fix: skipping label if it contains special symbol (#14068)

Co-authored-by: Vladyslav Diachenko <82767850+vlad-diachenko@users.noreply.github.com>

pkg/logql/log/parser.go

@@ -4,6 +4,7 @@ import (
 	"bytes"
 	"errors"
 	"fmt"
+	"strings"
 	"unicode/utf8"
 
 	"github.com/grafana/jsonparser"
@@ -39,6 +40,14 @@ var (
 	errMissingCapture       = errors.New("at least one named capture must be supplied")
 	errFoundAllLabels       = errors.New("found all required labels")
 	errLabelDoesNotMatch    = errors.New("found a label with a matcher that didn't match")
+
+	// the rune error replacement is rejected by Prometheus hence replacing them with space.
+	removeInvalidUtf = func(r rune) rune {
+		if r == utf8.RuneError {
+			return 32 // rune value for space
+		}
+		return r
+	}
 )
 
 type JSONParser struct {
@@ -200,12 +209,11 @@ func unescapeJSONString(b []byte) string {
 		return ""
 	}
 	res := string(bU)
-	// rune error is rejected by Prometheus
-	for _, r := range res {
-		if r == utf8.RuneError {
-			return ""
-		}
+
+	if strings.ContainsRune(res, utf8.RuneError) {
+		res = strings.Map(removeInvalidUtf, res)
 	}
+
 	return res
 }
 
@@ -339,9 +347,9 @@ func (l *LogfmtParser) Process(_ int64, line []byte, lbs *LabelsBuilder) ([]byte
 		}
 
 		val := l.dec.Value()
-		// the rune error replacement is rejected by Prometheus, so we skip it.
+
 		if bytes.ContainsRune(val, utf8.RuneError) {
-			val = nil
+			val = bytes.Map(removeInvalidUtf, val)
 		}
 
 		if !l.keepEmpty && len(val) == 0 {

pkg/logql/log/parser_test.go

@@ -83,7 +83,7 @@ func Test_jsonParser_Parse(t *testing.T) {
 			labels.EmptyLabels(),
 			labels.FromStrings("counter", "1",
 				"price__net_", "5.56909",
-				"foo", "",
+				"foo", " ",
 			),
 			NoParserHints(),
 		},
@@ -802,7 +802,7 @@ func TestLogfmtParser_parse(t *testing.T) {
 			"utf8 error rune",
 			[]byte(`buzz=foo bar=�f`),
 			labels.EmptyLabels(),
-			labels.FromStrings("buzz", "foo"),
+			labels.FromStrings("bar", " f", "buzz", "foo"),
 			nil,
 			NoParserHints(),
 		},
@@ -1037,15 +1037,15 @@ func TestLogfmtParser_keepEmpty(t *testing.T) {
 			false,
 			labels.FromStrings("foo", "bar"),
 			labels.FromStrings("foo", "bar",
-				"bar", "buzz"),
+				"bar", "buzz", "foo_extracted", "b r"),
 		},
 		{
 			"utf8 error rune with keep empty",
 			[]byte("foo=b�r bar=buzz"),
 			true,
 			labels.FromStrings("foo", "bar"),
 			labels.FromStrings("foo", "bar",
-				"foo_extracted", "",
+				"foo_extracted", "b r",
 				"bar", "buzz"),
 		},
 	}