<?
/*
(c) Copyright by Alexander Zhukov alex@veresk.ru. Released under LGPL
TODO: correct comment handling
*/
include("StateMachine.class.php");
include("AttributesParser.class.php");
include("HTMLSax.class.php");
// Sample HTML text
$html = "
<html>
<head>
<title>
my test title
</title>
</head>
<body topmargin=10>
<h1>A test text</h1>
example text1
example text2
<a href=http://test/?asd>test link</a>
<img src=\"/images/smile.gif\" width=100 height=100>
<table width=100% bgcolor=\"black\">
<tr color=#cceeff>
<td>cell1</td><td>cell2</td>
</tr>
</table>
</body>
</html>
";
class MySax extends HTMLSax
{
function MySax()
{
$this->HTMLSax();
$this->skipWhitespace = 1; // turn this on if you want to skip whitespace between tags
$this->trimDataNodes = 1; // turn this on if you want to trim the data nodes
}
function handle_data($data)
{
echo "data node \"$data\"<br>";
}
function handle_start_tag($tag,$attribs)
{
echo "start tag \"$tag\"<br>";
}
function handle_end_tag($tag)
{
echo "end tag \"$tag\"<br>";
}
}
$p = new MySax();
$p->parse($html);
?>
|