From f45a286b8d62f710b519a98c7d4b75a0c34d5d10 Mon Sep 17 00:00:00 2001
From: Andrew Dolgov <fox@bah.org.ru>
Date: Mon, 22 Jun 2009 13:56:49 +0400
Subject: strip_tags_long: use htmlpurifier to properly reformat html content

---
 .../library/HTMLPurifier/ChildDef/Custom.php       | 90 ++++++++++++++++++++++
 1 file changed, 90 insertions(+)
 create mode 100755 lib/htmlpurifier/library/HTMLPurifier/ChildDef/Custom.php

(limited to 'lib/htmlpurifier/library/HTMLPurifier/ChildDef/Custom.php')

diff --git a/lib/htmlpurifier/library/HTMLPurifier/ChildDef/Custom.php b/lib/htmlpurifier/library/HTMLPurifier/ChildDef/Custom.php
new file mode 100755
index 000000000..b68047b4b
--- /dev/null
+++ b/lib/htmlpurifier/library/HTMLPurifier/ChildDef/Custom.php
@@ -0,0 +1,90 @@
+<?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.
+ */
+class HTMLPurifier_ChildDef_Custom extends HTMLPurifier_ChildDef
+{
+    public $type = 'custom';
+    public $allow_empty = false;
+    /**
+     * Allowed child pattern as defined by the DTD
+     */
+    public $dtd_regex;
+    /**
+     * PCRE regex derived from $dtd_regex
+     * @private
+     */
+    private $_pcre_regex;
+    /**
+     * @param $dtd_regex Allowed child pattern from the DTD
+     */
+    public function __construct($dtd_regex) {
+        $this->dtd_regex = $dtd_regex;
+        $this->_compileRegex();
+    }
+    /**
+     * Compiles the PCRE regex from a DTD regex ($dtd_regex to $_pcre_regex)
+     */
+    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 = $reg;
+    }
+    public function validateChildren($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 instanceof HTMLPurifier_Token_Start) {
+                $nesting++;
+            } elseif ($token instanceof HTMLPurifier_Token_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;
+    }
+}
+
+// vim: et sw=4 sts=4
-- 
cgit v1.2.3