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         /**
  40          * Search all instances of a specific HTML tag node in the provided DOM document and replaces them with BBCode text nodes.
  41          *
  42          * @see HTML::tagToBBCodeSub()
  43          */
  44         private static function tagToBBCode(DOMDocument $doc, string $tag, array $attributes, string $startbb, string $endbb, bool $ignoreChildren = false)
  45         {
  46                 do {
  47                         $done = self::tagToBBCodeSub($doc, $tag, $attributes, $startbb, $endbb, $ignoreChildren);
  48                 } while ($done);
  49         }
  50
  51         /**
  52          * Search the first specific HTML tag node in the provided DOM document and replaces it with BBCode text nodes.
  53          *
  54          * @param DOMDocument $doc
  55          * @param string      $tag            HTML tag name
  56          * @param array       $attributes     Array of attributes to match and optionally use the value from
  57          * @param string      $startbb        BBCode tag opening
  58          * @param string      $endbb          BBCode tag closing
  59          * @param bool        $ignoreChildren If set to false, the HTML tag children will be appended as text inside the BBCode tag
  60          *                                    Otherwise, they will be entirely ignored. Useful for simple BBCode that draw their
  61          *                                    inner value from an attribute value and disregard the tag children.
  62          * @return bool Whether a replacement was done
  63          */
  64         private static function tagToBBCodeSub(DOMDocument $doc, string $tag, array $attributes, string $startbb, string $endbb, bool $ignoreChildren = false)
  65         {
  66                 $savestart = str_replace('$', '\x01', $startbb);
  67                 $replace = false;
  68
  69                 $xpath = new DOMXPath($doc);
  70
  71                 /** @var \DOMNode[] $list */
  72                 $list = $xpath->query("//" . $tag);
  73                 foreach ($list as $node) {
  74                         $attr = [];
  75                         if ($node->attributes->length) {
  76                                 foreach ($node->attributes as $attribute) {
  77                                         $attr[$attribute->name] = $attribute->value;
  78                                 }
  79                         }
  80
  81                         $replace = true;
  82
  83                         $startbb = $savestart;
  84
  85                         $i = 0;
  86
  87                         foreach ($attributes as $attribute => $value) {
  88                                 $startbb = str_replace('\x01' . ++$i, '$1', $startbb);
  89                                 if (strpos('*' . $startbb, '$1') > 0) {
  90                                         if ($replace && (@$attr[$attribute] != '')) {
  91                                                 $startbb = preg_replace($value, $startbb, $attr[$attribute], -1, $count);
  92
  93                                                 // If nothing could be changed
  94                                                 if ($count == 0) {
  95                                                         $replace = false;
  96                                                 }
  97                                         } else {
  98                                                 $replace = false;
  99                                         }
 100                                 } else {
 101                                         if (@$attr[$attribute] != $value) {
 102                                                 $replace = false;
 103                                         }
 104                                 }
 105                         }
 106
 107                         if ($replace) {
 108                                 $StartCode = $doc->createTextNode($startbb);
 109                                 $EndCode = $doc->createTextNode($endbb);
 110
 111                                 $node->parentNode->insertBefore($StartCode, $node);
 112
 113                                 if (!$ignoreChildren && $node->hasChildNodes()) {
 114                                         /** @var \DOMNode $child */
 115                                         foreach ($node->childNodes as $key => $child) {
 116                                                 /* Remove empty text nodes at the start or at the end of the children list */
 117                                                 if ($key > 0 && $key < $node->childNodes->length - 1 || $child->nodeName != '#text' || trim($child->nodeValue)) {
 118                                                         $newNode = $child->cloneNode(true);
 119                                                         $node->parentNode->insertBefore($newNode, $node);
 120                                                 }
 121                                         }
 122                                 }
 123
 124                                 $node->parentNode->insertBefore($EndCode, $node);
 125                                 $node->parentNode->removeChild($node);
 126                         }
 127                 }
 128
 129                 return $replace;
 130         }
 131
 132         /**
 133          * Converter for HTML to BBCode
 134          *
 135          * Made by: ike@piratenpartei.de
 136          * Originally made for the syncom project: http://wiki.piratenpartei.de/Syncom
 137          *                    https://github.com/annando/Syncom
 138          *
 139          * @param string $message
 140          * @param string $basepath
 141          * @return string
 142          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 143          */
 144         public static function toBBCode($message, $basepath = '')
 145         {
 146                 $message = str_replace("\r", "", $message);
 147
 148                 $message = Strings::performWithEscapedBlocks($message, '#<pre><code.*</code></pre>#iUs', function ($message) {
 149                         $message = str_replace(
 150                                 [
 151                                         "<li><p>",
 152                                         "</p></li>",
 153                                 ],
 154                                 [
 155                                         "<li>",
 156                                         "</li>",
 157                                 ],
 158                                 $message
 159                         );
 160
 161                         // remove namespaces
 162                         $message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
 163                         $message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
 164
 165                         $doc = new DOMDocument();
 166                         $doc->preserveWhiteSpace = false;
 167
 168                         $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 169
 170                         @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 171
 172                         XML::deleteNode($doc, 'style');
 173                         XML::deleteNode($doc, 'head');
 174                         XML::deleteNode($doc, 'title');
 175                         XML::deleteNode($doc, 'meta');
 176                         XML::deleteNode($doc, 'xml');
 177                         XML::deleteNode($doc, 'removeme');
 178
 179                         $xpath = new DomXPath($doc);
 180                         $list = $xpath->query("//pre");
 181                         foreach ($list as $node) {
 182                                 // Ensure to escape unescaped & - they will otherwise raise a warning
 183                                 $safe_value = preg_replace('/&(?!\w+;)/', '&amp;', $node->nodeValue);
 184                                 $node->nodeValue = str_replace("\n", "\r", $safe_value);
 185                         }
 186
 187                         $message = $doc->saveHTML();
 188                         $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br />", " ", ""], $message);
 189                         $message = preg_replace('= [\s]*=i', " ", $message);
 190
 191                         @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 192
 193                         self::tagToBBCode($doc, 'html', [], "", "");
 194                         self::tagToBBCode($doc, 'body', [], "", "");
 195
 196                         // Outlook-Quote - Variant 1
 197                         self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal', 'style' => 'margin-left:35.4pt'], '[quote]', '[/quote]');
 198
 199                         // Outlook-Quote - Variant 2
 200                         self::tagToBBCode(
 201                                 $doc,
 202                                 'div',
 203                                 ['style' => 'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'],
 204                                 '[quote]',
 205                                 '[/quote]'
 206                         );
 207
 208                         // MyBB-Stuff
 209                         self::tagToBBCode($doc, 'span', ['style' => 'text-decoration: underline;'], '[u]', '[/u]');
 210                         self::tagToBBCode($doc, 'span', ['style' => 'font-style: italic;'], '[i]', '[/i]');
 211                         self::tagToBBCode($doc, 'span', ['style' => 'font-weight: bold;'], '[b]', '[/b]');
 212
 213                         /* self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
 214                           self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
 215                           self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
 216                           self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
 217                           self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
 218                           self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
 219                           self::node2BBCode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
 220                          */
 221                         // Untested
 222                         //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
 223                         //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
 224                         //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
 225
 226                         self::tagToBBCode($doc, 'span', ['style' => '/.*color:\s*(.+?)[,;].*/'], '[color="$1"]', '[/color]');
 227
 228                         //self::node2BBCode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 229                         //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
 230                         //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
 231                         //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 232                         // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
 233                         // Test
 234                         //self::node2BBCode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
 235                         self::tagToBBCode($doc, 'span', ['class' => 'type-link'], '[class=type-link]', '[/class]');
 236                         self::tagToBBCode($doc, 'span', ['class' => 'type-video'], '[class=type-video]', '[/class]');
 237
 238                         self::tagToBBCode($doc, 'strong', [], '[b]', '[/b]');
 239                         self::tagToBBCode($doc, 'em', [], '[i]', '[/i]');
 240                         self::tagToBBCode($doc, 'b', [], '[b]', '[/b]');
 241                         self::tagToBBCode($doc, 'i', [], '[i]', '[/i]');
 242                         self::tagToBBCode($doc, 'u', [], '[u]', '[/u]');
 243                         self::tagToBBCode($doc, 's', [], '[s]', '[/s]');
 244                         self::tagToBBCode($doc, 'del', [], '[s]', '[/s]');
 245                         self::tagToBBCode($doc, 'strike', [], '[s]', '[/s]');
 246
 247                         self::tagToBBCode($doc, 'big', [], "[size=large]", "[/size]");
 248                         self::tagToBBCode($doc, 'small', [], "[size=small]", "[/size]");
 249
 250                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', '[/quote]');
 251
 252                         self::tagToBBCode($doc, 'br', [], "\n", '');
 253
 254                         self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal'], "\n", "");
 255                         self::tagToBBCode($doc, 'div', ['class' => 'MsoNormal'], "\r", "");
 256
 257                         self::tagToBBCode($doc, 'span', [], "", "");
 258
 259                         self::tagToBBCode($doc, 'span', [], "", "");
 260                         self::tagToBBCode($doc, 'pre', [], "", "");
 261
 262                         self::tagToBBCode($doc, 'div', [], "\r", "\r");
 263                         self::tagToBBCode($doc, 'p', [], "\n", "\n");
 264
 265                         self::tagToBBCode($doc, 'ul', [], "[list]", "[/list]");
 266                         self::tagToBBCode($doc, 'ol', [], "[list=1]", "[/list]");
 267                         self::tagToBBCode($doc, 'li', [], "[*]", "");
 268
 269                         self::tagToBBCode($doc, 'hr', [], "[hr]", "");
 270
 271                         self::tagToBBCode($doc, 'table', [], "[table]", "[/table]");
 272                         self::tagToBBCode($doc, 'th', [], "[th]", "[/th]");
 273                         self::tagToBBCode($doc, 'tr', [], "[tr]", "[/tr]");
 274                         self::tagToBBCode($doc, 'td', [], "[td]", "[/td]");
 275
 276                         self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 277                         self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 278                         self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 279                         self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 280                         self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 281                         self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 282
 283                         self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 284                         self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 285
 286                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'alt' => '/(.+)/'], '[img=$1]$2', '[/img]', true);
 287                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]', true);
 288                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]', true);
 289
 290
 291                         self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]', true);
 292                         self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]', true);
 293                         // Backward compatibility, [iframe] support has been removed in version 2020.12
 294                         self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[url]$1', '[/url]', true);
 295
 296                         self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 297                         self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 298
 299                         $message = $doc->saveHTML();
 300
 301                         // I'm removing something really disturbing
 302                         // Don't know exactly what it is
 303                         $message = str_replace(chr(194) . chr(160), ' ', $message);
 304
 305                         $message = str_replace("&nbsp;", " ", $message);
 306
 307                         // removing multiple DIVs
 308                         $message = preg_replace('=\r *\r=i', "\n", $message);
 309                         $message = str_replace("\r", "\n", $message);
 310
 311                         Hook::callAll('html2bbcode', $message);
 312
 313                         $message = strip_tags($message);
 314
 315                         $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 316
 317                         // remove quotes if they don't make sense
 318                         $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 319
 320                         $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 321                         $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 322
 323                         do {
 324                                 $oldmessage = $message;
 325                                 $message = str_replace("\n \n", "\n\n", $message);
 326                         } while ($oldmessage != $message);
 327
 328                         do {
 329                                 $oldmessage = $message;
 330                                 $message = str_replace("\n\n\n", "\n\n", $message);
 331                         } while ($oldmessage != $message);
 332
 333                         do {
 334                                 $oldmessage = $message;
 335                                 $message = str_replace(
 336                                         [
 337                                                 "[/size]\n\n",
 338                                                 "\n[hr]",
 339                                                 "[hr]\n",
 340                                                 "\n[list",
 341                                                 "[/list]\n",
 342                                                 "\n[/",
 343                                                 "[list]\n",
 344                                                 "[list=1]\n",
 345                                                 "\n[*]"],
 346                                         [
 347                                                 "[/size]\n",
 348                                                 "[hr]",
 349                                                 "[hr]",
 350                                                 "[list",
 351                                                 "[/list]",
 352                                                 "[/",
 353                                                 "[list]",
 354                                                 "[list=1]",
 355                                                 "[*]"],
 356                                         $message
 357                                 );
 358                         } while ($message != $oldmessage);
 359
 360                         $message = str_replace(
 361                                 ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 362                                 ['[b]', '[/b]', '[i]', '[/i]'],
 363                                 $message
 364                         );
 365
 366                         // Handling Yahoo style of mails
 367                         $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 368
 369                         return $message;
 370                 });
 371
 372                 $message = preg_replace_callback(
 373                         '#<pre><code(?: class="language-([^"]*)")?>(.*)</code></pre>#iUs',
 374                         function ($matches) {
 375                                 $prefix = '[code]';
 376                                 if ($matches[1] != '') {
 377                                         $prefix = '[code=' . $matches[1] . ']';
 378                                 }
 379
 380                                 return $prefix . PHP_EOL . trim($matches[2]) . PHP_EOL . '[/code]';
 381                         },
 382                         $message
 383                 );
 384
 385                 $message = trim($message);
 386
 387                 if ($basepath != '') {
 388                         $message = self::qualifyURLs($message, $basepath);
 389                 }
 390
 391                 return $message;
 392         }
 393
 394         /**
 395          * Sub function to complete incomplete URL
 396          *
 397          * @param array  $matches  Result of preg_replace_callback
 398          * @param string $basepath Basepath that is used to complete the URL
 399          *
 400          * @return string The expanded URL
 401          */
 402         private static function qualifyURLsSub($matches, $basepath)
 403         {
 404                 $base = parse_url($basepath);
 405                 unset($base['query']);
 406                 unset($base['fragment']);
 407
 408                 $link = $matches[0];
 409                 $url = $matches[1];
 410
 411                 if (empty($url) || empty(parse_url($url))) {
 412                         return $matches[0];
 413                 }
 414
 415                 $parts = array_merge($base, parse_url($url));
 416                 $url2 = Network::unparseURL($parts);
 417
 418                 return str_replace($url, $url2, $link);
 419         }
 420
 421         /**
 422          * Complete incomplete URLs in BBCode
 423          *
 424          * @param string $body     Body with URLs
 425          * @param string $basepath Base path that is used to complete the URL
 426          *
 427          * @return string Body with expanded URLs
 428          */
 429         private static function qualifyURLs($body, $basepath)
 430         {
 431                 $URLSearchString = "^\[\]";
 432
 433                 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
 434                         "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
 435                         "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 436                         "/\[img\](.*?)\[\/img\]/ism",
 437                         "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 438                         "/\[zmg\](.*?)\[\/zmg\]/ism",
 439                         "/\[video\](.*?)\[\/video\]/ism",
 440                         "/\[audio\](.*?)\[\/audio\]/ism",
 441                 ];
 442
 443                 foreach ($matches as $match) {
 444                         $body = preg_replace_callback(
 445                                 $match,
 446                                 function ($match) use ($basepath) {
 447                                         return self::qualifyURLsSub($match, $basepath);
 448                                 },
 449                                 $body
 450                         );
 451                 }
 452                 return $body;
 453         }
 454
 455         private static function breakLines($line, $level, $wraplength = 75)
 456         {
 457                 if ($wraplength == 0) {
 458                         $wraplength = 2000000;
 459                 }
 460
 461                 $wraplen = $wraplength - $level;
 462
 463                 $newlines = [];
 464
 465                 do {
 466                         $oldline = $line;
 467
 468                         $subline = substr($line, 0, $wraplen);
 469
 470                         $pos = strrpos($subline, ' ');
 471
 472                         if ($pos == 0) {
 473                                 $pos = strpos($line, ' ');
 474                         }
 475
 476                         if (($pos > 0) && strlen($line) > $wraplen) {
 477                                 $newline = trim(substr($line, 0, $pos));
 478                                 if ($level > 0) {
 479                                         $newline = str_repeat(">", $level) . ' ' . $newline;
 480                                 }
 481
 482                                 $newlines[] = $newline . " ";
 483                                 $line = substr($line, $pos + 1);
 484                         }
 485                 } while ((strlen($line) > $wraplen) && !($oldline == $line));
 486
 487                 if ($level > 0) {
 488                         $line = str_repeat(">", $level) . ' ' . $line;
 489                 }
 490
 491                 $newlines[] = $line;
 492
 493                 return implode("\n", $newlines);
 494         }
 495
 496         private static function quoteLevel($message, $wraplength = 75)
 497         {
 498                 $lines = explode("\n", $message);
 499
 500                 $newlines = [];
 501                 $level = 0;
 502                 foreach ($lines as $line) {
 503                         $line = trim($line);
 504                         $startquote = false;
 505                         while (strpos("*" . $line, '[quote]') > 0) {
 506                                 $level++;
 507                                 $pos = strpos($line, '[quote]');
 508                                 $line = substr($line, 0, $pos) . substr($line, $pos + 7);
 509                                 $startquote = true;
 510                         }
 511
 512                         $currlevel = $level;
 513
 514                         while (strpos("*" . $line, '[/quote]') > 0) {
 515                                 $level--;
 516                                 if ($level < 0) {
 517                                         $level = 0;
 518                                 }
 519
 520                                 $pos = strpos($line, '[/quote]');
 521                                 $line = substr($line, 0, $pos) . substr($line, $pos + 8);
 522                         }
 523
 524                         if (!$startquote || ($line != '')) {
 525                                 $newlines[] = self::breakLines($line, $currlevel, $wraplength);
 526                         }
 527                 }
 528
 529                 return implode("\n", $newlines);
 530         }
 531
 532         private static function collectURLs($message)
 533         {
 534                 $pattern = '/<a.*?href="(.*?)".*?>(.*?)<\/a>/is';
 535                 preg_match_all($pattern, $message, $result, PREG_SET_ORDER);
 536
 537                 $urls = [];
 538                 foreach ($result as $treffer) {
 539                         $ignore = false;
 540
 541                         // A list of some links that should be ignored
 542                         $list = ["/user/", "/tag/", "/group/", "/profile/", "/search?search=", "/search?tag=", "mailto:", "/u/", "/node/",
 543                                 "//plus.google.com/", "//twitter.com/"];
 544                         foreach ($list as $listitem) {
 545                                 if (strpos($treffer[1], $listitem) !== false) {
 546                                         $ignore = true;
 547                                 }
 548                         }
 549
 550                         if ((strpos($treffer[1], "//twitter.com/") !== false) && (strpos($treffer[1], "/status/") !== false)) {
 551                                 $ignore = false;
 552                         }
 553
 554                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/posts") !== false)) {
 555                                 $ignore = false;
 556                         }
 557
 558                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/photos") !== false)) {
 559                                 $ignore = false;
 560                         }
 561
 562                         $ignore = $ignore || strpos($treffer[1], '#') === 0;
 563
 564                         if (!$ignore) {
 565                                 $urls[$treffer[1]] = $treffer[1];
 566                         }
 567                 }
 568
 569                 return $urls;
 570         }
 571
 572         /**
 573          * @param string $html
 574          * @param int    $wraplength Ensures individual lines aren't longer than this many characters. Doesn't break words.
 575          * @param bool   $compact    True: Completely strips image tags; False: Keeps image URLs
 576          * @return string
 577          */
 578         public static function toPlaintext(string $html, $wraplength = 75, $compact = false)
 579         {
 580                 $message = str_replace("\r", "", $html);
 581
 582                 $doc = new DOMDocument();
 583                 $doc->preserveWhiteSpace = false;
 584
 585                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 586
 587                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 588
 589                 $message = $doc->saveHTML();
 590                 // Remove eventual UTF-8 BOM
 591                 $message = str_replace("\xC3\x82\xC2\xA0", "", $message);
 592
 593                 // Collecting all links
 594                 $urls = self::collectURLs($message);
 595
 596                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 597
 598                 self::tagToBBCode($doc, 'html', [], '', '');
 599                 self::tagToBBCode($doc, 'body', [], '', '');
 600
 601                 if ($compact) {
 602                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 603                 } else {
 604                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 605                 }
 606
 607                 self::tagToBBCode($doc, 'br', [], "\n", '');
 608
 609                 self::tagToBBCode($doc, 'span', [], "", "");
 610                 self::tagToBBCode($doc, 'pre', [], "", "");
 611                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 612                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 613
 614                 self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 615
 616                 self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
 617
 618                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 619                 self::tagToBBCode($doc, 'td', [], "\t", "");
 620
 621                 self::tagToBBCode($doc, 'h1', [], "\n\n*", "*\n");
 622                 self::tagToBBCode($doc, 'h2', [], "\n\n*", "*\n");
 623                 self::tagToBBCode($doc, 'h3', [], "\n\n*", "*\n");
 624                 self::tagToBBCode($doc, 'h4', [], "\n\n*", "*\n");
 625                 self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
 626                 self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 627
 628                 if (!$compact) {
 629                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
 630                 } else {
 631                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' ', ' ');
 632                 }
 633
 634                 // Backward compatibility, [iframe] support has been removed in version 2020.12
 635                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], ' $1 ', '');
 636
 637                 $message = $doc->saveHTML();
 638
 639                 if (!$compact) {
 640                         $message = str_replace("[img]", "", $message);
 641                         $message = str_replace("[/img]", "", $message);
 642                 }
 643
 644                 // was ersetze ich da?
 645                 // Irgendein stoerrisches UTF-Zeug
 646                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 647
 648                 $message = str_replace("&nbsp;", " ", $message);
 649
 650                 // Aufeinanderfolgende DIVs
 651                 $message = preg_replace('=\r *\r=i', "\n", $message);
 652                 $message = str_replace("\r", "\n", $message);
 653
 654                 $message = strip_tags($message);
 655
 656                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 657
 658                 if (!$compact && ($message != '')) {
 659                         foreach ($urls as $id => $url) {
 660                                 if ($url != '' && strpos($message, $url) === false) {
 661                                         $message .= "\n" . $url . ' ';
 662                                 }
 663                         }
 664                 }
 665
 666                 $message = str_replace("\n«", "«\n", $message);
 667                 $message = str_replace("»\n", "\n»", $message);
 668
 669                 do {
 670                         $oldmessage = $message;
 671                         $message = str_replace("\n\n\n", "\n\n", $message);
 672                 } while ($oldmessage != $message);
 673
 674                 $message = self::quoteLevel(trim($message), $wraplength);
 675
 676                 return trim($message);
 677         }
 678
 679         /**
 680          * Converts provided HTML code to Markdown. The hardwrap parameter maximizes
 681          * compatibility with Diaspora in spite of the Markdown standards.
 682          *
 683          * @param string $html
 684          * @return string
 685          */
 686         public static function toMarkdown($html)
 687         {
 688                 $converter = new HtmlConverter(['hard_break' => true]);
 689                 $markdown = $converter->convert($html);
 690
 691                 return $markdown;
 692         }
 693
 694         /**
 695          * Convert video HTML to BBCode tags
 696          *
 697          * @param string $s
 698          * @return string
 699          */
 700         public static function toBBCodeVideo($s)
 701         {
 702                 $s = preg_replace(
 703                         '#<object[^>]+>(.*?)https?://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+)(.*?)</object>#ism',
 704                         '[youtube]$2[/youtube]',
 705                         $s
 706                 );
 707
 708                 $s = preg_replace(
 709                         '#<iframe[^>](.*?)https?://www.youtube.com/embed/([A-Za-z0-9\-_=]+)(.*?)</iframe>#ism',
 710                         '[youtube]$2[/youtube]',
 711                         $s
 712                 );
 713
 714                 $s = preg_replace(
 715                         '#<iframe[^>](.*?)https?://player.vimeo.com/video/([0-9]+)(.*?)</iframe>#ism',
 716                         '[vimeo]$2[/vimeo]',
 717                         $s
 718                 );
 719
 720                 return $s;
 721         }
 722
 723         /**
 724          * transform link href and img src from relative to absolute
 725          *
 726          * @param string $text
 727          * @param string $base base url
 728          * @return string
 729          */
 730         public static function relToAbs($text, $base)
 731         {
 732                 if (empty($base)) {
 733                         return $text;
 734                 }
 735
 736                 $base = rtrim($base, '/');
 737
 738                 $base2 = $base . "/";
 739
 740                 // Replace links
 741                 $pattern = "/<a([^>]*) href=\"(?!http|https|\/)([^\"]*)\"/";
 742                 $replace = "<a\${1} href=\"" . $base2 . "\${2}\"";
 743                 $text = preg_replace($pattern, $replace, $text);
 744
 745                 $pattern = "/<a([^>]*) href=\"(?!http|https)([^\"]*)\"/";
 746                 $replace = "<a\${1} href=\"" . $base . "\${2}\"";
 747                 $text = preg_replace($pattern, $replace, $text);
 748
 749                 // Replace images
 750                 $pattern = "/<img([^>]*) src=\"(?!http|https|\/)([^\"]*)\"/";
 751                 $replace = "<img\${1} src=\"" . $base2 . "\${2}\"";
 752                 $text = preg_replace($pattern, $replace, $text);
 753
 754                 $pattern = "/<img([^>]*) src=\"(?!http|https)([^\"]*)\"/";
 755                 $replace = "<img\${1} src=\"" . $base . "\${2}\"";
 756                 $text = preg_replace($pattern, $replace, $text);
 757
 758
 759                 // Done
 760                 return $text;
 761         }
 762
 763         /**
 764          * return div element with class 'clear'
 765          * @return string
 766          * @deprecated
 767          */
 768         public static function clearDiv()
 769         {
 770                 return '<div class="clear"></div>';
 771         }
 772
 773         /**
 774          * Loader for infinite scrolling
 775          *
 776          * @return string html for loader
 777          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 778          */
 779         public static function scrollLoader()
 780         {
 781                 $tpl = Renderer::getMarkupTemplate("scroll_loader.tpl");
 782                 return Renderer::replaceMacros($tpl, [
 783                         'wait' => DI::l10n()->t('Loading more entries...'),
 784                         'end' => DI::l10n()->t('The end')
 785                 ]);
 786         }
 787
 788         /**
 789          * Get html for contact block.
 790          *
 791          * @deprecated since version 2019.03
 792          * @see ContactBlock::getHTML()
 793          * @return string
 794          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 795          * @throws \ImagickException
 796          */
 797         public static function contactBlock()
 798         {
 799                 $a = DI::app();
 800
 801                 return ContactBlock::getHTML($a->profile);
 802         }
 803
 804         /**
 805          * Format contacts as picture links or as text links
 806          *
 807          * @param array   $contact  Array with contacts which contains an array with
 808          *                          int 'id' => The ID of the contact
 809          *                          int 'uid' => The user ID of the user who owns this data
 810          *                          string 'name' => The name of the contact
 811          *                          string 'url' => The url to the profile page of the contact
 812          *                          string 'addr' => The webbie of the contact (e.g.) username@friendica.com
 813          *                          string 'network' => The network to which the contact belongs to
 814          *                          string 'thumb' => The contact picture
 815          *                          string 'click' => js code which is performed when clicking on the contact
 816          * @param boolean $redirect If true try to use the redir url if it's possible
 817          * @param string  $class    CSS class for the
 818          * @param boolean $textmode If true display the contacts as text links
 819          *                          if false display the contacts as picture links
 820          * @return string Formatted html
 821          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 822          * @throws \ImagickException
 823          */
 824         public static function micropro($contact, $redirect = false, $class = '', $textmode = false)
 825         {
 826                 // Use the contact URL if no address is available
 827                 if (empty($contact['addr'])) {
 828                         $contact["addr"] = $contact["url"];
 829                 }
 830
 831                 $url = $contact['url'];
 832                 $sparkle = '';
 833                 $redir = false;
 834
 835                 if ($redirect) {
 836                         $url = Contact::magicLink($contact['url']);
 837                         if (strpos($url, 'redir/') === 0) {
 838                                 $sparkle = ' sparkle';
 839                         }
 840                 }
 841
 842                 // If there is some js available we don't need the url
 843                 if (!empty($contact['click'])) {
 844                         $url = '';
 845                 }
 846
 847                 return Renderer::replaceMacros(Renderer::getMarkupTemplate($textmode ? 'micropro_txt.tpl' : 'micropro_img.tpl'), [
 848                         '$click' => $contact['click'] ?? '',
 849                         '$class' => $class,
 850                         '$url' => $url,
 851                         '$photo' => Contact::getThumb($contact),
 852                         '$name' => $contact['name'],
 853                         'title' => $contact['name'] . ' [' . $contact['addr'] . ']',
 854                         '$parkle' => $sparkle,
 855                         '$redir' => $redir
 856                 ]);
 857         }
 858
 859         /**
 860          * Search box.
 861          *
 862          * @param string $s     Search query.
 863          * @param string $id    HTML id
 864          * @param string $url   Search url.
 865          * @param bool   $aside Display the search widgit aside.
 866          *
 867          * @return string Formatted HTML.
 868          * @throws \Exception
 869          */
 870         public static function search($s, $id = 'search-box', $aside = true)
 871         {
 872                 $mode = 'text';
 873
 874                 if (strpos($s, '#') === 0) {
 875                         $mode = 'tag';
 876                 }
 877                 $save_label = $mode === 'text' ? DI::l10n()->t('Save') : DI::l10n()->t('Follow');
 878
 879                 $values = [
 880                         '$s'            => $s,
 881                         '$q'            => urlencode($s),
 882                         '$id'           => $id,
 883                         '$search_label' => DI::l10n()->t('Search'),
 884                         '$save_label'   => $save_label,
 885                         '$search_hint'  => DI::l10n()->t('@name, !forum, #tags, content'),
 886                         '$mode'         => $mode,
 887                         '$return_url'   => urlencode(Search::getSearchPath($s)),
 888                 ];
 889
 890                 if (!$aside) {
 891                         $values['$search_options'] = [
 892                                 'fulltext' => DI::l10n()->t('Full Text'),
 893                                 'tags'     => DI::l10n()->t('Tags'),
 894                                 'contacts' => DI::l10n()->t('Contacts')
 895                         ];
 896
 897                         if (DI::config()->get('system', 'poco_local_search')) {
 898                                 $values['$searchoption']['forums'] = DI::l10n()->t('Forums');
 899                         }
 900                 }
 901
 902                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 903         }
 904
 905         /**
 906          * Replace naked text hyperlink with HTML formatted hyperlink
 907          *
 908          * @param string $s
 909          * @return string
 910          */
 911         public static function toLink($s)
 912         {
 913                 $s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank" rel="noopener noreferrer">$1</a>', $s);
 914                 $s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
 915                 return $s;
 916         }
 917
 918         /**
 919          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
 920          *
 921          * Reasons are expected to have been translated already.
 922          *
 923          * @param string $html
 924          * @param array  $reasons
 925          * @return string
 926          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 927          */
 928         public static function applyContentFilter($html, array $reasons)
 929         {
 930                 if (count($reasons)) {
 931                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
 932                         $html = Renderer::replaceMacros($tpl, [
 933                                 '$reasons'   => $reasons,
 934                                 '$rnd'       => Strings::getRandomHex(8),
 935                                 '$openclose' => DI::l10n()->t('Click to open/close'),
 936                                 '$html'      => $html
 937                         ]);
 938                 }
 939
 940                 return $html;
 941         }
 942
 943         /**
 944          * replace html amp entity with amp char
 945          * @param string $s
 946          * @return string
 947          */
 948         public static function unamp($s)
 949         {
 950                 return str_replace('&amp;', '&', $s);
 951         }
 952 }