mirror/php-htmlpurifier
1
0
Fork 0
mirror of https://github.com/ezyang/htmlpurifier.git synced 2025-07-12 10:16:18 +02:00
Code Issues Releases Wiki Activity

Turn on entity parsing for the Lexers. Add PureHTMLDefinition and define removeForeignElements.

Browse Source 
git-svn-id: http://htmlpurifier.org/svnroot/html_purifier/trunk@31 48356398-32a2-884e-a903-53898d9a118a
This commit is contained in:
Edward Z. Yang
2006-04-16 03:00:05 +00:00
parent e7f5b1674d
commit c4b23cc775
5 changed files with 223 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
HTML_Lexer.php
View File

@ -75,7 +75,7 @@ class HTML_Lexer
if (!$inside_tag && $position_next_lt !== false) {
// We are not inside tag and there still is another tag to parse
$array[] = new MF_Text(substr($string, $cursor, $position_next_lt - $cursor));
$array[] = new MF_Text(html_entity_decode(substr($string, $cursor, $position_next_lt - $cursor)));
$cursor = $position_next_lt + 1;
$inside_tag = true;
continue;
@ -84,7 +84,7 @@ class HTML_Lexer
// If we're already at the end, break
if ($cursor === strlen($string)) break;
// Create Text of rest of string
$array[] = new MF_Text(substr($string, $cursor));
$array[] = new MF_Text(html_entity_decode(substr($string, $cursor)));
break;
} elseif ($inside_tag && $position_next_gt !== false) {
// We are in tag and it is well formed
@ -144,7 +144,7 @@ class HTML_Lexer
$inside_tag = false;
continue;
} else {
$array[] = new MF_Text('<' . substr($string, $cursor));
$array[] = new MF_Text('<' . html_entity_decode(substr($string, $cursor)));
break;
}
break;
@ -234,7 +234,7 @@ class HTML_Lexer
$value = substr($string, $position_next_quote + 1,
$position_end_quote - $position_next_quote - 1);
if ($key) {
$array[$key] = $value;
$array[$key] = html_entity_decode($value);
}
$cursor = $position_end_quote + 1;
} else {
@ -268,6 +268,7 @@ class HTML_Lexer_Sax extends HTML_Lexer
$parser->set_element_handler('openHandler','closeHandler');
$parser->set_data_handler('dataHandler');
$parser->set_escape_handler('escapeHandler');
$parser->set_option('XML_OPTION_ENTITIES_PARSED', 1);
$parser->parse($html);
return $this->tokens;
}