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