From 281976feaacff6ec0e3524527118997ea63b536a Mon Sep 17 00:00:00 2001
From: Ravishankar <ravishankar.gnanaprakasam@gmail.com>
Date: Mon, 9 Sep 2024 19:23:17 +0530
Subject: [PATCH] Check contains as per review comments

---
 pkg/logql/log/parser.go      | 26 ++++++++++++--------------
 pkg/logql/log/parser_test.go | 10 +++++-----
 2 files changed, 17 insertions(+), 19 deletions(-)

diff --git a/pkg/logql/log/parser.go b/pkg/logql/log/parser.go
index 19550778c2e4c..c8e65061ba41d 100644
--- a/pkg/logql/log/parser.go
+++ b/pkg/logql/log/parser.go
@@ -40,6 +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 {
@@ -202,14 +210,9 @@ func unescapeJSONString(b []byte) string {
 	}
 	res := string(bU)
 
-	// rune error is rejected by Prometheus hence replacing them with space
-	removeInvalidUtf := func(r rune) rune {
-		if r == utf8.RuneError {
-			return -1
-		}
-		return r
+	if strings.ContainsRune(res, utf8.RuneError) {
+		res = strings.Map(removeInvalidUtf, res)
 	}
-	res = strings.Map(removeInvalidUtf, res)
 
 	return res
 }
@@ -345,14 +348,9 @@ func (l *LogfmtParser) Process(_ int64, line []byte, lbs *LabelsBuilder) ([]byte
 
 		val := l.dec.Value()
 
-		// the rune error replacement is rejected by Prometheus hence skiping them.
-		removeInvalidUtf := func(r rune) rune {
-			if r == utf8.RuneError {
-				return -1
-			}
-			return r
+		if bytes.ContainsRune(val, utf8.RuneError) {
+			val = bytes.Map(removeInvalidUtf, val)
 		}
-		val = bytes.Map(removeInvalidUtf, val)
 
 		if !l.keepEmpty && len(val) == 0 {
 			continue
diff --git a/pkg/logql/log/parser_test.go b/pkg/logql/log/parser_test.go
index b0f61564c56cc..28989b7cb5fe8 100644
--- a/pkg/logql/log/parser_test.go
+++ b/pkg/logql/log/parser_test.go
@@ -79,11 +79,11 @@ func Test_jsonParser_Parse(t *testing.T) {
 		},
 		{
 			"utf8 error rune",
-			[]byte(`{"counter":1,"foo":"�f", "price": {"_net_":5.56909}}`),
+			[]byte(`{"counter":1,"foo":"�", "price": {"_net_":5.56909}}`),
 			labels.EmptyLabels(),
 			labels.FromStrings("counter", "1",
 				"price__net_", "5.56909",
-				"foo", "f",
+				"foo", " ",
 			),
 			NoParserHints(),
 		},
@@ -802,7 +802,7 @@ func TestLogfmtParser_parse(t *testing.T) {
 			"utf8 error rune",
 			[]byte(`buzz=foo bar=�f`),
 			labels.EmptyLabels(),
-			labels.FromStrings("bar", "f", "buzz", "foo"),
+			labels.FromStrings("bar", " f", "buzz", "foo"),
 			nil,
 			NoParserHints(),
 		},
@@ -1037,7 +1037,7 @@ func TestLogfmtParser_keepEmpty(t *testing.T) {
 			false,
 			labels.FromStrings("foo", "bar"),
 			labels.FromStrings("foo", "bar",
-				"bar", "buzz", "foo_extracted", "br"),
+				"bar", "buzz", "foo_extracted", "b r"),
 		},
 		{
 			"utf8 error rune with keep empty",
@@ -1045,7 +1045,7 @@ func TestLogfmtParser_keepEmpty(t *testing.T) {
 			true,
 			labels.FromStrings("foo", "bar"),
 			labels.FromStrings("foo", "bar",
-				"foo_extracted", "br",
+				"foo_extracted", "b r",
 				"bar", "buzz"),
 		},
 	}