From bb4e42068f195dc5672f02924583dcd4aed576e2 Mon Sep 17 00:00:00 2001
From: Ilya Margolin <ilya@jimdo.com>
Date: Fri, 15 Jul 2016 16:05:47 +0200
Subject: [PATCH] Skip metrics with invalid utf8

---
 bridge_test.go   | 31 +++++++++++++++++++++++++++++++
 exporter.go      |  7 ++++---
 exporter_test.go | 22 ++++++++++++++++++++++
 3 files changed, 57 insertions(+), 3 deletions(-)

diff --git a/bridge_test.go b/bridge_test.go
index 704ce70..312869e 100644
--- a/bridge_test.go
+++ b/bridge_test.go
@@ -136,6 +136,22 @@ func TestHandlePacket(t *testing.T) {
 					labels:     map[string]string{"tag1": "foo:bar"},
 				},
 			},
+		}, {
+			name: "datadog tag extension with invalid utf8 tag values",
+			in:   "foo:100|c|@0.1|#tag:\xc3\x28invalid",
+		}, {
+			name: "datadog tag extension with both valid and invalid utf8 tag values",
+			in:   "foo:100|c|@0.1|#tag1:valid,tag2:\xc3\x28invalid",
+		}, {
+			name: "multiple metrics with invalid datadog utf8 tag values",
+			in:   "foo:200|c|#tag:value\nfoo:300|c|#tag:\xc3\x28invalid",
+			out: Events{
+				&CounterEvent{
+					metricName: "foo",
+					value:      200,
+					labels:     map[string]string{"tag": "value"},
+				},
+			},
 		}, {
 			name: "combined multiline metrics",
 			in:   "foo:200|ms:300|ms:5|c|@0.1:6|g\nbar:1|c:5|ms",
@@ -212,6 +228,21 @@ func TestHandlePacket(t *testing.T) {
 			name: "empty component",
 			in:   "foo:1|c|",
 		},
+		{
+			name: "invalid utf8",
+			in:   "invalid\xc3\x28utf8:1|c",
+		},
+		{
+			name: "some invalid utf8",
+			in:   "valid_utf8:1|c\ninvalid\xc3\x28utf8:1|c",
+			out: Events{
+				&CounterEvent{
+					metricName: "valid_utf8",
+					value:      1,
+					labels:     map[string]string{},
+				},
+			},
+		},
 	}
 
 	l := StatsDListener{}
diff --git a/exporter.go b/exporter.go
index 7303f6b..ec8df97 100644
--- a/exporter.go
+++ b/exporter.go
@@ -22,6 +22,7 @@ import (
 	"regexp"
 	"strconv"
 	"strings"
+	"unicode/utf8"
 
 	"github.com/prometheus/client_golang/prometheus"
 	"github.com/prometheus/common/log"
@@ -350,9 +351,8 @@ func (l *StatsDListener) handlePacket(packet []byte, e chan<- Events) {
 		if line == "" {
 			continue
 		}
-
 		elements := strings.SplitN(line, ":", 2)
-		if len(elements) < 2 || len(elements[0]) == 0 {
+		if len(elements) < 2 || len(elements[0]) == 0 || !utf8.ValidString(line) {
 			networkStats.WithLabelValues("malformed_line").Inc()
 			log.Errorln("Bad line from StatsD:", line)
 			continue
@@ -365,7 +365,8 @@ func (l *StatsDListener) handlePacket(packet []byte, e chan<- Events) {
 		} else {
 			samples = strings.Split(elements[1], ":")
 		}
-		samples: for _, sample := range samples {
+	samples:
+		for _, sample := range samples {
 			components := strings.Split(sample, "|")
 			samplingFactor := 1.0
 			if len(components) < 2 || len(components) > 4 {
diff --git a/exporter_test.go b/exporter_test.go
index fb806e3..1eb4bf5 100644
--- a/exporter_test.go
+++ b/exporter_test.go
@@ -50,3 +50,25 @@ func TestNegativeCounter(t *testing.T) {
 
 	ex.Listen(events)
 }
+
+// TestInvalidUtf8InDatadogTagValue validates robustness of exporter listener
+// against datadog tags with invalid tag values.
+// It sends the same tags first with a valid value, then with an invalid one.
+// The exporter should not panic, but drop the invalid event
+func TestInvalidUtf8InDatadogTagValue(t *testing.T) {
+	l := StatsDListener{}
+	events := make(chan Events, 2)
+
+	l.handlePacket([]byte("bar:200|c|#tag:value"), events)
+	l.handlePacket([]byte("bar:200|c|#tag:\xc3\x28invalid"), events)
+
+	ex := NewExporter(&metricMapper{}, true)
+
+	// Close channel to signify we are done with the listener after a short period.
+	go func() {
+		time.Sleep(time.Millisecond * 100)
+		close(events)
+	}()
+
+	ex.Listen(events)
+}