MDL-22537 new html purifier backported to stable, unfortunately PHP5 only

This commit is contained in:
Petr Skoda
2010-05-21 11:39:22 +00:00
parent f8502768a2
commit db359e6369
330 changed files with 13929 additions and 10060 deletions
@@ -1,80 +1,76 @@
<?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;
public $type = 'custom';
public $allow_empty = false;
/**
* Allowed child pattern as defined by the DTD
*/
var $dtd_regex;
public $dtd_regex;
/**
* PCRE regex derived from $dtd_regex
* @private
*/
var $_pcre_regex;
private $_pcre_regex;
/**
* @param $dtd_regex Allowed child pattern from the DTD
*/
function HTMLPurifier_ChildDef_Custom($dtd_regex) {
public function __construct($dtd_regex) {
$this->dtd_regex = $dtd_regex;
$this->_compileRegex();
}
/**
* Compiles the PCRE regex from a DTD regex ($dtd_regex to $_pcre_regex)
*/
function _compileRegex() {
protected 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.
// collect all elements into the $elements array
preg_match_all("/$el/", $reg, $matches);
foreach ($matches[0] as $match) {
$this->elements[$match] = true;
}
// 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) {
public 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 ($token->type == 'start') {
if ($token instanceof HTMLPurifier_Token_Start) {
$nesting++;
} elseif ($token->type == 'end') {
} elseif ($token instanceof HTMLPurifier_Token_End) {
$nesting--;
}
if ($is_child) {
$list_of_children .= $token->name . ',';
}
@@ -86,8 +82,9 @@ class HTMLPurifier_ChildDef_Custom extends HTMLPurifier_ChildDef
'/^,?'.$this->_pcre_regex.'$/',
$list_of_children
);
return (bool) $okay;
}
}
// vim: et sw=4 sts=4