updated regex to properly handle more cases, extended tests and update pattern access

Signed-off-by: Yannick Alexander <yannick@alexanderdev.io>

pkg/registry/client.go

@@ -229,7 +229,7 @@ type (
 
 // hostRegex is a pretty naive regex for validating host urls. The goal of it is not to ultimately validate all possible valid hosts,
 // but to catch common user errors such as including a scheme or path in the host string.
-var hostRegex = regexp.MustCompile(`^(?P<scheme>[a-z]*:\/\/)?(?P<host>[a-zA-Z0-9-\.\:]+)(?P<path>\/.*)?$`)
+var hostRegex = regexp.MustCompile(`^(?P<scheme>[a-zA-Z][a-zA-Z0-9+\-.]*:\/\/)?(?P<host>[a-zA-Z0-9\-._:\[\]]+)(?P<path>\/.*)?$`)
 
 // validateHost checks that the host matches some required pre-checks e.g. does not contain a scheme or path.
 // While ORAS will also validate some of these things, the current errors are a bit opaque.
@@ -240,8 +240,8 @@ func validateHost(host string) error {
 		return fmt.Errorf("invalid host: %q", host)
 	}
 
-	scheme := matches[1]
-	path := matches[3]
+	scheme := matches[hostRegex.SubexpIndex("scheme")]
+	path := matches[hostRegex.SubexpIndex("path")]
 
 	if scheme != "" {
 		return fmt.Errorf("host should not contain a scheme (e.g. http://), found %q", scheme)

pkg/registry/client_test.go

@@ -169,6 +169,51 @@ func TestValidateHost(t *testing.T) {
 			host:    "oci://ghcr.io",
 			wantErr: true,
 		},
+		{
+			name:    "domain with uppercase scheme",
+			host:    "HTTPS://ghcr.io",
+			wantErr: true,
+		},
+		{
+			name:    "scheme with plus sign",
+			host:    "coap+tcp://ghcr.io",
+			wantErr: true,
+		},
+		{
+			name:    "scheme with dot and hyphen",
+			host:    "my.custom-scheme://ghcr.io",
+			wantErr: true,
+		},
+		{
+			name:    "IPv6 loopback with port",
+			host:    "[::1]:5000",
+			wantErr: false,
+		},
+		{
+			name:    "IPv6 full address with port",
+			host:    "[2001:db8::1]:443",
+			wantErr: false,
+		},
+		{
+			name:    "IPv4 address with port",
+			host:    "192.168.1.1:5000",
+			wantErr: false,
+		},
+		{
+			name:    "host with underscore",
+			host:    "my_registry.local",
+			wantErr: false,
+		},
+		{
+			name:    "scheme with path",
+			host:    "https://ghcr.io/myrepo",
+			wantErr: true,
+		},
+		{
+			name:    "empty string",
+			host:    "",
+			wantErr: true,
+		},
 	}
 
 	for _, tt := range tests {