add more tests with fixes

2 files changed, 108 insertions, 7 deletions

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'] : "");;
 }
 
 
diff --git a/tests/U_tests.json b/tests/U_tests.json
index a8b5688..3e1d092 100644
--- a/tests/U_tests.json
+++ b/tests/U_tests.json
@@ -1,4 +1,39 @@
 [
+	{
+		"args": ["/Zürich"],
+		"exp":  "/Z%C3%BCrich",
+		"name": "U() with utf8 umlaut"
+	},
+
+	{
+		"args": ["/path#Zürich"],
+		"exp":  "/path#Z%C3%BCrich",
+		"name": "U() with utf8 umlaut in fragment"
+	},
+
+	{
+		"args": ["/foo.html?"],
+		"exp":  "/foo.html",
+		"name": "U() removes ? at end"
+	},
+
+	{
+		"args": ["/foo.html#"],
+		"exp":  "/foo.html",
+		"name": "U() removes # at end"
+	},
+
+	{
+		"args": ["/foo.html?#"],
+		"exp":  "/foo.html",
+		"name": "U() removes ?# at end"
+	},
+
+	{
+		"args": ["/foo.html#?"],
+		"exp":  "/foo.html#?",
+		"name": "U() keeps #? at end"
+	},
 
 	{
 		"args": ["/foo.html", {"bar": {"gna": 42, "qux": {"quux": "<Zürich>", "gnöp": "fasel"}}}],
@@ -7,12 +42,30 @@
 	},
 
 	{
+		"args": ["/foo.html?", {"foo" : "bar[qux][gnöp]=fasel"}],
+		"exp":  "/foo.html?foo=bar%5Bqux%5D%5Bgn%C3%B6p%5D%3Dfasel",
+		"name": "U() with array syntax in value and ? at end of base"
+	},
+
+	{
 		"args": ["/foo.html?a=b&bar=qux&c=d", {"bar": "baz"}],
 		"exp":  "/foo.html?a=b&bar=baz&c=d",
 		"name": "U() args override get params in base url"
 	},
 
 	{
+		"args": ["/foo.html?a=b&bar=qux&c=d", {"bar": "baz", "gna": "gnaber"}],
+		"exp":  "/foo.html?a=b&bar=baz&c=d&gna=gnaber",
+		"name": "U() args override get params in base url and appends remaining params"
+	},
+
+	{
+		"args": ["/foo.html?gna=g&a=b&bar=qux&c=d", {"bar": "baz", "gna": "gnaber"}],
+		"exp":  "/foo.html?gna=gnaber&a=b&bar=baz&c=d",
+		"name": "U() args override get params in base url keeping same order"
+	},
+
+	{
 		"args": ["/foo.html?bar.qux=a.b", {"c.d": "e.f", "g h": "i j"}],
 		"exp":  "/foo.html?bar.qux=a.b&c.d=e.f&g+h=i+j",
 		"name": "U() dots and spaces in arg names are preserved"
@@ -43,8 +96,50 @@
 	},
 
 	{
+		"args": ["path#%% %1%x %1x%x1%xx%11%ff%FF%0f%0F %"],
+		"exp":  "path#%25%25%20%251%25x%20%251x%25x1%25xx%11%ff%FF%0f%0F%20%25",
+		"name": "U() quoting of % if not followed by 2 hex digits in fragment"
+	},
+
+	{
 		"args": ["a\\b"],
 		"exp":  "a/b",
 		"name": "U() converting of \\ to /"
+	},
+
+	{
+		"args": ["path/#a\\b"],
+		"exp":  "path/#a%5Cb",
+		"name": "U() encode \\ in fragment"
+	},
+
+	{
+		"args": ["foo.html?bar=\\gna"],
+		"exp":  "foo.html?bar=%5Cgna",
+		"name": "U() encode \\ in url parameter"
+	},
+
+	{
+		"args": ["path/#?a=b"],
+		"exp":  "path/#?a=b",
+		"name": "U() keep query syntax in fragment"
+	},
+
+	{
+		"args": ["path/#Jet d'eau"],
+		"exp":  "path/#Jet%20d%27eau",
+		"name": "U() with single quotes in fragment"
+	},
+
+	{
+		"args": ["//gna.ch"],
+		"exp":  "//gna.ch/",
+		"name": "U() add / for absolute url without path"
+	},
+
+	{
+		"args": ["http://gna.ch"],
+		"exp":  "http://gna.ch/",
+		"name": "U() add / for absolute http url without path"
 	}
 ]