U(): params can override base args, correctly add fragment to end of url if params are present, add some tests

2 files changed, 21 insertions, 6 deletions

diff --git a/it_html.class b/it_html.class
index 1ca2acf..ea8a159 100644
--- a/it_html.class
+++ b/it_html.class
@@ -542,6 +542,7 @@ static function U(/* ... */)
 
 	if (!($u = @parse_url($base)))
 		list($u['path'], $u['query']) = explode("?", $base, 2);
+	parse_str($u['query'], $u['params']);
 
 	$u['host'] = preg_match('/[^-_.0-9a-z]/i', $u['host']) && function_exists('idn_to_ascii') && ($idnahost = idn_to_ascii($GLOBALS['it_html']->p['charset'] == "iso-8859-1" ? utf8_encode($u['host']) : $u['host'])) ? $idnahost : $u['host'];	# Punycode hostname to include into webpage
 	$u['host'] = preg_replace_callback('/[^-_.0-9a-z\x80-\xff]/i', function($m) { return rawurlencode($m[0]); }, $u['host']); # Encode garbage chars in host
@@ -554,20 +555,23 @@ static function U(/* ... */)
 
 	$schemepart = $hostpart ? ($u['scheme'] ? $u['scheme'] . ":" : "") . "//$hostpart" : ($u['scheme'] == "mailto" ? $u['scheme'] . ":" : "");
 
-	$path = $u['path'] . ($u['query'] ? "?" . strtr($u['query'], array("?" => rawurlencode("?"))) : "") . ($u['fragment'] ? "#" . $u['fragment'] : "");
-
 	# hack: encode % if not followed by two hex digits
-	$parts = preg_split('/%([^%]{0,2})/', $path, -1, PREG_SPLIT_DELIM_CAPTURE);
+	$parts = preg_split('/%([^%]{0,2})/', $u['path'], -1, PREG_SPLIT_DELIM_CAPTURE);
 	for ($i = 1; $i < count($parts); $i+=2)
 		$parts[$i] = (preg_match('/[0-9a-f][0-9a-f]/i', $parts[$i]) ? "%" : "%25") . $parts[$i];
 	$path = join("", $parts);
 
+	# 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
-	$queryparams = it_url::params($params);
+	$queryparams = it_url::params($u['params'] + $params);
 	$separator = strpos($path, "?") === false ? "?" : "&";
 
-	return $schemepart . $path . ($queryparams ? "$separator$queryparams" : "");
+	return $schemepart . $path . ($queryparams ? "$separator$queryparams" : "") . ($u['fragment'] ? "#" . $u['fragment'] : "");;
 }
 
 
diff --git a/tests/it_html.t b/tests/it_html.t
index 2dc7be5..538a350 100755
--- a/tests/it_html.t
+++ b/tests/it_html.t
@@ -190,6 +190,17 @@ is(
 );
 
 is(
+	U("/foo.html?bar=qux", array('bar' => "baz")),
+	'/foo.html?bar=baz',
+	'U() args override get params in base url'
+);
+
+is(
+	U("/foo.html?bar=qux#frag=frog", array('baz' => "gna")),
+	'/foo.html?bar=qux&baz=gna#frag=frog',
+	'U() fragment after params'
+);
+is(
 	U("Jet d'eau"),
 	'Jet%20d%27eau',
 	'U() with single quotes in URL'
@@ -197,7 +208,7 @@ is(
 
 is(
 	U('/test.html?foo=bar?qux=gna', array('?q' => '?r')),
-	'/test.html?foo=bar%3Fqux=gna&%3Fq=%3Fr',
+	'/test.html?foo=bar%3Fqux%3Dgna&%3Fq=%3Fr',
 	'U() quoting of ? in args but not base'
 );
 is(