1
0
mirror of https://github.com/ezyang/htmlpurifier.git synced 2025-08-06 14:16:32 +02:00

Add CDATA support to the Lexers, as well as give PEARSax3 entity replacement.

git-svn-id: http://htmlpurifier.org/svnroot/htmlpurifier/trunk@106 48356398-32a2-884e-a903-53898d9a118a
This commit is contained in:
Edward Z. Yang
2006-07-23 23:04:34 +00:00
parent 5ce0ae7056
commit 609977f9f5
6 changed files with 165 additions and 65 deletions

View File

@@ -61,43 +61,6 @@ class HTMLPurifier_Lexer_DirectLex extends HTMLPurifier_Lexer
*/
var $_whitespace = "\x20\x09\x0D\x0A";
/**
* Decimal to parsed string conversion table for special entities.
* @protected
*/
var $_special_dec2str = array(
34 => '"', // quote
38 => '&', // ampersand
39 => "'", // apostrophe
60 => '<', // less than sign
62 => '>' // greater than sign
);
/**
* Stripped entity names to decimal conversion table for special entities.
* @protected
*/
var $_special_ent2dec = array(
'quot' => 34,
'amp' => 38,
'lt' => 60,
'gt' => 62,
);
/**
* Most common entity to raw value conversion table for special entities.
* @protected
*/
var $_special_entity2str = array(
'&quot;' => '"',
'&amp;' => '&',
'&lt;' => '<',
'&gt;' => '>',
'&#39;' => "'",
'&#039;' => "'",
'&#x27;' => "'",
);
/**
* Substitutes only special entities with their parsed equivalents.
*
@@ -153,6 +116,9 @@ class HTMLPurifier_Lexer_DirectLex extends HTMLPurifier_Lexer
$inside_tag = false; // whether or not we're parsing the inside of a tag
$array = array(); // result array
// escape CDATA
$string = $this->escapeCDATA($string);
// expand entities THAT AREN'T THE BIG FIVE
$string = $this->substituteNonSpecialEntities($string);