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', " ", $message);
 190
 191                         if (empty($message)) {
 192                                 return '';
 193                         }
 194
 195                         @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 196
 197                         self::tagToBBCode($doc, 'html', [], "", "");
 198                         self::tagToBBCode($doc, 'body', [], "", "");
 199
 200                         // Outlook-Quote - Variant 1
 201                         self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal', 'style' => 'margin-left:35.4pt'], '[quote]', '[/quote]');
 202
 203                         // Outlook-Quote - Variant 2
 204                         self::tagToBBCode(
 205                                 $doc,
 206                                 'div',
 207                                 ['style' => 'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'],
 208                                 '[quote]',
 209                                 '[/quote]'
 210                         );
 211
 212                         // MyBB-Stuff
 213                         self::tagToBBCode($doc, 'span', ['style' => 'text-decoration: underline;'], '[u]', '[/u]');
 214                         self::tagToBBCode($doc, 'span', ['style' => 'font-style: italic;'], '[i]', '[/i]');
 215                         self::tagToBBCode($doc, 'span', ['style' => 'font-weight: bold;'], '[b]', '[/b]');
 216
 217                         /* self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
 218                           self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
 219                           self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
 220                           self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
 221                           self::node2BBCode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
 222                           self::node2BBCode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
 223                           self::node2BBCode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
 224                          */
 225                         // Untested
 226                         //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
 227                         //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
 228                         //self::node2BBCode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
 229
 230                         self::tagToBBCode($doc, 'span', ['style' => '/.*color:\s*(.+?)[,;].*/'], '[color="$1"]', '[/color]');
 231
 232                         //self::node2BBCode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 233                         //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
 234                         //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
 235                         //self::node2BBCode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
 236                         // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
 237                         // Test
 238                         //self::node2BBCode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
 239                         self::tagToBBCode($doc, 'span', ['class' => 'type-link'], '[class=type-link]', '[/class]');
 240                         self::tagToBBCode($doc, 'span', ['class' => 'type-video'], '[class=type-video]', '[/class]');
 241
 242                         self::tagToBBCode($doc, 'strong', [], '[b]', '[/b]');
 243                         self::tagToBBCode($doc, 'em', [], '[i]', '[/i]');
 244                         self::tagToBBCode($doc, 'b', [], '[b]', '[/b]');
 245                         self::tagToBBCode($doc, 'i', [], '[i]', '[/i]');
 246                         self::tagToBBCode($doc, 'u', [], '[u]', '[/u]');
 247                         self::tagToBBCode($doc, 's', [], '[s]', '[/s]');
 248                         self::tagToBBCode($doc, 'del', [], '[s]', '[/s]');
 249                         self::tagToBBCode($doc, 'strike', [], '[s]', '[/s]');
 250
 251                         self::tagToBBCode($doc, 'big', [], "[size=large]", "[/size]");
 252                         self::tagToBBCode($doc, 'small', [], "[size=small]", "[/size]");
 253
 254                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', '[/quote]');
 255
 256                         self::tagToBBCode($doc, 'br', [], "\n", '');
 257
 258                         self::tagToBBCode($doc, 'p', ['class' => 'MsoNormal'], "\n", "");
 259                         self::tagToBBCode($doc, 'div', ['class' => 'MsoNormal'], "\r", "");
 260
 261                         self::tagToBBCode($doc, 'span', [], "", "");
 262
 263                         self::tagToBBCode($doc, 'span', [], "", "");
 264                         self::tagToBBCode($doc, 'pre', [], "", "");
 265
 266                         self::tagToBBCode($doc, 'div', [], "\r", "\r");
 267                         self::tagToBBCode($doc, 'p', [], "\n", "\n");
 268
 269                         self::tagToBBCode($doc, 'ul', [], "[list]", "[/list]");
 270                         self::tagToBBCode($doc, 'ol', [], "[list=1]", "[/list]");
 271                         self::tagToBBCode($doc, 'li', [], "[*]", "");
 272
 273                         self::tagToBBCode($doc, 'hr', [], "[hr]", "");
 274
 275                         self::tagToBBCode($doc, 'table', [], "[table]", "[/table]");
 276                         self::tagToBBCode($doc, 'th', [], "[th]", "[/th]");
 277                         self::tagToBBCode($doc, 'tr', [], "[tr]", "[/tr]");
 278                         self::tagToBBCode($doc, 'td', [], "[td]", "[/td]");
 279
 280                         self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 281                         self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 282                         self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 283                         self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 284                         self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 285                         self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 286
 287                         self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 288                         self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 289
 290                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'alt' => '/(.+)/'], '[img=$1]$2', '[/img]', true);
 291                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]', true);
 292                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]', true);
 293
 294
 295                         self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]', true);
 296                         self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]', true);
 297                         // Backward compatibility, [iframe] support has been removed in version 2020.12
 298                         self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[url]$1', '[/url]', true);
 299
 300                         self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 301                         self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 302
 303                         $message = $doc->saveHTML();
 304
 305                         // I'm removing something really disturbing
 306                         // Don't know exactly what it is
 307                         $message = str_replace(chr(194) . chr(160), ' ', $message);
 308
 309                         $message = str_replace("&nbsp;", " ", $message);
 310
 311                         // removing multiple DIVs
 312                         $message = preg_replace('=\r *\r=i', "\n", $message);
 313                         $message = str_replace("\r", "\n", $message);
 314
 315                         Hook::callAll('html2bbcode', $message);
 316
 317                         $message = strip_tags($message);
 318
 319                         $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 320
 321                         // remove quotes if they don't make sense
 322                         $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 323
 324                         $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 325                         $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 326
 327                         do {
 328                                 $oldmessage = $message;
 329                                 $message = str_replace("\n \n", "\n\n", $message);
 330                         } while ($oldmessage != $message);
 331
 332                         do {
 333                                 $oldmessage = $message;
 334                                 $message = str_replace("\n\n\n", "\n\n", $message);
 335                         } while ($oldmessage != $message);
 336
 337                         do {
 338                                 $oldmessage = $message;
 339                                 $message = str_replace(
 340                                         [
 341                                                 "[/size]\n\n",
 342                                                 "\n[hr]",
 343                                                 "[hr]\n",
 344                                                 "\n[list",
 345                                                 "[/list]\n",
 346                                                 "\n[/",
 347                                                 "[list]\n",
 348                                                 "[list=1]\n",
 349                                                 "\n[*]"],
 350                                         [
 351                                                 "[/size]\n",
 352                                                 "[hr]",
 353                                                 "[hr]",
 354                                                 "[list",
 355                                                 "[/list]",
 356                                                 "[/",
 357                                                 "[list]",
 358                                                 "[list=1]",
 359                                                 "[*]"],
 360                                         $message
 361                                 );
 362                         } while ($message != $oldmessage);
 363
 364                         $message = str_replace(
 365                                 ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 366                                 ['[b]', '[/b]', '[i]', '[/i]'],
 367                                 $message
 368                         );
 369
 370                         // Handling Yahoo style of mails
 371                         $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 372
 373                         return $message;
 374                 });
 375
 376                 $message = preg_replace_callback(
 377                         '#<pre><code(?: class="language-([^"]*)")?>(.*)</code></pre>#iUs',
 378                         function ($matches) {
 379                                 $prefix = '[code]';
 380                                 if ($matches[1] != '') {
 381                                         $prefix = '[code=' . $matches[1] . ']';
 382                                 }
 383
 384                                 return $prefix . PHP_EOL . trim($matches[2]) . PHP_EOL . '[/code]';
 385                         },
 386                         $message
 387                 );
 388
 389                 $message = trim($message);
 390
 391                 if ($basepath != '') {
 392                         $message = self::qualifyURLs($message, $basepath);
 393                 }
 394
 395                 return $message;
 396         }
 397
 398         /**
 399          * Sub function to complete incomplete URL
 400          *
 401          * @param array  $matches  Result of preg_replace_callback
 402          * @param string $basepath Basepath that is used to complete the URL
 403          *
 404          * @return string The expanded URL
 405          */
 406         private static function qualifyURLsSub($matches, $basepath)
 407         {
 408                 $base = parse_url($basepath);
 409                 unset($base['query']);
 410                 unset($base['fragment']);
 411
 412                 $link = $matches[0];
 413                 $url = $matches[1];
 414
 415                 if (empty($url) || empty(parse_url($url))) {
 416                         return $matches[0];
 417                 }
 418
 419                 $parts = array_merge($base, parse_url($url));
 420                 $url2 = Network::unparseURL($parts);
 421
 422                 return str_replace($url, $url2, $link);
 423         }
 424
 425         /**
 426          * Complete incomplete URLs in BBCode
 427          *
 428          * @param string $body     Body with URLs
 429          * @param string $basepath Base path that is used to complete the URL
 430          *
 431          * @return string Body with expanded URLs
 432          */
 433         private static function qualifyURLs($body, $basepath)
 434         {
 435                 $URLSearchString = "^\[\]";
 436
 437                 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
 438                         "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
 439                         "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 440                         "/\[img\](.*?)\[\/img\]/ism",
 441                         "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 442                         "/\[zmg\](.*?)\[\/zmg\]/ism",
 443                         "/\[video\](.*?)\[\/video\]/ism",
 444                         "/\[audio\](.*?)\[\/audio\]/ism",
 445                 ];
 446
 447                 foreach ($matches as $match) {
 448                         $body = preg_replace_callback(
 449                                 $match,
 450                                 function ($match) use ($basepath) {
 451                                         return self::qualifyURLsSub($match, $basepath);
 452                                 },
 453                                 $body
 454                         );
 455                 }
 456                 return $body;
 457         }
 458
 459         private static function breakLines($line, $level, $wraplength = 75)
 460         {
 461                 if ($wraplength == 0) {
 462                         $wraplength = 2000000;
 463                 }
 464
 465                 $wraplen = $wraplength - $level;
 466
 467                 $newlines = [];
 468
 469                 do {
 470                         $oldline = $line;
 471
 472                         $subline = substr($line, 0, $wraplen);
 473
 474                         $pos = strrpos($subline, ' ');
 475
 476                         if ($pos == 0) {
 477                                 $pos = strpos($line, ' ');
 478                         }
 479
 480                         if (($pos > 0) && strlen($line) > $wraplen) {
 481                                 $newline = trim(substr($line, 0, $pos));
 482                                 if ($level > 0) {
 483                                         $newline = str_repeat(">", $level) . ' ' . $newline;
 484                                 }
 485
 486                                 $newlines[] = $newline . " ";
 487                                 $line = substr($line, $pos + 1);
 488                         }
 489                 } while ((strlen($line) > $wraplen) && !($oldline == $line));
 490
 491                 if ($level > 0) {
 492                         $line = str_repeat(">", $level) . ' ' . $line;
 493                 }
 494
 495                 $newlines[] = $line;
 496
 497                 return implode("\n", $newlines);
 498         }
 499
 500         private static function quoteLevel($message, $wraplength = 75)
 501         {
 502                 $lines = explode("\n", $message);
 503
 504                 $newlines = [];
 505                 $level = 0;
 506                 foreach ($lines as $line) {
 507                         $line = trim($line);
 508                         $startquote = false;
 509                         while (strpos("*" . $line, '[quote]') > 0) {
 510                                 $level++;
 511                                 $pos = strpos($line, '[quote]');
 512                                 $line = substr($line, 0, $pos) . substr($line, $pos + 7);
 513                                 $startquote = true;
 514                         }
 515
 516                         $currlevel = $level;
 517
 518                         while (strpos("*" . $line, '[/quote]') > 0) {
 519                                 $level--;
 520                                 if ($level < 0) {
 521                                         $level = 0;
 522                                 }
 523
 524                                 $pos = strpos($line, '[/quote]');
 525                                 $line = substr($line, 0, $pos) . substr($line, $pos + 8);
 526                         }
 527
 528                         if (!$startquote || ($line != '')) {
 529                                 $newlines[] = self::breakLines($line, $currlevel, $wraplength);
 530                         }
 531                 }
 532
 533                 return implode("\n", $newlines);
 534         }
 535
 536         private static function collectURLs($message)
 537         {
 538                 $pattern = '/<a.*?href="(.*?)".*?>(.*?)<\/a>/is';
 539                 preg_match_all($pattern, $message, $result, PREG_SET_ORDER);
 540
 541                 $urls = [];
 542                 foreach ($result as $treffer) {
 543                         $ignore = false;
 544
 545                         // A list of some links that should be ignored
 546                         $list = ["/user/", "/tag/", "/group/", "/profile/", "/search?search=", "/search?tag=", "mailto:", "/u/", "/node/",
 547                                 "//plus.google.com/", "//twitter.com/"];
 548                         foreach ($list as $listitem) {
 549                                 if (strpos($treffer[1], $listitem) !== false) {
 550                                         $ignore = true;
 551                                 }
 552                         }
 553
 554                         if ((strpos($treffer[1], "//twitter.com/") !== false) && (strpos($treffer[1], "/status/") !== false)) {
 555                                 $ignore = false;
 556                         }
 557
 558                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/posts") !== false)) {
 559                                 $ignore = false;
 560                         }
 561
 562                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/photos") !== false)) {
 563                                 $ignore = false;
 564                         }
 565
 566                         $ignore = $ignore || strpos($treffer[1], '#') === 0;
 567
 568                         if (!$ignore) {
 569                                 $urls[$treffer[1]] = $treffer[1];
 570                         }
 571                 }
 572
 573                 return $urls;
 574         }
 575
 576         /**
 577          * @param string $html
 578          * @param int    $wraplength Ensures individual lines aren't longer than this many characters. Doesn't break words.
 579          * @param bool   $compact    True: Completely strips image tags; False: Keeps image URLs
 580          * @return string
 581          */
 582         public static function toPlaintext(string $html, $wraplength = 75, $compact = false)
 583         {
 584                 $message = str_replace("\r", "", $html);
 585
 586                 $doc = new DOMDocument();
 587                 $doc->preserveWhiteSpace = false;
 588
 589                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 590
 591                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 592
 593                 $message = $doc->saveHTML();
 594                 // Remove eventual UTF-8 BOM
 595                 $message = str_replace("\xC3\x82\xC2\xA0", "", $message);
 596
 597                 // Collecting all links
 598                 $urls = self::collectURLs($message);
 599
 600                 @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 601
 602                 self::tagToBBCode($doc, 'html', [], '', '');
 603                 self::tagToBBCode($doc, 'body', [], '', '');
 604
 605                 if ($compact) {
 606                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 607                 } else {
 608                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 609                 }
 610
 611                 self::tagToBBCode($doc, 'br', [], "\n", '');
 612
 613                 self::tagToBBCode($doc, 'span', [], "", "");
 614                 self::tagToBBCode($doc, 'pre', [], "", "");
 615                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 616                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 617
 618                 self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 619
 620                 self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
 621
 622                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 623                 self::tagToBBCode($doc, 'td', [], "\t", "");
 624
 625                 self::tagToBBCode($doc, 'h1', [], "\n\n*", "*\n");
 626                 self::tagToBBCode($doc, 'h2', [], "\n\n*", "*\n");
 627                 self::tagToBBCode($doc, 'h3', [], "\n\n*", "*\n");
 628                 self::tagToBBCode($doc, 'h4', [], "\n\n*", "*\n");
 629                 self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
 630                 self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 631
 632                 if (!$compact) {
 633                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
 634                 } else {
 635                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' ', ' ');
 636                 }
 637
 638                 // Backward compatibility, [iframe] support has been removed in version 2020.12
 639                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], ' $1 ', '');
 640
 641                 $message = $doc->saveHTML();
 642
 643                 if (!$compact) {
 644                         $message = str_replace("[img]", "", $message);
 645                         $message = str_replace("[/img]", "", $message);
 646                 }
 647
 648                 // was ersetze ich da?
 649                 // Irgendein stoerrisches UTF-Zeug
 650                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 651
 652                 $message = str_replace("&nbsp;", " ", $message);
 653
 654                 // Aufeinanderfolgende DIVs
 655                 $message = preg_replace('=\r *\r=i', "\n", $message);
 656                 $message = str_replace("\r", "\n", $message);
 657
 658                 $message = strip_tags($message);
 659
 660                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 661
 662                 if (!$compact && ($message != '')) {
 663                         foreach ($urls as $id => $url) {
 664                                 if ($url != '' && strpos($message, $url) === false) {
 665                                         $message .= "\n" . $url . ' ';
 666                                 }
 667                         }
 668                 }
 669
 670                 $message = str_replace("\n«", "«\n", $message);
 671                 $message = str_replace("»\n", "\n»", $message);
 672
 673                 do {
 674                         $oldmessage = $message;
 675                         $message = str_replace("\n\n\n", "\n\n", $message);
 676                 } while ($oldmessage != $message);
 677
 678                 $message = self::quoteLevel(trim($message), $wraplength);
 679
 680                 return trim($message);
 681         }
 682
 683         /**
 684          * Converts provided HTML code to Markdown. The hardwrap parameter maximizes
 685          * compatibility with Diaspora in spite of the Markdown standards.
 686          *
 687          * @param string $html
 688          * @return string
 689          */
 690         public static function toMarkdown($html)
 691         {
 692                 $converter = new HtmlConverter(['hard_break' => true]);
 693                 $markdown = $converter->convert($html);
 694
 695                 return $markdown;
 696         }
 697
 698         /**
 699          * Convert video HTML to BBCode tags
 700          *
 701          * @param string $s
 702          * @return string
 703          */
 704         public static function toBBCodeVideo($s)
 705         {
 706                 $s = preg_replace(
 707                         '#<object[^>]+>(.*?)https?://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+)(.*?)</object>#ism',
 708                         '[youtube]$2[/youtube]',
 709                         $s
 710                 );
 711
 712                 $s = preg_replace(
 713                         '#<iframe[^>](.*?)https?://www.youtube.com/embed/([A-Za-z0-9\-_=]+)(.*?)</iframe>#ism',
 714                         '[youtube]$2[/youtube]',
 715                         $s
 716                 );
 717
 718                 $s = preg_replace(
 719                         '#<iframe[^>](.*?)https?://player.vimeo.com/video/([0-9]+)(.*?)</iframe>#ism',
 720                         '[vimeo]$2[/vimeo]',
 721                         $s
 722                 );
 723
 724                 return $s;
 725         }
 726
 727         /**
 728          * transform link href and img src from relative to absolute
 729          *
 730          * @param string $text
 731          * @param string $base base url
 732          * @return string
 733          */
 734         public static function relToAbs($text, $base)
 735         {
 736                 if (empty($base)) {
 737                         return $text;
 738                 }
 739
 740                 $base = rtrim($base, '/');
 741
 742                 $base2 = $base . "/";
 743
 744                 // Replace links
 745                 $pattern = "/<a([^>]*) href=\"(?!http|https|\/)([^\"]*)\"/";
 746                 $replace = "<a\${1} href=\"" . $base2 . "\${2}\"";
 747                 $text = preg_replace($pattern, $replace, $text);
 748
 749                 $pattern = "/<a([^>]*) href=\"(?!http|https)([^\"]*)\"/";
 750                 $replace = "<a\${1} href=\"" . $base . "\${2}\"";
 751                 $text = preg_replace($pattern, $replace, $text);
 752
 753                 // Replace images
 754                 $pattern = "/<img([^>]*) src=\"(?!http|https|\/)([^\"]*)\"/";
 755                 $replace = "<img\${1} src=\"" . $base2 . "\${2}\"";
 756                 $text = preg_replace($pattern, $replace, $text);
 757
 758                 $pattern = "/<img([^>]*) src=\"(?!http|https)([^\"]*)\"/";
 759                 $replace = "<img\${1} src=\"" . $base . "\${2}\"";
 760                 $text = preg_replace($pattern, $replace, $text);
 761
 762
 763                 // Done
 764                 return $text;
 765         }
 766
 767         /**
 768          * return div element with class 'clear'
 769          * @return string
 770          * @deprecated
 771          */
 772         public static function clearDiv()
 773         {
 774                 return '<div class="clear"></div>';
 775         }
 776
 777         /**
 778          * Loader for infinite scrolling
 779          *
 780          * @return string html for loader
 781          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 782          */
 783         public static function scrollLoader()
 784         {
 785                 $tpl = Renderer::getMarkupTemplate("scroll_loader.tpl");
 786                 return Renderer::replaceMacros($tpl, [
 787                         'wait' => DI::l10n()->t('Loading more entries...'),
 788                         'end' => DI::l10n()->t('The end')
 789                 ]);
 790         }
 791
 792         /**
 793          * Get html for contact block.
 794          *
 795          * @deprecated since version 2019.03
 796          * @see ContactBlock::getHTML()
 797          * @return string
 798          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 799          * @throws \ImagickException
 800          */
 801         public static function contactBlock()
 802         {
 803                 $a = DI::app();
 804
 805                 return ContactBlock::getHTML($a->profile);
 806         }
 807
 808         /**
 809          * Format contacts as picture links or as text links
 810          *
 811          * @param array   $contact  Array with contacts which contains an array with
 812          *                          int 'id' => The ID of the contact
 813          *                          int 'uid' => The user ID of the user who owns this data
 814          *                          string 'name' => The name of the contact
 815          *                          string 'url' => The url to the profile page of the contact
 816          *                          string 'addr' => The webbie of the contact (e.g.) username@friendica.com
 817          *                          string 'network' => The network to which the contact belongs to
 818          *                          string 'thumb' => The contact picture
 819          *                          string 'click' => js code which is performed when clicking on the contact
 820          * @param boolean $redirect If true try to use the redir url if it's possible
 821          * @param string  $class    CSS class for the
 822          * @param boolean $textmode If true display the contacts as text links
 823          *                          if false display the contacts as picture links
 824          * @return string Formatted html
 825          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 826          * @throws \ImagickException
 827          */
 828         public static function micropro($contact, $redirect = false, $class = '', $textmode = false)
 829         {
 830                 // Use the contact URL if no address is available
 831                 if (empty($contact['addr'])) {
 832                         $contact["addr"] = $contact["url"];
 833                 }
 834
 835                 $url = $contact['url'];
 836                 $sparkle = '';
 837                 $redir = false;
 838
 839                 if ($redirect) {
 840                         $url = Contact::magicLink($contact['url']);
 841                         if (strpos($url, 'redir/') === 0) {
 842                                 $sparkle = ' sparkle';
 843                         }
 844                 }
 845
 846                 // If there is some js available we don't need the url
 847                 if (!empty($contact['click'])) {
 848                         $url = '';
 849                 }
 850
 851                 return Renderer::replaceMacros(Renderer::getMarkupTemplate($textmode ? 'micropro_txt.tpl' : 'micropro_img.tpl'), [
 852                         '$click' => $contact['click'] ?? '',
 853                         '$class' => $class,
 854                         '$url' => $url,
 855                         '$photo' => Contact::getThumb($contact),
 856                         '$name' => $contact['name'],
 857                         'title' => $contact['name'] . ' [' . $contact['addr'] . ']',
 858                         '$parkle' => $sparkle,
 859                         '$redir' => $redir
 860                 ]);
 861         }
 862
 863         /**
 864          * Search box.
 865          *
 866          * @param string $s     Search query.
 867          * @param string $id    HTML id
 868          * @param string $url   Search url.
 869          * @param bool   $aside Display the search widgit aside.
 870          *
 871          * @return string Formatted HTML.
 872          * @throws \Exception
 873          */
 874         public static function search($s, $id = 'search-box', $aside = true)
 875         {
 876                 $mode = 'text';
 877
 878                 if (strpos($s, '#') === 0) {
 879                         $mode = 'tag';
 880                 }
 881                 $save_label = $mode === 'text' ? DI::l10n()->t('Save') : DI::l10n()->t('Follow');
 882
 883                 $values = [
 884                         '$s'            => $s,
 885                         '$q'            => urlencode($s),
 886                         '$id'           => $id,
 887                         '$search_label' => DI::l10n()->t('Search'),
 888                         '$save_label'   => $save_label,
 889                         '$search_hint'  => DI::l10n()->t('@name, !forum, #tags, content'),
 890                         '$mode'         => $mode,
 891                         '$return_url'   => urlencode(Search::getSearchPath($s)),
 892                 ];
 893
 894                 if (!$aside) {
 895                         $values['$search_options'] = [
 896                                 'fulltext' => DI::l10n()->t('Full Text'),
 897                                 'tags'     => DI::l10n()->t('Tags'),
 898                                 'contacts' => DI::l10n()->t('Contacts')
 899                         ];
 900
 901                         if (DI::config()->get('system', 'poco_local_search')) {
 902                                 $values['$searchoption']['forums'] = DI::l10n()->t('Forums');
 903                         }
 904                 }
 905
 906                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 907         }
 908
 909         /**
 910          * Replace naked text hyperlink with HTML formatted hyperlink
 911          *
 912          * @param string $s
 913          * @return string
 914          */
 915         public static function toLink($s)
 916         {
 917                 $s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank" rel="noopener noreferrer">$1</a>', $s);
 918                 $s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
 919                 return $s;
 920         }
 921
 922         /**
 923          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
 924          *
 925          * Reasons are expected to have been translated already.
 926          *
 927          * @param string $html
 928          * @param array  $reasons
 929          * @return string
 930          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 931          */
 932         public static function applyContentFilter($html, array $reasons)
 933         {
 934                 if (count($reasons)) {
 935                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
 936                         $html = Renderer::replaceMacros($tpl, [
 937                                 '$reasons'   => $reasons,
 938                                 '$rnd'       => Strings::getRandomHex(8),
 939                                 '$openclose' => DI::l10n()->t('Click to open/close'),
 940                                 '$html'      => $html
 941                         ]);
 942                 }
 943
 944                 return $html;
 945         }
 946
 947         /**
 948          * replace html amp entity with amp char
 949          * @param string $s
 950          * @return string
 951          */
 952         public static function unamp($s)
 953         {
 954                 return str_replace('&amp;', '&', $s);
 955         }
 956 }