Allow colons in the prometheus metric name (#6751)

plugins/serializers/prometheus/collection.go

@@ -113,7 +113,7 @@ func (c *Collection) createLabels(metric telegraf.Metric) []LabelPair {
 			}
 		}
 
-		name, ok := SanitizeName(tag.Key)
+		name, ok := SanitizeLabelName(tag.Key)
 		if !ok {
 			continue
 		}
@@ -132,7 +132,7 @@ func (c *Collection) createLabels(metric telegraf.Metric) []LabelPair {
 			continue
 		}
 
-		name, ok := SanitizeName(field.Key)
+		name, ok := SanitizeLabelName(field.Key)
 		if !ok {
 			continue
 		}
@@ -161,7 +161,7 @@ func (c *Collection) Add(metric telegraf.Metric) {
 	labels := c.createLabels(metric)
 	for _, field := range metric.FieldList() {
 		metricName := MetricName(metric.Name(), field.Key, metric.Type())
-		metricName, ok := SanitizeName(metricName)
+		metricName, ok := SanitizeMetricName(metricName)
 		if !ok {
 			continue
 		}

plugins/serializers/prometheus/convert.go

@@ -8,26 +8,53 @@ import (
 	dto "github.com/prometheus/client_model/go"
 )
 
-var FirstTable = &unicode.RangeTable{
+type Table struct {
-	R16: []unicode.Range16{
+	First *unicode.RangeTable
-		{0x0041, 0x005A, 1}, // A-Z
+	Rest  *unicode.RangeTable
-		{0x005F, 0x005F, 1}, // _
-		{0x0061, 0x007A, 1}, // a-z
-	},
-	LatinOffset: 3,
 }
 
-var RestTable = &unicode.RangeTable{
+var MetricNameTable = Table{
-	R16: []unicode.Range16{
+	First: &unicode.RangeTable{
-		{0x0030, 0x0039, 1}, // 0-9
+		R16: []unicode.Range16{
-		{0x0041, 0x005A, 1}, // A-Z
+			{0x003A, 0x003A, 1}, // :
-		{0x005F, 0x005F, 1}, // _
+			{0x0041, 0x005A, 1}, // A-Z
-		{0x0061, 0x007A, 1}, // a-z
+			{0x005F, 0x005F, 1}, // _
+			{0x0061, 0x007A, 1}, // a-z
+		},
+		LatinOffset: 4,
+	},
+	Rest: &unicode.RangeTable{
+		R16: []unicode.Range16{
+			{0x0030, 0x003A, 1}, // 0-:
+			{0x0041, 0x005A, 1}, // A-Z
+			{0x005F, 0x005F, 1}, // _
+			{0x0061, 0x007A, 1}, // a-z
+		},
+		LatinOffset: 4,
 	},
-	LatinOffset: 4,
 }
 
-func isValid(name string) bool {
+var LabelNameTable = Table{
+	First: &unicode.RangeTable{
+		R16: []unicode.Range16{
+			{0x0041, 0x005A, 1}, // A-Z
+			{0x005F, 0x005F, 1}, // _
+			{0x0061, 0x007A, 1}, // a-z
+		},
+		LatinOffset: 3,
+	},
+	Rest: &unicode.RangeTable{
+		R16: []unicode.Range16{
+			{0x0030, 0x0039, 1}, // 0-9
+			{0x0041, 0x005A, 1}, // A-Z
+			{0x005F, 0x005F, 1}, // _
+			{0x0061, 0x007A, 1}, // a-z
+		},
+		LatinOffset: 4,
+	},
+}
+
+func isValid(name string, table Table) bool {
 	if name == "" {
 		return false
 	}
@@ -35,11 +62,11 @@ func isValid(name string) bool {
 	for i, r := range name {
 		switch {
 		case i == 0:
-			if !unicode.In(r, FirstTable) {
+			if !unicode.In(r, table.First) {
 				return false
 			}
 		default:
-			if !unicode.In(r, RestTable) {
+			if !unicode.In(r, table.Rest) {
 				return false
 			}
 		}
@@ -48,12 +75,11 @@ func isValid(name string) bool {
 	return true
 }
 
-// SanitizeName check if the name is a valid Prometheus metric name and label
+// Sanitize checks if the name is valid according to the table.  If not, it
-// name.  If not, it attempts to replaces invalid runes with an underscore to
+// attempts to replaces invalid runes with an underscore to create a valid
-// create a valid name.  Returns the metric name and true if the name is valid
+// name.
-// to use.
+func sanitize(name string, table Table) (string, bool) {
-func SanitizeName(name string) (string, bool) {
+	if isValid(name, table) {
-	if isValid(name) {
 		return name, true
 	}
 
@@ -62,11 +88,11 @@ func SanitizeName(name string) (string, bool) {
 	for i, r := range name {
 		switch {
 		case i == 0:
-			if unicode.In(r, FirstTable) {
+			if unicode.In(r, table.First) {
 				b.WriteRune(r)
 			}
 		default:
-			if unicode.In(r, RestTable) {
+			if unicode.In(r, table.Rest) {
 				b.WriteRune(r)
 			} else {
 				b.WriteString("_")
@@ -82,6 +108,20 @@ func SanitizeName(name string) (string, bool) {
 	return name, true
 }
 
+// SanitizeMetricName checks if the name is a valid Prometheus metric name.  If
+// not, it attempts to replaces invalid runes with an underscore to create a
+// valid name.
+func SanitizeMetricName(name string) (string, bool) {
+	return sanitize(name, MetricNameTable)
+}
+
+// SanitizeLabelName checks if the name is a valid Prometheus label name.  If
+// not, it attempts to replaces invalid runes with an underscore to create a
+// valid name.
+func SanitizeLabelName(name string) (string, bool) {
+	return sanitize(name, LabelNameTable)
+}
+
 // MetricName returns the Prometheus metric name.
 func MetricName(measurement, fieldKey string, valueType telegraf.ValueType) string {
 	switch valueType {

plugins/serializers/prometheus/prometheus_test.go

@@ -409,6 +409,42 @@ rpc_duration_seconds_count 2693
 # HELP cpu_time_idle Telegraf collected metric
 # TYPE cpu_time_idle untyped
 cpu_time_idle 43
+`),
+		},
+		{
+			name: "colons are not replaced in metric name from measurement",
+			metrics: []telegraf.Metric{
+				testutil.MustMetric(
+					"cpu::xyzzy",
+					map[string]string{},
+					map[string]interface{}{
+						"time_idle": 42.0,
+					},
+					time.Unix(0, 0),
+				),
+			},
+			expected: []byte(`
+# HELP cpu::xyzzy_time_idle Telegraf collected metric
+# TYPE cpu::xyzzy_time_idle untyped
+cpu::xyzzy_time_idle 42
+`),
+		},
+		{
+			name: "colons are not replaced in metric name from field",
+			metrics: []telegraf.Metric{
+				testutil.MustMetric(
+					"cpu",
+					map[string]string{},
+					map[string]interface{}{
+						"time:idle": 42.0,
+					},
+					time.Unix(0, 0),
+				),
+			},
+			expected: []byte(`
+# HELP cpu_time:idle Telegraf collected metric
+# TYPE cpu_time:idle untyped
+cpu_time:idle 42
 `),
 		},
 		{
@@ -429,6 +465,26 @@ cpu_time_idle 43
 # HELP cpu_time_idle Telegraf collected metric
 # TYPE cpu_time_idle untyped
 cpu_time_idle{host_name="example.org"} 42
+`),
+		},
+		{
+			name: "colons are replaced in label name",
+			metrics: []telegraf.Metric{
+				testutil.MustMetric(
+					"cpu",
+					map[string]string{
+						"host:name": "example.org",
+					},
+					map[string]interface{}{
+						"time_idle": 42.0,
+					},
+					time.Unix(0, 0),
+				),
+			},
+			expected: []byte(`
+# HELP cpu_time_idle Telegraf collected metric
+# TYPE cpu_time_idle untyped
+cpu_time_idle{host_name="example.org"} 42
 `),
 		},
 		{