422         private static function qualifyURLsSub($matches, $basepath)
 423         {
 424                 $base = parse_url($basepath);
 425                 unset($base['query']);
 426                 unset($base['fragment']);
 427
 428                 $link = $matches[0];
 429                 $url = $matches[1];
 430
 431                 if (empty($url) || empty(parse_url($url))) {
 432                         return $matches[0];
 433                 }
 434
 435                 $parts = array_merge($base, parse_url($url));
 436                 $url2 = Network::unparseURL($parts);
 437
 438                 return str_replace($url, $url2, $link);
 439         }
 440
 441         /**
 442          * @brief Complete incomplete URLs in BBCode
 443          *
 444          * @param string $body     Body with URLs
 445          * @param string $basepath Base path that is used to complete the URL
 446          *
 447          * @return string Body with expanded URLs
 448          */
 449         private static function qualifyURLs($body, $basepath)
 450         {
 451                 $URLSearchString = "^\[\]";
 452
 453                 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
 454                         "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
 455                         "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 456                         "/\[img\](.*?)\[\/img\]/ism",
 457                         "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 458                         "/\[zmg\](.*?)\[\/zmg\]/ism",
 459                         "/\[video\](.*?)\[\/video\]/ism",
 460                         "/\[audio\](.*?)\[\/audio\]/ism",
 461                 ];
 462
 463                 foreach ($matches as $match) {
 464                         $body = preg_replace_callback(
 465                                 $match,
 466                                 function ($match) use ($basepath) {
 467                                         return self::qualifyURLsSub($match, $basepath);
 468                                 },
 469                                 $body
 470                         );
 471                 }
 472                 return $body;
 473         }
 474
 475         private static function breakLines($line, $level, $wraplength = 75)
 476         {
 477                 if ($wraplength == 0) {
 478                         $wraplength = 2000000;
 479                 }
 480
 481                 $wraplen = $wraplength - $level;
 482
 483                 $newlines = [];
 484
 485                 do {
 486                         $oldline = $line;
 487
 488                         $subline = substr($line, 0, $wraplen);
 489
 490                         $pos = strrpos($subline, ' ');
 491
 492                         if ($pos == 0) {
 493                                 $pos = strpos($line, ' ');
 494                         }
 495
 496                         if (($pos > 0) && strlen($line) > $wraplen) {
 497                                 $newline = trim(substr($line, 0, $pos));
 498                                 if ($level > 0) {
 499                                         $newline = str_repeat(">", $level) . ' ' . $newline;
 500                                 }
 501
 502                                 $newlines[] = $newline . " ";
 503                                 $line = substr($line, $pos + 1);
 504                         }
 505                 } while ((strlen($line) > $wraplen) && !($oldline == $line));
 506
 507                 if ($level > 0) {
 508                         $line = str_repeat(">", $level) . ' ' . $line;
 509                 }
 510
 511                 $newlines[] = $line;
 512
 513                 return implode($newlines, "\n");
 514         }
 515
 516         private static function quoteLevel($message, $wraplength = 75)
 517         {
 518                 $lines = explode("\n", $message);
 519
 520                 $newlines = [];
 521                 $level = 0;
 522                 foreach ($lines as $line) {
 523                         $line = trim($line);
 524                         $startquote = false;
 525                         while (strpos("*" . $line, '[quote]') > 0) {
 526                                 $level++;
 527                                 $pos = strpos($line, '[quote]');
 528                                 $line = substr($line, 0, $pos) . substr($line, $pos + 7);
 529                                 $startquote = true;
 530                         }
 531
 532                         $currlevel = $level;
 533
 534                         while (strpos("*" . $line, '[/quote]') > 0) {
 535                                 $level--;
 536                                 if ($level < 0) {
 537                                         $level = 0;
 538                                 }
 539
 540                                 $pos = strpos($line, '[/quote]');
 541                                 $line = substr($line, 0, $pos) . substr($line, $pos + 8);
 542                         }
 543
 544                         if (!$startquote || ($line != '')) {
 545                                 $newlines[] = self::breakLines($line, $currlevel, $wraplength);
 546                         }
 547                 }
 548
 549                 return implode($newlines, "\n");
 550         }
 551
 552         private static function collectURLs($message)
 553         {
 554                 $pattern = '/<a.*?href="(.*?)".*?>(.*?)<\/a>/is';
 555                 preg_match_all($pattern, $message, $result, PREG_SET_ORDER);
 556
 557                 $urls = [];
 558                 foreach ($result as $treffer) {
 559                         $ignore = false;
 560
 561                         // A list of some links that should be ignored
 562                         $list = ["/user/", "/tag/", "/group/", "/profile/", "/search?search=", "/search?tag=", "mailto:", "/u/", "/node/",
 563                                 "//plus.google.com/", "//twitter.com/"];
 564                         foreach ($list as $listitem) {
 565                                 if (strpos($treffer[1], $listitem) !== false) {
 566                                         $ignore = true;
 567                                 }
 568                         }
 569
 570                         if ((strpos($treffer[1], "//twitter.com/") !== false) && (strpos($treffer[1], "/status/") !== false)) {
 571                                 $ignore = false;
 572                         }
 573
 574                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/posts") !== false)) {
 575                                 $ignore = false;
 576                         }
 577
 578                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/photos") !== false)) {
 579                                 $ignore = false;
 580                         }
 581
 582                         $ignore = $ignore || strpos($treffer[1], '#') === 0;
 583
 584                         if (!$ignore) {
 585                                 $urls[$treffer[1]] = $treffer[1];
 586                         }
 587                 }
 588
 589                 return $urls;
 590         }
 591
 592         /**
 593          * @param string $html
 594          * @param int    $wraplength Ensures individual lines aren't longer than this many characters. Doesn't break words.
 595          * @param bool   $compact    True: Completely strips image tags; False: Keeps image URLs
 596          * @return string
 597          */
 598         public static function toPlaintext(string $html, $wraplength = 75, $compact = false)
 599         {
 600                 $message = str_replace("\r", "", $html);
 601
 602                 $doc = new DOMDocument();
 603                 $doc->preserveWhiteSpace = false;
 604
 605                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 606
 607                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 608
 609                 $message = $doc->saveHTML();
 610                 // Remove eventual UTF-8 BOM
 611                 $message = str_replace("\xC3\x82\xC2\xA0", "", $message);
 612
 613                 // Collecting all links
 614                 $urls = self::collectURLs($message);
 615
 616                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 617
 618                 self::tagToBBCode($doc, 'html', [], '', '');
 619                 self::tagToBBCode($doc, 'body', [], '', '');
 620
 621                 if ($compact) {
 622                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 623                 } else {
 624                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 625                 }
 626
 627                 self::tagToBBCode($doc, 'br', [], "\n", '');
 628
 629                 self::tagToBBCode($doc, 'span', [], "", "");
 630                 self::tagToBBCode($doc, 'pre', [], "", "");
 631                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 632                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 633
 634                 self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 635
 636                 self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
 637
 638                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 639                 self::tagToBBCode($doc, 'td', [], "\t", "");
 640
 641                 self::tagToBBCode($doc, 'h1', [], "\n\n*", "*\n");
 642                 self::tagToBBCode($doc, 'h2', [], "\n\n*", "*\n");
 643                 self::tagToBBCode($doc, 'h3', [], "\n\n*", "*\n");
 644                 self::tagToBBCode($doc, 'h4', [], "\n\n*", "*\n");
 645                 self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
 646                 self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 647
 648                 if (!$compact) {
 649                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
 650                 } else {
 651                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' ', ' ');
 652                 }
 653
 654                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], ' $1 ', '');
 655
 656                 $message = $doc->saveHTML();
 657
 658                 if (!$compact) {
 659                         $message = str_replace("[img]", "", $message);
 660                         $message = str_replace("[/img]", "", $message);
 661                 }
 662
 663                 // was ersetze ich da?
 664                 // Irgendein stoerrisches UTF-Zeug
 665                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 666
 667                 $message = str_replace("&nbsp;", " ", $message);
 668
 669                 // Aufeinanderfolgende DIVs
 670                 $message = preg_replace('=\r *\r=i', "\n", $message);
 671                 $message = str_replace("\r", "\n", $message);
 672
 673                 $message = strip_tags($message);
 674
 675                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 676
 677                 if (!$compact && ($message != '')) {
 678                         foreach ($urls as $id => $url) {
 679                                 if ($url != '' && strpos($message, $url) === false) {
 680                                         $message .= "\n" . $url . ' ';
 681                                 }
 682                         }
 683                 }
 684
 685                 $message = str_replace("\n«", "«\n", $message);
 686                 $message = str_replace("»\n", "\n»", $message);
 687
 688                 do {
 689                         $oldmessage = $message;
 690                         $message = str_replace("\n\n\n", "\n\n", $message);
 691                 } while ($oldmessage != $message);
 692
 693                 $message = self::quoteLevel(trim($message), $wraplength);
 694
 695                 return trim($message);
 696         }
 697
 698         /**
 699          * Converts provided HTML code to Markdown. The hardwrap parameter maximizes
 700          * compatibility with Diaspora in spite of the Markdown standards.
 701          *
 702          * @param string $html
 703          * @return string
 704          */
 705         public static function toMarkdown($html)
 706         {
 707                 $converter = new HtmlConverter(['hard_break' => true]);
 708                 $markdown = $converter->convert($html);
 709
 710                 return $markdown;
 711         }
 712
 713         /**
 714          * @brief Convert video HTML to BBCode tags
 715          *
 716          * @param string $s
 717          * @return string
 718          */
 719         public static function toBBCodeVideo($s)
 720         {
 721                 $s = preg_replace(
 722                         '#<object[^>]+>(.*?)https?://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+)(.*?)</object>#ism',
 723                         '[youtube]$2[/youtube]',
 724                         $s
 725                 );
 726
 727                 $s = preg_replace(
 728                         '#<iframe[^>](.*?)https?://www.youtube.com/embed/([A-Za-z0-9\-_=]+)(.*?)</iframe>#ism',
 729                         '[youtube]$2[/youtube]',
 730                         $s
 731                 );
 732
 733                 $s = preg_replace(
 734                         '#<iframe[^>](.*?)https?://player.vimeo.com/video/([0-9]+)(.*?)</iframe>#ism',
 735                         '[vimeo]$2[/vimeo]',
 736                         $s
 737                 );
 738
 739                 return $s;
 740         }
 741
 742         /**
 743          * transform link href and img src from relative to absolute
 744          *
 745          * @param string $text
 746          * @param string $base base url
 747          * @return string
 748          */
 749         public static function relToAbs($text, $base)
 750         {
 751                 if (empty($base)) {
 752                         return $text;
 753                 }
 754
 755                 $base = rtrim($base, '/');
 756
 757                 $base2 = $base . "/";
 758
 759                 // Replace links
 760                 $pattern = "/<a([^>]*) href=\"(?!http|https|\/)([^\"]*)\"/";
 761                 $replace = "<a\${1} href=\"" . $base2 . "\${2}\"";
 762                 $text = preg_replace($pattern, $replace, $text);
 763
 764                 $pattern = "/<a([^>]*) href=\"(?!http|https)([^\"]*)\"/";
 765                 $replace = "<a\${1} href=\"" . $base . "\${2}\"";
 766                 $text = preg_replace($pattern, $replace, $text);
 767
 768                 // Replace images
 769                 $pattern = "/<img([^>]*) src=\"(?!http|https|\/)([^\"]*)\"/";
 770                 $replace = "<img\${1} src=\"" . $base2 . "\${2}\"";
 771                 $text = preg_replace($pattern, $replace, $text);
 772
 773                 $pattern = "/<img([^>]*) src=\"(?!http|https)([^\"]*)\"/";
 774                 $replace = "<img\${1} src=\"" . $base . "\${2}\"";
 775                 $text = preg_replace($pattern, $replace, $text);
 776
 777
 778                 // Done
 779                 return $text;
 780         }
 781
 782         /**
 783          * return div element with class 'clear'
 784          * @return string
 785          * @deprecated
 786          */
 787         public static function clearDiv()
 788         {
 789                 return '<div class="clear"></div>';
 790         }
 791
 792         /**
 793          * Loader for infinite scrolling
 794          *
 795          * @return string html for loader
 796          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 797          */
 798         public static function scrollLoader()
 799         {
 800                 $tpl = Renderer::getMarkupTemplate("scroll_loader.tpl");
 801                 return Renderer::replaceMacros($tpl, [
 802                         'wait' => L10n::t('Loading more entries...'),
 803                         'end' => L10n::t('The end')
 804                 ]);
 805         }
 806
 807         /**
 808          * Get html for contact block.
 809          *
 810          * @deprecated since version 2019.03
 811          * @see ContactBlock::getHTML()
 812          * @return string
 813          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 814          * @throws \ImagickException
 815          */
 816         public static function contactBlock()
 817         {
 818                 $a = \get_app();
 819
 820                 return ContactBlock::getHTML($a->profile);
 821         }
 822
 823         /**
 824          * @brief Format contacts as picture links or as text links
 825          *
 826          * @param array   $contact  Array with contacts which contains an array with
 827          *                          int 'id' => The ID of the contact
 828          *                          int 'uid' => The user ID of the user who owns this data
 829          *                          string 'name' => The name of the contact
 830          *                          string 'url' => The url to the profile page of the contact
 831          *                          string 'addr' => The webbie of the contact (e.g.) username@friendica.com
 832          *                          string 'network' => The network to which the contact belongs to
 833          *                          string 'thumb' => The contact picture
 834          *                          string 'click' => js code which is performed when clicking on the contact
 835          * @param boolean $redirect If true try to use the redir url if it's possible
 836          * @param string  $class    CSS class for the
 837          * @param boolean $textmode If true display the contacts as text links
 838          *                          if false display the contacts as picture links
 839          * @return string Formatted html
 840          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 841          * @throws \ImagickException
 842          */
 843         public static function micropro($contact, $redirect = false, $class = '', $textmode = false)
 844         {
 845                 // Use the contact URL if no address is available
 846                 if (empty($contact['addr'])) {
 847                         $contact["addr"] = $contact["url"];
 848                 }
 849
 850                 $url = $contact['url'];
 851                 $sparkle = '';
 852                 $redir = false;
 853
 854                 if ($redirect) {
 855                         $url = Contact::magicLink($contact['url']);
 856                         if (strpos($url, 'redir/') === 0) {
 857                                 $sparkle = ' sparkle';
 858                         }
 859                 }
 860
 861                 // If there is some js available we don't need the url
 862                 if (!empty($contact['click'])) {
 863                         $url = '';
 864                 }
 865
 866                 return Renderer::replaceMacros(Renderer::getMarkupTemplate($textmode ? 'micropro_txt.tpl' : 'micropro_img.tpl'), [
 867                         '$click' => $contact['click'] ?? '',
 868                         '$class' => $class,
 869                         '$url' => $url,
 870                         '$photo' => ProxyUtils::proxifyUrl($contact['thumb'], false, ProxyUtils::SIZE_THUMB),
 871                         '$name' => $contact['name'],
 872                         'title' => $contact['name'] . ' [' . $contact['addr'] . ']',
 873                         '$parkle' => $sparkle,
 874                         '$redir' => $redir
 875                 ]);
 876         }
 877
 878         /**
 879          * Search box.
 880          *
 881          * @param string $s     Search query.
 882          * @param string $id    HTML id
 883          * @param string $url   Search url.
 884          * @param bool   $aside Display the search widgit aside.
 885          *
 886          * @return string Formatted HTML.
 887          * @throws \Exception
 888          */
 889         public static function search($s, $id = 'search-box', $aside = true)
 890         {
 891                 $mode = 'text';
 892
 893                 if (strpos($s, '#') === 0) {
 894                         $mode = 'tag';
 895                 }
 896                 $save_label = $mode === 'text' ? L10n::t('Save') : L10n::t('Follow');
 897
 898                 $values = [
 899                         '$s'            => $s,
 900                         '$q'            => urlencode($s),
 901                         '$id'           => $id,
 902                         '$search_label' => L10n::t('Search'),
 903                         '$save_label'   => $save_label,
 904                         '$search_hint'  => L10n::t('@name, !forum, #tags, content'),
 905                         '$mode'         => $mode,
 906                         '$return_url'   => urlencode('search?q=' . urlencode($s)),
 907                 ];
 908
 909                 if (!$aside) {
 910                         $values['$search_options'] = [
 911                                 'fulltext' => L10n::t('Full Text'),
 912                                 'tags'     => L10n::t('Tags'),
 913                                 'contacts' => L10n::t('Contacts')
 914                         ];
 915
 916                         if (Config::get('system', 'poco_local_search')) {
 917                                 $values['$searchoption']['forums'] = L10n::t('Forums');
 918                         }
 919                 }
 920
 921                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 922         }
 923
 924         /**
 925          * Replace naked text hyperlink with HTML formatted hyperlink
 926          *
 927          * @param string $s
 928          * @return string
 929          */
 930         public static function toLink($s)
 931         {
 932                 $s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank">$1</a>', $s);
 933                 $s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
 934                 return $s;
 935         }
 936
 937         /**
 938          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
 939          *
 940          * Reasons are expected to have been translated already.
 941          *
 942          * @param string $html
 943          * @param array  $reasons
 944          * @return string
 945          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 946          */
 947         public static function applyContentFilter($html, array $reasons)
 948         {
 949                 if (count($reasons)) {
 950                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
 951                         $html = Renderer::replaceMacros($tpl, [
 952                                 '$reasons'   => $reasons,
 953                                 '$rnd'       => Strings::getRandomHex(8),
 954                                 '$openclose' => L10n::t('Click to open/close'),
 955                                 '$html'      => $html
 956                         ]);
 957                 }
 958
 959                 return $html;
 960         }
 961
 962         /**
 963          * replace html amp entity with amp char
 964          * @param string $s
 965          * @return string
 966          */
 967         public static function unamp($s)
 968         {
 969                 return str_replace('&amp;', '&', $s);
 970         }
 971 }