fix: accept unicode cypher symbols

Author: zinic

query/v2/query_test.go

@@ -248,6 +248,22 @@ func TestInvalidScopeAliasesReturnBuildErrors(t *testing.T) {
 	require.ErrorContains(t, err, `scope alias node has invalid symbol "bad name"`)
 }
 
+func TestUnicodeCypherSymbols(t *testing.T) {
+	scope := v2.NewScope("路径", "节点", "起点", "关系", "终点")
+
+	preparedQuery, err := scope.New().Where(
+		scope.Node().Property("name").Equals(v2.NamedParameter("名字", "alice")),
+	).Return(
+		v2.As(scope.Node().ID(), "标识"),
+	).Build()
+	require.NoError(t, err)
+
+	require.Equal(t, "match (节点) where 节点.name = $名字 return id(节点) as 标识", renderPrepared(t, preparedQuery))
+	require.Equal(t, map[string]any{
+		"名字": "alice",
+	}, preparedQuery.Parameters)
+}
+
 func TestInvalidRelationshipDirectionReturnsError(t *testing.T) {
 	_, err := v2.New().WithRelationshipDirection(graph.Direction(99)).Return(v2.Relationship()).Build()
 	require.ErrorContains(t, err, "unsupported relationship direction: invalid")

query/v2/util.go

@@ -7,6 +7,8 @@ import (
 	"sort"
 	"strconv"
 	"strings"
+	"unicode"
+	"unicode/utf8"
 
 	"github.com/specterops/dawgs/cypher/models/cypher"
 	"github.com/specterops/dawgs/cypher/models/walk"
@@ -264,25 +266,29 @@ func sortedPropertyKeys(properties map[string]any) []string {
 	return keys
 }
 
-func isCypherSymbolStart(char byte) bool {
-	return char == '_' || (char >= 'A' && char <= 'Z') || (char >= 'a' && char <= 'z')
+func isCypherSymbolStart(char rune) bool {
+	return char == '_' || unicode.IsLetter(char) || unicode.In(char, unicode.Nl, unicode.Pc)
 }
 
-func isCypherSymbolPart(char byte) bool {
-	return isCypherSymbolStart(char) || (char >= '0' && char <= '9')
+func isCypherSymbolPart(char rune) bool {
+	return isCypherSymbolStart(char) || unicode.IsDigit(char) || unicode.In(char, unicode.Mark, unicode.Sc)
 }
 
 func validateCypherSymbol(symbol, context string) error {
 	if strings.TrimSpace(symbol) == "" {
 		return fmt.Errorf("%s is empty", context)
 	}
 
-	if !isCypherSymbolStart(symbol[0]) {
+	if !utf8.ValidString(symbol) {
 		return fmt.Errorf("%s has invalid symbol %q", context, symbol)
 	}
 
-	for idx := 1; idx < len(symbol); idx++ {
-		if !isCypherSymbolPart(symbol[idx]) {
+	for idx, char := range symbol {
+		if idx == 0 {
+			if !isCypherSymbolStart(char) {
+				return fmt.Errorf("%s has invalid symbol %q", context, symbol)
+			}
+		} else if !isCypherSymbolPart(char) {
 			return fmt.Errorf("%s has invalid symbol %q", context, symbol)
 		}
 	}