chore(origindetection): implement LocalData struct for DogStatsD

comp/core/tagger/impl/tagger.go

@@ -470,7 +470,7 @@ func (t *TaggerWrapper) EnrichTags(tb tagset.TagsAccumulator, originInfo taggert
 		// Accumulate tags for pod UID
 		if originInfo.ExternalData.PodUID != "" {
 			if err := t.AccumulateTagsFor(types.NewEntityID(types.KubernetesPodUID, originInfo.ExternalData.PodUID), cardinality, tb); err != nil {
-				t.log.Tracef("Cannot get tags for entity %s: %s", originInfo.ContainerID, err)
+				t.log.Tracef("Cannot get tags for entity %s: %s", originInfo.ExternalData.PodUID, err)
 			}
 		}
 

comp/core/tagger/origindetection/origindetection.go

@@ -26,6 +26,14 @@ const (
 	// ProductOriginAPM is the ProductOrigin for APM.
 	ProductOriginAPM ProductOrigin = iota
 
+	// Local Data Prefixes
+	// These prefixes are used to build the Local Data list.
+
+	// LocalDataContainerIDPrefix is the prefix used for the Container ID sent in the Local Data list.
+	LocalDataContainerIDPrefix = "ci-"
+	// LocalDataInodePrefix is the prefix used for the Inode sent in the Local Data list.
+	LocalDataInodePrefix = "in-"
+
 	// External Data Prefixes
 	// These prefixes are used to build the External Data Environment Variable.
 
@@ -63,13 +71,49 @@ type ExternalData struct {
 // GenerateContainerIDFromExternalData generates a container ID from the external data.
 type GenerateContainerIDFromExternalData func(externalData ExternalData) (string, error)
 
+// ParseLocalData parses the local data string into a LocalData struct.
+func ParseLocalData(rawLocalData string) (LocalData, error) {
+	if rawLocalData == "" {
+		return LocalData{}, nil
+	}
+
+	var localData LocalData
+	var parsingError error
+
+	if strings.Contains(rawLocalData, ",") {
+		// The Local Data can contain a list.
+		items := strings.Split(rawLocalData, ",")
+		for _, item := range items {
+			if strings.HasPrefix(item, LocalDataContainerIDPrefix) {
+				localData.ContainerID = item[len(LocalDataContainerIDPrefix):]
+			} else if strings.HasPrefix(item, LocalDataInodePrefix) {
+				localData.Inode, parsingError = strconv.ParseUint(item[len(LocalDataInodePrefix):], 10, 64)
+			}
+		}
+	} else {
+		// The Local Data can contain a single value.
+		if strings.HasPrefix(rawLocalData, LocalDataContainerIDPrefix) {
+			localData.ContainerID = rawLocalData[len(LocalDataContainerIDPrefix):]
+		} else if strings.HasPrefix(rawLocalData, LocalDataInodePrefix) {
+			localData.Inode, parsingError = strconv.ParseUint(rawLocalData[len(LocalDataInodePrefix):], 10, 64)
+		} else {
+			// Container ID with old format: <container-id>
+			localData.ContainerID = rawLocalData
+		}
+	}
+
+	return localData, parsingError
+}
+
 // ParseExternalData parses the external data string into an ExternalData struct.
 func ParseExternalData(externalEnv string) (ExternalData, error) {
 	if externalEnv == "" {
 		return ExternalData{}, nil
 	}
+
 	var externalData ExternalData
 	var parsingError error
+
 	for _, item := range strings.Split(externalEnv, ",") {
 		switch {
 		case strings.HasPrefix(item, ExternalDataInitPrefix):
@@ -80,5 +124,6 @@ func ParseExternalData(externalEnv string) (ExternalData, error) {
 			externalData.PodUID = item[len(ExternalDataPodUIDPrefix):]
 		}
 	}
+
 	return externalData, parsingError
 }

comp/core/tagger/origindetection/origindetection_test.go

@@ -12,6 +12,65 @@ import (
 	"github.com/stretchr/testify/assert"
 )
 
+func TestParseLocalData(t *testing.T) {
+	tests := []struct {
+		name         string
+		rawLocalData string
+		expected     LocalData
+		expectError  bool
+	}{
+		{
+			name:         "Empty string",
+			rawLocalData: "",
+			expected:     LocalData{},
+			expectError:  false,
+		},
+		{
+			name:         "Single container ID",
+			rawLocalData: "ci-abc123",
+			expected:     LocalData{ContainerID: "abc123"},
+			expectError:  false,
+		},
+		{
+			name:         "Single inode",
+			rawLocalData: "in-12345",
+			expected:     LocalData{Inode: 12345},
+			expectError:  false,
+		},
+		{
+			name:         "Multiple values",
+			rawLocalData: "ci-abc123,in-12345",
+			expected:     LocalData{ContainerID: "abc123", Inode: 12345},
+			expectError:  false,
+		},
+		{
+			name:         "Invalid inode",
+			rawLocalData: "in-invalid",
+			expected:     LocalData{},
+			expectError:  true,
+		},
+		{
+			name:         "Old container format",
+			rawLocalData: "abc123",
+			expected:     LocalData{ContainerID: "abc123"},
+			expectError:  false,
+		},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			result, err := ParseLocalData(tc.rawLocalData)
+
+			if tc.expectError {
+				assert.Error(t, err)
+			} else {
+				assert.Equal(t, tc.expected, result)
+			}
+
+		})
+	}
+}
+
 func TestParseExternalData(t *testing.T) {
 	tests := []struct {
 		name          string

comp/dogstatsd/server/enrich.go

@@ -37,12 +37,13 @@ type enrichConfig struct {
 }
 
 // extractTagsMetadata returns tags (client tags + host tag) and information needed to query tagger (origins, cardinality).
-func extractTagsMetadata(tags []string, originFromUDS string, originFromMsg []byte, externalData origindetection.ExternalData, conf enrichConfig) ([]string, string, taggertypes.OriginInfo, metrics.MetricSource) {
+func extractTagsMetadata(tags []string, originFromUDS string, localData origindetection.LocalData, externalData origindetection.ExternalData, conf enrichConfig) ([]string, string, taggertypes.OriginInfo, metrics.MetricSource) {
 	host := conf.defaultHostname
 	metricSource := metrics.MetricSourceDogstatsd
 	origin := taggertypes.OriginInfo{
 		ContainerIDFromSocket: originFromUDS,
-		ContainerID:           string(originFromMsg),
+		ContainerID:           localData.ContainerID,
+		LocalData:             localData,
 		ExternalData:          externalData,
 		ProductOrigin:         origindetection.ProductOriginDogStatsD,
 	}
@@ -112,7 +113,7 @@ func tsToFloatForSamples(ts time.Time) float64 {
 
 func enrichMetricSample(dest []metrics.MetricSample, ddSample dogstatsdMetricSample, origin string, listenerID string, conf enrichConfig) []metrics.MetricSample {
 	metricName := ddSample.name
-	tags, hostnameFromTags, extractedOrigin, metricSource := extractTagsMetadata(ddSample.tags, origin, ddSample.containerID, ddSample.externalData, conf)
+	tags, hostnameFromTags, extractedOrigin, metricSource := extractTagsMetadata(ddSample.tags, origin, ddSample.localData, ddSample.externalData, conf)
 
 	if !isExcluded(metricName, conf.metricPrefix, conf.metricPrefixBlacklist) {
 		metricName = conf.metricPrefix + metricName
@@ -192,7 +193,7 @@ func enrichEventAlertType(dogstatsdAlertType alertType) metricsevent.AlertType {
 }
 
 func enrichEvent(event dogstatsdEvent, origin string, conf enrichConfig) *metricsevent.Event {
-	tags, hostnameFromTags, extractedOrigin, _ := extractTagsMetadata(event.tags, origin, event.containerID, event.externalData, conf)
+	tags, hostnameFromTags, extractedOrigin, _ := extractTagsMetadata(event.tags, origin, event.localData, event.externalData, conf)
 
 	enrichedEvent := &metricsevent.Event{
 		Title:          event.title,
@@ -229,7 +230,7 @@ func enrichServiceCheckStatus(status serviceCheckStatus) servicecheck.ServiceChe
 }
 
 func enrichServiceCheck(serviceCheck dogstatsdServiceCheck, origin string, conf enrichConfig) *servicecheck.ServiceCheck {
-	tags, hostnameFromTags, extractedOrigin, _ := extractTagsMetadata(serviceCheck.tags, origin, serviceCheck.containerID, serviceCheck.externalData, conf)
+	tags, hostnameFromTags, extractedOrigin, _ := extractTagsMetadata(serviceCheck.tags, origin, serviceCheck.localData, serviceCheck.externalData, conf)
 
 	enrichedServiceCheck := &servicecheck.ServiceCheck{
 		CheckName:  serviceCheck.name,

comp/dogstatsd/server/enrich_bench_test.go

@@ -35,7 +35,7 @@ func BenchmarkExtractTagsMetadata(b *testing.B) {
 			sb.ResetTimer()
 
 			for n := 0; n < sb.N; n++ {
-				tags, _, _, _ = extractTagsMetadata(baseTags, "", []byte{}, origindetection.ExternalData{}, conf)
+				tags, _, _, _ = extractTagsMetadata(baseTags, "", origindetection.LocalData{}, origindetection.ExternalData{}, conf)
 			}
 		})
 	}