[processor/elasticapmprocessor] set agentVersion, normalize dataset, service name values

internal/elasticattr/attributes.go

@@ -35,6 +35,7 @@ const (
 	CloudProjectName       = "cloud.project.name"
 	ContainerImageTag      = "container.image.tag"
 	DeviceManufacturer     = "device.manufacturer"
+	DataStreamType         = "data_stream.type"
 	DataStreamDataset      = "data_stream.dataset"
 	DataStreamNamespace    = "data_stream.namespace"
 	DestinationIP          = "destination.ip"

processor/elasticapmprocessor/README.md

@@ -24,6 +24,8 @@ The processor enriches traces, metrics, and logs with elastic specific requireme
 
 The processor configuration embeds all configuration options from [opentelemetry-lib enrichments config](https://github.com/elastic/opentelemetry-lib/tree/main/enrichments/config). All opentelemetry-lib enricher configuration fields are available and can be configured directly in the processor configuration.
 
+The resource enrichment config also includes `resource.default_deployment_environment.enabled`. This controls the fallback `deployment.environment=unset` behavior separately from `resource.deployment_environment.enabled`, which still handles copying `deployment.environment.name` to `deployment.environment`. The fallback is only applied for ECS mapping mode.
+
 ### Enable all enrichments
 
 ```yaml

processor/elasticapmprocessor/internal/ecs/ecs_translation.go

@@ -19,13 +19,12 @@ package ecs // import "github.com/elastic/opentelemetry-collector-components/pro
 
 import (
 	"strconv"
-	"strings"
 
+	"github.com/elastic/opentelemetry-collector-components/internal/elasticattr"
+	"github.com/elastic/opentelemetry-collector-components/processor/elasticapmprocessor/internal/sanitize"
 	"go.opentelemetry.io/collector/pdata/pcommon"
 	semconv26 "go.opentelemetry.io/otel/semconv/v1.26.0"
 	semconv "go.opentelemetry.io/otel/semconv/v1.27.0"
-
-	"github.com/elastic/opentelemetry-collector-components/internal/elasticattr"
 )
 
 // Supported ECS resource attributes
@@ -42,72 +41,21 @@ func TranslateResourceMetadata(resource pcommon.Resource) {
 	attributes := resource.Attributes()
 
 	attributes.Range(func(k string, v pcommon.Value) bool {
-		if isLabelAttribute(k) {
-			sanitized := sanitizeLabelAttributeKey(k)
+		if sanitize.IsLabelAttribute(k) {
+			sanitized := sanitize.HandleLabelAttributeKey(k)
 			if sanitized != k {
 				v.CopyTo(attributes.PutEmpty(sanitized))
 				attributes.Remove(k)
 			}
 		} else if !isSupportedAttribute(k) {
 			// Other attributes that are not supported by ECS are moved to labels with a "labels." prefix.
-			setLabelAttributeValue(attributes, sanitizeLabelKey(k), v)
+			setLabelAttributeValue(attributes, sanitize.HandleAttributeKey(k), v)
 			attributes.Remove(k)
 		}
 		return true
 	})
 }
 
-// sanitizeLabelAttributeKey sanitizes the key portion of a label attribute,
-// preserving the "labels." or "numeric_labels." prefix.
-func sanitizeLabelAttributeKey(attr string) string {
-	if strings.HasPrefix(attr, "labels.") {
-		return "labels." + sanitizeLabelKey(strings.TrimPrefix(attr, "labels."))
-	}
-	if strings.HasPrefix(attr, "numeric_labels.") {
-		return "numeric_labels." + sanitizeLabelKey(strings.TrimPrefix(attr, "numeric_labels."))
-	}
-	return attr
-}
-
-// sanitizeLabelKey sanitizes a label key, replacing the reserved characters
-// '.', '*' and '"' with '_'. This matches the apm-server behavior.
-// This matches the logic in the apm-data library here:
-// https://github.com/elastic/apm-data/blob/e3e170b/model/modeljson/labels.go.
-func sanitizeLabelKey(k string) string {
-	if strings.ContainsAny(k, ".*\"") {
-		return strings.Map(replaceReservedLabelKeyRune, k)
-	}
-	return k
-}
-
-func replaceReservedLabelKeyRune(r rune) rune {
-	switch r {
-	case '.', '*', '"':
-		return '_'
-	}
-	return r
-}
-
-// isLabelAttribute returns true if the resource attribute is already a prefixed label.
-// The elasticapmintake receiver moves labels and numeric_labels into attributes and
-// already prefixes those with "labels." and "numeric_labels." respectively and also does de-dotting.
-// So for those, we don't want to double prefix - we just leave them as is.
-func isLabelAttribute(attr string) bool {
-	return strings.HasPrefix(attr, "labels.") || strings.HasPrefix(attr, "numeric_labels.")
-}
-
-// truncate returns s truncated at n runes, and the number of runes in the resulting string (<= n).
-func truncate(s string) string {
-	var j int
-	for i := range s {
-		if j == keywordLength {
-			return s[:i]
-		}
-		j++
-	}
-	return s
-}
-
 // setLabelAttributeValue maps a value into labels.* / numeric_labels.*.
 // Elasticsearch label mappings only support flat scalar values and
 // homogeneous arrays thereof; Map, Bytes, and empty types cannot be
@@ -117,7 +65,7 @@ func truncate(s string) string {
 func setLabelAttributeValue(attributes pcommon.Map, key string, value pcommon.Value) {
 	switch value.Type() {
 	case pcommon.ValueTypeStr:
-		attributes.PutStr("labels."+key, truncate(value.Str()))
+		attributes.PutStr("labels."+key, sanitize.Truncate(value.Str(), keywordLength))
 	case pcommon.ValueTypeBool:
 		attributes.PutStr("labels."+key, strconv.FormatBool(value.Bool()))
 	case pcommon.ValueTypeInt:
@@ -135,7 +83,7 @@ func setLabelAttributeValue(attributes pcommon.Map, key string, value pcommon.Va
 			for i := 0; i < slice.Len(); i++ {
 				item := slice.At(i)
 				if item.Type() == pcommon.ValueTypeStr {
-					target.AppendEmpty().SetStr(truncate(item.Str()))
+					target.AppendEmpty().SetStr(sanitize.Truncate(item.Str(), keywordLength))
 				}
 			}
 		case pcommon.ValueTypeBool:

processor/elasticapmprocessor/internal/enrichments/config/config.go

@@ -30,12 +30,14 @@ type Config struct {
 
 // ResourceConfig configures the enrichment of resource attributes.
 type ResourceConfig struct {
-	AgentName             AttributeConfig `mapstructure:"agent_name"`
-	AgentVersion          AttributeConfig `mapstructure:"agent_version"`
-	OverrideHostName      AttributeConfig `mapstructure:"override_host_name"`
-	DeploymentEnvironment AttributeConfig `mapstructure:"deployment_environment"`
-	ServiceInstanceID     AttributeConfig `mapstructure:"service_instance_id"`
-	HostOSType            AttributeConfig `mapstructure:"host_os_type"`
+	AgentName                    AttributeConfig `mapstructure:"agent_name"`
+	AgentVersion                 AttributeConfig `mapstructure:"agent_version"`
+	OverrideHostName             AttributeConfig `mapstructure:"override_host_name"`
+	DeploymentEnvironment        AttributeConfig `mapstructure:"deployment_environment"`
+	DefaultDeploymentEnvironment AttributeConfig `mapstructure:"default_deployment_environment"`
+	ServiceInstanceID            AttributeConfig `mapstructure:"service_instance_id"`
+	ServiceName                  AttributeConfig `mapstructure:"service_name"`
+	HostOSType                   AttributeConfig `mapstructure:"host_os_type"`
 }
 
 // ScopeConfig configures the enrichment of scope attributes.

processor/elasticapmprocessor/internal/enrichments/config/config_test.go

@@ -45,9 +45,11 @@ func assertAttributeConfigDefaults(t *testing.T, cfg reflect.Value, expectDisabl
 
 	// Fields that are intentionally disabled by default
 	disabledByDefault := map[string]bool{
-		"ClearSpanID":   true,
-		"ClearSpanName": true,
-		"HostOSType":    true,
+		"ClearSpanID":                  true,
+		"ClearSpanName":                true,
+		"DefaultDeploymentEnvironment": true,
+		"ServiceName":                  true,
+		"HostOSType":                   true,
 	}
 
 	assertAttributeConfigDefaultsRecurse(t, cfg, disabled, disabledByDefault)

processor/elasticapmprocessor/internal/enrichments/resource.go

@@ -20,6 +20,7 @@ package enrichments // import "github.com/elastic/opentelemetry-collector-compon
 import (
 	"fmt"
 
+	"github.com/elastic/opentelemetry-collector-components/processor/elasticapmprocessor/internal/sanitize"
 	"go.opentelemetry.io/collector/pdata/pcommon"
 	semconv25 "go.opentelemetry.io/otel/semconv/v1.25.0"
 	semconv "go.opentelemetry.io/otel/semconv/v1.27.0"
@@ -49,6 +50,7 @@ type resourceEnrichmentContext struct {
 	deploymentEnvironmentName string
 
 	serviceInstanceID string
+	serviceName       string
 	containerID       string
 
 	osType string
@@ -78,6 +80,8 @@ func (s *resourceEnrichmentContext) Enrich(resource pcommon.Resource, cfg config
 			s.deploymentEnvironmentName = v.Str()
 		case string(semconv25.ServiceInstanceIDKey):
 			s.serviceInstanceID = v.Str()
+		case string(semconv.ServiceNameKey):
+			s.serviceName = v.Str()
 		case string(semconv.ContainerIDKey):
 			s.containerID = v.Str()
 		case string(semconv.OSTypeKey):
@@ -100,13 +104,21 @@ func (s *resourceEnrichmentContext) Enrich(resource pcommon.Resource, cfg config
 	if cfg.OverrideHostName.Enabled {
 		s.overrideHostNameWithK8sNodeName(resource)
 	}
+
 	if cfg.DeploymentEnvironment.Enabled {
 		s.setDeploymentEnvironment(resource)
 	}
 
+	if cfg.DefaultDeploymentEnvironment.Enabled {
+		s.setDefaultDeploymentEnvironment(resource)
+	}
+
 	if cfg.ServiceInstanceID.Enabled {
 		s.setServiceInstanceID(resource)
 	}
+	if cfg.ServiceName.Enabled {
+		s.sanitizeServiceName(resource)
+	}
 	if cfg.HostOSType.Enabled {
 		s.setHostOSType(resource)
 	}
@@ -117,13 +129,32 @@ func (s *resourceEnrichmentContext) Enrich(resource pcommon.Resource, cfg config
 // ES currently doesn't allow aliases with multiple targets, so if the new field name is used (SemConv v1.27+),
 // we duplicate the value and also send it with the old field name to make the alias work.
 func (s *resourceEnrichmentContext) setDeploymentEnvironment(resource pcommon.Resource) {
-	if s.deploymentEnvironmentName != "" && s.deploymentEnvironment == "" {
-		attribute.PutStr(
-			resource.Attributes(),
-			string(semconv25.DeploymentEnvironmentKey),
-			s.deploymentEnvironmentName,
-		)
+	if s.deploymentEnvironment != "" {
+		return
+	}
+	if s.deploymentEnvironmentName == "" {
+		return
+	}
+
+	attribute.PutStr(
+		resource.Attributes(),
+		string(semconv25.DeploymentEnvironmentKey),
+		s.deploymentEnvironmentName,
+	)
+}
+
+// setDefaultDeploymentEnvironment sets deployment.environment to "unset"
+// when neither deployment environment field is present, matching the
+// apm-data/MIS default only where explicitly configured.
+func (s *resourceEnrichmentContext) setDefaultDeploymentEnvironment(resource pcommon.Resource) {
+	if s.deploymentEnvironment != "" || s.deploymentEnvironmentName != "" {
+		return
 	}
+	attribute.PutStr(
+		resource.Attributes(),
+		string(semconv25.DeploymentEnvironmentKey),
+		"unset",
+	)
 }
 
 func (s *resourceEnrichmentContext) setAgentName(resource pcommon.Resource) {
@@ -225,3 +256,13 @@ func (s *resourceEnrichmentContext) setServiceInstanceID(resource pcommon.Resour
 	}
 	attribute.PutStr(resource.Attributes(), string(semconv25.ServiceInstanceIDKey), s.serviceInstanceID)
 }
+
+func (s *resourceEnrichmentContext) sanitizeServiceName(resource pcommon.Resource) {
+	if s.serviceName == "" {
+		return
+	}
+	cleaned := sanitize.CleanServiceName(s.serviceName)
+	if cleaned != s.serviceName {
+		resource.Attributes().PutStr(string(semconv.ServiceNameKey), cleaned)
+	}
+}