src/Content/Text/HTML.php

   1 <?php
   2 /**
   3  * @file src/Content/Text/HTML.php
   4  */
   5
   6 namespace Friendica\Content\Text;
   7
   8 use DOMDocument;
   9 use DOMXPath;
  10 use Friendica\Core\Hook;
  11 use Friendica\Core\L10n;
  12 use Friendica\Core\Config;
  13 use Friendica\Core\PConfig;
  14 use Friendica\Core\Protocol;
  15 use Friendica\Core\Renderer;
  16 use Friendica\Database\DBA;
  17 use Friendica\Model\Contact;
  18 use Friendica\Util\Network;
  19 use Friendica\Util\Proxy as ProxyUtils;
  20 use Friendica\Util\Strings;
  21 use Friendica\Util\XML;
  22 use League\HTMLToMarkdown\HtmlConverter;
  23
  24 class HTML
  25 {
  26         public static function sanitizeCSS($input)
  27         {
  28                 $cleaned = "";
  29
  30                 $input = strtolower($input);
  31
  32                 for ($i = 0; $i < strlen($input); $i++) {
  33                         $char = substr($input, $i, 1);
  34
  35                         if (($char >= "a") && ($char <= "z")) {
  36                                 $cleaned .= $char;
  37                         }
  38
  39                         if (!(strpos(" #;:0123456789-_.%", $char) === false)) {
  40                                 $cleaned .= $char;
  41                         }
  42                 }
  43
  44                 return $cleaned;
  45         }
  46
  47         private static function tagToBBCode(DOMDocument $doc, $tag, $attributes, $startbb, $endbb)
  48         {
  49                 do {
  50                         $done = self::tagToBBCodeSub($doc, $tag, $attributes, $startbb, $endbb);
  51                 } while ($done);
  52         }
  53
  54         private static function tagToBBCodeSub(DOMDocument $doc, $tag, $attributes, $startbb, $endbb)
  55         {
  56                 $savestart = str_replace('$', '\x01', $startbb);
  57                 $replace = false;
  58
  59                 $xpath = new DOMXPath($doc);
  60
  61                 $list = $xpath->query("//" . $tag);
  62                 foreach ($list as $node) {
  63                         $attr = [];
  64                         if ($node->attributes->length) {
  65                                 foreach ($node->attributes as $attribute) {
  66                                         $attr[$attribute->name] = $attribute->value;
  67                                 }
  68                         }
  69
  70                         $replace = true;
  71
  72                         $startbb = $savestart;
  73
  74                         $i = 0;
  75
  76                         foreach ($attributes as $attribute => $value) {
  77                                 $startbb = str_replace('\x01' . ++$i, '$1', $startbb);
  78                                 if (strpos('*' . $startbb, '$1') > 0) {
  79                                         if ($replace && (@$attr[$attribute] != '')) {
  80                                                 $startbb = preg_replace($value, $startbb, $attr[$attribute], -1, $count);
  81
  82                                                 // If nothing could be changed
  83                                                 if ($count == 0) {
  84                                                         $replace = false;
  85                                                 }
  86                                         } else {
  87                                                 $replace = false;
  88                                         }
  89                                 } else {
  90                                         if (@$attr[$attribute] != $value) {
  91                                                 $replace = false;
  92                                         }
  93                                 }
  94                         }
  95
  96                         if ($replace) {
  97                                 $StartCode = $doc->createTextNode($startbb);
  98                                 $EndCode = $doc->createTextNode($endbb);
  99
 100                                 $node->parentNode->insertBefore($StartCode, $node);
 101
 102                                 if ($node->hasChildNodes()) {
 103                                         foreach ($node->childNodes as $child) {
 104                                                 $newNode = $child->cloneNode(true);
 105                                                 $node->parentNode->insertBefore($newNode, $node);
 106                                         }
 107                                 }
 108
 109                                 $node->parentNode->insertBefore($EndCode, $node);
 110                                 $node->parentNode->removeChild($node);
 111                         }
 112                 }
 113
 114                 return $replace;
 115         }
 116
 117         /**
 118          * Made by: ike@piratenpartei.de
 119          * Originally made for the syncom project: http://wiki.piratenpartei.de/Syncom
 120          *                    https://github.com/annando/Syncom
 121          *
 122          * @brief Converter for HTML to BBCode
 123          * @param string $message
 124          * @param string $basepath
 125          * @return string
 126          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 127          */
 128         public static function toBBCode($message, $basepath = '')
 129         {
 130                 $message = str_replace("\r", "", $message);
 131
 132                 // Removing code blocks before the whitespace removal processing below
 133                 $codeblocks = [];
 134                 $message = preg_replace_callback(
 135                         '#<pre><code(?: class="language-([^"]*)")?>(.*)</code></pre>#iUs',
 136                         function ($matches) use (&$codeblocks) {
 137                                 $return = '[codeblock-' . count($codeblocks) . ']';
 138
 139                                 $prefix = '[code]';
 140                                 if ($matches[1] != '') {
 141                                         $prefix = '[code=' . $matches[1] . ']';
 142                                 }
 143
 144                                 $codeblocks[] = $prefix . PHP_EOL . trim($matches[2]) . PHP_EOL . '[/code]';
 145                                 return $return;
 146                         },
 147                         $message
 148                 );
 149
 150                 $message = str_replace(
 151                         [
 152                                 "<li><p>",
 153                                 "</p></li>",
 154                         ],
 155                         [
 156                                 "<li>",
 157                                 "</li>",
 158                         ],
 159                         $message
 160                 );
 161
 162                 // remove namespaces
 163                 $message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
 164                 $message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
 165
 166                 $doc = new DOMDocument();
 167                 $doc->preserveWhiteSpace = false;
 168
 169                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 170
 171                 @$doc->loadHTML($message);
 172
 173                 XML::deleteNode($doc, 'style');
 174                 XML::deleteNode($doc, 'head');
 175                 XML::deleteNode($doc, 'title');
 176                 XML::deleteNode($doc, 'meta');
 177                 XML::deleteNode($doc, 'xml');
 178                 XML::deleteNode($doc, 'removeme');
 179
 180                 $xpath = new DomXPath($doc);
 181                 $list = $xpath->query("//pre");
 182                 foreach ($list as $node) {
 183                         // Ensure to escape unescaped & - they will otherwise raise a warning
 184                         $safe_value = preg_replace('/&(?!\w+;)/', '&amp;', $node->nodeValue);
 185                         $node->nodeValue = str_replace("\n", "\r", $safe_value);
 186                 }
 187
 188                 $message = $doc->saveHTML();
 189                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br />", " ", ""], $message);
 190                 $message = preg_replace('= [\s]*=i', " ", $message);
 191                 @$doc->loadHTML($message);
 192
 193                 self::tagToBBCode($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($doc, 'table', [], "", "");
 272                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 273                 self::tagToBBCode($doc, 'td', [], "\t", "");
 274                 //self::node2BBCode($doc, 'table', array(), "[table]", "[/table]");
 275                 //self::node2BBCode($doc, 'th', array(), "[th]", "[/th]");
 276                 //self::node2BBCode($doc, 'tr', array(), "[tr]", "[/tr]");
 277                 //self::node2BBCode($doc, 'td', array(), "[td]", "[/td]");
 278                 //self::node2BBCode($doc, 'h1', array(), "\n\n[size=xx-large][b]", "[/b][/size]\n");
 279                 //self::node2BBCode($doc, 'h2', array(), "\n\n[size=x-large][b]", "[/b][/size]\n");
 280                 //self::node2BBCode($doc, 'h3', array(), "\n\n[size=large][b]", "[/b][/size]\n");
 281                 //self::node2BBCode($doc, 'h4', array(), "\n\n[size=medium][b]", "[/b][/size]\n");
 282                 //self::node2BBCode($doc, 'h5', array(), "\n\n[size=small][b]", "[/b][/size]\n");
 283                 //self::node2BBCode($doc, 'h6', array(), "\n\n[size=x-small][b]", "[/b][/size]\n");
 284
 285                 self::tagToBBCode($doc, 'h1', [], "[h1]", "[/h1]");
 286                 self::tagToBBCode($doc, 'h2', [], "[h2]", "[/h2]");
 287                 self::tagToBBCode($doc, 'h3', [], "[h3]", "[/h3]");
 288                 self::tagToBBCode($doc, 'h4', [], "[h4]", "[/h4]");
 289                 self::tagToBBCode($doc, 'h5', [], "[h5]", "[/h5]");
 290                 self::tagToBBCode($doc, 'h6', [], "[h6]", "[/h6]");
 291
 292                 self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
 293                 self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 294
 295                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]');
 296                 self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]');
 297
 298
 299                 self::tagToBBCode($doc, 'video', ['src' => '/(.+)/'], '[video]$1', '[/video]');
 300                 self::tagToBBCode($doc, 'audio', ['src' => '/(.+)/'], '[audio]$1', '[/audio]');
 301                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], '[iframe]$1', '[/iframe]');
 302
 303                 self::tagToBBCode($doc, 'key', [], '[code]', '[/code]');
 304                 self::tagToBBCode($doc, 'code', [], '[code]', '[/code]');
 305
 306                 $message = $doc->saveHTML();
 307
 308                 // I'm removing something really disturbing
 309                 // Don't know exactly what it is
 310                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 311
 312                 $message = str_replace("&nbsp;", " ", $message);
 313
 314                 // removing multiple DIVs
 315                 $message = preg_replace('=\r *\r=i', "\n", $message);
 316                 $message = str_replace("\r", "\n", $message);
 317
 318                 Hook::callAll('html2bbcode', $message);
 319
 320                 $message = strip_tags($message);
 321
 322                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 323
 324                 $message = str_replace(["<"], ["&lt;"], $message);
 325
 326                 // remove quotes if they don't make sense
 327                 $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
 328
 329                 $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
 330                 $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
 331
 332                 do {
 333                         $oldmessage = $message;
 334                         $message = str_replace("\n \n", "\n\n", $message);
 335                 } while ($oldmessage != $message);
 336
 337                 do {
 338                         $oldmessage = $message;
 339                         $message = str_replace("\n\n\n", "\n\n", $message);
 340                 } while ($oldmessage != $message);
 341
 342                 do {
 343                         $oldmessage = $message;
 344                         $message = str_replace(
 345                                 [
 346                                 "[/size]\n\n",
 347                                 "\n[hr]",
 348                                 "[hr]\n",
 349                                 "\n[list",
 350                                 "[/list]\n",
 351                                 "\n[/",
 352                                 "[list]\n",
 353                                 "[list=1]\n",
 354                                 "\n[*]"],
 355                                 [
 356                                 "[/size]\n",
 357                                 "[hr]",
 358                                 "[hr]",
 359                                 "[list",
 360                                 "[/list]",
 361                                 "[/",
 362                                 "[list]",
 363                                 "[list=1]",
 364                                 "[*]"],
 365                                 $message
 366                         );
 367                 } while ($message != $oldmessage);
 368
 369                 $message = str_replace(
 370                         ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
 371                         ['[b]', '[/b]', '[i]', '[/i]'],
 372                         $message
 373                 );
 374
 375                 // Handling Yahoo style of mails
 376                 $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
 377
 378                 // Restore code blocks
 379                 $message = preg_replace_callback(
 380                         '#\[codeblock-([0-9]+)\]#iU',
 381                         function ($matches) use ($codeblocks) {
 382                                 $return = '';
 383                                 if (isset($codeblocks[intval($matches[1])])) {
 384                                         $return = $codeblocks[$matches[1]];
 385                                 }
 386                                 return $return;
 387                         },
 388                         $message
 389                 );
 390
 391                 $message = trim($message);
 392
 393                 if ($basepath != '') {
 394                         $message = self::qualifyURLs($message, $basepath);
 395                 }
 396
 397                 return $message;
 398         }
 399
 400         /**
 401          * @brief Sub function to complete incomplete URL
 402          *
 403          * @param array  $matches  Result of preg_replace_callback
 404          * @param string $basepath Basepath that is used to complete the URL
 405          *
 406          * @return string The expanded URL
 407          */
 408         private static function qualifyURLsSub($matches, $basepath)
 409         {
 410                 $base = parse_url($basepath);
 411                 unset($base['query']);
 412                 unset($base['fragment']);
 413
 414                 $link = $matches[0];
 415                 $url = $matches[1];
 416
 417                 $parts = array_merge($base, parse_url($url));
 418                 $url2 = Network::unparseURL($parts);
 419
 420                 return str_replace($url, $url2, $link);
 421         }
 422
 423         /**
 424          * @brief Complete incomplete URLs in BBCode
 425          *
 426          * @param string $body     Body with URLs
 427          * @param string $basepath Base path that is used to complete the URL
 428          *
 429          * @return string Body with expanded URLs
 430          */
 431         private static function qualifyURLs($body, $basepath)
 432         {
 433                 $URLSearchString = "^\[\]";
 434
 435                 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
 436                         "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
 437                         "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 438                         "/\[img\](.*?)\[\/img\]/ism",
 439                         "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
 440                         "/\[zmg\](.*?)\[\/zmg\]/ism",
 441                         "/\[video\](.*?)\[\/video\]/ism",
 442                         "/\[audio\](.*?)\[\/audio\]/ism",
 443                 ];
 444
 445                 foreach ($matches as $match) {
 446                         $body = preg_replace_callback(
 447                                 $match,
 448                                 function ($match) use ($basepath) {
 449                                         return self::qualifyURLsSub($match, $basepath);
 450                                 },
 451                                 $body
 452                         );
 453                 }
 454                 return $body;
 455         }
 456
 457         private static function breakLines($line, $level, $wraplength = 75)
 458         {
 459                 if ($wraplength == 0) {
 460                         $wraplength = 2000000;
 461                 }
 462
 463                 $wraplen = $wraplength - $level;
 464
 465                 $newlines = [];
 466
 467                 do {
 468                         $oldline = $line;
 469
 470                         $subline = substr($line, 0, $wraplen);
 471
 472                         $pos = strrpos($subline, ' ');
 473
 474                         if ($pos == 0) {
 475                                 $pos = strpos($line, ' ');
 476                         }
 477
 478                         if (($pos > 0) && strlen($line) > $wraplen) {
 479                                 $newline = trim(substr($line, 0, $pos));
 480                                 if ($level > 0) {
 481                                         $newline = str_repeat(">", $level) . ' ' . $newline;
 482                                 }
 483
 484                                 $newlines[] = $newline . " ";
 485                                 $line = substr($line, $pos + 1);
 486                         }
 487                 } while ((strlen($line) > $wraplen) && !($oldline == $line));
 488
 489                 if ($level > 0) {
 490                         $line = str_repeat(">", $level) . ' ' . $line;
 491                 }
 492
 493                 $newlines[] = $line;
 494
 495                 return implode($newlines, "\n");
 496         }
 497
 498         private static function quoteLevel($message, $wraplength = 75)
 499         {
 500                 $lines = explode("\n", $message);
 501
 502                 $newlines = [];
 503                 $level = 0;
 504                 foreach ($lines as $line) {
 505                         $line = trim($line);
 506                         $startquote = false;
 507                         while (strpos("*" . $line, '[quote]') > 0) {
 508                                 $level++;
 509                                 $pos = strpos($line, '[quote]');
 510                                 $line = substr($line, 0, $pos) . substr($line, $pos + 7);
 511                                 $startquote = true;
 512                         }
 513
 514                         $currlevel = $level;
 515
 516                         while (strpos("*" . $line, '[/quote]') > 0) {
 517                                 $level--;
 518                                 if ($level < 0) {
 519                                         $level = 0;
 520                                 }
 521
 522                                 $pos = strpos($line, '[/quote]');
 523                                 $line = substr($line, 0, $pos) . substr($line, $pos + 8);
 524                         }
 525
 526                         if (!$startquote || ($line != '')) {
 527                                 $newlines[] = self::breakLines($line, $currlevel, $wraplength);
 528                         }
 529                 }
 530
 531                 return implode($newlines, "\n");
 532         }
 533
 534         private static function collectURLs($message)
 535         {
 536                 $pattern = '/<a.*?href="(.*?)".*?>(.*?)<\/a>/is';
 537                 preg_match_all($pattern, $message, $result, PREG_SET_ORDER);
 538
 539                 $urls = [];
 540                 foreach ($result as $treffer) {
 541                         $ignore = false;
 542
 543                         // A list of some links that should be ignored
 544                         $list = ["/user/", "/tag/", "/group/", "/profile/", "/search?search=", "/search?tag=", "mailto:", "/u/", "/node/",
 545                                 "//plus.google.com/", "//twitter.com/"];
 546                         foreach ($list as $listitem) {
 547                                 if (strpos($treffer[1], $listitem) !== false) {
 548                                         $ignore = true;
 549                                 }
 550                         }
 551
 552                         if ((strpos($treffer[1], "//twitter.com/") !== false) && (strpos($treffer[1], "/status/") !== false)) {
 553                                 $ignore = false;
 554                         }
 555
 556                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/posts") !== false)) {
 557                                 $ignore = false;
 558                         }
 559
 560                         if ((strpos($treffer[1], "//plus.google.com/") !== false) && (strpos($treffer[1], "/photos") !== false)) {
 561                                 $ignore = false;
 562                         }
 563
 564                         if (!$ignore) {
 565                                 $urls[$treffer[1]] = $treffer[1];
 566                         }
 567                 }
 568
 569                 return $urls;
 570         }
 571
 572         public static function toPlaintext($html, $wraplength = 75, $compact = false)
 573         {
 574                 $message = str_replace("\r", "", $html);
 575
 576                 $doc = new DOMDocument();
 577                 $doc->preserveWhiteSpace = false;
 578
 579                 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 580
 581                 @$doc->loadHTML($message);
 582
 583                 $xpath = new DOMXPath($doc);
 584                 $list = $xpath->query("//pre");
 585                 foreach ($list as $node) {
 586                         $node->nodeValue = str_replace("\n", "\r", $node->nodeValue);
 587                 }
 588
 589                 $message = $doc->saveHTML();
 590                 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br>", " ", ""], $message);
 591                 $message = preg_replace('= [\s]*=i', " ", $message);
 592
 593                 // Collecting all links
 594                 $urls = self::collectURLs($message);
 595
 596                 @$doc->loadHTML($message);
 597
 598                 self::tagToBBCode($doc, 'html', [], '', '');
 599                 self::tagToBBCode($doc, 'body', [], '', '');
 600
 601                 // MyBB-Auszeichnungen
 602                 /*
 603                   self::node2BBCode($doc, 'span', array('style'=>'text-decoration: underline;'), '_', '_');
 604                   self::node2BBCode($doc, 'span', array('style'=>'font-style: italic;'), '/', '/');
 605                   self::node2BBCode($doc, 'span', array('style'=>'font-weight: bold;'), '*', '*');
 606
 607                   self::node2BBCode($doc, 'strong', array(), '*', '*');
 608                   self::node2BBCode($doc, 'b', array(), '*', '*');
 609                   self::node2BBCode($doc, 'i', array(), '/', '/');
 610                   self::node2BBCode($doc, 'u', array(), '_', '_');
 611                  */
 612
 613                 if ($compact) {
 614                         self::tagToBBCode($doc, 'blockquote', [], "»", "«");
 615                 } else {
 616                         self::tagToBBCode($doc, 'blockquote', [], '[quote]', "[/quote]\n");
 617                 }
 618
 619                 self::tagToBBCode($doc, 'br', [], "\n", '');
 620
 621                 self::tagToBBCode($doc, 'span', [], "", "");
 622                 self::tagToBBCode($doc, 'pre', [], "", "");
 623                 self::tagToBBCode($doc, 'div', [], "\r", "\r");
 624                 self::tagToBBCode($doc, 'p', [], "\n", "\n");
 625
 626                 //self::node2BBCode($doc, 'ul', array(), "\n[list]", "[/list]\n");
 627                 //self::node2BBCode($doc, 'ol', array(), "\n[list=1]", "[/list]\n");
 628                 self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 629
 630                 self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
 631
 632                 self::tagToBBCode($doc, 'tr', [], "\n", "");
 633                 self::tagToBBCode($doc, 'td', [], "\t", "");
 634
 635                 self::tagToBBCode($doc, 'h1', [], "\n\n*", "*\n");
 636                 self::tagToBBCode($doc, 'h2', [], "\n\n*", "*\n");
 637                 self::tagToBBCode($doc, 'h3', [], "\n\n*", "*\n");
 638                 self::tagToBBCode($doc, 'h4', [], "\n\n*", "*\n");
 639                 self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
 640                 self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 641
 642                 // Problem: there is no reliable way to detect if it is a link to a tag or profile
 643                 //self::node2BBCode($doc, 'a', array('href'=>'/(.+)/'), ' $1 ', ' ', true);
 644                 //self::node2BBCode($doc, 'a', array('href'=>'/(.+)/', 'rel'=>'oembed'), ' $1 ', '', true);
 645                 //self::node2BBCode($doc, 'img', array('alt'=>'/(.+)/'), '$1', '');
 646                 //self::node2BBCode($doc, 'img', array('title'=>'/(.+)/'), '$1', '');
 647                 //self::node2BBCode($doc, 'img', array(), '', '');
 648                 if (!$compact) {
 649                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
 650                 } else {
 651                         self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' ', ' ');
 652                 }
 653
 654                 self::tagToBBCode($doc, 'iframe', ['src' => '/(.+)/'], ' $1 ', '');
 655
 656                 $message = $doc->saveHTML();
 657
 658                 if (!$compact) {
 659                         $message = str_replace("[img]", "", $message);
 660                         $message = str_replace("[/img]", "", $message);
 661                 }
 662
 663                 // was ersetze ich da?
 664                 // Irgendein stoerrisches UTF-Zeug
 665                 $message = str_replace(chr(194) . chr(160), ' ', $message);
 666
 667                 $message = str_replace("&nbsp;", " ", $message);
 668
 669                 // Aufeinanderfolgende DIVs
 670                 $message = preg_replace('=\r *\r=i', "\n", $message);
 671                 $message = str_replace("\r", "\n", $message);
 672
 673                 $message = strip_tags($message);
 674
 675                 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
 676
 677                 if (!$compact && ($message != '')) {
 678                         foreach ($urls as $id => $url) {
 679                                 if ($url != '' && strpos($message, $url) === false) {
 680                                         $message .= "\n" . $url . ' ';
 681                                 }
 682                         }
 683                 }
 684
 685                 $message = str_replace("\n«", "«\n", $message);
 686                 $message = str_replace("»\n", "\n»", $message);
 687
 688                 do {
 689                         $oldmessage = $message;
 690                         $message = str_replace("\n\n\n", "\n\n", $message);
 691                 } while ($oldmessage != $message);
 692
 693                 $message = self::quoteLevel(trim($message), $wraplength);
 694
 695                 return trim($message);
 696         }
 697
 698         /**
 699          * Converts provided HTML code to Markdown. The hardwrap parameter maximizes
 700          * compatibility with Diaspora in spite of the Markdown standards.
 701          *
 702          * @param string $html
 703          * @return string
 704          */
 705         public static function toMarkdown($html)
 706         {
 707                 $converter = new HtmlConverter(['hard_break' => true]);
 708                 $markdown = $converter->convert($html);
 709
 710                 return $markdown;
 711         }
 712
 713         /**
 714          * @brief Convert video HTML to BBCode tags
 715          *
 716          * @param string $s
 717          * @return string
 718          */
 719         public static function toBBCodeVideo($s)
 720         {
 721                 $s = preg_replace(
 722                         '#<object[^>]+>(.*?)https?://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+)(.*?)</object>#ism',
 723                         '[youtube]$2[/youtube]',
 724                         $s
 725                 );
 726
 727                 $s = preg_replace(
 728                         '#<iframe[^>](.*?)https?://www.youtube.com/embed/([A-Za-z0-9\-_=]+)(.*?)</iframe>#ism',
 729                         '[youtube]$2[/youtube]',
 730                         $s
 731                 );
 732
 733                 $s = preg_replace(
 734                         '#<iframe[^>](.*?)https?://player.vimeo.com/video/([0-9]+)(.*?)</iframe>#ism',
 735                         '[vimeo]$2[/vimeo]',
 736                         $s
 737                 );
 738
 739                 return $s;
 740         }
 741
 742         /**
 743          * transform link href and img src from relative to absolute
 744          *
 745          * @param string $text
 746          * @param string $base base url
 747          * @return string
 748          */
 749         public static function relToAbs($text, $base)
 750         {
 751                 if (empty($base)) {
 752                         return $text;
 753                 }
 754
 755                 $base = rtrim($base, '/');
 756
 757                 $base2 = $base . "/";
 758
 759                 // Replace links
 760                 $pattern = "/<a([^>]*) href=\"(?!http|https|\/)([^\"]*)\"/";
 761                 $replace = "<a\${1} href=\"" . $base2 . "\${2}\"";
 762                 $text = preg_replace($pattern, $replace, $text);
 763
 764                 $pattern = "/<a([^>]*) href=\"(?!http|https)([^\"]*)\"/";
 765                 $replace = "<a\${1} href=\"" . $base . "\${2}\"";
 766                 $text = preg_replace($pattern, $replace, $text);
 767
 768                 // Replace images
 769                 $pattern = "/<img([^>]*) src=\"(?!http|https|\/)([^\"]*)\"/";
 770                 $replace = "<img\${1} src=\"" . $base2 . "\${2}\"";
 771                 $text = preg_replace($pattern, $replace, $text);
 772
 773                 $pattern = "/<img([^>]*) src=\"(?!http|https)([^\"]*)\"/";
 774                 $replace = "<img\${1} src=\"" . $base . "\${2}\"";
 775                 $text = preg_replace($pattern, $replace, $text);
 776
 777
 778                 // Done
 779                 return $text;
 780         }
 781
 782         /**
 783          * return div element with class 'clear'
 784          * @return string
 785          * @deprecated
 786          */
 787         public static function clearDiv()
 788         {
 789                 return '<div class="clear"></div>';
 790         }
 791
 792         /**
 793          * Loader for infinite scrolling
 794          *
 795          * @return string html for loader
 796          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 797          */
 798         public static function scrollLoader()
 799         {
 800                 $tpl = Renderer::getMarkupTemplate("scroll_loader.tpl");
 801                 return Renderer::replaceMacros($tpl, [
 802                         'wait' => L10n::t('Loading more entries...'),
 803                         'end' => L10n::t('The end')
 804                 ]);
 805         }
 806
 807         /**
 808          * Get html for contact block.
 809          *
 810          * @template contact_block.tpl
 811          * @hook     contact_block_end (contacts=>array, output=>string)
 812          * @return string
 813          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 814          * @throws \ImagickException
 815          */
 816         public static function contactBlock()
 817         {
 818                 $o = '';
 819                 $a = \get_app();
 820
 821                 $shown = PConfig::get($a->profile['uid'], 'system', 'display_friend_count', 24);
 822                 if ($shown == 0) {
 823                         return;
 824                 }
 825
 826                 if (!is_array($a->profile) || $a->profile['hide-friends']) {
 827                         return $o;
 828                 }
 829
 830                 $r = q("SELECT COUNT(*) AS `total` FROM `contact`
 831                                 WHERE `uid` = %d AND NOT `self` AND NOT `blocked`
 832                                         AND NOT `pending` AND NOT `hidden` AND NOT `archive`
 833                                         AND `network` IN ('%s', '%s', '%s')",
 834                         intval($a->profile['uid']),
 835                         DBA::escape(Protocol::DFRN),
 836                         DBA::escape(Protocol::OSTATUS),
 837                         DBA::escape(Protocol::DIASPORA)
 838                 );
 839
 840                 if (DBA::isResult($r)) {
 841                         $total = intval($r[0]['total']);
 842                 }
 843
 844                 if (!$total) {
 845                         $contacts = L10n::t('No contacts');
 846                         $micropro = null;
 847                 } else {
 848                         // Splitting the query in two parts makes it much faster
 849                         $r = q("SELECT `id` FROM `contact`
 850                                         WHERE `uid` = %d AND NOT `self` AND NOT `blocked`
 851                                                 AND NOT `pending` AND NOT `hidden` AND NOT `archive`
 852                                                 AND `network` IN ('%s', '%s', '%s')
 853                                         ORDER BY RAND() LIMIT %d",
 854                                 intval($a->profile['uid']),
 855                                 DBA::escape(Protocol::DFRN),
 856                                 DBA::escape(Protocol::OSTATUS),
 857                                 DBA::escape(Protocol::DIASPORA),
 858                                 intval($shown)
 859                         );
 860
 861                         if (DBA::isResult($r)) {
 862                                 $contacts = [];
 863                                 foreach ($r as $contact) {
 864                                         $contacts[] = $contact["id"];
 865                                 }
 866
 867                                 $r = q("SELECT `id`, `uid`, `addr`, `url`, `name`, `thumb`, `network` FROM `contact` WHERE `id` IN (%s)",
 868                                         DBA::escape(implode(",", $contacts))
 869                                 );
 870
 871                                 if (DBA::isResult($r)) {
 872                                         $contacts = L10n::tt('%d Contact', '%d Contacts', $total);
 873                                         $micropro = [];
 874                                         foreach ($r as $rr) {
 875                                                 $micropro[] = self::micropro($rr, true, 'mpfriend');
 876                                         }
 877                                 }
 878                         }
 879                 }
 880
 881                 $tpl = Renderer::getMarkupTemplate('contact_block.tpl');
 882                 $o = Renderer::replaceMacros($tpl, [
 883                         '$contacts' => $contacts,
 884                         '$nickname' => $a->profile['nickname'],
 885                         '$viewcontacts' => L10n::t('View Contacts'),
 886                         '$micropro' => $micropro,
 887                 ]);
 888
 889                 $arr = ['contacts' => $r, 'output' => $o];
 890
 891                 Hook::callAll('contact_block_end', $arr);
 892
 893                 return $o;
 894         }
 895
 896         /**
 897          * @brief Format contacts as picture links or as text links
 898          *
 899          * @param array   $contact  Array with contacts which contains an array with
 900          *                          int 'id' => The ID of the contact
 901          *                          int 'uid' => The user ID of the user who owns this data
 902          *                          string 'name' => The name of the contact
 903          *                          string 'url' => The url to the profile page of the contact
 904          *                          string 'addr' => The webbie of the contact (e.g.) username@friendica.com
 905          *                          string 'network' => The network to which the contact belongs to
 906          *                          string 'thumb' => The contact picture
 907          *                          string 'click' => js code which is performed when clicking on the contact
 908          * @param boolean $redirect If true try to use the redir url if it's possible
 909          * @param string  $class    CSS class for the
 910          * @param boolean $textmode If true display the contacts as text links
 911          *                          if false display the contacts as picture links
 912          * @return string Formatted html
 913          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 914          * @throws \ImagickException
 915          */
 916         public static function micropro($contact, $redirect = false, $class = '', $textmode = false)
 917         {
 918                 // Use the contact URL if no address is available
 919                 if (empty($contact['addr'])) {
 920                         $contact["addr"] = $contact["url"];
 921                 }
 922
 923                 $url = $contact['url'];
 924                 $sparkle = '';
 925                 $redir = false;
 926
 927                 if ($redirect) {
 928                         $url = Contact::magicLink($contact['url']);
 929                         if (strpos($url, 'redir/') === 0) {
 930                                 $sparkle = ' sparkle';
 931                         }
 932                 }
 933
 934                 // If there is some js available we don't need the url
 935                 if (!empty($contact['click'])) {
 936                         $url = '';
 937                 }
 938
 939                 return Renderer::replaceMacros(Renderer::getMarkupTemplate(($textmode)?'micropro_txt.tpl':'micropro_img.tpl'), [
 940                         '$click' => defaults($contact, 'click', ''),
 941                         '$class' => $class,
 942                         '$url' => $url,
 943                         '$photo' => ProxyUtils::proxifyUrl($contact['thumb'], false, ProxyUtils::SIZE_THUMB),
 944                         '$name' => $contact['name'],
 945                         'title' => $contact['name'] . ' [' . $contact['addr'] . ']',
 946                         '$parkle' => $sparkle,
 947                         '$redir' => $redir
 948                 ]);
 949         }
 950
 951         /**
 952          * Search box.
 953          *
 954          * @param string $s     Search query.
 955          * @param string $id    HTML id
 956          * @param string $url   Search url.
 957          * @param bool   $aside Display the search widgit aside.
 958          *
 959          * @return string Formatted HTML.
 960          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 961          */
 962         public static function search($s, $id = 'search-box', $url = 'search', $aside = true)
 963         {
 964                 $mode = 'text';
 965
 966                 if (strpos($s, '#') === 0) {
 967                         $mode = 'tag';
 968                 }
 969                 $save_label = $mode === 'text' ? L10n::t('Save') : L10n::t('Follow');
 970
 971                 $values = [
 972                                 '$s' => $s,
 973                                 '$id' => $id,
 974                                 '$action_url' => $url,
 975                                 '$search_label' => L10n::t('Search'),
 976                                 '$save_label' => $save_label,
 977                                 '$savedsearch' => 'savedsearch',
 978                                 '$search_hint' => L10n::t('@name, !forum, #tags, content'),
 979                                 '$mode' => $mode
 980                         ];
 981
 982                 if (!$aside) {
 983                         $values['$searchoption'] = [
 984                                                 L10n::t("Full Text"),
 985                                                 L10n::t("Tags"),
 986                                                 L10n::t("Contacts")];
 987
 988                         if (Config::get('system', 'poco_local_search')) {
 989                                 $values['$searchoption'][] = L10n::t("Forums");
 990                         }
 991                 }
 992
 993                 return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 994         }
 995
 996         /**
 997          * Replace naked text hyperlink with HTML formatted hyperlink
 998          *
 999          * @param string $s
1000          * @return string
1001          */
1002         public static function toLink($s)
1003         {
1004                 $s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank">$1</a>', $s);
1005                 $s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
1006                 return $s;
1007         }
1008
1009         /**
1010          * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
1011          *
1012          * Reasons are expected to have been translated already.
1013          *
1014          * @param string $html
1015          * @param array  $reasons
1016          * @return string
1017          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1018          */
1019         public static function applyContentFilter($html, array $reasons)
1020         {
1021                 if (count($reasons)) {
1022                         $tpl = Renderer::getMarkupTemplate('wall/content_filter.tpl');
1023                         $html = Renderer::replaceMacros($tpl, [
1024                                 '$reasons'   => $reasons,
1025                                 '$rnd'       => Strings::getRandomHex(8),
1026                                 '$openclose' => L10n::t('Click to open/close'),
1027                                 '$html'      => $html
1028                         ]);
1029                 }
1030
1031                 return $html;
1032         }
1033
1034         /**
1035          * replace html amp entity with amp char
1036          * @param string $s
1037          * @return string
1038          */
1039         public static function unamp($s)
1040         {
1041                 return str_replace('&amp;', '&', $s);
1042         }
1043 }