DogStatsD: review changes

README.md

@@ -26,6 +26,16 @@ We recommend this only as an intermediate solution and recommend switching to
 [native Prometheus instrumentation](http://prometheus.io/docs/instrumenting/clientlibs/)
 in the long term.
 
+### DogStatsD extensions
+
+The exporter will convert DogStatsD-style tags to prometheus labels. See
+[Tags](http://docs.datadoghq.com/guides/dogstatsd/#tags) in the DogStatsD
+documentation for the concept description and
+[Datagram Format](http://docs.datadoghq.com/guides/dogstatsd/#datagram-format)
+for specifics. It boils down to appending
+`|#tag:value,another_tag:another_value` to the normal StatsD format.  Tags
+without values (`#some_tag`) are not supported.
+
 ## Building and Running
 
     $ go build

bridge_test.go

@@ -58,42 +58,82 @@ func TestHandlePacket(t *testing.T) {
 			},
 		}, {
 			name: "datadog tag extension",
-			in:   "foo:100|c|#tag1:bar,tag2:baz,tag3,tag4",
+			in:   "foo:100|c|#tag1:bar,tag2:baz",
 			out: Events{
 				&CounterEvent{
 					metricName: "foo",
 					value:      100,
-					labels:     map[string]string{"tag1": "bar", "tag2": "baz", "tag3": ".", "tag4": "."},
+					labels:     map[string]string{"tag1": "bar", "tag2": "baz"},
 				},
 			},
 		}, {
 			name: "datadog tag extension with # in all keys (as sent by datadog php client)",
-			in:   "foo:100|c|#tag1:bar,#tag2:baz,#tag3,#tag4",
+			in:   "foo:100|c|#tag1:bar,#tag2:baz",
 			out: Events{
 				&CounterEvent{
 					metricName: "foo",
 					value:      100,
-					labels:     map[string]string{"tag1": "bar", "tag2": "baz", "tag3": ".", "tag4": "."},
+					labels:     map[string]string{"tag1": "bar", "tag2": "baz"},
 				},
 			},
 		}, {
-			name: "datadog tag extension with tags unsupported by prometheus",
+			name: "datadog tag extension with tag keys unsupported by prometheus",
-			in:   "foo:100|c|#09digits:0,tag.with.dots,tag_with_empty_value:",
+			in:   "foo:100|c|#09digits:0,tag.with.dots:1",
 			out: Events{
 				&CounterEvent{
 					metricName: "foo",
 					value:      100,
-					labels:     map[string]string{"_09digits": "0", "tag_with_dots": ".", "tag_with_empty_value": "."},
+					labels:     map[string]string{"_09digits": "0", "tag_with_dots": "1"},
+				},
+			},
+		}, {
+			name: "datadog tag extension with valueless tags: ignored",
+			in:   "foo:100|c|#tag_without_a_value",
+			out: Events{
+				&CounterEvent{
+					metricName: "foo",
+					value:      100,
+					labels:     map[string]string{},
+				},
+			},
+		}, {
+			name: "datadog tag extension with valueless tags (edge case)",
+			in:   "foo:100|c|#tag_without_a_value,tag:value",
+			out: Events{
+				&CounterEvent{
+					metricName: "foo",
+					value:      100,
+					labels:     map[string]string{"tag": "value"},
+				},
+			},
+		}, {
+			name: "datadog tag extension with empty tags (edge case)",
+			in:   "foo:100|c|#tag:value,,",
+			out: Events{
+				&CounterEvent{
+					metricName: "foo",
+					value:      100,
+					labels:     map[string]string{"tag": "value"},
 				},
 			},
 		}, {
 			name: "datadog tag extension with sampling",
-			in:   "foo:100|c|@0.1|#tag1:bar,tag2,tag3:baz",
+			in:   "foo:100|c|@0.1|#tag1:bar,#tag2:baz",
 			out: Events{
 				&CounterEvent{
 					metricName: "foo",
 					value:      1000,
-					labels:     map[string]string{"tag1": "bar", "tag2": ".", "tag3": "baz"},
+					labels:     map[string]string{"tag1": "bar", "tag2": "baz"},
+				},
+			},
+		}, {
+			name: "datadog tag extension with multiple colons",
+			in:   "foo:100|c|@0.1|#tag1:foo:bar",
+			out: Events{
+				&CounterEvent{
+					metricName: "foo",
+					value:      1000,
+					labels:     map[string]string{"tag1": "foo:bar"},
 				},
 			},
 		}, {

exporter.go

@@ -303,6 +303,25 @@ func (l *StatsDListener) Listen(e chan<- Events) {
 	}
 }
 
+func parseDogStatsDTagsToLabels(component string) map[string]string {
+	labels := map[string]string{}
+	networkStats.WithLabelValues("dogstatsd_tags").Inc()
+	tags := strings.Split(component, ",")
+	for _, t := range tags {
+		t = strings.TrimPrefix(t, "#")
+		kv := strings.SplitN(t, ":", 2)
+
+		if len(kv) < 2 || len(kv[1]) == 0 {
+			networkStats.WithLabelValues("malformed_dogstatsd_tag").Inc()
+			log.Printf("Malformed or empty DogStatsD tag %s in component %s", t, component)
+			continue
+		}
+
+		labels[escapeMetricName(kv[0])] = kv[1]
+	}
+	return labels
+}
+
 func (l *StatsDListener) handlePacket(packet []byte, e chan<- Events) {
 	lines := strings.Split(string(packet), "\n")
 	events := Events{}
@@ -334,7 +353,6 @@ func (l *StatsDListener) handlePacket(packet []byte, e chan<- Events) {
 				continue
 			}
 			valueStr, statType := components[0], components[1]
-			labels := map[string]string{}
 			value, err := strconv.ParseFloat(valueStr, 64)
 			if err != nil {
 				log.Printf("Bad value %s on line: %s", valueStr, line)
@@ -342,6 +360,7 @@ func (l *StatsDListener) handlePacket(packet []byte, e chan<- Events) {
 				continue
 			}
 
+			labels := map[string]string{}
 			if len(components) >= 3 {
 				for _, component := range components[2:] {
 					switch component[0] {
@@ -360,26 +379,7 @@ func (l *StatsDListener) handlePacket(packet []byte, e chan<- Events) {
 						}
 						value /= samplingFactor
 					case '#':
-						networkStats.WithLabelValues("dogstatsd_tags").Inc()
+						labels = parseDogStatsDTagsToLabels(component)
-						tags := strings.Split(component, ",")
-						for _, t := range tags {
-							t = strings.TrimPrefix(t, "#")
-							kv := strings.Split(t, ":")
-							tag_key := kv[0]
-							tag_key = escapeMetricName(tag_key)
-
-							var tag_value string
-							if len(kv) == 2 {
-								if len(kv[1]) > 0 {
-									tag_value = kv[1]
-								} else {
-									tag_value = "."
-								}
-							} else if len(kv) == 1 {
-								tag_value = "."
-							}
-							labels[tag_key] = tag_value
-						}
 					default:
 						log.Printf("Invalid sampling factor or tag section %s on line %s", components[2], line)
 						networkStats.WithLabelValues("invalid_sample_factor").Inc()