From 80157030ad338d7a207b163bb9c8e0716d9068b3 Mon Sep 17 00:00:00 2001
From: Nathan Gass
Date: Tue, 28 Nov 2017 16:27:52 +0100
Subject: add more tests with fixes

---
 it_html.class | 20 +++++++++++++-------
 1 file changed, 13 insertions(+), 7 deletions(-)

(limited to 'it_html.class')

diff --git a/it_html.class b/it_html.class
index c7105b7..c75f05e 100644
--- a/it_html.class
+++ b/it_html.class
@@ -538,7 +538,7 @@ static function U(/* ... */)
 	if (!isset($base))
 		$base = preg_replace('/\?.*/', '', $_SERVER['REQUEST_URI']);
 
-	$base = preg_replace(array('|\0|', '/\\\\/'), array('', '/'), $base); # kill null chars, turn \ to /
+	$base = preg_replace('|\0|', '', $base); # kill null chars
 
 	if (!($u = @parse_url($base)))
 		list($u['path'], $u['query']) = explode("?", $base, 2);
@@ -555,20 +555,26 @@ static function U(/* ... */)
 
 	$schemepart = $hostpart ? ($u['scheme'] ? $u['scheme'] . ":" : "") . "//$hostpart" : ($u['scheme'] == "mailto" ? $u['scheme'] . ":" : "");
 
-	# hack: encode % if not followed by two hex digits
-	$path = preg_replace('/%(?![0-9a-f]{2})/i', '%25', $u['path']);
 
 	# if params replace url parameters that are present in base, replace them but keep their order
 	foreach ($u['params'] as $key => $dummy)
 		if (isset($params[$key]))
 			$u['params'][$key] = $params[$key];
 
-	$path = preg_replace_callback('|[^-a-z0-9_.+!*(),:?@&=/~$%#]|i', function($m) { return rawurlencode($m[0]); }, $path);
-	$path = preg_replace('|^([a-z0-9_]+:)?//[^/]*$|', '$0/', $path);	# Add slash if absolute url without a path, e.g. http://gna.ch
+	$u['path'] = preg_replace('|\\\\|', '/', $u['path']); # turn \ to /
+	foreach (['path', 'fragment'] as $key)
+		$u[$key] = preg_replace_callback(
+			'/[^-a-z0-9_.+!*(),:?@&=\/~$%#]|%(?![0-9a-f]{2})/i',
+			function($m) { return rawurlencode($m[0]); },
+			$u[$key]
+		);
+
+	if (!$u['path'] && $hostpart)
+		$u['path'] = '/';
+
 	$queryparams = it_url::params($u['params'] + $params);
-	$separator = strpos($path, "?") === false ? "?" : "&";
 
-	return $schemepart . $path . ($queryparams ? "$separator$queryparams" : "") . ($u['fragment'] ? "#" . $u['fragment'] : "");;
+	return $schemepart . $u['path'] . ($queryparams ? "?" . $queryparams : "") . ($u['fragment'] ? "#" . $u['fragment'] : "");;
 }
 
 
-- 
cgit v1.2.3