library/HTMLPurifier/Lexer.php

   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') &&
 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     /**
 173      * Parses special entities into the proper characters.
 174      *
 175      * This string will translate escaped versions of the special characters
 176      * into the correct ones.
 177      *
 178      * @warning
 179      * You should be able to treat the output of this function as
 180      * completely parsed, but that's only because all other entities should
 181      * have been handled previously in substituteNonSpecialEntities()
 182      *
 183      * @param string $string String character data to be parsed.
 184      * @return string Parsed character data.
 185      */
 186     public function parseData($string)
 187     {
 188         // following functions require at least one character
 189         if ($string === '') {
 190             return '';
 191         }
 192
 193         // subtracts amps that cannot possibly be escaped
 194         $num_amp = substr_count($string, '&') - substr_count($string, '& ') -
 195             ($string[strlen($string) - 1] === '&' ? 1 : 0);
 196
 197         if (!$num_amp) {
 198             return $string;
 199         } // abort if no entities
 200         $num_esc_amp = substr_count($string, '&amp;');
 201         $string = strtr($string, $this->_special_entity2str);
 202
 203         // code duplication for sake of optimization, see above
 204         $num_amp_2 = substr_count($string, '&') - substr_count($string, '& ') -
 205             ($string[strlen($string) - 1] === '&' ? 1 : 0);
 206
 207         if ($num_amp_2 <= $num_esc_amp) {
 208             return $string;
 209         }
 210
 211         // hmm... now we have some uncommon entities. Use the callback.
 212         $string = $this->_entity_parser->substituteSpecialEntities($string);
 213         return $string;
 214     }
 215
 216     /**
 217      * Lexes an HTML string into tokens.
 218      * @param $string String HTML.
 219      * @param HTMLPurifier_Config $config
 220      * @param HTMLPurifier_Context $context
 221      * @return HTMLPurifier_Token[] array representation of HTML.
 222      */
 223     public function tokenizeHTML($string, $config, $context)
 224     {
 225         trigger_error('Call to abstract class', E_USER_ERROR);
 226     }
 227
 228     /**
 229      * Translates CDATA sections into regular sections (through escaping).
 230      * @param string $string HTML string to process.
 231      * @return string HTML with CDATA sections escaped.
 232      */
 233     protected static function escapeCDATA($string)
 234     {
 235         return preg_replace_callback(
 236             '/<!\[CDATA\[(.+?)\]\]>/s',
 237             array('HTMLPurifier_Lexer', 'CDATACallback'),
 238             $string
 239         );
 240     }
 241
 242     /**
 243      * Special CDATA case that is especially convoluted for <script>
 244      * @param string $string HTML string to process.
 245      * @return string HTML with CDATA sections escaped.
 246      */
 247     protected static function escapeCommentedCDATA($string)
 248     {
 249         return preg_replace_callback(
 250             '#<!--//--><!\[CDATA\[//><!--(.+?)//--><!\]\]>#s',
 251             array('HTMLPurifier_Lexer', 'CDATACallback'),
 252             $string
 253         );
 254     }
 255
 256     /**
 257      * Special Internet Explorer conditional comments should be removed.
 258      * @param string $string HTML string to process.
 259      * @return string HTML with conditional comments removed.
 260      */
 261     protected static function removeIEConditional($string)
 262     {
 263         return preg_replace(
 264             '#<!--\[if [^>]+\]>.*?<!\[endif\]-->#si', // probably should generalize for all strings
 265             '',
 266             $string
 267         );
 268     }
 269
 270     /**
 271      * Callback function for escapeCDATA() that does the work.
 272      *
 273      * @warning Though this is public in order to let the callback happen,
 274      *          calling it directly is not recommended.
 275      * @param array $matches PCRE matches array, with index 0 the entire match
 276      *                  and 1 the inside of the CDATA section.
 277      * @return string Escaped internals of the CDATA section.
 278      */
 279     protected static function CDATACallback($matches)
 280     {
 281         // not exactly sure why the character set is needed, but whatever
 282         return htmlspecialchars($matches[1], ENT_COMPAT, 'UTF-8');
 283     }
 284
 285     /**
 286      * Takes a piece of HTML and normalizes it by converting entities, fixing
 287      * encoding, extracting bits, and other good stuff.
 288      * @param string $html HTML.
 289      * @param HTMLPurifier_Config $config
 290      * @param HTMLPurifier_Context $context
 291      * @return string
 292      * @todo Consider making protected
 293      */
 294     public function normalize($html, $config, $context)
 295     {
 296         // normalize newlines to \n
 297         if ($config->get('Core.NormalizeNewlines')) {
 298             $html = str_replace("\r\n", "\n", $html);
 299             $html = str_replace("\r", "\n", $html);
 300         }
 301
 302         if ($config->get('HTML.Trusted')) {
 303             // escape convoluted CDATA
 304             $html = $this->escapeCommentedCDATA($html);
 305         }
 306
 307         // escape CDATA
 308         $html = $this->escapeCDATA($html);
 309
 310         $html = $this->removeIEConditional($html);
 311
 312         // extract body from document if applicable
 313         if ($config->get('Core.ConvertDocumentToFragment')) {
 314             $e = false;
 315             if ($config->get('Core.CollectErrors')) {
 316                 $e =& $context->get('ErrorCollector');
 317             }
 318             $new_html = $this->extractBody($html);
 319             if ($e && $new_html != $html) {
 320                 $e->send(E_WARNING, 'Lexer: Extracted body');
 321             }
 322             $html = $new_html;
 323         }
 324
 325         // expand entities that aren't the big five
 326         $html = $this->_entity_parser->substituteNonSpecialEntities($html);
 327
 328         // clean into wellformed UTF-8 string for an SGML context: this has
 329         // to be done after entity expansion because the entities sometimes
 330         // represent non-SGML characters (horror, horror!)
 331         $html = HTMLPurifier_Encoder::cleanUTF8($html);
 332
 333         // if processing instructions are to removed, remove them now
 334         if ($config->get('Core.RemoveProcessingInstructions')) {
 335             $html = preg_replace('#<\?.+?\?>#s', '', $html);
 336         }
 337
 338         return $html;
 339     }
 340
 341     /**
 342      * Takes a string of HTML (fragment or document) and returns the content
 343      * @todo Consider making protected
 344      */
 345     public function extractBody($html)
 346     {
 347         $matches = array();
 348         $result = preg_match('!<body[^>]*>(.*)</body>!is', $html, $matches);
 349         if ($result) {
 350             return $matches[1];
 351         } else {
 352             return $html;
 353         }
 354     }
 355 }
 356
 357 // vim: et sw=4 sts=4