0
0
mirror of https://github.com/ezyang/htmlpurifier.git synced 2024-12-24 01:01:53 +00:00
htmlpurifier/library/HTMLPurifier/ChildDef/Custom.php

88 lines
2.6 KiB
PHP

<?php
require_once 'HTMLPurifier/ChildDef.php';
/**
* Custom validation class, accepts DTD child definitions
*
* @warning Currently this class is an all or nothing proposition, that is,
* it will only give a bool return value.
* @note This class is currently not used by any code, although it is unit
* tested.
*/
class HTMLPurifier_ChildDef_Custom extends HTMLPurifier_ChildDef
{
var $type = 'custom';
var $allow_empty = false;
/**
* Allowed child pattern as defined by the DTD
*/
var $dtd_regex;
/**
* PCRE regex derived from $dtd_regex
* @private
*/
var $_pcre_regex;
/**
* @param $dtd_regex Allowed child pattern from the DTD
*/
function HTMLPurifier_ChildDef_Custom($dtd_regex) {
$this->dtd_regex = $dtd_regex;
$this->_compileRegex();
}
/**
* Compiles the PCRE regex from a DTD regex ($dtd_regex to $_pcre_regex)
*/
function _compileRegex() {
$raw = str_replace(' ', '', $this->dtd_regex);
if ($raw{0} != '(') {
$raw = "($raw)";
}
$el = '[#a-zA-Z0-9_.-]+';
$reg = $raw;
// COMPLICATED! AND MIGHT BE BUGGY! I HAVE NO CLUE WHAT I'M
// DOING! Seriously: if there's problems, please report them.
// setup all elements as parentheticals with leading commas
$reg = preg_replace("/$el/", '(,\\0)', $reg);
// remove commas when they were not solicited
$reg = preg_replace("/([^,(|]\(+),/", '\\1', $reg);
// remove all non-paranthetical commas: they are handled by first regex
$reg = preg_replace("/,\(/", '(', $reg);
$this->_pcre_regex = $reg;
}
function validateChildren($tokens_of_children, $config, &$context) {
$list_of_children = '';
$nesting = 0; // depth into the nest
foreach ($tokens_of_children as $token) {
if (!empty($token->is_whitespace)) continue;
$is_child = ($nesting == 0); // direct
if ($token->type == 'start') {
$nesting++;
} elseif ($token->type == 'end') {
$nesting--;
}
if ($is_child) {
$list_of_children .= $token->name . ',';
}
}
// add leading comma to deal with stray comma declarations
$list_of_children = ',' . rtrim($list_of_children, ',');
$okay =
preg_match(
'/^,?'.$this->_pcre_regex.'$/',
$list_of_children
);
return (bool) $okay;
}
}
?>