Refactor parseEntry

- add value/attribute checker(check missing space)
- allow multiple spaces
- sort attributes
- improve readablity

main.go

@@ -29,6 +29,12 @@ const (
 	RuleTypeInclude    string = "include"
 )
 
+var (
+	TypeChecker  = regexp.MustCompile(`^(domain|full|keyword|regexp|include)$`)
+	ValueChecker = regexp.MustCompile(`^[a-z0-9!\.-]+$`)
+	AttrChecker  = regexp.MustCompile(`^[a-z0-9!-]+$`)
+)
+
 type Entry struct {
 	Type  string
 	Value string
@@ -79,11 +85,6 @@ func (l *ParsedList) toProto() (*router.GeoSite, error) {
 			})
 
 		case RuleTypeRegexp:
-			// check regexp validity to avoid runtime error
-			_, err := regexp.Compile(entry.Value)
-			if err != nil {
-				return nil, fmt.Errorf("invalid regexp in list %s: %s", l.Name, entry.Value)
-			}
 			site.Domain = append(site.Domain, &router.Domain{
 				Type:      router.Domain_Regex,
 				Value:     entry.Value,
@@ -103,9 +104,6 @@ func (l *ParsedList) toProto() (*router.GeoSite, error) {
 				Value:     entry.Value,
 				Attribute: entry.Attrs,
 			})
-
-		default:
-			return nil, fmt.Errorf("unknown domain type: %s", entry.Type)
 		}
 	}
 	return site, nil
@@ -123,60 +121,56 @@ func exportPlainTextList(list []string, refName string, pl *ParsedList) {
 	}
 }
 
-func parseDomain(domain string, entry *Entry) error {
-	kv := strings.Split(domain, ":")
+func parseEntry(line string) (Entry, error) {
+	var entry Entry
+	parts := strings.Fields(line)
+
+	// Parse/Check type and value
+	rawTypeVal := parts[0]
+	kv := strings.Split(rawTypeVal, ":")
 	if len(kv) == 1 {
-		entry.Type = RuleTypeDomain
-		entry.Value = strings.ToLower(kv[0])
-		return nil
-	}
-
-	if len(kv) == 2 {
+		entry.Type = RuleTypeDomain // Default type
+		entry.Value = strings.ToLower(rawTypeVal)
+	} else if len(kv) == 2 {
 		entry.Type = strings.ToLower(kv[0])
-
-		if strings.EqualFold(entry.Type, RuleTypeRegexp) {
+		if entry.Type == RuleTypeRegexp {
 			entry.Value = kv[1]
 		} else {
 			entry.Value = strings.ToLower(kv[1])
 		}
-
-		return nil
+	} else {
+		return entry, fmt.Errorf("invalid format: %s", line)
 	}
-
-	return fmt.Errorf("invalid format: %s", domain)
-}
-
-func parseAttribute(attr string) (*router.Domain_Attribute, error) {
-	var attribute router.Domain_Attribute
-	if len(attr) == 0 || attr[0] != '@' {
-		return &attribute, fmt.Errorf("invalid attribute: %s", attr)
+	if !TypeChecker.MatchString(entry.Type) {
+		return entry, fmt.Errorf("invalid type: %s", entry.Type)
 	}
-
-	attribute.Key = strings.ToLower(attr[1:]) // Trim attribute prefix `@` character
-	attribute.TypedValue = &router.Domain_Attribute_BoolValue{BoolValue: true}
-	return &attribute, nil
-}
-
-func parseEntry(line string) (Entry, error) {
-	parts := strings.Split(line, " ")
-
-	var entry Entry
-	if len(parts) == 0 {
-		return entry, fmt.Errorf("empty entry")
-	}
-
-	if err := parseDomain(parts[0], &entry); err != nil {
-		return entry, err
-	}
-
-	for i := 1; i < len(parts); i++ {
-		attr, err := parseAttribute(parts[i])
-		if err != nil {
-			return entry, err
+	if entry.Type == RuleTypeRegexp {
+		if _, err := regexp.Compile(entry.Value); err != nil {
+			return entry, fmt.Errorf("invalid regexp: %s", entry.Value)
 		}
-		entry.Attrs = append(entry.Attrs, attr)
+	} else if !ValueChecker.MatchString(entry.Value) {
+		return entry, fmt.Errorf("invalid value: %s", entry.Value)
 	}
 
+	// Parse/Check attributes
+	for _, part := range parts[1:] {
+		if !strings.HasPrefix(part, "@") {
+			return entry, fmt.Errorf("invalid attribute: %s", part)
+		}
+		attrKey := strings.ToLower(part[1:]) // Trim attribute prefix `@` character
+		if !AttrChecker.MatchString(attrKey) {
+			return entry, fmt.Errorf("invalid attribute key: %s", attrKey)
+		}
+		entry.Attrs = append(entry.Attrs, &router.Domain_Attribute{
+			Key:        attrKey,
+			TypedValue: &router.Domain_Attribute_BoolValue{BoolValue: true},
+		})
+	}
+	// Sort attributes
+	sort.Slice(entry.Attrs, func(i, j int) bool {
+		return entry.Attrs[i].Key < entry.Attrs[j].Key
+	})
+
 	return entry, nil
 }