refactor: use url.Parse instead of regex

operator/pkg/utils.go

@@ -1,10 +1,9 @@
 package operator
 
 import (
-	"fmt"
 	"github.com/yetanotherco/aligned_layer/common"
 	"math/big"
-	"regexp"
+	"net/url"
 )
 
 func IsVerifierDisabled(disabledVerifiersBitmap *big.Int, verifierId common.ProvingSystemId) bool {
@@ -17,18 +16,11 @@ func IsVerifierDisabled(disabledVerifiersBitmap *big.Int, verifierId common.Prov
 }
 
 func BaseUrlOnly(input string) (string, error) {
-	// Define a regex pattern to match the URL format
-	// The pattern captures the scheme, host, and path
-	pattern := `^(?P<scheme>[^:]+)://(?P<host>[^/]+)(?P<path>/.*)?$`
-	re := regexp.MustCompile(pattern)
-
-	matches := re.FindStringSubmatch(input)
-
-	if matches == nil {
-		return "", fmt.Errorf("invalid URL: %s", input)
+	// https://gobyexample.com/url-parsing
+	u, err := url.Parse(input)
+	if err != nil {
+		return "", err
 	}
 
-	host := matches[2]
-
-	return host, nil
+	return u.Host, nil
 }

operator/pkg/utils_test.go

@@ -64,6 +64,13 @@ func TestBaseUrlOnlyHappyPath(t *testing.T) {
 		{"https://a.b.c.d/1234", "a.b.c.d"},
 		{"https://a.b.c.d/1234/5678", "a.b.c.d"},
 		{"https://a.b.c.d.e/1234/", "a.b.c.d.e"},
+		{"https://a.b.c.d?e=1234", "a.b.c.d"},
+		{"https://a.b.c.d/rpc?e=1234", "a.b.c.d"},
+		{"wss://a.b.c.d/1234", "a.b.c.d"},
+		{"wss://a.b.c.d/1234/5678", "a.b.c.d"},
+		{"wss://a.b.c.d.e/1234/", "a.b.c.d.e"},
+		{"wss://a.b.c.d?e=1234", "a.b.c.d"},
+		{"wss://a.b.c.d/rpc?e=1234", "a.b.c.d"},
 	}
 
 	for _, pair := range urls {
@@ -81,19 +88,3 @@ func TestBaseUrlOnlyHappyPath(t *testing.T) {
 		}
 	}
 }
-
-func TestBaseUrlOnlyFailureCases(t *testing.T) {
-
-	urls := [...]string{
-		"localhost:8545/asdfoij2a7831has89%342jddav98j2748",
-		"this-is-all-wrong",
-	}
-
-	for _, url := range urls {
-		baseUrl, err := BaseUrlOnly(url)
-
-		if err == nil {
-			t.Errorf("An error was expected, but received %s", baseUrl)
-		}
-	}
-}