From d0ccb60ce955f674208db5e29b03c9f80fd03eed Mon Sep 17 00:00:00 2001
From: Eric Yap <eric.yap@elastic.co>
Date: Mon, 27 Jan 2025 16:05:20 +0800
Subject: [PATCH] fix: Truncate string slice attributes

---
 input/otlp/metadata.go      |   2 +-
 input/otlp/metadata_test.go | 117 ++++++++++++++++++++++--------------
 2 files changed, 72 insertions(+), 47 deletions(-)

diff --git a/input/otlp/metadata.go b/input/otlp/metadata.go
index 2a1e854c..279356d4 100644
--- a/input/otlp/metadata.go
+++ b/input/otlp/metadata.go
@@ -516,7 +516,7 @@ func setLabel(key string, event *modelpb.APMEvent, v pcommon.Value) {
 			for i := 0; i < s.Len(); i++ {
 				r := s.At(i)
 				if r.Type() == pcommon.ValueTypeStr {
-					result = append(result, r.Str())
+					result = append(result, truncate(r.Str()))
 				}
 			}
 			modelpb.Labels(event.Labels).SetSlice(key, result)
diff --git a/input/otlp/metadata_test.go b/input/otlp/metadata_test.go
index db5a48cf..e8bb364e 100644
--- a/input/otlp/metadata_test.go
+++ b/input/otlp/metadata_test.go
@@ -18,6 +18,7 @@
 package otlp_test
 
 import (
+	"strings"
 	"testing"
 
 	"github.com/google/go-cmp/cmp"
@@ -350,53 +351,77 @@ func TestResourceConventions(t *testing.T) {
 // This test ensures that the values are properly translated,
 // and that heterogeneous array elements are dropped without a panic.
 func TestResourceLabels(t *testing.T) {
-	metadata := transformResourceMetadata(t, map[string]interface{}{
-		"string_value": "abc",
-		"bool_value":   true,
-		"int_value":    123,
-		"float_value":  1.23,
-		"string_array": []interface{}{"abc", "def", true, 123, 1.23, nil},
-		"bool_array":   []interface{}{true, false, "true", 123, 1.23, nil},
-		"int_array":    []interface{}{123, 456, "abc", true, 1.23, nil},
-		"float_array":  []interface{}{1.23, 4.56, "abc", true, 123, nil},
-		"empty_array":  []interface{}{}, // Ensure that an empty array is ignored.
+	t.Run("drop invalid", func(t *testing.T) {
+		metadata := transformResourceMetadata(t, map[string]interface{}{
+			"string_value": "abc",
+			"bool_value":   true,
+			"int_value":    123,
+			"float_value":  1.23,
+			"string_array": []interface{}{"abc", "def", true, 123, 1.23, nil},
+			"bool_array":   []interface{}{true, false, "true", 123, 1.23, nil},
+			"int_array":    []interface{}{123, 456, "abc", true, 1.23, nil},
+			"float_array":  []interface{}{1.23, 4.56, "abc", true, 123, nil},
+			"empty_array":  []interface{}{}, // Ensure that an empty array is ignored.
+		})
+		assert.Equal(t, modelpb.Labels{
+			"string_value": {
+				Global: true,
+				Value:  "abc",
+			},
+			"bool_value": {
+				Global: true,
+				Value:  "true",
+			},
+			"string_array": {
+				Global: true,
+				Values: []string{"abc", "def"},
+			},
+			"bool_array": {
+				Global: true,
+				Values: []string{"true", "false"},
+			},
+		}, modelpb.Labels(metadata.Labels))
+		assert.Equal(t, modelpb.NumericLabels{
+			"int_value": {
+				Global: true,
+				Value:  123,
+			},
+			"float_value": {
+				Global: true,
+				Value:  1.23,
+			},
+			"int_array": {
+				Global: true,
+				Values: []float64{123, 456},
+			},
+			"float_array": {
+				Global: true,
+				Values: []float64{1.23, 4.56},
+			},
+		}, modelpb.NumericLabels(metadata.NumericLabels))
+	})
+
+	t.Run("string truncate", func(t *testing.T) {
+		metadata := transformResourceMetadata(t, map[string]interface{}{
+			"string_value": strings.Repeat("a", 2000),
+			"string_array": []interface{}{strings.Repeat("a", 2000)},
+		})
+
+		assert.Equal(t,
+			// Expect to truncate to keywordLength
+			modelpb.Labels{
+				"string_value": {
+					Global: true,
+					Value:  strings.Repeat("a", 1024),
+				},
+				"string_array": {
+					Global: true,
+					Values: []string{strings.Repeat("a", 1024)},
+				},
+			},
+			modelpb.Labels(metadata.Labels),
+		)
 	})
-	assert.Equal(t, modelpb.Labels{
-		"string_value": {
-			Global: true,
-			Value:  "abc",
-		},
-		"bool_value": {
-			Global: true,
-			Value:  "true",
-		},
-		"string_array": {
-			Global: true,
-			Values: []string{"abc", "def"},
-		},
-		"bool_array": {
-			Global: true,
-			Values: []string{"true", "false"},
-		},
-	}, modelpb.Labels(metadata.Labels))
-	assert.Equal(t, modelpb.NumericLabels{
-		"int_value": {
-			Global: true,
-			Value:  123,
-		},
-		"float_value": {
-			Global: true,
-			Value:  1.23,
-		},
-		"int_array": {
-			Global: true,
-			Values: []float64{123, 456},
-		},
-		"float_array": {
-			Global: true,
-			Values: []float64{1.23, 4.56},
-		},
-	}, modelpb.NumericLabels(metadata.NumericLabels))
 }
 
 func transformResourceMetadata(t *testing.T, resourceAttrs map[string]interface{}) *modelpb.APMEvent {