Add regex matching to resource attribute filters (#535)

exporter/collector/config.go

@@ -18,6 +18,7 @@ import (
 	"context"
 	"errors"
 	"fmt"
+	"regexp"
 	"strings"
 	"time"
 
@@ -138,6 +139,8 @@ type ImpersonateConfig struct {
 type ResourceFilter struct {
 	// Match resource keys by prefix
 	Prefix string `mapstructure:"prefix"`
+	// Match resource keys by regex
+	Regex string `mapstructure:"regex"`
 }
 
 type LogConfig struct {
@@ -185,6 +188,15 @@ func ValidateConfig(cfg Config) error {
 		}
 		seenReplacements[mapping.Replacement] = struct{}{}
 	}
+
+	for _, resourceFilter := range cfg.MetricConfig.ResourceFilters {
+		if len(resourceFilter.Regex) == 0 {
+			continue
+		}
+		if _, err := regexp.Compile(resourceFilter.Regex); err != nil {
+			return fmt.Errorf("unable to parse resource filter regex: %s", err.Error())
+		}
+	}
 	return nil
 }
 

exporter/collector/config_test.go

@@ -66,6 +66,15 @@ func TestValidateConfig(t *testing.T) {
 			},
 			expectedErr: true,
 		},
+		{
+			desc: "Invalid resource filter regex",
+			input: Config{
+				MetricConfig: MetricConfig{
+					ResourceFilters: []ResourceFilter{{Regex: "*"}},
+				},
+			},
+			expectedErr: true,
+		},
 	} {
 		t.Run(tc.desc, func(t *testing.T) {
 			err := ValidateConfig(tc.input)

exporter/collector/logs.go

@@ -211,7 +211,7 @@ func (l logMapper) createEntries(ld plog.Logs) ([]*logpb.LogEntry, error) {
 	for i := 0; i < ld.ResourceLogs().Len(); i++ {
 		rl := ld.ResourceLogs().At(i)
 		mr := defaultResourceToMonitoredResource(rl.Resource())
-		extraResourceLabels := resourceToLabels(rl.Resource(), false, l.cfg.LogConfig.ResourceFilters)
+		extraResourceLabels := resourceToLabels(rl.Resource(), false, l.cfg.LogConfig.ResourceFilters, l.obs.log)
 		projectID := l.cfg.ProjectID
 		// override project ID with gcp.project.id, if present
 		if projectFromResource, found := rl.Resource().Attributes().Get(resourcemapping.ProjectIDAttributeKey); found {

exporter/collector/metrics.go

@@ -174,7 +174,7 @@ func (me *MetricsExporter) PushMetrics(ctx context.Context, m pmetric.Metrics) e
 	for i := 0; i < rms.Len(); i++ {
 		rm := rms.At(i)
 		monitoredResource := me.cfg.MetricConfig.MapMonitoredResource(rm.Resource())
-		extraResourceLabels := resourceToLabels(rm.Resource(), me.cfg.MetricConfig.ServiceResourceLabels, me.cfg.MetricConfig.ResourceFilters)
+		extraResourceLabels := resourceToLabels(rm.Resource(), me.cfg.MetricConfig.ServiceResourceLabels, me.cfg.MetricConfig.ResourceFilters, me.obs.log)
 		projectID := me.cfg.ProjectID
 		// override project ID with gcp.project.id, if present
 		if projectFromResource, found := rm.Resource().Attributes().Get(resourcemapping.ProjectIDAttributeKey); found {

exporter/collector/monitoredresource.go

@@ -15,8 +15,11 @@
 package collector
 
 import (
+	"regexp"
 	"strings"
 
+	"go.uber.org/zap"
+
 	"go.opentelemetry.io/collector/pdata/pcommon"
 	semconv "go.opentelemetry.io/collector/semconv/v1.8.0"
 	monitoredrespb "google.golang.org/genproto/googleapis/api/monitoredres"
@@ -58,6 +61,7 @@ func resourceToLabels(
 	resource pcommon.Resource,
 	serviceResourceLabels bool,
 	resourceFilters []ResourceFilter,
+	log *zap.Logger,
 ) labels {
 	attrs := pcommon.NewMap()
 	resource.Attributes().Range(func(k string, v pcommon.Value) bool {
@@ -71,7 +75,7 @@ func resourceToLabels(
 		}
 		// Matches one of the resource filters
 		for _, resourceFilter := range resourceFilters {
-			if strings.HasPrefix(k, resourceFilter.Prefix) {
+			if match, _ := regexp.Match(resourceFilter.Regex, []byte(k)); strings.HasPrefix(k, resourceFilter.Prefix) && match {
 				v.CopyTo(attrs.PutEmpty(k))
 				return true
 			}

exporter/collector/monitoredresource_test.go

@@ -456,6 +456,27 @@ func TestResourceToMetricLabels(t *testing.T) {
 				"service_namespace": "myservicenamespace",
 			},
 		},
+		{
+			name: "Resource filter regex config + prefix",
+			updateMapper: func(mapper *metricMapper) {
+				mapper.cfg.MetricConfig.ResourceFilters = []ResourceFilter{
+					{Prefix: "cloud.", Regex: ".*availability.*"},
+					{Regex: "host.*"},
+				}
+			},
+			resourceLabels: map[string]string{
+				"cloud.platform":            "gcp_compute_engine",
+				"cloud.availability_zone":   "us-central1",
+				"cloud.availability_region": "us-central",
+				"host.id":                   "abc123",
+				"random.fake":               "woot",
+			},
+			expectExtraLabels: labels{
+				"cloud_availability_zone":   "us-central1",
+				"cloud_availability_region": "us-central",
+				"host_id":                   "abc123",
+			},
+		},
 	}
 
 	for _, test := range tests {
@@ -468,7 +489,7 @@ func TestResourceToMetricLabels(t *testing.T) {
 			for k, v := range test.resourceLabels {
 				r.Attributes().PutStr(k, v)
 			}
-			extraLabels := resourceToLabels(r, mapper.cfg.MetricConfig.ServiceResourceLabels, mapper.cfg.MetricConfig.ResourceFilters)
+			extraLabels := resourceToLabels(r, mapper.cfg.MetricConfig.ServiceResourceLabels, mapper.cfg.MetricConfig.ResourceFilters, nil)
 			assert.Equal(t, test.expectExtraLabels, extraLabels)
 		})
 	}
@@ -516,6 +537,6 @@ func TestResourceMetricsToMonitoredResourceUTF8(t *testing.T) {
 	}
 	mr := defaultResourceToMonitoredResource(r)
 	assert.Equal(t, expectMr, mr)
-	extraLabels := resourceToLabels(r, mapper.cfg.MetricConfig.ServiceResourceLabels, mapper.cfg.MetricConfig.ResourceFilters)
+	extraLabels := resourceToLabels(r, mapper.cfg.MetricConfig.ServiceResourceLabels, mapper.cfg.MetricConfig.ResourceFilters, nil)
 	assert.Equal(t, expectExtraLabels, extraLabels)
 }