src/Content/Text/HTML.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2020, Friendica
   4  *
   5  * @license GNU AGPL version 3 or any later version
   6  *
   7  * This program is free software: you can redistribute it and/or modify
   8  * it under the terms of the GNU Affero General Public License as
   9  * published by the Free Software Foundation, either version 3 of the
  10  * License, or (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU Affero General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Affero General Public License
  18  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  19  *
  20  */
  21
  22 namespace Friendica\Content\Text;
  23
  24 use DOMDocument;
  25 use DOMXPath;
  26 use Friendica\Content\Widget\ContactBlock;
  27 use Friendica\Core\Hook;
  28 use Friendica\Core\Renderer;
  29 use Friendica\Core\Search;
  30 use Friendica\DI;
  31 use Friendica\Model\Contact;
  32 use Friendica\Util\Network;
  33 use Friendica\Util\Proxy as ProxyUtils;
  34 use Friendica\Util\Strings;
  35 use Friendica\Util\XML;
  36 use League\HTMLToMarkdown\HtmlConverter;
  37
  38 class HTML
  39 {
  40         public static function sanitizeCSS($input)
  41         {
  42                 $cleaned = "";
  43
  44                 $input = strtolower($input);
  45
  46                 for ($i = 0; $i < strlen($input); $i++) {
  47                         $char = substr($input, $i, 1);
  48
  49                         if (($char >= "a") && ($char <= "z")) {
  50                                 $cleaned .= $char;
  51                         }
  52
  53                         if (!(strpos(" #;:0123456789-_.%", $char) === false)) {
  54                                 $cleaned .= $char;
  55                         }
  56                 }
  57
  58                 return $cleaned;
  59         }
  60
  61         /**
  62          * Search all instances of a specific HTML tag node in the provided DOM document and replaces them with BBCode text nodes.
  63          *
  64          * @see HTML::tagToBBCodeSub()
  65          */
  66         private static function tagToBBCode(DOMDocument $doc, string $tag, array $attributes, string $startbb, string $endbb, bool $ignoreChildren = false)
  67         {
  68                 do {
  69                         $done = self::tagToBBCodeSub($doc, $tag, $attributes, $startbb, $endbb, $ignoreChildren);
  70                 } while ($done);
  71         }
  72
  73         /**
  74          * Search the first specific HTML tag node in the provided DOM document and replaces it with BBCode text nodes.
  75          *
  76          * @param DOMDocument $doc
  77          * @param string      $tag            HTML tag name
  78          * @param array       $attributes     Array of attributes to match and optionally use the value from
  79          * @param string      $startbb        BBCode tag opening
  80          * @param string      $endbb          BBCode tag closing
  81          * @param bool        $ignoreChildren If set to false, the HTML tag children will be appended as text inside the BBCode tag
  82          *                                    Otherwise, they will be entirely ignored. Useful for simple BBCode that draw their
  83          *                                    inner value from an attribute value and disregard the tag children.
  84          * @return bool Whether a replacement was done
  85          */
  86         private static function tagToBBCodeSub(DOMDocument $doc, string $tag, array $attributes, string $startbb, string $endbb, bool $ignoreChildren = false)
  87         {
  88                 $savestart = str_replace('$', '\x01', $startbb);
  89                 $replace = false;
  90
  91                 $xpath = new DOMXPath($doc);
  92
  93                 /** @var \DOMNode[] $list */
  94                 $list = $xpath->query("//" . $tag);
  95                 foreach ($list as $node) {
  96                         $attr = [];
  97                         if ($node->attributes->length) {
  98                                 foreach ($node->attributes as $attribute) {
  99                                         $attr[$attribute->name] = $attribute->value;
 100                                 }
 101                         }
 102
 103                         $replace = true;
 104
 105                         $startbb = $savestart;
 106
 107                         $i = 0;
 108
 109                         foreach ($attributes as $attribute => $value) {
 110                                 $startbb = str_replace('\x01' . ++$i, '$1', $startbb);
 111                                 if (strpos('*' . $startbb, '$1') > 0) {
 112                                         if ($replace && (@$attr[$attribute] != '')) {
 113                                                 $startbb = preg_replace($value, $startbb, $attr[$attribute], -1, $count);
 114
 115                                                 // If nothing could be changed
 116                                                 if ($count == 0) {
 117                                                         $replace = false;
 118                                                 }
 119                                         } else {
 120                                                 $replace = false;
 121                                         }
 122                                 } else {
 123                                         if (@$attr[$attribute] != $value) {
 124                                                 $replace = false;
 125                                         }
 126                                 }
 127                         }
 128
 129                         if ($replace) {
 130                                 $StartCode = $doc->createTextNode($startbb);
 131                                 $EndCode = $doc->createTextNode($endbb);
 132
 133                                 $node->parentNode->insertBefore($StartCode, $node);
 134
 135                                 if (!$ignoreChildren && $node->hasChildNodes()) {
 136                                         /** @var \DOMNode $child */
 137                                         foreach ($node->childNodes as $key => $child) {
 138                                                 /* Remove empty text nodes at the start or at the end of the children list */
 139                                                 if ($key > 0 && $key < $node->childNodes->length - 1 || $child->nodeName != '#text' || trim($child->nodeValue)) {
 140                                                         $newNode = $child->cloneNode(true);
 141                                                         $node->parentNode->insertBefore($newNode, $node);
 142                                                 }
 143                                         }
 144                                 }
 145
 146                                 $node->parentNode->insertBefore($EndCode, $node);
 147                                 $node->parentNode->removeChild($node);
 148                         }
 149                 }
 150
 151                 return $replace;
 152         }
 153
 154         /**
 155          * Converter for HTML to BBCode
 156          *
 157          * Made by: ike@piratenpartei.de
 158          * Originally made for the syncom project: http://wiki.piratenpartei.de/Syncom
 159          *                    https://github.com/annando/Syncom
 160          *
 161          * @param string $message
 162          * @param string $basepath
 163          * @return string
 164          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 165          */
 166         public static function toBBCode($message, $basepath = '')
 167         {
 168                 $message = str_replace("\r", "", $message);
 169
 170                 // Removing code blocks before the whitespace removal processing below
 171                 $codeblocks = [];
 172                 $message = preg_replace_callback(
 173                         '#<pre><code(?: class="language-([^"]*)")?>(.*)</code></pre>#iUs',
 174                         function ($matches) use (&$codeblocks) {
 175                                 $return = '[codeblock-' . count($codeblocks) . ']';
 176
 177                                 $prefix = '[code]';
 178                                 if ($matches[1] != '') {
 179                                         $prefix = '[code=' . $matches[1] . ']';
 180                                 }
 181
 182                                 $codeblocks[] = $prefix . PHP_EOL . trim($matches[2]) . PHP_EOL . '[/code]';
 183                                 return $return;
 184                         },
 185                         $message
 186                 );
 187
 188                 $message = str_replace(
 189                         [
 190                                 "<li><p>",
 191                                 "</p></li>",
 192                         ],
 193                         [
 194                                 "<li>",
 195                                 "</li>",
 196                         ],
 197                         $message
 198                 );
 199
 200                 // remove namespaces
 201                 $message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
 202                 $message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
 203
 204                 $doc = new DOMDocument();
 205                 $doc->preserveWhiteSpace = false;
 206
 207                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 208
 209                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 210
 211                 XML::deleteNode($doc, 'style');
 212                 XML::deleteNode($doc, 'head');
 213                 XML::deleteNode($doc, 'title');
 214                 XML::deleteNode($doc, 'meta');
 215                 XML::deleteNode($doc, 'xml');
 216                 XML::deleteNode($doc, 'removeme');
 217
 218                 $xpath = new DomXPath($doc);
 219                 $list = $xpath->query("//pre");
 220                 foreach ($list as $node) {
 221                         // Ensure to escape unescaped & - they will otherwise raise a warning
 222                         $safe_value = preg_replace('/&(?!\w+;)/', '&amp;', $node->nodeValue);
 223                         $node->nodeValue = str_replace("\n", "\r", $safe_value);
 224                 }
 225
 226                 $message = $doc->saveHTML();
 227                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br />", " ", ""], $message);
 228                 $message = preg_replace('= [\s]*=i', " ", $message);
 229
 230                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 231
 232                 self::tagToBBCode($doc, 'html', [], "", "");
 233                 self::tagToBBCode($doc, 'body', [], "", "");
 234
 235                 // Outlook-Quote - Variant 1
 236                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal', 'style' => 'margin-left:35.4pt'], '[quote]', '[/quote]');
 237
 238                 // Outlook-Quote - Variant 2
 239                 self::tagToBBCode(
 240                         $doc,
 241                         'div',
 242                         ['style' => 'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'],
 243                         '[quote]',
 244                         '[/quote]'
 245                 );
 246
 247                 // MyBB-Stuff
 248                 self::tagToBBCode($doc, 'span', ['style' => 'text-decoration: underline;'], '[u]', '[/u]');
 249                 self::tagToBBCode($doc, 'span', ['style' => 'font-style: italic;'], '[i]', '[/i]');
 250                 self::tagToBBCode($doc, 'span', ['style' => 'font-weight: bold;'], '[b]', '[/b]');
 251
 252                 /* self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
 253                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
 254                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
 255                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
 256                   self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
 257                   self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
 258                   self::node2BBCode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
 259                  */
 260                 // Untested
 261                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
 262                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
 263                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
 264
 265                 self::tagToBBCode($doc, 'span', ['style' => '/.*color:\s*(.+?)[,;].*/'], '[color="$1"]', '[/color]');
 266
 267                 //self::node2BBCode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 268                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
 269                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
 270                 //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 271                 // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
 272                 // Test
 273                 //self::node2BBCode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
 274                 self::tagToBBCode($doc, 'span', ['class' => 'type-link'], '[class=type-link]', '[/class]');
 275                 self::tagToBBCode($doc, 'span', ['class' => 'type-video'], '[class=type-video]', '[/class]');
 276
 277                 self::tagToBBCode($doc, 'strong', [], '[b]', '[/b]');
 278                 self::tagToBBCode($doc, 'em', [], '[i]', '[/i]');
 279                 self::tagToBBCode($doc, 'b', [], '[b]', '[/b]');
 280                 self::tagToBBCode($doc, 'i', [], '[i]', '[/i]');
 281                 self::tagToBBCode($doc, 'u', [], '[u]', '[/u]');
 282                 self::tagToBBCode($doc, 's', [], '[s]', '[/s]');
 283                 self::tagToBBCode($doc, 'del', [], '[s]', '[/s]');
 284                 self::tagToBBCode($doc, 'strike', [], '[s]', '[/s]');
 285
 286                 self::tagToBBCode($doc, 'big', [], "[size=large]", "[/size]");
 287                 self::tagToBBCode($doc, 'small', [], "[size=small]", "[/size]");
 288
 289                 self::tagToBBCode($doc, 'blockquote', [], '[quote]', '[/quote]');
 290
 291                 self::tagToBBCode($doc, 'br', [], "\n", '');
 292
 293                 self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal'], "\n", "");
 294                 self::tagToBBCode($doc, 'div', ['class' => 'MsoNormal'], "\r", "");
 295
 296                 self::tagToBBCode($doc, 'span', [], "", "");
 297
 298                 self::tagToBBCode($doc, 'span', [], "", "");
 299                 self::tagToBBCode($doc, 'pre', [], "", "");
 300
 301                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 302                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 303
 304                 self::tagToBBCode($doc, 'ul', [], "[list]", "[/list]");
 305                 self::tagToBBCode($doc, 'ol', [], "[list=1]", "[/list]");
 306                 self::tagToBBCode($doc, 'li', [], "[*]", "");
 307
 308                 self::tagToBBCode($doc, 'hr', [], "[hr]", "");
 309
 310                 self::tagToBBCode($doc, 'table', [], "[table]", "[/table]");
 311                 self::tagToBBCode($doc, 'th', [], "[th]", "[/th]");
 312                 self::tagToBBCode($doc, 'tr', [], "[tr]", "[/tr]");
 313                 self::tagToBBCode($doc, 'td', [], "[td]", "[/td]");
 314
 315                 self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 316                 self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 317                 self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 318                 self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 319                 self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 320                 self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 321
 322                 self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 323                 self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 324
 325                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'alt' => '/(.+)/'], '[img=$1]$2', '[/img]', true);
 326                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]', true);
 327                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]', true);
 328
 329
 330                 self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]', true);
 331                 self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]', true);
 332                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[iframe]$1', '[/iframe]', true);
 333
 334                 self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 335                 self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 336
 337                 $message = $doc->saveHTML();
 338
 339                 // I'm removing something really disturbing
 340                 // Don't know exactly what it is
 341                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 342
 343                 $message = str_replace("&nbsp;", " ", $message);
 344
 345                 // removing multiple DIVs
 346                 $message = preg_replace('=\r *\r=i', "\n", $message);
 347                 $message = str_replace("\r", "\n", $message);
 348
 349                 Hook::callAll('html2bbcode', $message);
 350
 351                 $message = strip_tags($message);
 352
 353                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 354
 355                 // remove quotes if they don't make sense
 356                 $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 357
 358                 $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 359                 $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 360
 361                 do {
 362                         $oldmessage = $message;
 363                         $message = str_replace("\n \n", "\n\n", $message);
 364                 } while ($oldmessage != $message);
 365
 366                 do {
 367                         $oldmessage = $message;
 368                         $message = str_replace("\n\n\n", "\n\n", $message);
 369                 } while ($oldmessage != $message);
 370
 371                 do {
 372                         $oldmessage = $message;
 373                         $message = str_replace(
 374                                 [
 375                                 "[/size]\n\n",
 376                                 "\n[hr]",
 377                                 "[hr]\n",
 378                                 "\n[list",
 379                                 "[/list]\n",
 380                                 "\n[/",
 381                                 "[list]\n",
 382                                 "[list=1]\n",
 383                                 "\n[*]"],
 384                                 [
 385                                 "[/size]\n",
 386                                 "[hr]",
 387                                 "[hr]",
 388                                 "[list",
 389                                 "[/list]",
 390                                 "[/",
 391                                 "[list]",
 392                                 "[list=1]",
 393                                 "[*]"],
 394                                 $message
 395                         );
 396                 } while ($message != $oldmessage);
 397
 398                 $message = str_replace(
 399                         ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 400                         ['[b]', '[/b]', '[i]', '[/i]'],
 401                         $message
 402                 );
 403
 404                 // Handling Yahoo style of mails
 405                 $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 406
 407                 // Restore code blocks
 408                 $message = preg_replace_callback(
 409                         '#\[codeblock-([0-9]+)\]#iU',
 410                         function ($matches) use ($codeblocks) {
 411                                 $return = '';
 412                                 if (isset($codeblocks[intval($matches[1])])) {
 413                                         $return = $codeblocks[$matches[1]];
 414                                 }
 415                                 return $return;
 416                         },
 417                         $message
 418                 );
 419
 420                 $message = trim($message);
 421
 422                 if ($basepath != '') {
 423                         $message = self::qualifyURLs($message, $basepath);
 424                 }
 425
 426                 return $message;
 427         }
 428
 429         /**
 430          * Sub function to complete incomplete URL
 431          *
 432          * @param array  $matches  Result of preg_replace_callback
 433          * @param string $basepath Basepath that is used to complete the URL
 434          *
 435          * @return string The expanded URL
 436          */
 437         private static function qualifyURLsSub($matches, $basepath)
 438         {
 439                 $base = parse_url($basepath);
 440                 unset($base['query']);
 441                 unset($base['fragment']);
 442
 443                 $link = $matches[0];
 444                 $url = $matches[1];
 445
 446                 if (empty($url) || empty(parse_url($url))) {
 447                         return $matches[0];
 448                 }
 449
 450                 $parts = array_merge($base, parse_url($url));
 451                 $url2 = Network::unparseURL($parts);
 452
 453                 return str_replace($url, $url2, $link);
 454         }
 455
 456         /**
 457          * Complete incomplete URLs in BBCode
 458          *
 459          * @param string $body     Body with URLs
 460          * @param string $basepath Base path that is used to complete the URL
 461          *
 462          * @return string Body with expanded URLs
 463          */
 464         private static function qualifyURLs($body, $basepath)
 465         {
 466                 $URLSearchString = "^\[\]";
 467
 468                 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
 469                         "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
 470                         "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 471                         "/\[img\](.*?)\[\/img\]/ism",
 472                         "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 473                         "/\[zmg\](.*?)\[\/zmg\]/ism",
 474                         "/\[video\](.*?)\[\/video\]/ism",
 475                         "/\[audio\](.*?)\[\/audio\]/ism",
 476                 ];
 477
 478                 foreach ($matches as $match) {
 479                         $body = preg_replace_callback(
 480                                 $match,
 481                                 function ($match) use ($basepath) {
 482                                         return self::qualifyURLsSub($match, $basepath);
 483                                 },
 484                                 $body
 485                         );
 486                 }
 487                 return $body;
 488         }
 489
 490         private static function breakLines($line, $level, $wraplength = 75)
 491         {
 492                 if ($wraplength == 0) {
 493                         $wraplength = 2000000;
 494                 }
 495
 496                 $wraplen = $wraplength - $level;
 497
 498                 $newlines = [];
 499
 500                 do {
 501                         $oldline = $line;
 502
 503                         $subline = substr($line, 0, $wraplen);
 504
 505                         $pos = strrpos($subline, ' ');
 506
 507                         if ($pos == 0) {
 508                                 $pos = strpos($line, ' ');
 509                         }
 510
 511                         if (($pos > 0) && strlen($line) > $wraplen) {
 512                                 $newline = trim(substr($line, 0, $pos));
 513                                 if ($level > 0) {
 514                                         $newline = str_repeat(">", $level) . ' ' . $newline;
 515                                 }
 516
 517                                 $newlines[] = $newline . " ";
 518                                 $line = substr($line, $pos + 1);
 519                         }
 520                 } while ((strlen($line) > $wraplen) && !($oldline == $line));
 521
 522                 if ($level > 0) {
 523                         $line = str_repeat(">", $level) . ' ' . $line;
 524                 }
 525
 526                 $newlines[] = $line;
 527
 528                 return implode("\n", $newlines);
 529         }
 530
 531         private static function quoteLevel($message, $wraplength = 75)
 532         {
 533                 $lines = explode("\n", $message);
 534
 535                 $newlines = [];
 536                 $level = 0;
 537                 foreach ($lines as $line) {
 538                         $line = trim($line);
 539                         $startquote = false;
 540                         while (strpos("*" . $line, '[quote]') > 0) {
 541                                 $level++;
 542                                 $pos = strpos($line, '[quote]');
 543                                 $line = substr($line, 0, $pos) . substr($line, $pos + 7);
 544                                 $startquote = true;
 545                         }
 546
 547                         $currlevel = $level;
 548
 549                         while (strpos("*" . $line, '[/quote]') > 0) {
 550                                 $level--;
 551                                 if ($level < 0) {
 552                                         $level = 0;
 553                                 }
 554
 555                                 $pos = strpos($line, '[/quote]');
 556                                 $line = substr($line, 0, $pos) . substr($line, $pos + 8);
 557                         }
 558
 559                         if (!$startquote || ($line != '')) {
 560                                 $newlines[] = self::breakLines($line, $currlevel, $wraplength);
 561                         }
 562                 }
 563
 564                 return implode("\n", $newlines);
 565         }
 566
 567         private static function collectURLs($message)
 568         {
 569                 $pattern = '/<a.*?href="(.*?)".*?>(.*?)<\/a>/is';
 570                 preg_match_all($pattern, $message, $result, PREG_SET_ORDER);
 571
 572                 $urls = [];
 573                 foreach ($result as $treffer) {
 574                         $ignore = false;
 575
 576                         // A list of some links that should be ignored
 577                         $list = ["/user/", "/tag/", "/group/", "/profile/", "/search?search=", "/search?tag=", "mailto:", "/u/", "/node/",
 578                                 "//plus.google.com/", "//twitter.com/"];
 579                         foreach ($list as $listitem) {
 580                                 if (strpos($treffer[1], $listitem) !== false) {
 581                                         $ignore = true;
 582                                 }
 583                         }
 584
 585                         if ((strpos($treffer[1], "//twitter.com/") !== false) && (strpos($treffer[1], "/status/") !== false)) {
 586                                 $ignore = false;
 587                         }
 588
 589                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/posts") !== false)) {
 590                                 $ignore = false;
 591                         }
 592
 593                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/photos") !== false)) {
 594                                 $ignore = false;
 595                         }
 596
 597                         $ignore = $ignore || strpos($treffer[1], '#') === 0;
 598
 599                         if (!$ignore) {
 600                                 $urls[$treffer[1]] = $treffer[1];
 601                         }
 602                 }
 603
 604                 return $urls;
 605         }
 606
 607         /**
 608          * @param string $html
 609          * @param int    $wraplength Ensures individual lines aren't longer than this many characters. Doesn't break words.
 610          * @param bool   $compact    True: Completely strips image tags; False: Keeps image URLs
 611          * @return string
 612          */
 613         public static function toPlaintext(string $html, $wraplength = 75, $compact = false)
 614         {
 615                 $message = str_replace("\r", "", $html);
 616
 617                 $doc = new DOMDocument();
 618                 $doc->preserveWhiteSpace = false;
 619
 620                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 621
 622                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 623
 624                 $message = $doc->saveHTML();
 625                 // Remove eventual UTF-8 BOM
 626                 $message = str_replace("\xC3\x82\xC2\xA0", "", $message);
 627
 628                 // Collecting all links
 629                 $urls = self::collectURLs($message);
 630
 631                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 632
 633                 self::tagToBBCode($doc, 'html', [], '', '');
 634                 self::tagToBBCode($doc, 'body', [], '', '');
 635
 636                 if ($compact) {
 637                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 638                 } else {
 639                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 640                 }
 641
 642                 self::tagToBBCode($doc, 'br', [], "\n", '');
 643
 644                 self::tagToBBCode($doc, 'span', [], "", "");
 645                 self::tagToBBCode($doc, 'pre', [], "", "");
 646                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 647                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 648
 649                 self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 650
 651                 self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
 652
 653                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 654                 self::tagToBBCode($doc, 'td', [], "\t", "");
 655
 656                 self::tagToBBCode($doc, 'h1', [], "\n\n*", "*\n");
 657                 self::tagToBBCode($doc, 'h2', [], "\n\n*", "*\n");
 658                 self::tagToBBCode($doc, 'h3', [], "\n\n*", "*\n");
 659                 self::tagToBBCode($doc, 'h4', [], "\n\n*", "*\n");
 660                 self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
 661                 self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 662
 663                 if (!$compact) {
 664                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
 665                 } else {
 666                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' ', ' ');
 667                 }
 668
 669                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], ' $1 ', '');
 670
 671                 $message = $doc->saveHTML();
 672
 673                 if (!$compact) {
 674                         $message = str_replace("[img]", "", $message);
 675                         $message = str_replace("[/img]", "", $message);
 676                 }
 677
 678                 // was ersetze ich da?
 679                 // Irgendein stoerrisches UTF-Zeug
 680                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 681
 682                 $message = str_replace("&nbsp;", " ", $message);
 683
 684                 // Aufeinanderfolgende DIVs
 685                 $message = preg_replace('=\r *\r=i', "\n", $message);
 686                 $message = str_replace("\r", "\n", $message);
 687
 688                 $message = strip_tags($message);
 689
 690                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 691
 692                 if (!$compact && ($message != '')) {
 693                         foreach ($urls as $id => $url) {
 694                                 if ($url != '' && strpos($message, $url) === false) {
 695                                         $message .= "\n" . $url . ' ';
 696                                 }
 697                         }
 698                 }
 699
 700                 $message = str_replace("\n«", "«\n", $message);
 701                 $message = str_replace("»\n", "\n»", $message);
 702
 703                 do {
 704                         $oldmessage = $message;
 705                         $message = str_replace("\n\n\n", "\n\n", $message);
 706                 } while ($oldmessage != $message);
 707
 708                 $message = self::quoteLevel(trim($message), $wraplength);
 709
 710                 return trim($message);
 711         }
 712
 713         /**
 714          * Converts provided HTML code to Markdown. The hardwrap parameter maximizes
 715          * compatibility with Diaspora in spite of the Markdown standards.
 716          *
 717          * @param string $html
 718          * @return string
 719          */
 720         public static function toMarkdown($html)
 721         {
 722                 $converter = new HtmlConverter(['hard_break' => true]);
 723                 $markdown = $converter->convert($html);
 724
 725                 return $markdown;
 726         }
 727
 728         /**
 729          * Convert video HTML to BBCode tags
 730          *
 731          * @param string $s
 732          * @return string
 733          */
 734         public static function toBBCodeVideo($s)
 735         {
 736                 $s = preg_replace(
 737                         '#<object[^>]+>(.*?)https?://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+)(.*?)</object>#ism',
 738                         '[youtube]$2[/youtube]',
 739                         $s
 740                 );
 741
 742                 $s = preg_replace(
 743                         '#<iframe[^>](.*?)https?://www.youtube.com/embed/([A-Za-z0-9\-_=]+)(.*?)</iframe>#ism',
 744                         '[youtube]$2[/youtube]',
 745                         $s
 746                 );
 747
 748                 $s = preg_replace(
 749                         '#<iframe[^>](.*?)https?://player.vimeo.com/video/([0-9]+)(.*?)</iframe>#ism',
 750                         '[vimeo]$2[/vimeo]',
 751                         $s
 752                 );
 753
 754                 return $s;
 755         }
 756
 757         /**
 758          * transform link href and img src from relative to absolute
 759          *
 760          * @param string $text
 761          * @param string $base base url
 762          * @return string
 763          */
 764         public static function relToAbs($text, $base)
 765         {
 766                 if (empty($base)) {
 767                         return $text;
 768                 }
 769
 770                 $base = rtrim($base, '/');
 771
 772                 $base2 = $base . "/";
 773
 774                 // Replace links
 775                 $pattern = "/<a([^>]*) href=\"(?!http|https|\/)([^\"]*)\"/";
 776                 $replace = "<a\${1} href=\"" . $base2 . "\${2}\"";
 777                 $text = preg_replace($pattern, $replace, $text);
 778
 779                 $pattern = "/<a([^>]*) href=\"(?!http|https)([^\"]*)\"/";
 780                 $replace = "<a\${1} href=\"" . $base . "\${2}\"";
 781                 $text = preg_replace($pattern, $replace, $text);
 782
 783                 // Replace images
 784                 $pattern = "/<img([^>]*) src=\"(?!http|https|\/)([^\"]*)\"/";
 785                 $replace = "<img\${1} src=\"" . $base2 . "\${2}\"";
 786                 $text = preg_replace($pattern, $replace, $text);
 787
 788                 $pattern = "/<img([^>]*) src=\"(?!http|https)([^\"]*)\"/";
 789                 $replace = "<img\${1} src=\"" . $base . "\${2}\"";
 790                 $text = preg_replace($pattern, $replace, $text);
 791
 792
 793                 // Done
 794                 return $text;
 795         }
 796
 797         /**
 798          * return div element with class 'clear'
 799          * @return string
 800          * @deprecated
 801          */
 802         public static function clearDiv()
 803         {
 804                 return '<div class="clear"></div>';
 805         }
 806
 807         /**
 808          * Loader for infinite scrolling
 809          *
 810          * @return string html for loader
 811          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 812          */
 813         public static function scrollLoader()
 814         {
 815                 $tpl = Renderer::getMarkupTemplate("scroll_loader.tpl");
 816                 return Renderer::replaceMacros($tpl, [
 817                         'wait' => DI::l10n()->t('Loading more entries...'),
 818                         'end' => DI::l10n()->t('The end')
 819                 ]);
 820         }
 821
 822         /**
 823          * Get html for contact block.
 824          *
 825          * @deprecated since version 2019.03
 826          * @see ContactBlock::getHTML()
 827          * @return string
 828          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 829          * @throws \ImagickException
 830          */
 831         public static function contactBlock()
 832         {
 833                 $a = DI::app();
 834
 835                 return ContactBlock::getHTML($a->profile);
 836         }
 837
 838         /**
 839          * Format contacts as picture links or as text links
 840          *
 841          * @param array   $contact  Array with contacts which contains an array with
 842          *                          int 'id' => The ID of the contact
 843          *                          int 'uid' => The user ID of the user who owns this data
 844          *                          string 'name' => The name of the contact
 845          *                          string 'url' => The url to the profile page of the contact
 846          *                          string 'addr' => The webbie of the contact (e.g.) username@friendica.com
 847          *                          string 'network' => The network to which the contact belongs to
 848          *                          string 'thumb' => The contact picture
 849          *                          string 'click' => js code which is performed when clicking on the contact
 850          * @param boolean $redirect If true try to use the redir url if it's possible
 851          * @param string  $class    CSS class for the
 852          * @param boolean $textmode If true display the contacts as text links
 853          *                          if false display the contacts as picture links
 854          * @return string Formatted html
 855          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 856          * @throws \ImagickException
 857          */
 858         public static function micropro($contact, $redirect = false, $class = '', $textmode = false)
 859         {
 860                 // Use the contact URL if no address is available
 861                 if (empty($contact['addr'])) {
 862                         $contact["addr"] = $contact["url"];
 863                 }
 864
 865                 $url = $contact['url'];
 866                 $sparkle = '';
 867                 $redir = false;
 868
 869                 if ($redirect) {
 870                         $url = Contact::magicLink($contact['url']);
 871                         if (strpos($url, 'redir/') === 0) {
 872                                 $sparkle = ' sparkle';
 873                         }
 874                 }
 875
 876                 // If there is some js available we don't need the url
 877                 if (!empty($contact['click'])) {
 878                         $url = '';
 879                 }
 880
 881                 return Renderer::replaceMacros(Renderer::getMarkupTemplate($textmode ? 'micropro_txt.tpl' : 'micropro_img.tpl'), [
 882                         '$click' => $contact['click'] ?? '',
 883                         '$class' => $class,
 884                         '$url' => $url,
 885                         '$photo' => ProxyUtils::proxifyUrl($contact['thumb'], false, ProxyUtils::SIZE_THUMB),
 886                         '$name' => $contact['name'],
 887                         'title' => $contact['name'] . ' [' . $contact['addr'] . ']',
 888                         '$parkle' => $sparkle,
 889                         '$redir' => $redir
 890                 ]);
 891         }
 892
 893         /**
 894          * Search box.
 895          *
 896          * @param string $s     Search query.
 897          * @param string $id    HTML id
 898          * @param string $url   Search url.
 899          * @param bool   $aside Display the search widgit aside.
 900          *
 901          * @return string Formatted HTML.
 902          * @throws \Exception
 903          */
 904         public static function search($s, $id = 'search-box', $aside = true)
 905         {
 906                 $mode = 'text';
 907
 908                 if (strpos($s, '#') === 0) {
 909                         $mode = 'tag';
 910                 }
 911                 $save_label = $mode === 'text' ? DI::l10n()->t('Save') : DI::l10n()->t('Follow');
 912
 913                 $values = [
 914                         '$s'            => $s,
 915                         '$q'            => urlencode($s),
 916                         '$id'           => $id,
 917                         '$search_label' => DI::l10n()->t('Search'),
 918                         '$save_label'   => $save_label,
 919                         '$search_hint'  => DI::l10n()->t('@name, !forum, #tags, content'),
 920                         '$mode'         => $mode,
 921                         '$return_url'   => urlencode(Search::getSearchPath($s)),
 922                 ];
 923
 924                 if (!$aside) {
 925                         $values['$search_options'] = [
 926                                 'fulltext' => DI::l10n()->t('Full Text'),
 927                                 'tags'     => DI::l10n()->t('Tags'),
 928                                 'contacts' => DI::l10n()->t('Contacts')
 929                         ];
 930
 931                         if (DI::config()->get('system', 'poco_local_search')) {
 932                                 $values['$searchoption']['forums'] = DI::l10n()->t('Forums');
 933                         }
 934                 }
 935
 936                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 937         }
 938
 939         /**
 940          * Replace naked text hyperlink with HTML formatted hyperlink
 941          *
 942          * @param string $s
 943          * @return string
 944          */
 945         public static function toLink($s)
 946         {
 947                 $s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank" rel="noopener noreferrer">$1</a>', $s);
 948                 $s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
 949                 return $s;
 950         }
 951
 952         /**
 953          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
 954          *
 955          * Reasons are expected to have been translated already.
 956          *
 957          * @param string $html
 958          * @param array  $reasons
 959          * @return string
 960          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 961          */
 962         public static function applyContentFilter($html, array $reasons)
 963         {
 964                 if (count($reasons)) {
 965                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
 966                         $html = Renderer::replaceMacros($tpl, [
 967                                 '$reasons'   => $reasons,
 968                                 '$rnd'       => Strings::getRandomHex(8),
 969                                 '$openclose' => DI::l10n()->t('Click to open/close'),
 970                                 '$html'      => $html
 971                         ]);
 972                 }
 973
 974                 return $html;
 975         }
 976
 977         /**
 978          * replace html amp entity with amp char
 979          * @param string $s
 980          * @return string
 981          */
 982         public static function unamp($s)
 983         {
 984                 return str_replace('&amp;', '&', $s);
 985         }
 986 }