Add derivative and nonNegativeDerivative functions

docs/graphite.md

@@ -36,6 +36,7 @@ Here are the currently included functions:
 | averageSeries(seriesLists) series                              | avg         | Stable     |
 | consolidateBy(seriesList, func) seriesList                     |             | Stable     |
 | countSeries(seriesLists) series                                |             | Stable     |
+| derivative(seriesLists) series                                 |             | Stable     |
 | diffSeries(seriesLists) series                                 |             | Stable     |
 | divideSeries(dividend, divisor) seriesList                     |             | Stable     |
 | divideSeriesLists(dividends, divisors) seriesList              |             | Stable     |
@@ -55,6 +56,7 @@ Here are the currently included functions:
 | minSeries(seriesList) series                                   | min         | Stable     |
 | multiplySeries(seriesList) series                              |             | Stable     |
 | movingAverage(seriesLists, windowSize) seriesList              |             | Unstable   |
+| nonNegatievDerivative(seriesList, maxValue) seriesList         |             | Stable     |
 | perSecond(seriesLists) seriesList                              |             | Stable     |
 | rangeOfSeries(seriesList) series                               |             | Stable     |
 | removeAboveValue(seriesList, n) seriesList                     |             | Stable     |

expr/func_derivative.go

@@ -0,0 +1,63 @@
+package expr
+
+import (
+	"fmt"
+	"math"
+
+	"github.com/grafana/metrictank/api/models"
+	schema "gopkg.in/raintank/schema.v1"
+)
+
+type FuncDerivative struct {
+	in GraphiteFunc
+}
+
+func NewDerivative() GraphiteFunc {
+	return &FuncDerivative{}
+}
+
+func (s *FuncDerivative) Signature() ([]Arg, []Arg) {
+	return []Arg{
+		ArgSeriesList{val: &s.in}}, []Arg{ArgSeriesList{}}
+}
+
+func (s *FuncDerivative) Context(context Context) Context {
+	return context
+}
+
+func (s *FuncDerivative) Exec(cache map[Req][]models.Series) ([]models.Series, error) {
+	series, err := s.in.Exec(cache)
+	if err != nil {
+		return nil, err
+	}
+
+	outSeries := make([]models.Series, len(series))
+	for i, serie := range series {
+		serie.Target = fmt.Sprintf("derivative(%s)", serie.Target)
+		serie.QueryPatt = fmt.Sprintf("derivative(%s)", serie.QueryPatt)
+		out := pointSlicePool.Get().([]schema.Point)
+
+		newTags := make(map[string]string, len(serie.Tags)+1)
+		for k, v := range serie.Tags {
+			newTags[k] = v
+		}
+		newTags["derivative"] = "1"
+		serie.Tags = newTags
+
+		prev := math.NaN()
+		for _, p := range serie.Datapoints {
+			val := p.Val
+			if math.IsNaN(prev) || math.IsNaN(val) {
+				p.Val = math.NaN()
+			} else {
+				p.Val -= prev
+			}
+			prev = val
+			out = append(out, p)
+		}
+		serie.Datapoints = out
+		outSeries[i] = serie
+	}
+	cache[Req{}] = append(cache[Req{}], outSeries...)
+	return outSeries, nil
+}

expr/func_derivative_test.go

@@ -0,0 +1,182 @@
+package expr
+
+import (
+	"math"
+	"strconv"
+	"testing"
+
+	"github.com/grafana/metrictank/api/models"
+	"github.com/grafana/metrictank/test"
+	"gopkg.in/raintank/schema.v1"
+)
+
+func TestDerivativeNoMax(t *testing.T) {
+	testDerivative(
+		"derivative",
+		[]models.Series{
+			{
+				Interval:   10,
+				QueryPatt:  "abcd",
+				Target:     "a",
+				Datapoints: getCopy(a),
+			},
+			{
+				Interval:   10,
+				QueryPatt:  "abcd",
+				Target:     "b",
+				Datapoints: getCopy(b),
+			},
+			{
+				Interval:   10,
+				QueryPatt:  "abcd",
+				Target:     "c",
+				Datapoints: getCopy(c),
+			},
+			{
+				Interval:   10,
+				QueryPatt:  "abcd",
+				Target:     "d",
+				Datapoints: getCopy(d),
+			},
+		},
+		[]models.Series{
+			{
+				Interval:  10,
+				QueryPatt: "derivative(abcd)",
+				Datapoints: []schema.Point{
+					{Val: math.NaN(), Ts: 10},
+					{Val: 0, Ts: 20},
+					{Val: 5.5, Ts: 30},
+					{Val: math.NaN(), Ts: 40},
+					{Val: math.NaN(), Ts: 50},
+					{Val: math.NaN(), Ts: 60},
+				},
+			},
+			{
+				Interval:  10,
+				QueryPatt: "derivative(abcd)",
+				Datapoints: []schema.Point{
+					{Val: math.NaN(), Ts: 10},
+					{Val: math.MaxFloat64, Ts: 20},
+					{Val: 0, Ts: 30},
+					{Val: math.NaN(), Ts: 40},
+					{Val: math.NaN(), Ts: 50},
+					{Val: math.NaN(), Ts: 60},
+				},
+			},
+			{
+				Interval:  10,
+				QueryPatt: "derivative(abcd)",
+				Datapoints: []schema.Point{
+					{Val: math.NaN(), Ts: 10},
+					{Val: 0, Ts: 20},
+					{Val: 1, Ts: 30},
+					{Val: 1, Ts: 40},
+					{Val: 1, Ts: 50},
+					{Val: 1, Ts: 60},
+				},
+			},
+			{
+				Interval:  10,
+				QueryPatt: "derivative(abcd)",
+				Datapoints: []schema.Point{
+					{Val: math.NaN(), Ts: 10},
+					{Val: 33, Ts: 20},
+					{Val: 166, Ts: 30},
+					{Val: -170, Ts: 40},
+					{Val: 51, Ts: 50},
+					{Val: 170, Ts: 60},
+				},
+			},
+		},
+		t,
+	)
+}
+
+func testDerivative(name string, in []models.Series, out []models.Series, t *testing.T) {
+	f := NewDerivative()
+	f.(*FuncDerivative).in = NewMock(in)
+	gots, err := f.Exec(make(map[Req][]models.Series))
+	if err != nil {
+		t.Fatalf("case %q: err should be nil. got %q", name, err)
+	}
+	if len(gots) != len(out) {
+		t.Fatalf("case %q: isNonNull len output expected %d, got %d", name, len(out), len(gots))
+	}
+	for i, g := range gots {
+		exp := out[i]
+		if g.QueryPatt != exp.QueryPatt {
+			t.Fatalf("case %q: expected target %q, got %q", name, exp.QueryPatt, g.QueryPatt)
+		}
+		if len(g.Datapoints) != len(exp.Datapoints) {
+			t.Fatalf("case %q len output expected %d, got %d", name, len(exp.Datapoints), len(g.Datapoints))
+		}
+		for j, p := range g.Datapoints {
+			bothNaN := math.IsNaN(p.Val) && math.IsNaN(exp.Datapoints[j].Val)
+			if (bothNaN || p.Val == exp.Datapoints[j].Val) && p.Ts == exp.Datapoints[j].Ts {
+				continue
+			}
+			t.Fatalf("case %q: output point %d - expected %v got %v", name, j, exp.Datapoints[j], p)
+		}
+	}
+}
+func BenchmarkDerivative10k_1NoNulls(b *testing.B) {
+	benchmarkDerivative(b, 1, test.RandFloats10k, test.RandFloats10k)
+}
+func BenchmarkDerivative10k_10NoNulls(b *testing.B) {
+	benchmarkDerivative(b, 10, test.RandFloats10k, test.RandFloats10k)
+}
+func BenchmarkDerivative10k_100NoNulls(b *testing.B) {
+	benchmarkDerivative(b, 100, test.RandFloats10k, test.RandFloats10k)
+}
+func BenchmarkDerivative10k_1000NoNulls(b *testing.B) {
+	benchmarkDerivative(b, 1000, test.RandFloats10k, test.RandFloats10k)
+}
+func BenchmarkDerivative10k_1SomeSeriesHalfNulls(b *testing.B) {
+	benchmarkDerivative(b, 1, test.RandFloats10k, test.RandFloatsWithNulls10k)
+}
+func BenchmarkDerivative10k_10SomeSeriesHalfNulls(b *testing.B) {
+	benchmarkDerivative(b, 10, test.RandFloats10k, test.RandFloatsWithNulls10k)
+}
+func BenchmarkDerivative10k_100SomeSeriesHalfNulls(b *testing.B) {
+	benchmarkDerivative(b, 100, test.RandFloats10k, test.RandFloatsWithNulls10k)
+}
+func BenchmarkDerivative10k_1000SomeSeriesHalfNulls(b *testing.B) {
+	benchmarkDerivative(b, 1000, test.RandFloats10k, test.RandFloatsWithNulls10k)
+}
+func BenchmarkDerivative10k_1AllSeriesHalfNulls(b *testing.B) {
+	benchmarkDerivative(b, 1, test.RandFloatsWithNulls10k, test.RandFloatsWithNulls10k)
+}
+func BenchmarkDerivative10k_10AllSeriesHalfNulls(b *testing.B) {
+	benchmarkDerivative(b, 10, test.RandFloatsWithNulls10k, test.RandFloatsWithNulls10k)
+}
+func BenchmarkDerivative10k_100AllSeriesHalfNulls(b *testing.B) {
+	benchmarkDerivative(b, 100, test.RandFloatsWithNulls10k, test.RandFloatsWithNulls10k)
+}
+func BenchmarkDerivative10k_1000AllSeriesHalfNulls(b *testing.B) {
+	benchmarkDerivative(b, 1000, test.RandFloatsWithNulls10k, test.RandFloatsWithNulls10k)
+}
+func benchmarkDerivative(b *testing.B, numSeries int, fn0, fn1 func() []schema.Point) {
+	var input []models.Series
+	for i := 0; i < numSeries; i++ {
+		series := models.Series{
+			QueryPatt: strconv.Itoa(i),
+		}
+		if i%2 == 0 {
+			series.Datapoints = fn0()
+		} else {
+			series.Datapoints = fn1()
+		}
+		input = append(input, series)
+	}
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		f := NewDerivative()
+		f.(*FuncDerivative).in = NewMock(input)
+		got, err := f.Exec(make(map[Req][]models.Series))
+		if err != nil {
+			b.Fatalf("%s", err)
+		}
+		results = got
+	}
+}

expr/func_nonnegativederivative.go

@@ -0,0 +1,84 @@
+package expr
+
+import (
+	"fmt"
+	"math"
+
+	"github.com/grafana/metrictank/api/models"
+	schema "gopkg.in/raintank/schema.v1"
+)
+
+type FuncNonNegativeDerivative struct {
+	in       GraphiteFunc
+	maxValue float64
+}
+
+func NewNonNegativeDerivative() GraphiteFunc {
+	return &FuncNonNegativeDerivative{maxValue: math.NaN()}
+}
+
+func (s *FuncNonNegativeDerivative) Signature() ([]Arg, []Arg) {
+	return []Arg{
+		ArgSeriesList{val: &s.in},
+		ArgFloat{
+			key: "maxValue",
+			opt: true,
+			val: &s.maxValue}}, []Arg{ArgSeriesList{}}
+}
+
+func (s *FuncNonNegativeDerivative) Context(context Context) Context {
+	return context
+}
+
+func (s *FuncNonNegativeDerivative) Exec(cache map[Req][]models.Series) ([]models.Series, error) {
+	series, err := s.in.Exec(cache)
+	if err != nil {
+		return nil, err
+	}
+
+	outSeries := make([]models.Series, len(series))
+	for i, serie := range series {
+		serie.Target = fmt.Sprintf("nonNegativeDerivative(%s)", serie.Target)
+		serie.QueryPatt = fmt.Sprintf("nonNegativeDerivative(%s)", serie.QueryPatt)
+		out := pointSlicePool.Get().([]schema.Point)
+
+		newTags := make(map[string]string, len(serie.Tags)+1)
+		for k, v := range serie.Tags {
+			newTags[k] = v
+		}
+		newTags["nonNegativeDerivative"] = "1"
+		serie.Tags = newTags
+
+		prev := math.NaN()
+		for _, p := range serie.Datapoints {
+			var delta float64
+			delta, prev = nonNegativeDelta(p.Val, prev, s.maxValue)
+			p.Val = delta
+			out = append(out, p)
+		}
+		serie.Datapoints = out
+		outSeries[i] = serie
+	}
+	cache[Req{}] = append(cache[Req{}], outSeries...)
+	return outSeries, nil
+}
+
+func nonNegativeDelta(val, prev, maxValue float64) (float64, float64) {
+	if val > maxValue {
+		return math.NaN(), math.NaN()
+	}
+
+	if math.IsNaN(prev) || math.IsNaN(val) {
+		return math.NaN(), val
+	}
+
+	if val >= prev {
+		return val - prev, val
+	}
+
+	if !math.IsNaN(maxValue) {
+		return maxValue + 1 + val - prev, val
+	}
+
+	return math.NaN(), val
+}