[Summary view] [Print] [Text view]

   1  <?php
   2  
   3  /**
   4   * Forgivingly lexes HTML (SGML-style) markup into tokens.
   5   *
   6   * A lexer parses a string of SGML-style markup and converts them into
   7   * corresponding tokens.  It doesn't check for well-formedness, although its
   8   * internal mechanism may make this automatic (such as the case of
   9   * HTMLPurifier_Lexer_DOMLex).  There are several implementations to choose
  10   * from.
  11   *
  12   * A lexer is HTML-oriented: it might work with XML, but it's not
  13   * recommended, as we adhere to a subset of the specification for optimization
  14   * reasons. This might change in the future. Also, most tokenizers are not
  15   * expected to handle DTDs or PIs.
  16   *
  17   * This class should not be directly instantiated, but you may use create() to
  18   * retrieve a default copy of the lexer.  Being a supertype, this class
  19   * does not actually define any implementation, but offers commonly used
  20   * convenience functions for subclasses.
  21   *
  22   * @note The unit tests will instantiate this class for testing purposes, as
  23   *       many of the utility functions require a class to be instantiated.
  24   *       This means that, even though this class is not runnable, it will
  25   *       not be declared abstract.
  26   *
  27   * @par
  28   *
  29   * @note
  30   * We use tokens rather than create a DOM representation because DOM would:
  31   *
  32   * @par
  33   *  -# Require more processing and memory to create,
  34   *  -# Is not streamable, and
  35   *  -# Has the entire document structure (html and body not needed).
  36   *
  37   * @par
  38   * However, DOM is helpful in that it makes it easy to move around nodes
  39   * without a lot of lookaheads to see when a tag is closed. This is a
  40   * limitation of the token system and some workarounds would be nice.
  41   */
  42  class HTMLPurifier_Lexer
  43  {
  44  
  45      /**
  46       * Whether or not this lexer implements line-number/column-number tracking.
  47       * If it does, set to true.
  48       */
  49      public $tracksLineNumbers = false;
  50  
  51      // -- STATIC ----------------------------------------------------------
  52  
  53      /**
  54       * Retrieves or sets the default Lexer as a Prototype Factory.
  55       *
  56       * By default HTMLPurifier_Lexer_DOMLex will be returned. There are
  57       * a few exceptions involving special features that only DirectLex
  58       * implements.
  59       *
  60       * @note The behavior of this class has changed, rather than accepting
  61       *       a prototype object, it now accepts a configuration object.
  62       *       To specify your own prototype, set %Core.LexerImpl to it.
  63       *       This change in behavior de-singletonizes the lexer object.
  64       *
  65       * @param HTMLPurifier_Config $config
  66       * @return HTMLPurifier_Lexer
  67       * @throws HTMLPurifier_Exception
  68       */
  69      public static function create($config)
  70      {
  71          if (!($config instanceof HTMLPurifier_Config)) {
  72              $lexer = $config;
  73              trigger_error(
  74                  "Passing a prototype to
  75                  HTMLPurifier_Lexer::create() is deprecated, please instead
  76                  use %Core.LexerImpl",
  77                  E_USER_WARNING
  78              );
  79          } else {
  80              $lexer = $config->get('Core.LexerImpl');
  81          }
  82  
  83          $needs_tracking =
  84              $config->get('Core.MaintainLineNumbers') ||
  85              $config->get('Core.CollectErrors');
  86  
  87          $inst = null;
  88          if (is_object($lexer)) {
  89              $inst = $lexer;
  90          } else {
  91              if (is_null($lexer)) {
  92                  do {
  93                      // auto-detection algorithm
  94                      if ($needs_tracking) {
  95                          $lexer = 'DirectLex';
  96                          break;
  97                      }
  98  
  99                      if (class_exists('DOMDocument', false) &&
 100                          method_exists('DOMDocument', 'loadHTML') &&
 101                          !extension_loaded('domxml')
 102                      ) {
 103                          // check for DOM support, because while it's part of the
 104                          // core, it can be disabled compile time. Also, the PECL
 105                          // domxml extension overrides the default DOM, and is evil
 106                          // and nasty and we shan't bother to support it
 107                          $lexer = 'DOMLex';
 108                      } else {
 109                          $lexer = 'DirectLex';
 110                      }
 111                  } while (0);
 112              } // do..while so we can break
 113  
 114              // instantiate recognized string names
 115              switch ($lexer) {
 116                  case 'DOMLex':
 117                      $inst = new HTMLPurifier_Lexer_DOMLex();
 118                      break;
 119                  case 'DirectLex':
 120                      $inst = new HTMLPurifier_Lexer_DirectLex();
 121                      break;
 122                  case 'PH5P':
 123                      $inst = new HTMLPurifier_Lexer_PH5P();
 124                      break;
 125                  default:
 126                      throw new HTMLPurifier_Exception(
 127                          "Cannot instantiate unrecognized Lexer type " .
 128                          htmlspecialchars($lexer)
 129                      );
 130              }
 131          }
 132  
 133          if (!$inst) {
 134              throw new HTMLPurifier_Exception('No lexer was instantiated');
 135          }
 136  
 137          // once PHP DOM implements native line numbers, or we
 138          // hack out something using XSLT, remove this stipulation
 139          if ($needs_tracking && !$inst->tracksLineNumbers) {
 140              throw new HTMLPurifier_Exception(
 141                  'Cannot use lexer that does not support line numbers with ' .
 142                  'Core.MaintainLineNumbers or Core.CollectErrors (use DirectLex instead)'
 143              );
 144          }
 145  
 146          return $inst;
 147  
 148      }
 149  
 150      // -- CONVENIENCE MEMBERS ---------------------------------------------
 151  
 152      public function __construct()
 153      {
 154          $this->_entity_parser = new HTMLPurifier_EntityParser();
 155      }
 156  
 157      /**
 158       * Most common entity to raw value conversion table for special entities.
 159       * @type array
 160       */
 161      protected $_special_entity2str =
 162          array(
 163              '&quot;' => '"',
 164              '&amp;' => '&',
 165              '&lt;' => '<',
 166              '&gt;' => '>',
 167              '&#39;' => "'",
 168              '&#039;' => "'",
 169              '&#x27;' => "'"
 170          );
 171  
 172      public function parseText($string, $config) {
 173          return $this->parseData($string, false, $config);
 174      }
 175  
 176      public function parseAttr($string, $config) {
 177          return $this->parseData($string, true, $config);
 178      }
 179  
 180      /**
 181       * Parses special entities into the proper characters.
 182       *
 183       * This string will translate escaped versions of the special characters
 184       * into the correct ones.
 185       *
 186       * @param string $string String character data to be parsed.
 187       * @return string Parsed character data.
 188       */
 189      public function parseData($string, $is_attr, $config)
 190      {
 191          // following functions require at least one character
 192          if ($string === '') {
 193              return '';
 194          }
 195  
 196          // subtracts amps that cannot possibly be escaped
 197          $num_amp = substr_count($string, '&') - substr_count($string, '& ') -
 198              ($string[strlen($string) - 1] === '&' ? 1 : 0);
 199  
 200          if (!$num_amp) {
 201              return $string;
 202          } // abort if no entities
 203          $num_esc_amp = substr_count($string, '&amp;');
 204          $string = strtr($string, $this->_special_entity2str);
 205  
 206          // code duplication for sake of optimization, see above
 207          $num_amp_2 = substr_count($string, '&') - substr_count($string, '& ') -
 208              ($string[strlen($string) - 1] === '&' ? 1 : 0);
 209  
 210          if ($num_amp_2 <= $num_esc_amp) {
 211              return $string;
 212          }
 213  
 214          // hmm... now we have some uncommon entities. Use the callback.
 215          if ($config->get('Core.LegacyEntityDecoder')) {
 216              $string = $this->_entity_parser->substituteSpecialEntities($string);
 217          } else {
 218              if ($is_attr) {
 219                  $string = $this->_entity_parser->substituteAttrEntities($string);
 220              } else {
 221                  $string = $this->_entity_parser->substituteTextEntities($string);
 222              }
 223          }
 224          return $string;
 225      }
 226  
 227      /**
 228       * Lexes an HTML string into tokens.
 229       * @param $string String HTML.
 230       * @param HTMLPurifier_Config $config
 231       * @param HTMLPurifier_Context $context
 232       * @return HTMLPurifier_Token[] array representation of HTML.
 233       */
 234      public function tokenizeHTML($string, $config, $context)
 235      {
 236          trigger_error('Call to abstract class', E_USER_ERROR);
 237      }
 238  
 239      /**
 240       * Translates CDATA sections into regular sections (through escaping).
 241       * @param string $string HTML string to process.
 242       * @return string HTML with CDATA sections escaped.
 243       */
 244      protected static function escapeCDATA($string)
 245      {
 246          return preg_replace_callback(
 247              '/<!\[CDATA\[(.+?)\]\]>/s',
 248              array('HTMLPurifier_Lexer', 'CDATACallback'),
 249              $string
 250          );
 251      }
 252  
 253      /**
 254       * Special CDATA case that is especially convoluted for <script>
 255       * @param string $string HTML string to process.
 256       * @return string HTML with CDATA sections escaped.
 257       */
 258      protected static function escapeCommentedCDATA($string)
 259      {
 260          return preg_replace_callback(
 261              '#<!--//--><!\[CDATA\[//><!--(.+?)//--><!\]\]>#s',
 262              array('HTMLPurifier_Lexer', 'CDATACallback'),
 263              $string
 264          );
 265      }
 266  
 267      /**
 268       * Special Internet Explorer conditional comments should be removed.
 269       * @param string $string HTML string to process.
 270       * @return string HTML with conditional comments removed.
 271       */
 272      protected static function removeIEConditional($string)
 273      {
 274          return preg_replace(
 275              '#<!--\[if [^>]+\]>.*?<!\[endif\]-->#si', // probably should generalize for all strings
 276              '',
 277              $string
 278          );
 279      }
 280  
 281      /**
 282       * Callback function for escapeCDATA() that does the work.
 283       *
 284       * @warning Though this is public in order to let the callback happen,
 285       *          calling it directly is not recommended.
 286       * @param array $matches PCRE matches array, with index 0 the entire match
 287       *                  and 1 the inside of the CDATA section.
 288       * @return string Escaped internals of the CDATA section.
 289       */
 290      protected static function CDATACallback($matches)
 291      {
 292          // not exactly sure why the character set is needed, but whatever
 293          return htmlspecialchars($matches[1], ENT_COMPAT, 'UTF-8');
 294      }
 295  
 296      /**
 297       * Takes a piece of HTML and normalizes it by converting entities, fixing
 298       * encoding, extracting bits, and other good stuff.
 299       * @param string $html HTML.
 300       * @param HTMLPurifier_Config $config
 301       * @param HTMLPurifier_Context $context
 302       * @return string
 303       * @todo Consider making protected
 304       */
 305      public function normalize($html, $config, $context)
 306      {
 307          // normalize newlines to \n
 308          if ($config->get('Core.NormalizeNewlines')) {
 309              $html = str_replace("\r\n", "\n", $html);
 310              $html = str_replace("\r", "\n", $html);
 311          }
 312  
 313          if ($config->get('HTML.Trusted')) {
 314              // escape convoluted CDATA
 315              $html = $this->escapeCommentedCDATA($html);
 316          }
 317  
 318          // escape CDATA
 319          $html = $this->escapeCDATA($html);
 320  
 321          $html = $this->removeIEConditional($html);
 322  
 323          // extract body from document if applicable
 324          if ($config->get('Core.ConvertDocumentToFragment')) {
 325              $e = false;
 326              if ($config->get('Core.CollectErrors')) {
 327                  $e =& $context->get('ErrorCollector');
 328              }
 329              $new_html = $this->extractBody($html);
 330              if ($e && $new_html != $html) {
 331                  $e->send(E_WARNING, 'Lexer: Extracted body');
 332              }
 333              $html = $new_html;
 334          }
 335  
 336          // expand entities that aren't the big five
 337          if ($config->get('Core.LegacyEntityDecoder')) {
 338              $html = $this->_entity_parser->substituteNonSpecialEntities($html);
 339          }
 340  
 341          // clean into wellformed UTF-8 string for an SGML context: this has
 342          // to be done after entity expansion because the entities sometimes
 343          // represent non-SGML characters (horror, horror!)
 344          $html = HTMLPurifier_Encoder::cleanUTF8($html);
 345  
 346          // if processing instructions are to removed, remove them now
 347          if ($config->get('Core.RemoveProcessingInstructions')) {
 348              $html = preg_replace('#<\?.+?\?>#s', '', $html);
 349          }
 350  
 351          $hidden_elements = $config->get('Core.HiddenElements');
 352          if ($config->get('Core.AggressivelyRemoveScript') &&
 353              !($config->get('HTML.Trusted') || !$config->get('Core.RemoveScriptContents')
 354              || empty($hidden_elements["script"]))) {
 355              $html = preg_replace('#<script[^>]*>.*?</script>#i', '', $html);
 356          }
 357  
 358          return $html;
 359      }
 360  
 361      /**
 362       * Takes a string of HTML (fragment or document) and returns the content
 363       * @todo Consider making protected
 364       */
 365      public function extractBody($html)
 366      {
 367          $matches = array();
 368          $result = preg_match('|(.*?)<body[^>]*>(.*)</body>|is', $html, $matches);
 369          if ($result) {
 370              // Make sure it's not in a comment
 371              $comment_start = strrpos($matches[1], '<!--');
 372              $comment_end   = strrpos($matches[1], '-->');
 373              if ($comment_start === false ||
 374                  ($comment_end !== false && $comment_end > $comment_start)) {
 375                  return $matches[2];
 376              }
 377          }
 378          return $html;
 379      }
 380  }
 381  
 382  // vim: et sw=4 sts=4