feat: only store bucket values for histogram

The previous implementation stored every single recorded value, causing
massive memory usage over time.

collector.v

@@ -15,27 +15,26 @@ pub mut:
 struct Histogram {
 	metric Metric
 pub mut:
+	total_count   u64
+	sum           f64
 	buckets       []f64
-}
-
-struct FloatSeries {
-	metric Metric
-pub mut:
-	data []f64
+	bucket_counts []u64
 }
 
 [heap]
 struct DefaultCollector {
 mut:
+	buckets    map[string][]f64
 	counters   shared map[string]&Counter
-	histograms shared map[string]&FloatSeries
+	histograms shared map[string]&Histogram
 	gauges     shared map[string]&Gauge
 }
 
 pub fn new_default_collector() &DefaultCollector {
 	return &DefaultCollector{
+		buckets: map[string][]f64{}
 		counters: map[string]&Counter{}
-		histograms: map[string]&FloatSeries{}
+		histograms: map[string]&Histogram{}
 		gauges: map[string]&Gauge{}
 	}
 }
@@ -92,29 +91,52 @@ pub fn (c &DefaultCollector) counters() []Metric {
 	return metrics
 }
 
-pub fn (c &DefaultCollector) histogram_record(value f64, metric Metric) {
+pub fn (mut c DefaultCollector) histogram_buckets_set(name string, buckets []f64) {
+	lock c.histograms {
+		c.buckets[name] = buckets
+	}
+}
+
+pub fn (mut c DefaultCollector) histogram_record(value f64, metric Metric) {
 	lock c.histograms {
 		mut entry := c.histograms[metric.str()] or {
-			hist := &FloatSeries{
+			buckets := c.buckets[metric.name] or { [] }
+			hist := &Histogram{
 				metric: metric
-				data: []f64{}
+				buckets: buckets
+				bucket_counts: []u64{len: buckets.len, init: 0}
 			}
 			c.histograms[metric.str()] = hist
 
 			hist
 		}
 
-		entry.data << value
+		entry.sum += value
+		entry.total_count += 1
+
+		mut i := entry.buckets.len - 1
+
+		for i >= 0 && value <= entry.buckets[i] {
+			entry.bucket_counts[i]++
+
+			i -= 1
+		}
 	}
 }
 
-pub fn (c &DefaultCollector) histogram_get(metric Metric) ?[]f64 {
+pub fn (c &DefaultCollector) histogram_get(metric Metric) ?Histogram {
 	return rlock c.histograms {
 		entry := c.histograms[metric.str()] or { return none }
 
 		// Return a clone of the data to prevent user from altering
 		// internal structure
-		entry.data.clone()
+		Histogram{
+			metric: metric
+			total_count: entry.total_count
+			sum: entry.sum
+			buckets: entry.buckets.clone()
+			bucket_counts: entry.bucket_counts.clone()
+		}
 	}
 }
 

collector_test.v

@@ -32,10 +32,28 @@ fn test_histogram() {
 	mut m := new_default_collector()
 
 	m.histogram_record(5.0, name: 'test')
-	assert m.histogram_get(name: 'test')? == [5.0]
+
+	assert m.histogram_get(name: 'test')? == Histogram{
+		metric: Metric{
+			name: 'test'
+		}
+		total_count: 1
+		sum: 5.0
+		buckets: []
+		bucket_counts: []
+	}
 
 	m.histogram_record(7.0, name: 'test')
-	assert m.histogram_get(name: 'test')? == [5.0, 7.0]
+
+	assert m.histogram_get(name: 'test')? == Histogram{
+		metric: Metric{
+			name: 'test'
+		}
+		total_count: 2
+		sum: 12.0
+		buckets: []
+		bucket_counts: []
+	}
 
 	// Test with labels
 	metric := Metric{
@@ -43,11 +61,26 @@ fn test_histogram() {
 		labels: [['hi', 'label']!, ['hi2', 'label2']!]
 	}
 
+	m.histogram_buckets_set('test2', [10.0])
 	m.histogram_record(5.0, metric)
-	assert m.histogram_get(metric)? == [5.0]
+
+	assert m.histogram_get(metric)? == Histogram{
+		metric: metric
+		total_count: 1
+		sum: 5.0
+		buckets: [10.0]
+		bucket_counts: [u64(1)]
+	}
 
 	m.histogram_record(7.0, metric)
-	assert m.histogram_get(metric)? == [5.0, 7.0]
+
+	assert m.histogram_get(metric)? == Histogram{
+		metric: metric
+		total_count: 2
+		sum: 12.0
+		buckets: [10.0]
+		bucket_counts: [u64(2)]
+	}
 }
 
 fn test_gauge_add() {

metrics.v

@@ -16,7 +16,7 @@ pub fn (m &Metric) str() string {
 pub interface MetricsCollector {
 	counter_get(metric Metric) ?u64
 	counters() []Metric
-	histogram_get(metric Metric) ?[]f64
+	histogram_get(metric Metric) ?Histogram
 	histograms() []Metric
 	gauge_get(metric Metric) ?f64
 	gauges() []Metric

null.v

@@ -21,7 +21,7 @@ pub fn (c &NullCollector) counters() []Metric {
 
 pub fn (c &NullCollector) histogram_record(value f64, metric Metric) {}
 
-pub fn (c &NullCollector) histogram_get(metric Metric) ?[]f64 {
+pub fn (c &NullCollector) histogram_get(metric Metric) ?Histogram {
 	return none
 }
 

prometheus.v

@@ -2,19 +2,15 @@ module metrics
 
 import strings
 import io
-import arrays
 
 pub struct PrometheusExporter {
-	buckets []f64
 mut:
 	prefix    string
 	collector &MetricsCollector = unsafe { nil }
 }
 
-pub fn new_prometheus_exporter(buckets []f64) PrometheusExporter {
-	return PrometheusExporter{
-		buckets: buckets
-	}
+pub fn new_prometheus_exporter() PrometheusExporter {
+	return PrometheusExporter{}
 }
 
 pub fn (mut e PrometheusExporter) load(prefix string, collector &MetricsCollector) {
@@ -61,45 +57,25 @@ pub fn (mut e PrometheusExporter) export_to_writer(mut writer io.Writer) ! {
 	}
 
 	for hist in e.collector.histograms() {
-		data := e.collector.histogram_get(hist) or { return error("This can't happen.") }
-
-		sum := arrays.sum(data) or { 0.0 }
-		total_count := data.len
-
-		mut bucket_counts := []u64{len: e.buckets.len}
-
-		mut i := bucket_counts.len - 1
-
-		// For each data point, increment all buckets that the value is
-		// contained in. Because the buckets are sorted, we can stop once we
-		// encounter one that it doesn't fit in
-		for val in data {
-			for i >= 0 && val <= e.buckets[i] {
-				bucket_counts[i]++
-
-				i -= 1
-			}
-
-			i = bucket_counts.len - 1
-		}
+		hist_data := e.collector.histogram_get(hist) or { return error("This can't happen.") }
 
 		mut m := Metric{
 			...hist
 			name: '${hist.name}_count'
 		}
-		writer.write('${e.serialize_metric(m)} $total_count\n'.bytes())!
+		writer.write('${e.serialize_metric(m)} $hist_data.total_count\n'.bytes())!
 
 		m = Metric{
 			...hist
 			name: '${hist.name}_sum'
 		}
-		writer.write('${e.serialize_metric(m)} $sum\n'.bytes())!
+		writer.write('${e.serialize_metric(m)} $hist_data.sum\n'.bytes())!
 
 		mut le_labels := [][2]string{}
 		le_labels.prepend(hist.labels)
 		le_labels << ['le', '']!
 
-		for j, bucket in e.buckets {
+		for j, bucket in hist_data.buckets {
 			le_labels[le_labels.len - 1][1] = bucket.str()
 
 			m = Metric{
@@ -107,17 +83,19 @@ pub fn (mut e PrometheusExporter) export_to_writer(mut writer io.Writer) ! {
 				labels: le_labels
 			}
 
-			writer.write('${e.serialize_metric(m)} ${bucket_counts[j]}\n'.bytes())!
+			writer.write('${e.serialize_metric(m)} ${hist_data.bucket_counts[j]}\n'.bytes())!
 		}
 
 		// Always output the +Inf bucket
 		le_labels[le_labels.len - 1][1] = '+Inf'
 
+		if hist_data.buckets.len > 0 {
 			m = Metric{
 				name: '${hist.name}_bucket'
 				labels: le_labels
 			}
 
-		writer.write('${e.serialize_metric(m)} $total_count\n'.bytes())!
+			writer.write('${e.serialize_metric(m)} $hist_data.total_count\n'.bytes())!
+		}
 	}
 }

prometheus_test.v

@@ -4,7 +4,7 @@ fn test_only_counters() {
 	mut m := new_default_collector()
 	m.counter_increment(name: 'test')
 
-	mut e := new_prometheus_exporter([])
+	mut e := new_prometheus_exporter()
 	e.load('hi_', m)
 
 	assert e.export_to_string()! == 'hi_test 1\n'
@@ -23,7 +23,7 @@ fn test_only_gauges() {
 	mut m := new_default_collector()
 	m.gauge_set(3.25, name: 'test')
 
-	mut e := new_prometheus_exporter([])
+	mut e := new_prometheus_exporter()
 	e.load('hi_', m)
 
 	assert e.export_to_string()! == 'hi_test 3.25\n'
@@ -40,10 +40,11 @@ fn test_only_gauges() {
 fn test_single_histogram() {
 	mut m := new_default_collector()
 
+    m.histogram_buckets_set('test', [0.5, 5.0])
 	m.histogram_record(5.0, name: 'test')
 	m.histogram_record(7.0, name: 'test')
 
-	mut e := new_prometheus_exporter([0.5, 5.0])
+	mut e := new_prometheus_exporter()
 	e.load('hi_', m)
 
 	assert e.export_to_string()! == 'hi_test_count 2\nhi_test_sum 12.0\nhi_test_bucket{le="0.5"} 0\nhi_test_bucket{le="5.0"} 1\nhi_test_bucket{le="+Inf"} 2\n'