dont urlencode IDN host parts in U()

1 files changed, 19 insertions, 7 deletions

diff --git a/it_html.class b/it_html.class
index 15442c3..5569309 100644
--- a/it_html.class
+++ b/it_html.class
@@ -442,20 +442,32 @@ function u(/* ... */)
 	if (!isset($base))
 		$base = preg_replace('/\?.*/', '', $_SERVER['REQUEST_URI']);
 
-	$base = preg_replace(array('|\0|', '/\\\\/'), array('', '/'), $base);
+	$base = preg_replace(array('|\0|', '/\\\\/'), array('', '/'), $base); # kill null chars, turn \ to /
+
+	$u = parse_url($base);
+
+	# handle scheme, user, password, host
+	$hostpart = 
+		($u['user'] ? $u['user'] . ($u['pass'] ? ":" . $u['pass'] : "") . "@" : "") .
+		($u['host'] ? $u['host'] : "") .
+		($u['port'] ? ":" . intval($u['port']) : "");
+
+	$schemepart = $hostpart ? ($u['scheme'] ? $u['scheme'] . ":" : "") . "//$hostpart" : "";
+
+	$path = $u['path'] . ($u['query'] ? "?" . $u['query'] : "") . ($u['anchor'] ? "#" . $u['anchor'] : "");
 
 	# hack: encode % if not followed by two hex digits
-	$parts = preg_split('/%([^%]{0,2})/', $base, -1, PREG_SPLIT_DELIM_CAPTURE);
+	$parts = preg_split('/%([^%]{0,2})/', $path, -1, PREG_SPLIT_DELIM_CAPTURE);
 	for ($i = 1; $i < count($parts); $i+=2)
 		$parts[$i] = (preg_match('/[0-9a-f][0-9a-f]/i', $parts[$i]) ? "%" : "%25") . $parts[$i];
-	$base = join("", $parts);
+	$path = join("", $parts);
 
-	$base = preg_replace('|[^-\w.+!*(),:?@&=/~$%]|e', 'urlencode(stripslashes("$0"))', $base);	# Single quotes are escaped with slash by preg_replace, remove it for urlencode
-	$base = preg_replace('|^(\w+:)?//[^/]*$|', '$0/', $base);	# Add slash if absolute url without a path, e.g. http://gna.ch
+	$path = preg_replace('|[^-\w.+!*(),:?@&=/~$%]|e', 'urlencode(stripslashes("$0"))', $path);	# Single quotes are escaped with slash by preg_replace, remove it for urlencode
+	$path = preg_replace('|^(\w+:)?//[^/]*$|', '$0/', $path);	# Add slash if absolute url without a path, e.g. http://gna.ch
 	$queryparams = it_url::params($params);
-	$separator = strpos($base, "?") === false ? "?" : "&";
+	$separator = strpos($path, "?") === false ? "?" : "&";
 
-	return $base . ($queryparams ? "$separator$queryparams" : "");
+	return $schemepart . $path . ($queryparams ? "$separator$queryparams" : "");
 }