library/HTMLPurifier/ChildDef/Custom.php

   1 <?php
   2
   3 /**
   4  * Custom validation class, accepts DTD child definitions
   5  *
   6  * @warning Currently this class is an all or nothing proposition, that is,
   7  *          it will only give a bool return value.
   8  */
   9 class HTMLPurifier_ChildDef_Custom extends HTMLPurifier_ChildDef
  10 {
  11     /**
  12      * @type string
  13      */
  14     public $type = 'custom';
  15
  16     /**
  17      * @type bool
  18      */
  19     public $allow_empty = false;
  20
  21     /**
  22      * Allowed child pattern as defined by the DTD.
  23      * @type string
  24      */
  25     public $dtd_regex;
  26
  27     /**
  28      * PCRE regex derived from $dtd_regex.
  29      * @type string
  30      */
  31     private $_pcre_regex;
  32
  33     /**
  34      * @param $dtd_regex Allowed child pattern from the DTD
  35      */
  36     public function __construct($dtd_regex)
  37     {
  38         $this->dtd_regex = $dtd_regex;
  39         $this->_compileRegex();
  40     }
  41
  42     /**
  43      * Compiles the PCRE regex from a DTD regex ($dtd_regex to $_pcre_regex)
  44      */
  45     protected function _compileRegex()
  46     {
  47         $raw = str_replace(' ', '', $this->dtd_regex);
  48         if ($raw{0} != '(') {
  49             $raw = "($raw)";
  50         }
  51         $el = '[#a-zA-Z0-9_.-]+';
  52         $reg = $raw;
  53
  54         // COMPLICATED! AND MIGHT BE BUGGY! I HAVE NO CLUE WHAT I'M
  55         // DOING! Seriously: if there's problems, please report them.
  56
  57         // collect all elements into the $elements array
  58         preg_match_all("/$el/", $reg, $matches);
  59         foreach ($matches[0] as $match) {
  60             $this->elements[$match] = true;
  61         }
  62
  63         // setup all elements as parentheticals with leading commas
  64         $reg = preg_replace("/$el/", '(,\\0)', $reg);
  65
  66         // remove commas when they were not solicited
  67         $reg = preg_replace("/([^,(|]\(+),/", '\\1', $reg);
  68
  69         // remove all non-paranthetical commas: they are handled by first regex
  70         $reg = preg_replace("/,\(/", '(', $reg);
  71
  72         $this->_pcre_regex = $reg;
  73     }
  74
  75     /**
  76      * @param HTMLPurifier_Node[] $children
  77      * @param HTMLPurifier_Config $config
  78      * @param HTMLPurifier_Context $context
  79      * @return bool
  80      */
  81     public function validateChildren($children, $config, $context)
  82     {
  83         $list_of_children = '';
  84         $nesting = 0; // depth into the nest
  85         foreach ($children as $node) {
  86             if (!empty($node->is_whitespace)) {
  87                 continue;
  88             }
  89             $list_of_children .= $node->name . ',';
  90         }
  91         // add leading comma to deal with stray comma declarations
  92         $list_of_children = ',' . rtrim($list_of_children, ',');
  93         $okay =
  94             preg_match(
  95                 '/^,?' . $this->_pcre_regex . '$/',
  96                 $list_of_children
  97             );
  98         return (bool)$okay;
  99     }
 100 }
 101
 102 // vim: et sw=4 sts=4