Fix: reject malformed domain rules (#3917)

Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: Loyalsoldier <10487845+Loyalsoldier@users.noreply.github.com>

cmd/datdump/main.go

@@ -83,7 +83,7 @@ func exportSite(name string, gs *GeoSites) error {
 	}
 	defer file.Close()
 	w := bufio.NewWriter(file)
-	fmt.Fprintf(w, "%s:\n", name)
+	fmt.Fprintf(w, "%q:\n", name)
 	var b strings.Builder
 	b.Grow(64)
 	for _, vdomain := range vDomains {

main.go

@@ -66,6 +66,9 @@ const (
 	ModeAll       string = "all"
 	ModeAllowlist string = "allowlist"
 	ModeDenylist  string = "denylist"
+
+	maxDomainLen int = 253 // Maximum length of a domain name
+	maxLabelLen  int = 63  // Maximum length of a label of a domain name
 )
 
 func makeProtoList(listName string, entries []*Entry) *router.GeoSite {
@@ -208,10 +211,15 @@ func parseEntry(typ, rule string) (*Entry, []string, error) {
 			return entry, nil, fmt.Errorf("invalid regexp %q: %w", parts[0], err)
 		}
 		entry.Value = parts[0]
-	case dlc.RuleTypeDomain, dlc.RuleTypeFullDomain, dlc.RuleTypeKeyword:
+	case dlc.RuleTypeDomain, dlc.RuleTypeFullDomain:
+		entry.Value = strings.ToLower(parts[0])
+		if !validateDomainName(entry.Value) {
+			return entry, nil, fmt.Errorf("invalid domain: %q", entry.Value)
+		}
+	case dlc.RuleTypeKeyword:
 		entry.Value = strings.ToLower(parts[0])
 		if !validateDomainChars(entry.Value) {
-			return entry, nil, fmt.Errorf("invalid domain: %q", entry.Value)
+			return entry, nil, fmt.Errorf("invalid keyword: %q", entry.Value)
 		}
 	default:
 		return entry, nil, fmt.Errorf("unknown rule type: %q", entry.Type)
@@ -310,6 +318,21 @@ func validateDomainChars(domain string) bool {
 	return true
 }
 
+// validateDomainName reports whether the domain is a valid domain name, so that
+// typos like "example..com" or "-example.com" would not be silently built into
+// rules which can never match any domain.
+func validateDomainName(domain string) bool {
+	if !validateDomainChars(domain) || len(domain) > maxDomainLen {
+		return false
+	}
+	for label := range strings.SplitSeq(domain, ".") {
+		if label == "" || len(label) > maxLabelLen || label[0] == '-' || label[len(label)-1] == '-' {
+			return false
+		}
+	}
+	return true
+}
+
 func validateAttrChars(attr string) bool {
 	if attr == "" {
 		return false

main_test.go

@@ -22,10 +22,19 @@ func TestParseEntry(t *testing.T) {
 		{name: "duplicated attrs", typ: "domain", rule: "example.com @ads @ads", wantPlain: "domain:example.com:@ads"},
 		{name: "affiliations", typ: "domain", rule: "example.com &other @ads", wantPlain: "domain:example.com:@ads", wantAffs: []string{"OTHER"}},
 		{name: "regexp", typ: "regexp", rule: `^example\.com$`, wantPlain: `regexp:^example\.com$`},
+		{name: "tld", typ: "domain", rule: "google", wantPlain: "domain:google"},
+		{name: "hyphen inside", typ: "domain", rule: "my-example.com", wantPlain: "domain:my-example.com"},
+		{name: "keyword allows partial", typ: "keyword", rule: "-ads-", wantPlain: "keyword:-ads-"},
 		{name: "invalid regexp", typ: "regexp", rule: "^example(", wantErr: true},
 		{name: "empty rule", typ: "domain", rule: "   ", wantErr: true},
 		{name: "unknown type", typ: "prefix", rule: "example.com", wantErr: true},
 		{name: "invalid domain", typ: "domain", rule: "exa_mple.com @ads", wantErr: true},
+		{name: "empty label", typ: "domain", rule: "example..com", wantErr: true},
+		{name: "trailing dot", typ: "full", rule: "example.com.", wantErr: true},
+		{name: "leading dot", typ: "domain", rule: ".example.com", wantErr: true},
+		{name: "leading hyphen", typ: "domain", rule: "-example.com", wantErr: true},
+		{name: "trailing hyphen", typ: "full", rule: "example-.com", wantErr: true},
+		{name: "overlong label", typ: "domain", rule: strings.Repeat("a", 64) + ".com", wantErr: true},
 		{name: "empty attr", typ: "domain", rule: "example.com @", wantErr: true},
 		{name: "empty affiliation", typ: "domain", rule: "example.com &", wantErr: true},
 		{name: "unknown field", typ: "domain", rule: "example.com ads", wantErr: true},