gistfile1.php

<?

function replace_h($matches) {
	print_r($matches); die;
	return $matches[1] . "\n" . str_repeat('-', length($matches[1]));
}

function replace_a($matches) {
	return $matches[2] . " (" . $matches[1] . ") ";
}

$html = $_POST['html'];

$clean = strip_tags($html, "<h1><h2><h3><h4><p><a><td>");
$clean = preg_replace("/&nbsp;/", "", $clean);
$clean = preg_replace("/[\n\t]/", "", $clean);
$clean = html_entity_decode($clean);

$clean = preg_replace_callback("|<h([1-4]).*?>(.+?)</h\1>|", 'replace_h', $clean);
$clean = preg_replace_callback('|<a [^>]*?href="(.+?)"[^>]*?>(.+?)</a>|', 'replace_a', $clean);
$clean = preg_replace(':<(p|td)[^>]*?>(.+?)</\1>:', "\$2\n", $clean);

# Anything the stupid regexes missed.
$clean = strip_tags($clean);
$clean = preg_replace('/^\s+/m', '', $clean);
$clean = preg_replace("/\n/", "\n\n", $clean);
$clean = wordwrap($clean, 80);

header('Content-Type: text/plain');
print trim($clean);