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($doc, 'table', [], "", "");
 294                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 295                 self::tagToBBCode($doc, 'td', [], "\t", "");
 296                 //self::node2BBCode($doc, 'table', array(), "[table]", "[/table]");
 297                 //self::node2BBCode($doc, 'th', array(), "[th]", "[/th]");
 298                 //self::node2BBCode($doc, 'tr', array(), "[tr]", "[/tr]");
 299                 //self::node2BBCode($doc, 'td', array(), "[td]", "[/td]");
 300                 //self::node2BBCode($doc, 'h1', array(), "\n\n[size=xx-large][b]", "[/b][/size]\n");
 301                 //self::node2BBCode($doc, 'h2', array(), "\n\n[size=x-large][b]", "[/b][/size]\n");
 302                 //self::node2BBCode($doc, 'h3', array(), "\n\n[size=large][b]", "[/b][/size]\n");
 303                 //self::node2BBCode($doc, 'h4', array(), "\n\n[size=medium][b]", "[/b][/size]\n");
 304                 //self::node2BBCode($doc, 'h5', array(), "\n\n[size=small][b]", "[/b][/size]\n");
 305                 //self::node2BBCode($doc, 'h6', array(), "\n\n[size=x-small][b]", "[/b][/size]\n");
 306
 307                 self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 308                 self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 309                 self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 310                 self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 311                 self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 312                 self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 313
 314                 self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 315                 self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 316
 317                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'alt' => '/(.+)/'], '[img=$1]$2', '[/img]', true);
 318                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]', true);
 319                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]', true);
 320
 321
 322                 self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]', true);
 323                 self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]', true);
 324                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[iframe]$1', '[/iframe]', true);
 325
 326                 self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 327                 self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 328
 329                 $message = $doc->saveHTML();
 330
 331                 // I'm removing something really disturbing
 332                 // Don't know exactly what it is
 333                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 334
 335                 $message = str_replace("&nbsp;", " ", $message);
 336
 337                 // removing multiple DIVs
 338                 $message = preg_replace('=\r *\r=i', "\n", $message);
 339                 $message = str_replace("\r", "\n", $message);
 340
 341                 Hook::callAll('html2bbcode', $message);
 342
 343                 $message = strip_tags($message);
 344
 345                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 346
 347                 $message = str_replace(["<"], ["&lt;"], $message);
 348
 349                 // remove quotes if they don't make sense
 350                 $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 351
 352                 $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 353                 $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 354
 355                 do {
 356                         $oldmessage = $message;
 357                         $message = str_replace("\n \n", "\n\n", $message);
 358                 } while ($oldmessage != $message);
 359
 360                 do {
 361                         $oldmessage = $message;
 362                         $message = str_replace("\n\n\n", "\n\n", $message);
 363                 } while ($oldmessage != $message);
 364
 365                 do {
 366                         $oldmessage = $message;
 367                         $message = str_replace(
 368                                 [
 369                                 "[/size]\n\n",
 370                                 "\n[hr]",
 371                                 "[hr]\n",
 372                                 "\n[list",
 373                                 "[/list]\n",
 374                                 "\n[/",
 375                                 "[list]\n",
 376                                 "[list=1]\n",
 377                                 "\n[*]"],
 378                                 [
 379                                 "[/size]\n",
 380                                 "[hr]",
 381                                 "[hr]",
 382                                 "[list",
 383                                 "[/list]",
 384                                 "[/",
 385                                 "[list]",
 386                                 "[list=1]",
 387                                 "[*]"],
 388                                 $message
 389                         );
 390                 } while ($message != $oldmessage);
 391
 392                 $message = str_replace(
 393                         ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 394                         ['[b]', '[/b]', '[i]', '[/i]'],
 395                         $message
 396                 );
 397
 398                 // Handling Yahoo style of mails
 399                 $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 400
 401                 // Restore code blocks
 402                 $message = preg_replace_callback(
 403                         '#\[codeblock-([0-9]+)\]#iU',
 404                         function ($matches) use ($codeblocks) {
 405                                 $return = '';
 406                                 if (isset($codeblocks[intval($matches[1])])) {
 407                                         $return = $codeblocks[$matches[1]];
 408                                 }
 409                                 return $return;
 410                         },
 411                         $message
 412                 );
 413
 414                 $message = trim($message);
 415
 416                 if ($basepath != '') {
 417                         $message = self::qualifyURLs($message, $basepath);
 418                 }
 419
 420                 return $message;
 421         }
 422
 423         /**
 424          * @brief Sub function to complete incomplete URL
 425          *
 426          * @param array  $matches  Result of preg_replace_callback
 427          * @param string $basepath Basepath that is used to complete the URL
 428          *
 429          * @return string The expanded URL
 430          */
 431         private static function qualifyURLsSub($matches, $basepath)
 432         {
 433                 $base = parse_url($basepath);
 434                 unset($base['query']);
 435                 unset($base['fragment']);
 436
 437                 $link = $matches[0];
 438                 $url = $matches[1];
 439
 440                 if (empty($url) || empty(parse_url($url))) {
 441                         return $matches[0];
 442                 }
 443
 444                 $parts = array_merge($base, parse_url($url));
 445                 $url2 = Network::unparseURL($parts);
 446
 447                 return str_replace($url, $url2, $link);
 448         }
 449
 450         /**
 451          * @brief Complete incomplete URLs in BBCode
 452          *
 453          * @param string $body     Body with URLs
 454          * @param string $basepath Base path that is used to complete the URL
 455          *
 456          * @return string Body with expanded URLs
 457          */
 458         private static function qualifyURLs($body, $basepath)
 459         {
 460                 $URLSearchString = "^\[\]";
 461
 462                 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
 463                         "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
 464                         "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 465                         "/\[img\](.*?)\[\/img\]/ism",
 466                         "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 467                         "/\[zmg\](.*?)\[\/zmg\]/ism",
 468                         "/\[video\](.*?)\[\/video\]/ism",
 469                         "/\[audio\](.*?)\[\/audio\]/ism",
 470                 ];
 471
 472                 foreach ($matches as $match) {
 473                         $body = preg_replace_callback(
 474                                 $match,
 475                                 function ($match) use ($basepath) {
 476                                         return self::qualifyURLsSub($match, $basepath);
 477                                 },
 478                                 $body
 479                         );
 480                 }
 481                 return $body;
 482         }
 483
 484         private static function breakLines($line, $level, $wraplength = 75)
 485         {
 486                 if ($wraplength == 0) {
 487                         $wraplength = 2000000;
 488                 }
 489
 490                 $wraplen = $wraplength - $level;
 491
 492                 $newlines = [];
 493
 494                 do {
 495                         $oldline = $line;
 496
 497                         $subline = substr($line, 0, $wraplen);
 498
 499                         $pos = strrpos($subline, ' ');
 500
 501                         if ($pos == 0) {
 502                                 $pos = strpos($line, ' ');
 503                         }
 504
 505                         if (($pos > 0) && strlen($line) > $wraplen) {
 506                                 $newline = trim(substr($line, 0, $pos));
 507                                 if ($level > 0) {
 508                                         $newline = str_repeat(">", $level) . ' ' . $newline;
 509                                 }
 510
 511                                 $newlines[] = $newline . " ";
 512                                 $line = substr($line, $pos + 1);
 513                         }
 514                 } while ((strlen($line) > $wraplen) && !($oldline == $line));
 515
 516                 if ($level > 0) {
 517                         $line = str_repeat(">", $level) . ' ' . $line;
 518                 }
 519
 520                 $newlines[] = $line;
 521
 522                 return implode($newlines, "\n");
 523         }
 524
 525         private static function quoteLevel($message, $wraplength = 75)
 526         {
 527                 $lines = explode("\n", $message);
 528
 529                 $newlines = [];
 530                 $level = 0;
 531                 foreach ($lines as $line) {
 532                         $line = trim($line);
 533                         $startquote = false;
 534                         while (strpos("*" . $line, '[quote]') > 0) {
 535                                 $level++;
 536                                 $pos = strpos($line, '[quote]');
 537                                 $line = substr($line, 0, $pos) . substr($line, $pos + 7);
 538                                 $startquote = true;
 539                         }
 540
 541                         $currlevel = $level;
 542
 543                         while (strpos("*" . $line, '[/quote]') > 0) {
 544                                 $level--;
 545                                 if ($level < 0) {
 546                                         $level = 0;
 547                                 }
 548
 549                                 $pos = strpos($line, '[/quote]');
 550                                 $line = substr($line, 0, $pos) . substr($line, $pos + 8);
 551                         }
 552
 553                         if (!$startquote || ($line != '')) {
 554                                 $newlines[] = self::breakLines($line, $currlevel, $wraplength);
 555                         }
 556                 }
 557
 558                 return implode($newlines, "\n");
 559         }
 560
 561         private static function collectURLs($message)
 562         {
 563                 $pattern = '/<a.*?href="(.*?)".*?>(.*?)<\/a>/is';
 564                 preg_match_all($pattern, $message, $result, PREG_SET_ORDER);
 565
 566                 $urls = [];
 567                 foreach ($result as $treffer) {
 568                         $ignore = false;
 569
 570                         // A list of some links that should be ignored
 571                         $list = ["/user/", "/tag/", "/group/", "/profile/", "/search?search=", "/search?tag=", "mailto:", "/u/", "/node/",
 572                                 "//plus.google.com/", "//twitter.com/"];
 573                         foreach ($list as $listitem) {
 574                                 if (strpos($treffer[1], $listitem) !== false) {
 575                                         $ignore = true;
 576                                 }
 577                         }
 578
 579                         if ((strpos($treffer[1], "//twitter.com/") !== false) && (strpos($treffer[1], "/status/") !== false)) {
 580                                 $ignore = false;
 581                         }
 582
 583                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/posts") !== false)) {
 584                                 $ignore = false;
 585                         }
 586
 587                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/photos") !== false)) {
 588                                 $ignore = false;
 589                         }
 590
 591                         $ignore = $ignore || strpos($treffer[1], '#') === 0;
 592
 593                         if (!$ignore) {
 594                                 $urls[$treffer[1]] = $treffer[1];
 595                         }
 596                 }
 597
 598                 return $urls;
 599         }
 600
 601         /**
 602          * @param string $html
 603          * @param int    $wraplength Ensures individual lines aren't longer than this many characters. Doesn't break words.
 604          * @param bool   $compact    True: Completely strips image tags; False: Keeps image URLs
 605          * @return string
 606          */
 607         public static function toPlaintext(string $html, $wraplength = 75, $compact = false)
 608         {
 609                 $message = str_replace("\r", "", $html);
 610
 611                 $doc = new DOMDocument();
 612                 $doc->preserveWhiteSpace = false;
 613
 614                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 615
 616                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 617
 618                 $message = $doc->saveHTML();
 619                 // Remove eventual UTF-8 BOM
 620                 $message = str_replace("\xC3\x82\xC2\xA0", "", $message);
 621
 622                 // Collecting all links
 623                 $urls = self::collectURLs($message);
 624
 625                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 626
 627                 self::tagToBBCode($doc, 'html', [], '', '');
 628                 self::tagToBBCode($doc, 'body', [], '', '');
 629
 630                 if ($compact) {
 631                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 632                 } else {
 633                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 634                 }
 635
 636                 self::tagToBBCode($doc, 'br', [], "\n", '');
 637
 638                 self::tagToBBCode($doc, 'span', [], "", "");
 639                 self::tagToBBCode($doc, 'pre', [], "", "");
 640                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 641                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 642
 643                 self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 644
 645                 self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
 646
 647                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 648                 self::tagToBBCode($doc, 'td', [], "\t", "");
 649
 650                 self::tagToBBCode($doc, 'h1', [], "\n\n*", "*\n");
 651                 self::tagToBBCode($doc, 'h2', [], "\n\n*", "*\n");
 652                 self::tagToBBCode($doc, 'h3', [], "\n\n*", "*\n");
 653                 self::tagToBBCode($doc, 'h4', [], "\n\n*", "*\n");
 654                 self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
 655                 self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 656
 657                 if (!$compact) {
 658                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
 659                 } else {
 660                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' ', ' ');
 661                 }
 662
 663                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], ' $1 ', '');
 664
 665                 $message = $doc->saveHTML();
 666
 667                 if (!$compact) {
 668                         $message = str_replace("[img]", "", $message);
 669                         $message = str_replace("[/img]", "", $message);
 670                 }
 671
 672                 // was ersetze ich da?
 673                 // Irgendein stoerrisches UTF-Zeug
 674                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 675
 676                 $message = str_replace("&nbsp;", " ", $message);
 677
 678                 // Aufeinanderfolgende DIVs
 679                 $message = preg_replace('=\r *\r=i', "\n", $message);
 680                 $message = str_replace("\r", "\n", $message);
 681
 682                 $message = strip_tags($message);
 683
 684                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 685
 686                 if (!$compact && ($message != '')) {
 687                         foreach ($urls as $id => $url) {
 688                                 if ($url != '' && strpos($message, $url) === false) {
 689                                         $message .= "\n" . $url . ' ';
 690                                 }
 691                         }
 692                 }
 693
 694                 $message = str_replace("\n«", "«\n", $message);
 695                 $message = str_replace("»\n", "\n»", $message);
 696
 697                 do {
 698                         $oldmessage = $message;
 699                         $message = str_replace("\n\n\n", "\n\n", $message);
 700                 } while ($oldmessage != $message);
 701
 702                 $message = self::quoteLevel(trim($message), $wraplength);
 703
 704                 return trim($message);
 705         }
 706
 707         /**
 708          * Converts provided HTML code to Markdown. The hardwrap parameter maximizes
 709          * compatibility with Diaspora in spite of the Markdown standards.
 710          *
 711          * @param string $html
 712          * @return string
 713          */
 714         public static function toMarkdown($html)
 715         {
 716                 $converter = new HtmlConverter(['hard_break' => true]);
 717                 $markdown = $converter->convert($html);
 718
 719                 return $markdown;
 720         }
 721
 722         /**
 723          * @brief Convert video HTML to BBCode tags
 724          *
 725          * @param string $s
 726          * @return string
 727          */
 728         public static function toBBCodeVideo($s)
 729         {
 730                 $s = preg_replace(
 731                         '#<object[^>]+>(.*?)https?://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+)(.*?)</object>#ism',
 732                         '[youtube]$2[/youtube]',
 733                         $s
 734                 );
 735
 736                 $s = preg_replace(
 737                         '#<iframe[^>](.*?)https?://www.youtube.com/embed/([A-Za-z0-9\-_=]+)(.*?)</iframe>#ism',
 738                         '[youtube]$2[/youtube]',
 739                         $s
 740                 );
 741
 742                 $s = preg_replace(
 743                         '#<iframe[^>](.*?)https?://player.vimeo.com/video/([0-9]+)(.*?)</iframe>#ism',
 744                         '[vimeo]$2[/vimeo]',
 745                         $s
 746                 );
 747
 748                 return $s;
 749         }
 750
 751         /**
 752          * transform link href and img src from relative to absolute
 753          *
 754          * @param string $text
 755          * @param string $base base url
 756          * @return string
 757          */
 758         public static function relToAbs($text, $base)
 759         {
 760                 if (empty($base)) {
 761                         return $text;
 762                 }
 763
 764                 $base = rtrim($base, '/');
 765
 766                 $base2 = $base . "/";
 767
 768                 // Replace links
 769                 $pattern = "/<a([^>]*) href=\"(?!http|https|\/)([^\"]*)\"/";
 770                 $replace = "<a\${1} href=\"" . $base2 . "\${2}\"";
 771                 $text = preg_replace($pattern, $replace, $text);
 772
 773                 $pattern = "/<a([^>]*) href=\"(?!http|https)([^\"]*)\"/";
 774                 $replace = "<a\${1} href=\"" . $base . "\${2}\"";
 775                 $text = preg_replace($pattern, $replace, $text);
 776
 777                 // Replace images
 778                 $pattern = "/<img([^>]*) src=\"(?!http|https|\/)([^\"]*)\"/";
 779                 $replace = "<img\${1} src=\"" . $base2 . "\${2}\"";
 780                 $text = preg_replace($pattern, $replace, $text);
 781
 782                 $pattern = "/<img([^>]*) src=\"(?!http|https)([^\"]*)\"/";
 783                 $replace = "<img\${1} src=\"" . $base . "\${2}\"";
 784                 $text = preg_replace($pattern, $replace, $text);
 785
 786
 787                 // Done
 788                 return $text;
 789         }
 790
 791         /**
 792          * return div element with class 'clear'
 793          * @return string
 794          * @deprecated
 795          */
 796         public static function clearDiv()
 797         {
 798                 return '<div class="clear"></div>';
 799         }
 800
 801         /**
 802          * Loader for infinite scrolling
 803          *
 804          * @return string html for loader
 805          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 806          */
 807         public static function scrollLoader()
 808         {
 809                 $tpl = Renderer::getMarkupTemplate("scroll_loader.tpl");
 810                 return Renderer::replaceMacros($tpl, [
 811                         'wait' => L10n::t('Loading more entries...'),
 812                         'end' => L10n::t('The end')
 813                 ]);
 814         }
 815
 816         /**
 817          * Get html for contact block.
 818          *
 819          * @deprecated since version 2019.03
 820          * @see ContactBlock::getHTML()
 821          * @return string
 822          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 823          * @throws \ImagickException
 824          */
 825         public static function contactBlock()
 826         {
 827                 $a = \get_app();
 828
 829                 return ContactBlock::getHTML($a->profile);
 830         }
 831
 832         /**
 833          * @brief Format contacts as picture links or as text links
 834          *
 835          * @param array   $contact  Array with contacts which contains an array with
 836          *                          int 'id' => The ID of the contact
 837          *                          int 'uid' => The user ID of the user who owns this data
 838          *                          string 'name' => The name of the contact
 839          *                          string 'url' => The url to the profile page of the contact
 840          *                          string 'addr' => The webbie of the contact (e.g.) username@friendica.com
 841          *                          string 'network' => The network to which the contact belongs to
 842          *                          string 'thumb' => The contact picture
 843          *                          string 'click' => js code which is performed when clicking on the contact
 844          * @param boolean $redirect If true try to use the redir url if it's possible
 845          * @param string  $class    CSS class for the
 846          * @param boolean $textmode If true display the contacts as text links
 847          *                          if false display the contacts as picture links
 848          * @return string Formatted html
 849          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 850          * @throws \ImagickException
 851          */
 852         public static function micropro($contact, $redirect = false, $class = '', $textmode = false)
 853         {
 854                 // Use the contact URL if no address is available
 855                 if (empty($contact['addr'])) {
 856                         $contact["addr"] = $contact["url"];
 857                 }
 858
 859                 $url = $contact['url'];
 860                 $sparkle = '';
 861                 $redir = false;
 862
 863                 if ($redirect) {
 864                         $url = Contact::magicLink($contact['url']);
 865                         if (strpos($url, 'redir/') === 0) {
 866                                 $sparkle = ' sparkle';
 867                         }
 868                 }
 869
 870                 // If there is some js available we don't need the url
 871                 if (!empty($contact['click'])) {
 872                         $url = '';
 873                 }
 874
 875                 return Renderer::replaceMacros(Renderer::getMarkupTemplate($textmode ? 'micropro_txt.tpl' : 'micropro_img.tpl'), [
 876                         '$click' => $contact['click'] ?? '',
 877                         '$class' => $class,
 878                         '$url' => $url,
 879                         '$photo' => ProxyUtils::proxifyUrl($contact['thumb'], false, ProxyUtils::SIZE_THUMB),
 880                         '$name' => $contact['name'],
 881                         'title' => $contact['name'] . ' [' . $contact['addr'] . ']',
 882                         '$parkle' => $sparkle,
 883                         '$redir' => $redir
 884                 ]);
 885         }
 886
 887         /**
 888          * Search box.
 889          *
 890          * @param string $s     Search query.
 891          * @param string $id    HTML id
 892          * @param string $url   Search url.
 893          * @param bool   $aside Display the search widgit aside.
 894          *
 895          * @return string Formatted HTML.
 896          * @throws \Exception
 897          */
 898         public static function search($s, $id = 'search-box', $aside = true)
 899         {
 900                 $mode = 'text';
 901
 902                 if (strpos($s, '#') === 0) {
 903                         $mode = 'tag';
 904                 }
 905                 $save_label = $mode === 'text' ? L10n::t('Save') : L10n::t('Follow');
 906
 907                 $values = [
 908                         '$s'            => $s,
 909                         '$q'            => urlencode($s),
 910                         '$id'           => $id,
 911                         '$search_label' => L10n::t('Search'),
 912                         '$save_label'   => $save_label,
 913                         '$search_hint'  => L10n::t('@name, !forum, #tags, content'),
 914                         '$mode'         => $mode,
 915                         '$return_url'   => urlencode('search?q=' . urlencode($s)),
 916                 ];
 917
 918                 if (!$aside) {
 919                         $values['$search_options'] = [
 920                                 'fulltext' => L10n::t('Full Text'),
 921                                 'tags'     => L10n::t('Tags'),
 922                                 'contacts' => L10n::t('Contacts')
 923                         ];
 924
 925                         if (Config::get('system', 'poco_local_search')) {
 926                                 $values['$searchoption']['forums'] = L10n::t('Forums');
 927                         }
 928                 }
 929
 930                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 931         }
 932
 933         /**
 934          * Replace naked text hyperlink with HTML formatted hyperlink
 935          *
 936          * @param string $s
 937          * @return string
 938          */
 939         public static function toLink($s)
 940         {
 941                 $s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank">$1</a>', $s);
 942                 $s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
 943                 return $s;
 944         }
 945
 946         /**
 947          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
 948          *
 949          * Reasons are expected to have been translated already.
 950          *
 951          * @param string $html
 952          * @param array  $reasons
 953          * @return string
 954          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 955          */
 956         public static function applyContentFilter($html, array $reasons)
 957         {
 958                 if (count($reasons)) {
 959                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
 960                         $html = Renderer::replaceMacros($tpl, [
 961                                 '$reasons'   => $reasons,
 962                                 '$rnd'       => Strings::getRandomHex(8),
 963                                 '$openclose' => L10n::t('Click to open/close'),
 964                                 '$html'      => $html
 965                         ]);
 966                 }
 967
 968                 return $html;
 969         }
 970
 971         /**
 972          * replace html amp entity with amp char
 973          * @param string $s
 974          * @return string
 975          */
 976         public static function unamp($s)
 977         {
 978                 return str_replace('&amp;', '&', $s);
 979         }
 980 }