Merge branch 'main' into fix-multi-target-missing-metrics

Author: deniszh

expr/expr_test.go

@@ -172,6 +172,20 @@ func TestEvalExpr(t *testing.T) {
 			from:     1437127020,
 			until:    1437127140,
 		},
+		"metric unicode name inside a function": {
+			metric:  "1Метрика",
+			request: "summarize(1Метрика,'1min')",
+			metricRequest: parser.MetricRequest{
+				Metric: "1Метрика",
+				From:   1437127020,
+				Until:  1437127140,
+			},
+			values:   []float64{343, 407, 385},
+			isAbsent: []bool{false, false, false},
+			stepTime: 60,
+			from:     1437127020,
+			until:    1437127140,
+		},
 	}
 
 	parser.RangeTables = append(parser.RangeTables, unicode.Cyrillic)

pkg/parser/parser.go

@@ -847,12 +847,16 @@ var disallowedCharactersInMetricName = map[rune]struct{}{
 	'\'': struct{}{},
 	' ':  struct{}{},
 	'/':  struct{}{},
+	'|':  struct{}{},
 }
 
 func unicodeRuneAllowedInName(r rune) bool {
 	if _, ok := disallowedCharactersInMetricName[r]; ok {
 		return false
 	}
+	if unicode.IsSpace(r) {
+		return false
+	}
 
 	return true
 }
@@ -920,6 +924,10 @@ FOR:
 				if len(s) < i+2 || s[i+1] == '=' || s[i+1] == ',' || s[i+1] == ')' {
 					continue
 				}
+				// we are here: `key=value`
+				//                  ^
+				// even if '=' is a valid character for metric name, it's a separator for a named arg or tag
+				break FOR
 			}
 			fallthrough
 		/* */
@@ -929,6 +937,10 @@ FOR:
 		if isDefault {
 			r, w = utf8.DecodeRuneInString(s[i:])
 			if unicodeRuneAllowedInName(r) && unicode.In(r, RangeTables...) {
+				// this may be a multi-byte rune, but we only wrote the first byte
+				if w > 1 {
+					buf.WriteString(s[i+1 : i+w])
+				}
 				continue
 			}
 			if !isEscape {

pkg/parser/parser_test.go

@@ -2,6 +2,7 @@ package parser
 
 import (
 	"testing"
+	"unicode"
 
 	"github.com/stretchr/testify/assert"
 )
@@ -39,13 +40,24 @@ func TestSkipWhitespace(t *testing.T) {
 }
 
 func TestParseExpr(t *testing.T) {
+	// Widen RangeTables to every script so non-ASCII names are allowed.
+	// The grammar must still recognize '|', '\n', '=', etc. as special characters depending on context.
+	orig := RangeTables
+	defer func() { RangeTables = orig }()
+	for _, t := range unicode.Scripts {
+		RangeTables = append(RangeTables, t)
+	}
+
 	tests := []struct {
 		s string
 		e *expr
 	}{
 		{"metric=",
 			&expr{target: "metric="},
 		},
+		{"métric.ñame",
+			&expr{target: "métric.ñame"},
+		},
 		{"metric",
 			&expr{target: "metric"},
 		},
@@ -181,6 +193,18 @@ func TestParseExpr(t *testing.T) {
 				argString: "metric, key='value'",
 			},
 		},
+		{
+			"func(metric, true\n)",
+			&expr{
+				target: "func",
+				etype:  EtFunc,
+				args: []*expr{
+					{target: "metric"},
+					{etype: EtBool, target: "true", valStr: "true"},
+				},
+				argString: "metric, true\n",
+			},
+		},
 		{
 			"func(metric, key=true)",
 			&expr{