library/HTMLPurifier/ChildDef/Custom.php

   1 <?php
   2
   3 require_once 'HTMLPurifier/ChildDef.php';
   4
   5 /**
   6  * Custom validation class, accepts DTD child definitions
   7  *
   8  * @warning Currently this class is an all or nothing proposition, that is,
   9  *          it will only give a bool return value.
  10  * @note This class is currently not used by any code, although it is unit
  11  *       tested.
  12  */
  13 class HTMLPurifier_ChildDef_Custom extends HTMLPurifier_ChildDef
  14 {
  15     public $type = 'custom';
  16     public $allow_empty = false;
  17     /**
  18      * Allowed child pattern as defined by the DTD
  19      */
  20     public $dtd_regex;
  21     /**
  22      * PCRE regex derived from $dtd_regex
  23      * @private
  24      */
  25     private $_pcre_regex;
  26     /**
  27      * @param $dtd_regex Allowed child pattern from the DTD
  28      */
  29     public function __construct($dtd_regex) {
  30         $this->dtd_regex = $dtd_regex;
  31         $this->_compileRegex();
  32     }
  33     /**
  34      * Compiles the PCRE regex from a DTD regex ($dtd_regex to $_pcre_regex)
  35      */
  36     protected function _compileRegex() {
  37         $raw = str_replace(' ', '', $this->dtd_regex);
  38         if ($raw{0} != '(') {
  39             $raw = "($raw)";
  40         }
  41         $el = '[#a-zA-Z0-9_.-]+';
  42         $reg = $raw;
  43
  44         // COMPLICATED! AND MIGHT BE BUGGY! I HAVE NO CLUE WHAT I'M
  45         // DOING! Seriously: if there's problems, please report them.
  46
  47         // collect all elements into the $elements array
  48         preg_match_all("/$el/", $reg, $matches);
  49         foreach ($matches[0] as $match) {
  50             $this->elements[$match] = true;
  51         }
  52
  53         // setup all elements as parentheticals with leading commas
  54         $reg = preg_replace("/$el/", '(,\\0)', $reg);
  55
  56         // remove commas when they were not solicited
  57         $reg = preg_replace("/([^,(|]\(+),/", '\\1', $reg);
  58
  59         // remove all non-paranthetical commas: they are handled by first regex
  60         $reg = preg_replace("/,\(/", '(', $reg);
  61
  62         $this->_pcre_regex = $reg;
  63     }
  64     public function validateChildren($tokens_of_children, $config, $context) {
  65         $list_of_children = '';
  66         $nesting = 0; // depth into the nest
  67         foreach ($tokens_of_children as $token) {
  68             if (!empty($token->is_whitespace)) continue;
  69
  70             $is_child = ($nesting == 0); // direct
  71
  72             if ($token instanceof HTMLPurifier_Token_Start) {
  73                 $nesting++;
  74             } elseif ($token instanceof HTMLPurifier_Token_End) {
  75                 $nesting--;
  76             }
  77
  78             if ($is_child) {
  79                 $list_of_children .= $token->name . ',';
  80             }
  81         }
  82         // add leading comma to deal with stray comma declarations
  83         $list_of_children = ',' . rtrim($list_of_children, ',');
  84         $okay =
  85             preg_match(
  86                 '/^,?'.$this->_pcre_regex.'$/',
  87                 $list_of_children
  88             );
  89
  90         return (bool) $okay;
  91     }
  92 }
  93