typo3/sysext/core/Classes/Html/RteHtmlParser.php

   1 <?php
   2 namespace TYPO3\CMS\Core\Html;
   3
   4 /*
   5  * This file is part of the TYPO3 CMS project.
   6  *
   7  * It is free software; you can redistribute it and/or modify it under
   8  * the terms of the GNU General Public License, either version 2
   9  * of the License, or any later version.
  10  *
  11  * For the full copyright and license information, please read the
  12  * LICENSE.txt file that was distributed with this source code.
  13  *
  14  * The TYPO3 project - inspiring people to share!
  15  */
  16
  17 use TYPO3\CMS\Backend\Utility\BackendUtility;
  18 use TYPO3\CMS\Core\Utility\GeneralUtility;
  19 use TYPO3\CMS\Core\Utility\StringUtility;
  20 use TYPO3\CMS\Core\Resource;
  21 use TYPO3\CMS\Frontend\Service\TypoLinkCodecService;
  22
  23 /**
  24  * Class for parsing HTML for the Rich Text Editor. (also called transformations)
  25  */
  26 class RteHtmlParser extends \TYPO3\CMS\Core\Html\HtmlParser
  27 {
  28     /**
  29      * @var string
  30      */
  31     public $blockElementList = 'PRE,UL,OL,H1,H2,H3,H4,H5,H6,ADDRESS,DL,DD,HEADER,SECTION,FOOTER,NAV,ARTICLE,ASIDE';
  32
  33     /**
  34      * Set this to the pid of the record manipulated by the class.
  35      *
  36      * @var int
  37      */
  38     public $recPid = 0;
  39
  40     /**
  41      * Element reference [table]:[field], eg. "tt_content:bodytext"
  42      *
  43      * @var string
  44      */
  45     public $elRef = '';
  46
  47     /**
  48      * Relative path
  49      *
  50      * @var string
  51      */
  52     public $relPath = '';
  53
  54     /**
  55      * Relative back-path
  56      *
  57      * @var string
  58      */
  59     public $relBackPath = '';
  60
  61     /**
  62      * Current Page TSConfig
  63      *
  64      * @var array
  65      */
  66     public $tsConfig = array();
  67
  68     /**
  69      * Set to the TSconfig options coming from Page TSconfig
  70      *
  71      * @var array
  72      */
  73     public $procOptions = array();
  74
  75     /**
  76      * Run-away brake for recursive calls.
  77      *
  78      * @var int
  79      */
  80     public $TS_transform_db_safecounter = 100;
  81
  82     /**
  83      * Parameters from TCA types configuration related to the RTE
  84      *
  85      * @var string
  86      */
  87     public $rte_p = '';
  88
  89     /**
  90      * Data caching for processing function
  91      *
  92      * @var array
  93      */
  94     public $getKeepTags_cache = array();
  95
  96     /**
  97      * Storage of the allowed CSS class names in the RTE
  98      *
  99      * @var array
 100      */
 101     public $allowedClasses = array();
 102
 103     /**
 104      * Set to tags to preserve from Page TSconfig configuration
 105      *
 106      * @var string
 107      */
 108     public $preserveTags = '';
 109
 110     /**
 111      * Initialize, setting element reference and record PID
 112      *
 113      * @param string $elRef Element reference, eg "tt_content:bodytext
 114      * @param int $recPid PID of the record (page id)
 115      * @return void
 116      */
 117     public function init($elRef = '', $recPid = 0)
 118     {
 119         $this->recPid = $recPid;
 120         $this->elRef = $elRef;
 121     }
 122
 123     /**
 124      * Setting the ->relPath and ->relBackPath to proper values so absolute references to links and images can be converted to relative dittos.
 125      * This is used when editing files with the RTE
 126      *
 127      * @param string $path The relative path from PATH_site to the place where the file being edited is. Eg. "fileadmin/static".
 128      * @return void There is no output, it is set in internal variables. With the above example of "fileadmin/static" as input this will yield ->relPath to be "fileadmin/static/" and ->relBackPath to be "../../
 129      * @TODO: Check if relPath and relBackPath are used for anything useful after removal of "static file edit" with #63818
 130      */
 131     public function setRelPath($path)
 132     {
 133         $path = trim($path);
 134         $path = preg_replace('/^\\//', '', $path);
 135         $path = preg_replace('/\\/$/', '', $path);
 136         if ($path) {
 137             $this->relPath = $path;
 138             $this->relBackPath = '';
 139             $partsC = count(explode('/', $this->relPath));
 140             for ($a = 0; $a < $partsC; $a++) {
 141                 $this->relBackPath .= '../';
 142             }
 143             $this->relPath .= '/';
 144         }
 145     }
 146
 147     /**
 148      * Evaluate the environment for editing a staticFileEdit file.
 149      * Called for almost all fields being saved in the database. Is called without
 150      * an instance of \TYPO3\CMS\Core\Html\RteHtmlParser::evalWriteFile()
 151      *
 152      * @param array $pArr Parameters for the current field as found in types-config
 153      * @param array $currentRecord Current record we are editing.
 154      * @return mixed On success an array with various information is returned, otherwise a string with an error message
 155      * @deprecated since TYPO3 CMS 7, will be removed in TYPO3 CMS 8
 156      */
 157     public static function evalWriteFile($pArr, $currentRecord)
 158     {
 159         GeneralUtility::logDeprecatedFunction();
 160     }
 161
 162     /**********************************************
 163      *
 164      * Main function
 165      *
 166      **********************************************/
 167     /**
 168      * Transform value for RTE based on specConf in the direction specified by $direction (rte/db)
 169      * This is the main function called from tcemain and transfer data classes
 170      *
 171      * @param string Input value
 172      * @param array Special configuration for a field; This is coming from the types-configuration of the field in the TCA. In the types-configuration you can setup features for the field rendering and in particular the RTE takes al its major configuration options from there!
 173      * @param string Direction of the transformation. Two keywords are allowed; "db" or "rte". If "db" it means the transformation will clean up content coming from the Rich Text Editor and goes into the database. The other direction, "rte", is of course when content is coming from database and must be transformed to fit the RTE.
 174      * @param array Parsed TypoScript content configuring the RTE, probably coming from Page TSconfig.
 175      * @return string Output value
 176      */
 177     public function RTE_transform($value, $specConf, $direction = 'rte', $thisConfig = array())
 178     {
 179         // Init:
 180         $this->tsConfig = $thisConfig;
 181         $this->procOptions = (array)$thisConfig['proc.'];
 182         $this->preserveTags = strtoupper(implode(',', GeneralUtility::trimExplode(',', $this->procOptions['preserveTags'])));
 183         // dynamic configuration of blockElementList
 184         if ($this->procOptions['blockElementList']) {
 185             $this->blockElementList = $this->procOptions['blockElementList'];
 186         }
 187         // Get parameters for rte_transformation:
 188         $p = ($this->rte_p = BackendUtility::getSpecConfParametersFromArray($specConf['rte_transform']['parameters']));
 189         // Setting modes:
 190         if ((string)$this->procOptions['overruleMode'] !== '') {
 191             $modes = array_unique(GeneralUtility::trimExplode(',', $this->procOptions['overruleMode']));
 192         } else {
 193             $modes = array_unique(GeneralUtility::trimExplode('-', $p['mode']));
 194         }
 195         $revmodes = array_flip($modes);
 196         // Find special modes and extract them:
 197         if (isset($revmodes['ts'])) {
 198             $modes[$revmodes['ts']] = 'ts_transform,ts_preserve,ts_images,ts_links';
 199         }
 200         // Find special modes and extract them:
 201         if (isset($revmodes['ts_css'])) {
 202             $modes[$revmodes['ts_css']] = 'css_transform,ts_images,ts_links';
 203         }
 204         // Make list unique
 205         $modes = array_unique(GeneralUtility::trimExplode(',', implode(',', $modes), true));
 206         // Reverse order if direction is "rte"
 207         if ($direction == 'rte') {
 208             $modes = array_reverse($modes);
 209         }
 210         // Getting additional HTML cleaner configuration. These are applied either before or after the main transformation is done and is thus totally independent processing options you can set up:
 211         $entry_HTMLparser = $this->procOptions['entryHTMLparser_' . $direction] ? $this->HTMLparserConfig($this->procOptions['entryHTMLparser_' . $direction . '.']) : '';
 212         $exit_HTMLparser = $this->procOptions['exitHTMLparser_' . $direction] ? $this->HTMLparserConfig($this->procOptions['exitHTMLparser_' . $direction . '.']) : '';
 213         // Line breaks of content is unified into char-10 only (removing char 13)
 214         if (!$this->procOptions['disableUnifyLineBreaks']) {
 215             $value = str_replace(CRLF, LF, $value);
 216         }
 217         // In an entry-cleaner was configured, pass value through the HTMLcleaner with that:
 218         if (is_array($entry_HTMLparser)) {
 219             $value = $this->HTMLcleaner($value, $entry_HTMLparser[0], $entry_HTMLparser[1], $entry_HTMLparser[2], $entry_HTMLparser[3]);
 220         }
 221         // Traverse modes:
 222         foreach ($modes as $cmd) {
 223             // ->DB
 224             if ($direction == 'db') {
 225                 // Checking for user defined transformation:
 226                 if ($_classRef = $GLOBALS['TYPO3_CONF_VARS']['SC_OPTIONS']['t3lib/class.t3lib_parsehtml_proc.php']['transformation'][$cmd]) {
 227                     $_procObj = GeneralUtility::getUserObj($_classRef);
 228                     $_procObj->pObj = $this;
 229                     $_procObj->transformationKey = $cmd;
 230                     $value = $_procObj->transform_db($value, $this);
 231                 } else {
 232                     // ... else use defaults:
 233                     switch ($cmd) {
 234                         case 'ts_images':
 235                             $value = $this->TS_images_db($value);
 236                             break;
 237                         case 'ts_reglinks':
 238                             $value = $this->TS_reglinks($value, 'db');
 239                             break;
 240                         case 'ts_links':
 241                             $value = $this->TS_links_db($value);
 242                             break;
 243                         case 'ts_preserve':
 244                             $value = $this->TS_preserve_db($value);
 245                             break;
 246                         case 'ts_transform':
 247
 248                         case 'css_transform':
 249                             $this->allowedClasses = GeneralUtility::trimExplode(',', $this->procOptions['allowedClasses'], true);
 250                             // CR has a very disturbing effect, so just remove all CR and rely on LF
 251                             $value = str_replace(CR, '', $value);
 252                             // Transform empty paragraphs into spacing paragraphs
 253                             $value = str_replace('<p></p>', '<p>&nbsp;</p>', $value);
 254                             // Double any trailing spacing paragraph so that it does not get removed by divideIntoLines()
 255                             $value = preg_replace('/<p>&nbsp;<\/p>$/', '<p>&nbsp;</p>' . '<p>&nbsp;</p>', $value);
 256                             $value = $this->TS_transform_db($value, $cmd == 'css_transform');
 257                             break;
 258                         case 'ts_strip':
 259                             $value = $this->TS_strip_db($value);
 260                             break;
 261                         default:
 262                             // Do nothing
 263                     }
 264                 }
 265             }
 266             // ->RTE
 267             if ($direction == 'rte') {
 268                 // Checking for user defined transformation:
 269                 if ($_classRef = $GLOBALS['TYPO3_CONF_VARS']['SC_OPTIONS']['t3lib/class.t3lib_parsehtml_proc.php']['transformation'][$cmd]) {
 270                     $_procObj = GeneralUtility::getUserObj($_classRef);
 271                     $_procObj->pObj = $this;
 272                     $value = $_procObj->transform_rte($value, $this);
 273                 } else {
 274                     // ... else use defaults:
 275                     switch ($cmd) {
 276                         case 'ts_images':
 277                             $value = $this->TS_images_rte($value);
 278                             break;
 279                         case 'ts_reglinks':
 280                             $value = $this->TS_reglinks($value, 'rte');
 281                             break;
 282                         case 'ts_links':
 283                             $value = $this->TS_links_rte($value);
 284                             break;
 285                         case 'ts_preserve':
 286                             $value = $this->TS_preserve_rte($value);
 287                             break;
 288                         case 'ts_transform':
 289
 290                         case 'css_transform':
 291                             // Has a very disturbing effect, so just remove all '13' - depend on '10'
 292                             $value = str_replace(CR, '', $value);
 293                             $value = $this->TS_transform_rte($value, $cmd == 'css_transform');
 294                             break;
 295                         default:
 296                             // Do nothing
 297                     }
 298                 }
 299             }
 300         }
 301         // In an exit-cleaner was configured, pass value through the HTMLcleaner with that:
 302         if (is_array($exit_HTMLparser)) {
 303             $value = $this->HTMLcleaner($value, $exit_HTMLparser[0], $exit_HTMLparser[1], $exit_HTMLparser[2], $exit_HTMLparser[3]);
 304         }
 305         // Final clean up of linebreaks:
 306         if (!$this->procOptions['disableUnifyLineBreaks']) {
 307             // Make sure no \r\n sequences has entered in the meantime...
 308             $value = str_replace(CRLF, LF, $value);
 309             // ... and then change all \n into \r\n
 310             $value = str_replace(LF, CRLF, $value);
 311         }
 312         // Return value:
 313         return $value;
 314     }
 315
 316     /************************************
 317      *
 318      * Specific RTE TRANSFORMATION functions
 319      *
 320      *************************************/
 321     /**
 322      * Transformation handler: 'ts_images' / direction: "db"
 323      * Processing images inserted in the RTE.
 324      * This is used when content goes from the RTE to the database.
 325      * Images inserted in the RTE has an absolute URL applied to the src attribute. This URL is converted to a relative URL
 326      * If it turns out that the URL is from another website than the current the image is read from that external URL and moved to the local server.
 327      * Also "magic" images are processed here.
 328      *
 329      * @param string $value The content from RTE going to Database
 330      * @return string Processed content
 331      */
 332     public function TS_images_db($value)
 333     {
 334         // Split content by <img> tags and traverse the resulting array for processing:
 335         $imgSplit = $this->splitTags('img', $value);
 336         if (count($imgSplit) > 1) {
 337             $siteUrl = $this->siteUrl();
 338             $sitePath = str_replace(GeneralUtility::getIndpEnv('TYPO3_REQUEST_HOST'), '', $siteUrl);
 339             /** @var $resourceFactory Resource\ResourceFactory */
 340             $resourceFactory = Resource\ResourceFactory::getInstance();
 341             /** @var $magicImageService Resource\Service\MagicImageService */
 342             $magicImageService = GeneralUtility::makeInstance(\TYPO3\CMS\Core\Resource\Service\MagicImageService::class);
 343             $magicImageService->setMagicImageMaximumDimensions($this->tsConfig);
 344             foreach ($imgSplit as $k => $v) {
 345                 // Image found, do processing:
 346                 if ($k % 2) {
 347                     // Get attributes
 348                     $attribArray = $this->get_tag_attributes_classic($v, 1);
 349                     // It's always an absolute URL coming from the RTE into the Database.
 350                     $absoluteUrl = trim($attribArray['src']);
 351                     // Make path absolute if it is relative and we have a site path which is not '/'
 352                     $pI = pathinfo($absoluteUrl);
 353                     if ($sitePath && !$pI['scheme'] && GeneralUtility::isFirstPartOfStr($absoluteUrl, $sitePath)) {
 354                         // If site is in a subpath (eg. /~user_jim/) this path needs to be removed because it will be added with $siteUrl
 355                         $absoluteUrl = substr($absoluteUrl, strlen($sitePath));
 356                         $absoluteUrl = $siteUrl . $absoluteUrl;
 357                     }
 358                     // Image dimensions set in the img tag, if any
 359                     $imgTagDimensions = $this->getWHFromAttribs($attribArray);
 360                     if ($imgTagDimensions[0]) {
 361                         $attribArray['width'] = $imgTagDimensions[0];
 362                     }
 363                     if ($imgTagDimensions[1]) {
 364                         $attribArray['height'] = $imgTagDimensions[1];
 365                     }
 366                     $originalImageFile = null;
 367                     if ($attribArray['data-htmlarea-file-uid']) {
 368                         // An original image file uid is available
 369                         try {
 370                             /** @var $originalImageFile Resource\File */
 371                             $originalImageFile = $resourceFactory->getFileObject(intval($attribArray['data-htmlarea-file-uid']));
 372                         } catch (Resource\Exception\FileDoesNotExistException $fileDoesNotExistException) {
 373                             // Log the fact the file could not be retrieved.
 374                             $message = sprintf('Could not find file with uid "%s"', $attribArray['data-htmlarea-file-uid']);
 375                             $this->getLogger()->error($message);
 376                         }
 377                     }
 378                     if ($originalImageFile instanceof Resource\File) {
 379                         // Public url of local file is relative to the site url, absolute otherwise
 380                         if ($absoluteUrl == $originalImageFile->getPublicUrl() || $absoluteUrl == $siteUrl . $originalImageFile->getPublicUrl()) {
 381                             // This is a plain image, i.e. reference to the original image
 382                             if ($this->procOptions['plainImageMode']) {
 383                                 // "plain image mode" is configured
 384                                 // Find the dimensions of the original image
 385                                 $imageInfo = array(
 386                                     $originalImageFile->getProperty('width'),
 387                                     $originalImageFile->getProperty('height')
 388                                 );
 389                                 if (!$imageInfo[0] || !$imageInfo[1]) {
 390                                     $filePath = $originalImageFile->getForLocalProcessing(false);
 391                                     $imageInfo = @getimagesize($filePath);
 392                                 }
 393                                 $attribArray = $this->applyPlainImageModeSettings($imageInfo, $attribArray);
 394                             }
 395                         } else {
 396                             // Magic image case: get a processed file with the requested configuration
 397                             $imageConfiguration = array(
 398                                 'width' => $imgTagDimensions[0],
 399                                 'height' => $imgTagDimensions[1]
 400                             );
 401                             $magicImage = $magicImageService->createMagicImage($originalImageFile, $imageConfiguration);
 402                             $attribArray['width'] = $magicImage->getProperty('width');
 403                             $attribArray['height'] = $magicImage->getProperty('height');
 404                             $attribArray['src'] = $magicImage->getPublicUrl();
 405                         }
 406                     } elseif (!GeneralUtility::isFirstPartOfStr($absoluteUrl, $siteUrl) && !$this->procOptions['dontFetchExtPictures'] && TYPO3_MODE === 'BE') {
 407                         // External image from another URL: in that case, fetch image, unless the feature is disabled or we are not in backend mode
 408                         // Fetch the external image
 409                         $externalFile = $this->getUrl($absoluteUrl);
 410                         if ($externalFile) {
 411                             $pU = parse_url($absoluteUrl);
 412                             $pI = pathinfo($pU['path']);
 413                             if (GeneralUtility::inList('gif,png,jpeg,jpg', strtolower($pI['extension']))) {
 414                                 $fileName = GeneralUtility::shortMD5($absoluteUrl) . '.' . $pI['extension'];
 415                                 // We insert this image into the user default upload folder
 416                                 list($table, $field) = explode(':', $this->elRef);
 417                                 $folder = $GLOBALS['BE_USER']->getDefaultUploadFolder($this->recPid, $table, $field);
 418                                 $fileObject = $folder->createFile($fileName)->setContents($externalFile);
 419                                 $imageConfiguration = array(
 420                                     'width' => $attribArray['width'],
 421                                     'height' => $attribArray['height']
 422                                 );
 423                                 $magicImage = $magicImageService->createMagicImage($fileObject, $imageConfiguration);
 424                                 $attribArray['width'] = $magicImage->getProperty('width');
 425                                 $attribArray['height'] = $magicImage->getProperty('height');
 426                                 $attribArray['data-htmlarea-file-uid'] = $fileObject->getUid();
 427                                 $attribArray['src'] = $magicImage->getPublicUrl();
 428                             }
 429                         }
 430                     } elseif (GeneralUtility::isFirstPartOfStr($absoluteUrl, $siteUrl)) {
 431                         // Finally, check image as local file (siteURL equals the one of the image)
 432                         // Image has no data-htmlarea-file-uid attribute
 433                         // Relative path, rawurldecoded for special characters.
 434                         $path = rawurldecode(substr($absoluteUrl, strlen($siteUrl)));
 435                         // Absolute filepath, locked to relative path of this project
 436                         $filepath = GeneralUtility::getFileAbsFileName($path);
 437                         // Check file existence (in relative directory to this installation!)
 438                         if ($filepath && @is_file($filepath)) {
 439                             // Treat it as a plain image
 440                             if ($this->procOptions['plainImageMode']) {
 441                                 // If "plain image mode" has been configured
 442                                 // Find the original dimensions of the image
 443                                 $imageInfo = @getimagesize($filepath);
 444                                 $attribArray = $this->applyPlainImageModeSettings($imageInfo, $attribArray);
 445                             }
 446                             // Let's try to find a file uid for this image
 447                             try {
 448                                 $fileOrFolderObject = $resourceFactory->retrieveFileOrFolderObject($path);
 449                                 if ($fileOrFolderObject instanceof Resource\FileInterface) {
 450                                     $fileIdentifier = $fileOrFolderObject->getIdentifier();
 451                                     $fileObject = $fileOrFolderObject->getStorage()->getFile($fileIdentifier);
 452                                     // @todo if the retrieved file is a processed file, get the original file...
 453                                     $attribArray['data-htmlarea-file-uid'] = $fileObject->getUid();
 454                                 }
 455                             } catch (Resource\Exception\ResourceDoesNotExistException $resourceDoesNotExistException) {
 456                                 // Nothing to be done if file/folder not found
 457                             }
 458                         }
 459                     }
 460                     // Remove width and height from style attribute
 461                     $attribArray['style'] = preg_replace('/((?:^|)\\s*(?:width|height)\\s*:[^;]*(?:$|;))/si', '', $attribArray['style']);
 462                     // Must have alt attribute
 463                     if (!isset($attribArray['alt'])) {
 464                         $attribArray['alt'] = '';
 465                     }
 466                     // Convert absolute to relative url
 467                     if (GeneralUtility::isFirstPartOfStr($attribArray['src'], $siteUrl)) {
 468                         $attribArray['src'] = $this->relBackPath . substr($attribArray['src'], strlen($siteUrl));
 469                     }
 470                     $imgSplit[$k] = '<img ' . GeneralUtility::implodeAttributes($attribArray, 1, 1) . ' />';
 471                 }
 472             }
 473         }
 474         return implode('', $imgSplit);
 475     }
 476
 477     /**
 478      * Transformation handler: 'ts_images' / direction: "rte"
 479      * Processing images from database content going into the RTE.
 480      * Processing includes converting the src attribute to an absolute URL.
 481      *
 482      * @param string $value Content input
 483      * @return string Content output
 484      */
 485     public function TS_images_rte($value)
 486     {
 487         // Split content by <img> tags and traverse the resulting array for processing:
 488         $imgSplit = $this->splitTags('img', $value);
 489         if (count($imgSplit) > 1) {
 490             $siteUrl = $this->siteUrl();
 491             $sitePath = str_replace(GeneralUtility::getIndpEnv('TYPO3_REQUEST_HOST'), '', $siteUrl);
 492             foreach ($imgSplit as $k => $v) {
 493                 // Image found
 494                 if ($k % 2) {
 495                     // Get the attributes of the img tag
 496                     $attribArray = $this->get_tag_attributes_classic($v, 1);
 497                     $absoluteUrl = trim($attribArray['src']);
 498                     // Transform the src attribute into an absolute url, if it not already
 499                     if (strtolower(substr($absoluteUrl, 0, 4)) !== 'http') {
 500                         $attribArray['src'] = substr($attribArray['src'], strlen($this->relBackPath));
 501                         // If site is in a subpath (eg. /~user_jim/) this path needs to be removed because it will be added with $siteUrl
 502                         $attribArray['src'] = preg_replace('#^' . preg_quote($sitePath, '#') . '#', '', $attribArray['src']);
 503                         $attribArray['src'] = $siteUrl . $attribArray['src'];
 504                     }
 505                     // Must have alt attribute
 506                     if (!isset($attribArray['alt'])) {
 507                         $attribArray['alt'] = '';
 508                     }
 509                     $imgSplit[$k] = '<img ' . GeneralUtility::implodeAttributes($attribArray, 1, 1) . ' />';
 510                 }
 511             }
 512         }
 513         // Return processed content:
 514         return implode('', $imgSplit);
 515     }
 516
 517     /**
 518      * Transformation handler: 'ts_reglinks' / direction: "db"+"rte" depending on $direction variable.
 519      * Converting <A>-tags to/from abs/rel
 520      *
 521      * @param string $value Content input
 522      * @param string $direction Direction of conversion; "rte" (from database to RTE) or "db" (from RTE to database)
 523      * @return string Content output
 524      */
 525     public function TS_reglinks($value, $direction)
 526     {
 527         $retVal = '';
 528         switch ($direction) {
 529             case 'rte':
 530                 $retVal = $this->TS_AtagToAbs($value, 1);
 531                 break;
 532             case 'db':
 533                 $siteURL = $this->siteUrl();
 534                 $blockSplit = $this->splitIntoBlock('A', $value);
 535                 foreach ($blockSplit as $k => $v) {
 536                     // Block
 537                     if ($k % 2) {
 538                         $attribArray = $this->get_tag_attributes_classic($this->getFirstTag($v), 1);
 539                         // If the url is local, remove url-prefix
 540                         if ($siteURL && substr($attribArray['href'], 0, strlen($siteURL)) == $siteURL) {
 541                             $attribArray['href'] = $this->relBackPath . substr($attribArray['href'], strlen($siteURL));
 542                         }
 543                         $bTag = '<a ' . GeneralUtility::implodeAttributes($attribArray, 1) . '>';
 544                         $eTag = '</a>';
 545                         $blockSplit[$k] = $bTag . $this->TS_reglinks($this->removeFirstAndLastTag($blockSplit[$k]), $direction) . $eTag;
 546                     }
 547                 }
 548                 $retVal = implode('', $blockSplit);
 549                 break;
 550         }
 551         return $retVal;
 552     }
 553
 554     /**
 555      * Transformation handler: 'ts_links' / direction: "db"
 556      * Converting <A>-tags to <link tags>
 557      *
 558      * @param string $value Content input
 559      * @return string Content output
 560      * @see TS_links_rte()
 561      */
 562     public function TS_links_db($value)
 563     {
 564         $conf = array();
 565         // Split content into <a> tag blocks and process:
 566         $blockSplit = $this->splitIntoBlock('A', $value);
 567         foreach ($blockSplit as $k => $v) {
 568             // If an A-tag was found:
 569             if ($k % 2) {
 570                 $attribArray = $this->get_tag_attributes_classic($this->getFirstTag($v), 1);
 571                 $info = $this->urlInfoForLinkTags($attribArray['href']);
 572                 // Check options:
 573                 $attribArray_copy = $attribArray;
 574                 unset($attribArray_copy['href']);
 575                 unset($attribArray_copy['target']);
 576                 unset($attribArray_copy['class']);
 577                 unset($attribArray_copy['title']);
 578                 unset($attribArray_copy['data-htmlarea-external']);
 579                 // Unset "rteerror" and "style" attributes if "rteerror" is set!
 580                 if ($attribArray_copy['rteerror']) {
 581                     unset($attribArray_copy['style']);
 582                     unset($attribArray_copy['rteerror']);
 583                 }
 584                 // Remove additional parameters
 585                 if (isset($GLOBALS['TYPO3_CONF_VARS']['SC_OPTIONS']['t3lib/class.t3lib_parsehtml_proc.php']['removeParams_PostProc']) && is_array($GLOBALS['TYPO3_CONF_VARS']['SC_OPTIONS']['t3lib/class.t3lib_parsehtml_proc.php']['removeParams_PostProc'])) {
 586                     $parameters = array(
 587                         'conf' => &$conf,
 588                         'aTagParams' => &$attribArray_copy
 589                     );
 590                     foreach ($GLOBALS['TYPO3_CONF_VARS']['SC_OPTIONS']['t3lib/class.t3lib_parsehtml_proc.php']['removeParams_PostProc'] as $objRef) {
 591                         $processor = GeneralUtility::getUserObj($objRef);
 592                         $attribArray_copy = $processor->removeParams($parameters, $this);
 593                     }
 594                 }
 595                 // Only if href, target, class and tile are the only attributes, we can alter the link!
 596                 if (empty($attribArray_copy)) {
 597                     // Quoting class and title attributes if they contain spaces
 598                     $attribArray['class'] = preg_match('/ /', $attribArray['class']) ? '"' . $attribArray['class'] . '"' : $attribArray['class'];
 599                     $attribArray['title'] = preg_match('/ /', $attribArray['title']) ? '"' . $attribArray['title'] . '"' : $attribArray['title'];
 600                     // Creating the TYPO3 pseudo-tag "<LINK>" for the link (includes href/url, target and class attributes):
 601                     // If data-htmlarea-external attribute is set, keep the href unchanged
 602                     if ($attribArray['data-htmlarea-external']) {
 603                         $href = $attribArray['href'];
 604                     } else {
 605                         $href = $info['url'] . ($info['query'] ? ',0,' . $info['query'] : '');
 606                     }
 607                     $bTag = '<link ' . $href . ($attribArray['target'] ? ' ' . $attribArray['target'] : ($attribArray['class'] || $attribArray['title'] ? ' -' : '')) . ($attribArray['class'] ? ' ' . $attribArray['class'] : ($attribArray['title'] ? ' -' : '')) . ($attribArray['title'] ? ' ' . $attribArray['title'] : '') . '>';
 608                     $eTag = '</link>';
 609                     // Modify parameters
 610                     if (isset($GLOBALS['TYPO3_CONF_VARS']['SC_OPTIONS']['t3lib/class.t3lib_parsehtml_proc.php']['modifyParams_LinksDb_PostProc']) && is_array($GLOBALS['TYPO3_CONF_VARS']['SC_OPTIONS']['t3lib/class.t3lib_parsehtml_proc.php']['modifyParams_LinksDb_PostProc'])) {
 611                         $parameters = array(
 612                             'conf' => &$conf,
 613                             'currentBlock' => $v,
 614                             'url' => $href,
 615                             'attributes' => $attribArray
 616                         );
 617                         foreach ($GLOBALS['TYPO3_CONF_VARS']['SC_OPTIONS']['t3lib/class.t3lib_parsehtml_proc.php']['modifyParams_LinksDb_PostProc'] as $objRef) {
 618                             $processor = GeneralUtility::getUserObj($objRef);
 619                             $blockSplit[$k] = $processor->modifyParamsLinksDb($parameters, $this);
 620                         }
 621                     } else {
 622                         $blockSplit[$k] = $bTag . $this->TS_links_db($this->removeFirstAndLastTag($blockSplit[$k])) . $eTag;
 623                     }
 624                 } else {
 625                     // ... otherwise store the link as a-tag.
 626                     // Unsetting 'rtekeep' attribute if that had been set.
 627                     unset($attribArray['rtekeep']);
 628                     if (!$attribArray['data-htmlarea-external']) {
 629                         $siteURL = $this->siteUrl();
 630                         // If the url is local, remove url-prefix
 631                         if ($siteURL && substr($attribArray['href'], 0, strlen($siteURL)) == $siteURL) {
 632                             $attribArray['href'] = $this->relBackPath . substr($attribArray['href'], strlen($siteURL));
 633                         }
 634                         // Check for FAL link-handler keyword
 635                         list($linkHandlerKeyword, $linkHandlerValue) = explode(':', $attribArray['href'], 2);
 636                         if ($linkHandlerKeyword === '?file') {
 637                             try {
 638                                 $fileOrFolderObject = \TYPO3\CMS\Core\Resource\ResourceFactory::getInstance()->retrieveFileOrFolderObject(rawurldecode($linkHandlerValue));
 639                                 if ($fileOrFolderObject instanceof \TYPO3\CMS\Core\Resource\FileInterface || $fileOrFolderObject instanceof \TYPO3\CMS\Core\Resource\Folder) {
 640                                     $attribArray['href'] = $fileOrFolderObject->getPublicUrl();
 641                                 }
 642                             } catch (\TYPO3\CMS\Core\Resource\Exception\ResourceDoesNotExistException $resourceDoesNotExistException) {
 643                                 // The indentifier inserted in the RTE is already gone...
 644                             }
 645                         }
 646                     }
 647                     unset($attribArray['data-htmlarea-external']);
 648                     $bTag = '<a ' . GeneralUtility::implodeAttributes($attribArray, 1) . '>';
 649                     $eTag = '</a>';
 650                     $blockSplit[$k] = $bTag . $this->TS_links_db($this->removeFirstAndLastTag($blockSplit[$k])) . $eTag;
 651                 }
 652             }
 653         }
 654         return implode('', $blockSplit);
 655     }
 656
 657     /**
 658      * Transformation handler: 'ts_links' / direction: "rte"
 659      * Converting <link tags> to <A>-tags
 660      *
 661      * @param string $value Content input
 662      * @return string Content output
 663      * @see TS_links_rte()
 664      */
 665     public function TS_links_rte($value)
 666     {
 667         $conf = array();
 668         $value = $this->TS_AtagToAbs($value);
 669         // Split content by the TYPO3 pseudo tag "<link>":
 670         $blockSplit = $this->splitIntoBlock('link', $value, 1);
 671         $siteUrl = $this->siteUrl();
 672         foreach ($blockSplit as $k => $v) {
 673             $error = '';
 674             $external = false;
 675             // Block
 676             if ($k % 2) {
 677                 // split away the first "<link" part
 678                 $typolink = explode(' ', substr($this->getFirstTag($v), 0, -1), 2)[1];
 679                 $tagCode = GeneralUtility::makeInstance(TypoLinkCodecService::class)->decode($typolink);
 680
 681                 $link_param = $tagCode['url'];
 682                 // Parsing the typolink data. This parsing is roughly done like in \TYPO3\CMS\Frontend\ContentObject->typoLink()
 683                 // Parse URL:
 684                 $pU = parse_url($link_param);
 685                 if (strstr($link_param, '@') && (!$pU['scheme'] || $pU['scheme'] == 'mailto')) {
 686                     // mailadr
 687                     $href = 'mailto:' . preg_replace('/^mailto:/i', '', $link_param);
 688                 } elseif ($link_param[0] === '#') {
 689                     // check if anchor
 690                     $href = $siteUrl . $link_param;
 691                 } else {
 692                     // Check for FAL link-handler keyword:
 693                     list($linkHandlerKeyword, $linkHandlerValue) = explode(':', trim($link_param), 2);
 694                     if ($linkHandlerKeyword === 'file' && !StringUtility::beginsWith($link_param, 'file://')) {
 695                         $href = $siteUrl . '?' . $linkHandlerKeyword . ':' . rawurlencode($linkHandlerValue);
 696                     } else {
 697                         $fileChar = (int)strpos($link_param, '/');
 698                         $urlChar = (int)strpos($link_param, '.');
 699                         // Detects if a file is found in site-root.
 700                         list($rootFileDat) = explode('?', $link_param);
 701                         $rFD_fI = pathinfo($rootFileDat);
 702                         if (trim($rootFileDat) && !strstr($link_param, '/') && (@is_file((PATH_site . $rootFileDat)) || GeneralUtility::inList('php,html,htm', strtolower($rFD_fI['extension'])))) {
 703                             $href = $siteUrl . $link_param;
 704                         } elseif (
 705                             (
 706                                 $pU['scheme']
 707                                 && !isset($GLOBALS['TYPO3_CONF_VARS']['SC_OPTIONS']['tslib/class.tslib_content.php']['typolinkLinkHandler'][$pU['scheme']])
 708                             )
 709                             || $urlChar && (!$fileChar || $urlChar < $fileChar)
 710                         ) {
 711                             // url (external): if has scheme or if a '.' comes before a '/'.
 712                             $href = $link_param;
 713                             if (!$pU['scheme']) {
 714                                 $href = 'http://' . $href;
 715                             }
 716                             $external = true;
 717                         } elseif ($fileChar) {
 718                             // It is an internal file or folder
 719                             // Try to transform the href into a FAL reference
 720                             try {
 721                                 $fileOrFolderObject = \TYPO3\CMS\Core\Resource\ResourceFactory::getInstance()->retrieveFileOrFolderObject($link_param);
 722                             } catch (\TYPO3\CMS\Core\Resource\Exception $exception) {
 723                                 // Nothing to be done if file/folder not found or path invalid
 724                                 $fileOrFolderObject = null;
 725                             }
 726                             if ($fileOrFolderObject instanceof \TYPO3\CMS\Core\Resource\Folder) {
 727                                 // It's a folder
 728                                 $folderIdentifier = $fileOrFolderObject->getIdentifier();
 729                                 $href = $siteUrl . '?file:' . rawurlencode($folderIdentifier);
 730                             } elseif ($fileOrFolderObject instanceof \TYPO3\CMS\Core\Resource\FileInterface) {
 731                                 // It's a file
 732                                 $fileIdentifier = $fileOrFolderObject->getIdentifier();
 733                                 $fileObject = $fileOrFolderObject->getStorage()->getFile($fileIdentifier);
 734                                 $href = $siteUrl . '?file:' . $fileObject->getUid();
 735                             } else {
 736                                 $href = $siteUrl . $link_param;
 737                             }
 738                         } else {
 739                             // integer or alias (alias is without slashes or periods or commas, that is 'nospace,alphanum_x,lower,unique' according to tables.php!!)
 740                             // Splitting the parameter by ',' and if the array counts more than 1 element it's an id/type/parameters triplet
 741                             $pairParts = GeneralUtility::trimExplode(',', $link_param, true);
 742                             $idPart = $pairParts[0];
 743                             $link_params_parts = explode('#', $idPart);
 744                             $idPart = trim($link_params_parts[0]);
 745                             $sectionMark = trim($link_params_parts[1]);
 746                             if ((string)$idPart === '') {
 747                                 $idPart = $this->recPid;
 748                             }
 749                             // If no id or alias is given, set it to class record pid
 750                             // Checking if the id-parameter is an alias.
 751                             if (!\TYPO3\CMS\Core\Utility\MathUtility::canBeInterpretedAsInteger($idPart)) {
 752                                 list($idPartR) = BackendUtility::getRecordsByField('pages', 'alias', $idPart);
 753                                 $idPart = (int)$idPartR['uid'];
 754                             }
 755                             $page = BackendUtility::getRecord('pages', $idPart);
 756                             if (is_array($page)) {
 757                                 // Page must exist...
 758                                 $href = $siteUrl . '?id=' . $idPart . ($pairParts[2] ? $pairParts[2] : '') . ($sectionMark ? '#' . $sectionMark : '');
 759                             } elseif (isset($GLOBALS['TYPO3_CONF_VARS']['SC_OPTIONS']['tslib/class.tslib_content.php']['typolinkLinkHandler'][array_shift(explode(':', $link_param))])) {
 760                                 $href = $link_param;
 761                             } else {
 762                                 $href = $siteUrl . '?id=' . $link_param;
 763                                 $error = 'No page found: ' . $idPart;
 764                             }
 765                         }
 766                     }
 767                 }
 768                 // Setting the A-tag:
 769                 $bTag = '<a href="' . htmlspecialchars($href) . '"'
 770                     . ($tagCode['target'] ? ' target="' . htmlspecialchars($tagCode['target']) . '"' : '')
 771                     . ($tagCode['class'] ? ' class="' . htmlspecialchars($tagCode['class']) . '"' : '')
 772                     . ($tagCode['title'] ? ' title="' . htmlspecialchars($tagCode['title']) . '"' : '')
 773                     . ($external ? ' data-htmlarea-external="1"' : '')
 774                     . ($error ? ' rteerror="' . htmlspecialchars($error) . '" style="background-color: yellow; border:2px red solid; color: black;"' : '') . '>';
 775                 $eTag = '</a>';
 776                 // Modify parameters
 777                 if (isset($GLOBALS['TYPO3_CONF_VARS']['SC_OPTIONS']['t3lib/class.t3lib_parsehtml_proc.php']['modifyParams_LinksRte_PostProc']) && is_array($GLOBALS['TYPO3_CONF_VARS']['SC_OPTIONS']['t3lib/class.t3lib_parsehtml_proc.php']['modifyParams_LinksRte_PostProc'])) {
 778                     $parameters = array(
 779                         'conf' => &$conf,
 780                         'currentBlock' => $v,
 781                         'url' => $href,
 782                         'tagCode' => $tagCode,
 783                         'external' => $external,
 784                         'error' => $error
 785                     );
 786                     foreach ($GLOBALS['TYPO3_CONF_VARS']['SC_OPTIONS']['t3lib/class.t3lib_parsehtml_proc.php']['modifyParams_LinksRte_PostProc'] as $objRef) {
 787                         $processor = GeneralUtility::getUserObj($objRef);
 788                         $blockSplit[$k] = $processor->modifyParamsLinksRte($parameters, $this);
 789                     }
 790                 } else {
 791                     $blockSplit[$k] = $bTag . $this->TS_links_rte($this->removeFirstAndLastTag($blockSplit[$k])) . $eTag;
 792                 }
 793             }
 794         }
 795         // Return content:
 796         return implode('', $blockSplit);
 797     }
 798
 799     /**
 800      * Preserve special tags
 801      *
 802      * @param string $value Content input
 803      * @return string Content output
 804      */
 805     public function TS_preserve_db($value)
 806     {
 807         if (!$this->preserveTags) {
 808             return $value;
 809         }
 810         // Splitting into blocks for processing (span-tags are used for special tags)
 811         $blockSplit = $this->splitIntoBlock('span', $value);
 812         foreach ($blockSplit as $k => $v) {
 813             // Block
 814             if ($k % 2) {
 815                 $attribArray = $this->get_tag_attributes_classic($this->getFirstTag($v));
 816                 if ($attribArray['specialtag']) {
 817                     $theTag = rawurldecode($attribArray['specialtag']);
 818                     $theTagName = $this->getFirstTagName($theTag);
 819                     $blockSplit[$k] = $theTag . $this->removeFirstAndLastTag($blockSplit[$k]) . '</' . $theTagName . '>';
 820                 }
 821             }
 822         }
 823         return implode('', $blockSplit);
 824     }
 825
 826     /**
 827      * Preserve special tags
 828      *
 829      * @param string $value Content input
 830      * @return string Content output
 831      */
 832     public function TS_preserve_rte($value)
 833     {
 834         if (!$this->preserveTags) {
 835             return $value;
 836         }
 837         $blockSplit = $this->splitIntoBlock($this->preserveTags, $value);
 838         foreach ($blockSplit as $k => $v) {
 839             // Block
 840             if ($k % 2) {
 841                 $blockSplit[$k] = '<span specialtag="' . rawurlencode($this->getFirstTag($v)) . '">' . $this->removeFirstAndLastTag($blockSplit[$k]) . '</span>';
 842             }
 843         }
 844         return implode('', $blockSplit);
 845     }
 846
 847     /**
 848      * Transformation handler: 'ts_transform' + 'css_transform' / direction: "db"
 849      * Cleaning (->db) for standard content elements (ts)
 850      *
 851      * @param string $value Content input
 852      * @param bool $css If TRUE, the transformation was "css_transform", otherwise "ts_transform
 853      * @return string Content output
 854      * @see TS_transform_rte()
 855      */
 856     public function TS_transform_db($value, $css = false)
 857     {
 858         // Safety... so forever loops are avoided (they should not occur, but an error would potentially do this...)
 859         $this->TS_transform_db_safecounter--;
 860         if ($this->TS_transform_db_safecounter < 0) {
 861             return $value;
 862         }
 863         // Split the content from RTE by the occurrence of these blocks:
 864         $blockSplit = $this->splitIntoBlock('TABLE,BLOCKQUOTE,' . ($this->procOptions['preserveDIVSections'] ? 'DIV,' : '') . $this->blockElementList, $value);
 865         $cc = 0;
 866         $aC = count($blockSplit);
 867         // Avoid superfluous linebreaks by transform_db after ending headListTag
 868         while ($aC && trim($blockSplit[$aC - 1]) === '') {
 869             unset($blockSplit[$aC - 1]);
 870             $aC = count($blockSplit);
 871         }
 872         // Traverse the blocks
 873         foreach ($blockSplit as $k => $v) {
 874             $cc++;
 875             $lastBR = $cc == $aC ? '' : LF;
 876             if ($k % 2) {
 877                 // Inside block:
 878                 // Init:
 879                 $tag = $this->getFirstTag($v);
 880                 $tagName = strtolower($this->getFirstTagName($v));
 881                 // Process based on the tag:
 882                 switch ($tagName) {
 883                     case 'blockquote':
 884
 885                     case 'dd':
 886
 887                     case 'div':
 888
 889                     case 'header':
 890
 891                     case 'section':
 892
 893                     case 'footer':
 894
 895                     case 'nav':
 896
 897                     case 'article':
 898
 899                     case 'aside':
 900                         $blockSplit[$k] = $tag . $this->TS_transform_db($this->removeFirstAndLastTag($blockSplit[$k]), $css) . '</' . $tagName . '>' . $lastBR;
 901                         break;
 902                     case 'ol':
 903
 904                     case 'ul':
 905                         if ($css) {
 906                             $blockSplit[$k] = preg_replace(('/[' . LF . CR . ']+/'), ' ', $this->transformStyledATags($blockSplit[$k])) . $lastBR;
 907                         }
 908                         break;
 909                     case 'table':
 910                         // Tables are NOT allowed in any form (unless preserveTables is set or CSS is the mode)
 911                         if (!$this->procOptions['preserveTables'] && !$css) {
 912                             $blockSplit[$k] = $this->TS_transform_db($this->removeTables($blockSplit[$k]));
 913                         } else {
 914                             $blockSplit[$k] = preg_replace(('/[' . LF . CR . ']+/'), ' ', $this->transformStyledATags($blockSplit[$k])) . $lastBR;
 915                         }
 916                         break;
 917                     case 'h1':
 918
 919                     case 'h2':
 920
 921                     case 'h3':
 922
 923                     case 'h4':
 924
 925                     case 'h5':
 926
 927                     case 'h6':
 928                         if (!$css) {
 929                             $attribArray = $this->get_tag_attributes_classic($tag);
 930                             // Processing inner content here:
 931                             $innerContent = $this->HTMLcleaner_db($this->removeFirstAndLastTag($blockSplit[$k]));
 932                             $blockSplit[$k] = '<' . $tagName . ($attribArray['align'] ? ' align="' . htmlspecialchars($attribArray['align']) . '"' : '') . ($attribArray['class'] ? ' class="' . htmlspecialchars($attribArray['class']) . '"' : '') . '>' . $innerContent . '</' . $tagName . '>' . $lastBR;
 933                         } else {
 934                             // Eliminate true linebreaks inside Hx tags
 935                             $blockSplit[$k] = preg_replace(('/[' . LF . CR . ']+/'), ' ', $this->transformStyledATags($blockSplit[$k])) . $lastBR;
 936                         }
 937                         break;
 938                     default:
 939                         // Eliminate true linebreaks inside other headlist tags
 940                         $blockSplit[$k] = preg_replace(('/[' . LF . CR . ']+/'), ' ', $this->transformStyledATags($blockSplit[$k])) . $lastBR;
 941                 }
 942             } else {
 943                 // NON-block:
 944                 if (trim($blockSplit[$k]) !== '') {
 945                     $blockSplit[$k] = preg_replace('/<hr\\/>/', '<hr />', $blockSplit[$k]);
 946                     // Remove linebreaks preceding hr tags
 947                     $blockSplit[$k] = preg_replace('/[' . LF . CR . ']+<(hr)(\\s[^>\\/]*)?[[:space:]]*\\/?>/', '<$1$2/>', $blockSplit[$k]);
 948                     // Remove linebreaks following hr tags
 949                     $blockSplit[$k] = preg_replace('/<(hr)(\\s[^>\\/]*)?[[:space:]]*\\/?>[' . LF . CR . ']+/', '<$1$2/>', $blockSplit[$k]);
 950                     // Replace other linebreaks with space
 951                     $blockSplit[$k] = preg_replace('/[' . LF . CR . ']+/', ' ', $blockSplit[$k]);
 952                     $blockSplit[$k] = $this->divideIntoLines($blockSplit[$k]) . $lastBR;
 953                     $blockSplit[$k] = $this->transformStyledATags($blockSplit[$k]);
 954                 } else {
 955                     unset($blockSplit[$k]);
 956                 }
 957             }
 958         }
 959         $this->TS_transform_db_safecounter++;
 960         return implode('', $blockSplit);
 961     }
 962
 963     /**
 964      * Wraps a-tags that contain a style attribute with a span-tag
 965      *
 966      * @param string $value Content input
 967      * @return string Content output
 968      */
 969     public function transformStyledATags($value)
 970     {
 971         $blockSplit = $this->splitIntoBlock('A', $value);
 972         foreach ($blockSplit as $k => $v) {
 973             // If an A-tag was found
 974             if ($k % 2) {
 975                 $attribArray = $this->get_tag_attributes_classic($this->getFirstTag($v), 1);
 976                 // If "style" attribute is set and rteerror is not set!
 977                 if ($attribArray['style'] && !$attribArray['rteerror']) {
 978                     $attribArray_copy['style'] = $attribArray['style'];
 979                     unset($attribArray['style']);
 980                     $bTag = '<span ' . GeneralUtility::implodeAttributes($attribArray_copy, 1) . '><a ' . GeneralUtility::implodeAttributes($attribArray, 1) . '>';
 981                     $eTag = '</a></span>';
 982                     $blockSplit[$k] = $bTag . $this->removeFirstAndLastTag($blockSplit[$k]) . $eTag;
 983                 }
 984             }
 985         }
 986         return implode('', $blockSplit);
 987     }
 988
 989     /**
 990      * Transformation handler: 'ts_transform' + 'css_transform' / direction: "rte"
 991      * Set (->rte) for standard content elements (ts)
 992      *
 993      * @param string Content input
 994      * @param bool If TRUE, the transformation was "css_transform", otherwise "ts_transform
 995      * @return string Content output
 996      * @see TS_transform_db()
 997      */
 998     public function TS_transform_rte($value, $css = 0)
 999     {
1000         // Split the content from database by the occurrence of the block elements
1001         $blockElementList = 'TABLE,BLOCKQUOTE,' . ($this->procOptions['preserveDIVSections'] ? 'DIV,' : '') . $this->blockElementList;
1002         $blockSplit = $this->splitIntoBlock($blockElementList, $value);
1003         // Traverse the blocks
1004         foreach ($blockSplit as $k => $v) {
1005             if ($k % 2) {
1006                 // Inside one of the blocks:
1007                 // Init:
1008                 $tag = $this->getFirstTag($v);
1009                 $tagName = strtolower($this->getFirstTagName($v));
1010                 $attribArray = $this->get_tag_attributes_classic($tag);
1011                 // Based on tagname, we do transformations:
1012                 switch ($tagName) {
1013                     case 'blockquote':
1014
1015                     case 'dd':
1016
1017                     case 'div':
1018
1019                     case 'header':
1020
1021                     case 'section':
1022
1023                     case 'footer':
1024
1025                     case 'nav':
1026
1027                     case 'article':
1028
1029                     case 'aside':
1030                         $blockSplit[$k] = $tag . $this->TS_transform_rte($this->removeFirstAndLastTag($blockSplit[$k]), $css) . '</' . $tagName . '>';
1031                         break;
1032                 }
1033                 $blockSplit[$k + 1] = preg_replace('/^[ ]*' . LF . '/', '', $blockSplit[$k + 1]);
1034             } else {
1035                 // NON-block:
1036                 $nextFTN = $this->getFirstTagName($blockSplit[$k + 1]);
1037                 $onlyLineBreaks = (preg_match('/^[ ]*' . LF . '+[ ]*$/', $blockSplit[$k]) == 1);
1038                 // If the line is followed by a block or is the last line:
1039                 if (GeneralUtility::inList($blockElementList, $nextFTN) || !isset($blockSplit[$k + 1])) {
1040                     // If the line contains more than just linebreaks, reduce the number of trailing linebreaks by 1
1041                     if (!$onlyLineBreaks) {
1042                         $blockSplit[$k] = preg_replace('/(' . LF . '*)' . LF . '[ ]*$/', '$1', $blockSplit[$k]);
1043                     } else {
1044                         // If the line contains only linebreaks, remove the leading linebreak
1045                         $blockSplit[$k] = preg_replace('/^[ ]*' . LF . '/', '', $blockSplit[$k]);
1046                     }
1047                 }
1048                 // If $blockSplit[$k] is blank then unset the line, unless the line only contained linebreaks
1049                 if ((string)$blockSplit[$k] === '' && !$onlyLineBreaks) {
1050                     unset($blockSplit[$k]);
1051                 } else {
1052                     $blockSplit[$k] = $this->setDivTags($blockSplit[$k], $this->procOptions['useDIVasParagraphTagForRTE'] ? 'div' : 'p');
1053                 }
1054             }
1055         }
1056         return implode(LF, $blockSplit);
1057     }
1058
1059     /**
1060      * Transformation handler: 'ts_strip' / direction: "db"
1061      * Removing all non-allowed tags
1062      *
1063      * @param string $value Content input
1064      * @return string Content output
1065      */
1066     public function TS_strip_db($value)
1067     {
1068         $value = strip_tags($value, '<' . implode('><', explode(',', 'b,i,u,a,img,br,div,center,pre,font,hr,sub,sup,p,strong,em,li,ul,ol,blockquote')) . '>');
1069         return $value;
1070     }
1071
1072     /***************************************************************
1073      *
1074      * Generic RTE transformation, analysis and helper functions
1075      *
1076      **************************************************************/
1077     /**
1078      * Reads the file or url $url and returns the content
1079      *
1080      * @param string $url Filepath/URL to read
1081      * @return string The content from the resource given as input.
1082      * @see \TYPO3\CMS\Core\Utility\GeneralUtility::getUrl()
1083      */
1084     public function getUrl($url)
1085     {
1086         return GeneralUtility::getUrl($url);
1087     }
1088
1089     /**
1090      * Function for cleaning content going into the database.
1091      * Content is cleaned eg. by removing unallowed HTML and ds-HSC content
1092      * It is basically calling HTMLcleaner from the parent class with some preset configuration specifically set up for cleaning content going from the RTE into the db
1093      *
1094      * @param string $content Content to clean up
1095      * @param string $tagList Comma list of tags to specifically allow. Default comes from getKeepTags and is
1096      * @return string Clean content
1097      * @see getKeepTags()
1098      */
1099     public function HTMLcleaner_db($content, $tagList = '')
1100     {
1101         if (!$tagList) {
1102             $keepTags = $this->getKeepTags('db');
1103         } else {
1104             $keepTags = $this->getKeepTags('db', $tagList);
1105         }
1106         // Default: remove unknown tags.
1107         $kUknown = $this->procOptions['dontRemoveUnknownTags_db'] ? 1 : 0;
1108         // Default: re-convert literals to characters (that is &lt; to <)
1109         $hSC = $this->procOptions['dontUndoHSC_db'] ? 0 : -1;
1110         // Create additional configuration in order to honor the setting RTE.default.proc.HTMLparser_db.xhtml_cleaning=1
1111         $addConfig = array();
1112         if (is_array($this->procOptions['HTMLparser_db.']) && $this->procOptions['HTMLparser_db.']['xhtml_cleaning'] || is_array($this->procOptions['entryHTMLparser_db.']) && $this->procOptions['entryHTMLparser_db.']['xhtml_cleaning'] || is_array($this->procOptions['exitHTMLparser_db.']) && $this->procOptions['exitHTMLparser_db.']['xhtml_cleaning']) {
1113             $addConfig['xhtml'] = 1;
1114         }
1115         return $this->HTMLcleaner($content, $keepTags, $kUknown, $hSC, $addConfig);
1116     }
1117
1118     /**
1119      * Creates an array of configuration for the HTMLcleaner function based on whether content go TO or FROM the Rich Text Editor ($direction)
1120      * Unless "tagList" is given, the function will cache the configuration for next time processing goes on. (In this class that is the case only if we are processing a bulletlist)
1121      *
1122      * @param string $direction The direction of the content being processed by the output configuration; "db" (content going into the database FROM the rte) or "rte" (content going into the form)
1123      * @param string $tagList Comma list of tags to keep (overriding default which is to keep all + take notice of internal configuration)
1124      * @return array Configuration array
1125      * @see HTMLcleaner_db()
1126      */
1127     public function getKeepTags($direction = 'rte', $tagList = '')
1128     {
1129         if (!is_array($this->getKeepTags_cache[$direction]) || $tagList) {
1130             // Setting up allowed tags:
1131             // If the $tagList input var is set, this will take precedence
1132             if ((string)$tagList !== '') {
1133                 $keepTags = array_flip(GeneralUtility::trimExplode(',', $tagList, true));
1134             } else {
1135                 // Default is to get allowed/denied tags from internal array of processing options:
1136                 // Construct default list of tags to keep:
1137                 $typoScript_list = 'b,i,u,a,img,br,div,center,pre,font,hr,sub,sup,p,strong,em,li,ul,ol,blockquote,strike,span';
1138                 $keepTags = array_flip(GeneralUtility::trimExplode(',', $typoScript_list . ',' . strtolower($this->procOptions['allowTags']), true));
1139                 // For tags to deny, remove them from $keepTags array:
1140                 $denyTags = GeneralUtility::trimExplode(',', $this->procOptions['denyTags'], true);
1141                 foreach ($denyTags as $dKe) {
1142                     unset($keepTags[$dKe]);
1143                 }
1144             }
1145             // Based on the direction of content, set further options:
1146             switch ($direction) {
1147                 case 'rte':
1148                     if (!isset($this->procOptions['transformBoldAndItalicTags']) || $this->procOptions['transformBoldAndItalicTags']) {
1149                         // Transform bold/italics tags to strong/em
1150                         if (isset($keepTags['b'])) {
1151                             $keepTags['b'] = array('remap' => 'STRONG');
1152                         }
1153                         if (isset($keepTags['i'])) {
1154                             $keepTags['i'] = array('remap' => 'EM');
1155                         }
1156                     }
1157                     // Transforming keepTags array so it can be understood by the HTMLcleaner function. This basically converts the format of the array from TypoScript (having .'s) to plain multi-dimensional array.
1158                     list($keepTags) = $this->HTMLparserConfig($this->procOptions['HTMLparser_rte.'], $keepTags);
1159                     break;
1160                 case 'db':
1161                     if (!isset($this->procOptions['transformBoldAndItalicTags']) || $this->procOptions['transformBoldAndItalicTags']) {
1162                         // Transform strong/em back to bold/italics:
1163                         if (isset($keepTags['strong'])) {
1164                             $keepTags['strong'] = array('remap' => 'b');
1165                         }
1166                         if (isset($keepTags['em'])) {
1167                             $keepTags['em'] = array('remap' => 'i');
1168                         }
1169                     }
1170                     // Setting up span tags if they are allowed:
1171                     if (isset($keepTags['span'])) {
1172                         $classes = array_merge(array(''), $this->allowedClasses);
1173                         $keepTags['span'] = array(
1174                             'allowedAttribs' => 'id,class,style,title,lang,xml:lang,dir,itemscope,itemtype,itemprop',
1175                             'fixAttrib' => array(
1176                                 'class' => array(
1177                                     'list' => $classes,
1178                                     'removeIfFalse' => 1
1179                                 )
1180                             ),
1181                             'rmTagIfNoAttrib' => 1
1182                         );
1183                         if (!$this->procOptions['allowedClasses']) {
1184                             unset($keepTags['span']['fixAttrib']['class']['list']);
1185                         }
1186                     }
1187                     // Setting up font tags if they are allowed:
1188                     if (isset($keepTags['font'])) {
1189                         $colors = array_merge(array(''), GeneralUtility::trimExplode(',', $this->procOptions['allowedFontColors'], true));
1190                         $keepTags['font'] = array(
1191                             'allowedAttribs' => 'face,color,size',
1192                             'fixAttrib' => array(
1193                                 'face' => array(
1194                                     'removeIfFalse' => 1
1195                                 ),
1196                                 'color' => array(
1197                                     'removeIfFalse' => 1,
1198                                     'list' => $colors
1199                                 ),
1200                                 'size' => array(
1201                                     'removeIfFalse' => 1
1202                                 )
1203                             ),
1204                             'rmTagIfNoAttrib' => 1
1205                         );
1206                         if (!$this->procOptions['allowedFontColors']) {
1207                             unset($keepTags['font']['fixAttrib']['color']['list']);
1208                         }
1209                     }
1210                     // Setting further options, getting them from the processiong options:
1211                     $TSc = $this->procOptions['HTMLparser_db.'];
1212                     if (!$TSc['globalNesting']) {
1213                         $TSc['globalNesting'] = 'b,i,u,a,center,font,sub,sup,strong,em,strike,span';
1214                     }
1215                     if (!$TSc['noAttrib']) {
1216                         $TSc['noAttrib'] = 'b,i,u,br,center,hr,sub,sup,strong,em,li,ul,ol,blockquote,strike';
1217                     }
1218                     // Transforming the array from TypoScript to regular array:
1219                     list($keepTags) = $this->HTMLparserConfig($TSc, $keepTags);
1220                     break;
1221             }
1222             // Caching (internally, in object memory) the result unless tagList is set:
1223             if (!$tagList) {
1224                 $this->getKeepTags_cache[$direction] = $keepTags;
1225             } else {
1226                 return $keepTags;
1227             }
1228         }
1229         // Return result:
1230         return $this->getKeepTags_cache[$direction];
1231     }
1232
1233     /**
1234      * This resolves the $value into parts based on <div></div>-sections and <p>-sections and <br />-tags. These are returned as lines separated by LF.
1235      * This point is to resolve the HTML-code returned from RTE into ordinary lines so it's 'human-readable'
1236      * The function ->setDivTags does the opposite.
1237      * This function processes content to go into the database.
1238      *
1239      * @param string $value Value to process.
1240      * @param int $count Recursion brake. Decremented on each recursion down to zero. Default is 5 (which equals the allowed nesting levels of p/div tags).
1241      * @param bool $returnArray If TRUE, an array with the lines is returned, otherwise a string of the processed input value.
1242      * @return string Processed input value.
1243      * @see setDivTags()
1244      */
1245     public function divideIntoLines($value, $count = 5, $returnArray = false)
1246     {
1247         // Setting configuration for processing:
1248         $allowTagsOutside = GeneralUtility::trimExplode(',', strtolower($this->procOptions['allowTagsOutside'] ? 'hr,' . $this->procOptions['allowTagsOutside'] : 'hr,img'), true);
1249         $remapParagraphTag = strtoupper($this->procOptions['remapParagraphTag']);
1250         $divSplit = $this->splitIntoBlock('div,p', $value, 1);
1251         // Setting the third param to 1 will eliminate false end-tags. Maybe this is a good thing to do...?
1252         if ($this->procOptions['keepPDIVattribs']) {
1253             $keepAttribListArr = GeneralUtility::trimExplode(',', strtolower($this->procOptions['keepPDIVattribs']), true);
1254         } else {
1255             $keepAttribListArr = array();
1256         }
1257         // Returns plainly the value if there was no div/p sections in it
1258         if (count($divSplit) <= 1 || $count <= 0) {
1259             // Wrap hr tags with LF's
1260             $newValue = preg_replace('/<(hr)(\\s[^>\\/]*)?[[:space:]]*\\/?>/i', LF . '<$1$2/>' . LF, $value);
1261             $newValue = preg_replace('/' . LF . LF . '/i', LF, $newValue);
1262             $newValue = preg_replace('/(^' . LF . ')|(' . LF . '$)/i', '', $newValue);
1263             return $newValue;
1264         }
1265         // Traverse the splitted sections:
1266         foreach ($divSplit as $k => $v) {
1267             if ($k % 2) {
1268                 // Inside
1269                 $v = $this->removeFirstAndLastTag($v);
1270                 // Fetching 'sub-lines' - which will explode any further p/div nesting...
1271                 $subLines = $this->divideIntoLines($v, $count - 1, 1);
1272                 // So, if there happend to be sub-nesting of p/div, this is written directly as the new content of THIS section. (This would be considered 'an error')
1273                 if (is_array($subLines)) {
1274                 } else {
1275                     //... but if NO subsection was found, we process it as a TRUE line without erronous content:
1276                     $subLines = array($subLines);
1277                     // process break-tags, if configured for. Simply, the breaktags will here be treated like if each was a line of content...
1278                     if (!$this->procOptions['dontConvBRtoParagraph']) {
1279                         $subLines = preg_split('/<br[[:space:]]*[\\/]?>/i', $v);
1280                     }
1281                     // Traverse sublines (there is typically one, except if <br/> has been converted to lines as well!)
1282                     foreach ($subLines as $sk => $value) {
1283                         // Clear up the subline for DB.
1284                         $subLines[$sk] = $this->HTMLcleaner_db($subLines[$sk]);
1285                         // Get first tag, attributes etc:
1286                         $fTag = $this->getFirstTag($divSplit[$k]);
1287                         $tagName = strtolower($this->getFirstTagName($divSplit[$k]));
1288                         $attribs = $this->get_tag_attributes($fTag);
1289                         // Keep attributes (lowercase)
1290                         $newAttribs = array();
1291                         if (!empty($keepAttribListArr)) {
1292                             foreach ($keepAttribListArr as $keepA) {
1293                                 if (isset($attribs[0][$keepA])) {
1294                                     $newAttribs[$keepA] = $attribs[0][$keepA];
1295                                 }
1296                             }
1297                         }
1298                         // ALIGN attribute:
1299                         if (!$this->procOptions['skipAlign'] && trim($attribs[0]['align']) !== '' && strtolower($attribs[0]['align']) != 'left') {
1300                             // Set to value, but not 'left'
1301                             $newAttribs['align'] = strtolower($attribs[0]['align']);
1302                         }
1303                         // CLASS attribute:
1304                         // Set to whatever value
1305                         if (!$this->procOptions['skipClass'] && trim($attribs[0]['class']) !== '') {
1306                             if (empty($this->allowedClasses) || in_array($attribs[0]['class'], $this->allowedClasses)) {
1307                                 $newAttribs['class'] = $attribs[0]['class'];
1308                             } else {
1309                                 $classes = GeneralUtility::trimExplode(' ', $attribs[0]['class'], true);
1310                                 $newClasses = array();
1311                                 foreach ($classes as $class) {
1312                                     if (in_array($class, $this->allowedClasses)) {
1313                                         $newClasses[] = $class;
1314                                     }
1315                                 }
1316                                 if (!empty($newClasses)) {
1317                                     $newAttribs['class'] = implode(' ', $newClasses);
1318                                 }
1319                             }
1320                         }
1321                         // Remove any line break char (10 or 13)
1322                         $subLines[$sk] = preg_replace('/' . LF . '|' . CR . '/', '', $subLines[$sk]);
1323                         // If there are any attributes or if we are supposed to remap the tag, then do so:
1324                         if (!empty($newAttribs) && $remapParagraphTag !== '1') {
1325                             if ($remapParagraphTag === 'P') {
1326                                 $tagName = 'p';
1327                             }
1328                             if ($remapParagraphTag === 'DIV') {
1329                                 $tagName = 'div';
1330                             }
1331                             $subLines[$sk] = '<' . trim($tagName . ' ' . $this->compileTagAttribs($newAttribs)) . '>' . $subLines[$sk] . '</' . $tagName . '>';
1332                         }
1333                     }
1334                 }
1335                 // Add the processed line(s)
1336                 $divSplit[$k] = implode(LF, $subLines);
1337                 // If it turns out the line is just blank (containing a &nbsp; possibly) then just make it pure blank.
1338                 // But, prevent filtering of lines that are blank in sense above, but whose tags contain attributes.
1339                 // Those attributes should have been filtered before; if they are still there they must be considered as possible content.
1340                 if (trim(strip_tags($divSplit[$k])) == '&nbsp;' && !preg_match('/\\<(img)(\\s[^>]*)?\\/?>/si', $divSplit[$k]) && !preg_match('/\\<([^>]*)?( align| class| style| id| title| dir| lang| xml:lang)([^>]*)?>/si', trim($divSplit[$k]))) {
1341                     $divSplit[$k] = '';
1342                 }
1343             } else {
1344                 // outside div:
1345                 // Remove positions which are outside div/p tags and without content
1346                 $divSplit[$k] = trim(strip_tags($divSplit[$k], '<' . implode('><', $allowTagsOutside) . '>'));
1347                 // Wrap hr tags with LF's
1348                 $divSplit[$k] = preg_replace('/<(hr)(\\s[^>\\/]*)?[[:space:]]*\\/?>/i', LF . '<$1$2/>' . LF, $divSplit[$k]);
1349                 $divSplit[$k] = preg_replace('/' . LF . LF . '/i', LF, $divSplit[$k]);
1350                 $divSplit[$k] = preg_replace('/(^' . LF . ')|(' . LF . '$)/i', '', $divSplit[$k]);
1351                 if ((string)$divSplit[$k] === '') {
1352                     unset($divSplit[$k]);
1353                 }
1354             }
1355         }
1356         // Return value:
1357         return $returnArray ? $divSplit : implode(LF, $divSplit);
1358     }
1359
1360     /**
1361      * Converts all lines into <div></div>/<p></p>-sections (unless the line is a div-section already)
1362      * For processing of content going FROM database TO RTE.
1363      *
1364      * @param string $value Value to convert
1365      * @param string $dT Tag to wrap with. Either "p" or "div" should it be. Lowercase preferably.
1366      * @return string Processed value.
1367      * @see divideIntoLines()
1368      */
1369     public function setDivTags($value, $dT = 'p')
1370     {
1371         // First, setting configuration for the HTMLcleaner function. This will process each line between the <div>/<p> section on their way to the RTE
1372         $keepTags = $this->getKeepTags('rte');
1373         // Default: remove unknown tags.
1374         $kUknown = $this->procOptions['dontProtectUnknownTags_rte'] ? 0 : 'protect';
1375         // Default: re-convert literals to characters (that is &lt; to <)
1376         $hSC = $this->procOptions['dontHSC_rte'] ? 0 : 1;
1377         $convNBSP = !$this->procOptions['dontConvAmpInNBSP_rte'] ? 1 : 0;
1378         // Divide the content into lines, based on LF:
1379         $parts = explode(LF, $value);
1380         foreach ($parts as $k => $v) {
1381             // Processing of line content:
1382             // If the line is blank, set it to &nbsp;
1383             if (trim($parts[$k]) === '') {
1384                 $parts[$k] = '&nbsp;';
1385             } else {
1386                 // Clean the line content:
1387                 $parts[$k] = $this->HTMLcleaner($parts[$k], $keepTags, $kUknown, $hSC);
1388                 if ($convNBSP) {
1389                     $parts[$k] = str_replace('&amp;nbsp;', '&nbsp;', $parts[$k]);
1390                 }
1391             }
1392             // Wrapping the line in <$dT> if not already wrapped and does not contain an hr tag
1393             if (!preg_match('/<(hr)(\\s[^>\\/]*)?[[:space:]]*\\/?>/i', $parts[$k])) {
1394                 $testStr = strtolower(trim($parts[$k]));
1395                 if (substr($testStr, 0, 4) != '<div' || substr($testStr, -6) != '</div>') {
1396                     if (substr($testStr, 0, 2) != '<p' || substr($testStr, -4) != '</p>') {
1397                         // Only set p-tags if there is not already div or p tags:
1398                         $parts[$k] = '<' . $dT . '>' . $parts[$k] . '</' . $dT . '>';
1399                     }
1400                 }
1401             }
1402         }
1403         // Implode result:
1404         return implode(LF, $parts);
1405     }
1406
1407
1408     /**
1409      * Returns SiteURL based on thisScript.
1410      *
1411      * @return string Value of GeneralUtility::getIndpEnv('TYPO3_SITE_URL');
1412      * @see \TYPO3\CMS\Core\Utility\GeneralUtility::getIndpEnv()
1413      */
1414     public function siteUrl()
1415     {
1416         return GeneralUtility::getIndpEnv('TYPO3_SITE_URL');
1417     }
1418
1419     /**
1420      * Remove all tables from incoming code
1421      * The function is trying to to this is some more or less respectfull way. The approach is to resolve each table cells content and implode it all by <br /> chars. Thus at least the content is preserved in some way.
1422      *
1423      * @param string $value Input value
1424      * @param string $breakChar Break character to use for linebreaks.
1425      * @return string Output value
1426      */
1427     public function removeTables($value, $breakChar = '<br />')
1428     {
1429         // Splitting value into table blocks:
1430         $tableSplit = $this->splitIntoBlock('table', $value);
1431         // Traverse blocks of tables:
1432         foreach ($tableSplit as $k => $v) {
1433             if ($k % 2) {
1434                 $tableSplit[$k] = '';
1435                 $rowSplit = $this->splitIntoBlock('tr', $v);
1436                 foreach ($rowSplit as $k2 => $v2) {
1437                     if ($k2 % 2) {
1438                         $cellSplit = $this->getAllParts($this->splitIntoBlock('td', $v2), 1, 0);
1439                         foreach ($cellSplit as $k3 => $v3) {
1440                             $tableSplit[$k] .= $v3 . $breakChar;
1441                         }
1442                     }
1443                 }
1444             }
1445         }
1446         // Implode it all again:
1447         return implode($breakChar, $tableSplit);
1448     }
1449
1450     /**
1451      * Default tag mapping for TS
1452      *
1453      * @param string $code Input code to process
1454      * @param string $direction Direction To databsae (db) or from database to RTE (rte)
1455      * @return string Processed value
1456      */
1457     public function defaultTStagMapping($code, $direction = 'rte')
1458     {
1459         if ($direction == 'db') {
1460             $code = $this->mapTags($code, array(
1461                 // Map tags
1462                 'strong' => 'b',
1463                 'em' => 'i'
1464             ));
1465         }
1466         if ($direction == 'rte') {
1467             $code = $this->mapTags($code, array(
1468                 // Map tags
1469                 'b' => 'strong',
1470                 'i' => 'em'
1471             ));
1472         }
1473         return $code;
1474     }
1475
1476     /**
1477      * Finds width and height from attrib-array
1478      * If the width and height is found in the style-attribute, use that!
1479      *
1480      * @param array $attribArray Array of attributes from tag in which to search. More specifically the content of the key "style" is used to extract "width:xxx / height:xxx" information
1481      * @return array Integer w/h in key 0/1. Zero is returned if not found.
1482      */
1483     public function getWHFromAttribs($attribArray)
1484     {
1485         $style = trim($attribArray['style']);
1486         if ($style) {
1487             $regex = '[[:space:]]*:[[:space:]]*([0-9]*)[[:space:]]*px';
1488             // Width
1489             $reg = array();
1490             preg_match('/width' . $regex . '/i', $style, $reg);
1491             $w = (int)$reg[1];
1492             // Height
1493             preg_match('/height' . $regex . '/i', $style, $reg);
1494             $h = (int)$reg[1];
1495         }
1496         if (!$w) {
1497             $w = $attribArray['width'];
1498         }
1499         if (!$h) {
1500             $h = $attribArray['height'];
1501         }
1502         return array((int)$w, (int)$h);
1503     }
1504
1505     /**
1506      * Parse <A>-tag href and return status of email,external,file or page
1507      *
1508      * @param string $url URL to analyse.
1509      * @return array Information in an array about the URL
1510      */
1511     public function urlInfoForLinkTags($url)
1512     {
1513         $info = array();
1514         $url = trim($url);
1515         if (substr(strtolower($url), 0, 7) == 'mailto:') {
1516             $info['url'] = trim(substr($url, 7));
1517             $info['type'] = 'email';
1518         } elseif (strpos($url, '?file:') !== false) {
1519             $info['type'] = 'file';
1520             $info['url'] = rawurldecode(substr($url, strpos($url, '?file:') + 1));
1521         } else {
1522             $curURL = $this->siteUrl();
1523             $urlLength = strlen($url);
1524             for ($a = 0; $a < $urlLength; $a++) {
1525                 if ($url[$a] != $curURL[$a]) {
1526                     break;
1527                 }
1528             }
1529             $info['relScriptPath'] = substr($curURL, $a);
1530             $info['relUrl'] = substr($url, $a);
1531             $info['url'] = $url;
1532             $info['type'] = 'ext';
1533             $siteUrl_parts = parse_url($url);
1534             $curUrl_parts = parse_url($curURL);
1535             // Hosts should match
1536             if ($siteUrl_parts['host'] == $curUrl_parts['host'] && (!$info['relScriptPath'] || defined('TYPO3_mainDir') && substr($info['relScriptPath'], 0, strlen(TYPO3_mainDir)) == TYPO3_mainDir)) {
1537                 // If the script path seems to match or is empty (FE-EDIT)
1538                 // New processing order 100502
1539                 $uP = parse_url($info['relUrl']);
1540                 if ($info['relUrl'] === '#' . $siteUrl_parts['fragment']) {
1541                     $info['url'] = $info['relUrl'];
1542                     $info['type'] = 'anchor';
1543                 } elseif (!trim($uP['path']) || $uP['path'] === 'index.php') {
1544                     // URL is a page (id parameter)
1545                     $pp = preg_split('/^id=/', $uP['query']);
1546                     $pp[1] = preg_replace('/&id=[^&]*/', '', $pp[1]);
1547                     $parameters = explode('&', $pp[1]);
1548                     $id = array_shift($parameters);
1549                     if ($id) {
1550                         $info['pageid'] = $id;
1551                         $info['cElement'] = $uP['fragment'];
1552                         $info['url'] = $id . ($info['cElement'] ? '#' . $info['cElement'] : '');
1553                         $info['type'] = 'page';
1554                         $info['query'] = $parameters[0] ? '&' . implode('&', $parameters) : '';
1555                     }
1556                 } else {
1557                     $info['url'] = $info['relUrl'];
1558                     $info['type'] = 'file';
1559                 }
1560             } else {
1561                 unset($info['relScriptPath']);
1562                 unset($info['relUrl']);
1563             }
1564         }
1565         return $info;
1566     }
1567
1568     /**
1569      * Converting <A>-tags to absolute URLs (+ setting rtekeep attribute)
1570      *
1571      * @param string $value Content input
1572      * @param bool $dontSetRTEKEEP If TRUE, then the "rtekeep" attribute will not be set.
1573      * @return string Content output
1574      */
1575     public function TS_AtagToAbs($value, $dontSetRTEKEEP = false)
1576     {
1577         $blockSplit = $this->splitIntoBlock('A', $value);
1578         foreach ($blockSplit as $k => $v) {
1579             // Block
1580             if ($k % 2) {
1581                 $attribArray = $this->get_tag_attributes_classic($this->getFirstTag($v), 1);
1582                 // Checking if there is a scheme, and if not, prepend the current url.
1583                 // ONLY do this if href has content - the <a> tag COULD be an anchor and if so, it should be preserved...
1584                 if ($attribArray['href'] !== '') {
1585                     $uP = parse_url(strtolower($attribArray['href']));
1586                     if (!$uP['scheme']) {
1587                         $attribArray['href'] = $this->siteUrl() . substr($attribArray['href'], strlen($this->relBackPath));
1588                     } elseif ($uP['scheme'] != 'mailto') {
1589                         $attribArray['data-htmlarea-external'] = 1;
1590                     }
1591                 } else {
1592                     $attribArray['rtekeep'] = 1;
1593                 }
1594                 if (!$dontSetRTEKEEP) {
1595                     $attribArray['rtekeep'] = 1;
1596                 }
1597                 $bTag = '<a ' . GeneralUtility::implodeAttributes($attribArray, 1) . '>';
1598                 $eTag = '</a>';
1599                 $blockSplit[$k] = $bTag . $this->TS_AtagToAbs($this->removeFirstAndLastTag($blockSplit[$k])) . $eTag;
1600             }
1601         }
1602         return implode('', $blockSplit);
1603     }
1604
1605     /**
1606      * Apply plain image settings to the dimensions of the image
1607      *
1608      * @param array $imageInfo: info array of the image
1609      * @param array $attribArray: array of attributes of an image tag
1610      *
1611      * @return array a modified attributes array
1612      */
1613     protected function applyPlainImageModeSettings($imageInfo, $attribArray)
1614     {
1615         if ($this->procOptions['plainImageMode']) {
1616             // Perform corrections to aspect ratio based on configuration
1617             switch ((string)$this->procOptions['plainImageMode']) {
1618                 case 'lockDimensions':
1619                     $attribArray['width'] = $imageInfo[0];
1620                     $attribArray['height'] = $imageInfo[1];
1621                     break;
1622                 case 'lockRatioWhenSmaller':
1623                     if ($attribArray['width'] > $imageInfo[0]) {
1624                         $attribArray['width'] = $imageInfo[0];
1625                     }
1626                 case 'lockRatio':
1627                     if ($imageInfo[0] > 0) {
1628                         $attribArray['height'] = round($attribArray['width'] * ($imageInfo[1] / $imageInfo[0]));
1629                     }
1630                     break;
1631             }
1632         }
1633         return $attribArray;
1634     }
1635
1636     /**
1637     * @return \TYPO3\CMS\Core\Log\Logger
1638     */
1639     protected function getLogger()
1640     {
1641         /** @var $logManager \TYPO3\CMS\Core\Log\LogManager */
1642         $logManager = GeneralUtility::makeInstance(\TYPO3\CMS\Core\Log\LogManager::class);
1643
1644         return $logManager->getLogger(get_class($this));
1645     }
1646 }