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