Add histogram plot type, display bar chart labels on the right instead of overlapping.

6 files changed, 113 insertions, 60 deletions

diff --git a/cmd/jp/bar.go b/cmd/jp/bar.go
index 15dd196..cfc39c2 100644
--- a/cmd/jp/bar.go
+++ b/cmd/jp/bar.go
@@ -2,37 +2,38 @@ package main
 
 import (
 	"fmt"
+	"log"
 	"reflect"
 
+	"github.com/sgreben/jp/pkg/data"
 	"github.com/sgreben/jp/pkg/draw"
 	"github.com/sgreben/jp/pkg/plot"
 )
 
-func barPlotData(xvv, yvv [][]reflect.Value) (x []string, y []float64) {
-	for _, xv := range xvv {
-		for i := range xv {
-			if xv[i].IsValid() && xv[i].CanInterface() {
-				x = append(x, fmt.Sprint(xv[i].Interface()))
-			}
+func barPlotData(xv, yv []reflect.Value) (x []string, y []float64) {
+	for i := range xv {
+		if xv[i].IsValid() && xv[i].CanInterface() {
+			x = append(x, fmt.Sprint(xv[i].Interface()))
 		}
 	}
-	for _, yv := range yvv {
-		for i := range yv {
-			if yv[i].IsValid() && yv[i].CanInterface() {
-				yvi, ok := yv[i].Interface().(float64)
-				if ok {
-					y = append(y, yvi)
-				}
+	for i := range yv {
+		if yv[i].IsValid() && yv[i].CanInterface() {
+			yvi, ok := yv[i].Interface().(float64)
+			if ok {
+				y = append(y, yvi)
 			}
 		}
 	}
 	return
 }
 
-func barPlot(xvv, yvv [][]reflect.Value, c draw.Canvas) string {
-	groups, y := barPlotData(xvv, yvv)
+func barPlot(xv, yv []reflect.Value, c draw.Canvas) string {
+	groups, y := barPlotData(xv, yv)
 	chart := plot.NewBarChart(c)
-	data := new(plot.DataTable)
+	data := new(data.Table)
+	if len(y) == 0 {
+		log.Fatal("no valid y values given")
+	}
 	if len(groups) != len(y) {
 		for i := range y {
 			data.AddColumn(fmt.Sprint(i))
diff --git a/cmd/jp/hist.go b/cmd/jp/hist.go
new file mode 100644
index 0000000..77c04e6
--- /dev/null
+++ b/cmd/jp/hist.go
@@ -0,0 +1,50 @@
+package main
+
+import (
+	"log"
+	"reflect"
+
+	"github.com/sgreben/jp/pkg/data"
+	"github.com/sgreben/jp/pkg/draw"
+	"github.com/sgreben/jp/pkg/plot"
+)
+
+func histogramData(xv []reflect.Value, nbins uint) (groups []string, counts []float64) {
+	var x []float64
+	for i := range xv {
+		if xv[i].IsValid() && xv[i].CanInterface() {
+			xvi, ok := xv[i].Interface().(float64)
+			if ok {
+				x = append(x, xvi)
+			}
+		}
+	}
+	if len(x) == 0 {
+		log.Fatal("no valid x values given")
+	}
+	bins := data.NewBins(x)
+	bins.Number = int(nbins)
+	if nbins == 0 {
+		bins.ChooseSturges()
+	}
+	hist := data.Histogram(x, bins)
+	groups = make([]string, len(hist))
+	counts = make([]float64, len(hist))
+	for i, b := range hist {
+		groups[i] = b.String()
+		counts[i] = float64(b.Count)
+	}
+	return
+}
+
+func histogram(xv []reflect.Value, c draw.Canvas, nbins uint) string {
+	groups, counts := histogramData(xv, nbins)
+	chart := plot.NewBarChart(c)
+	chart.BarPaddingX = 0
+	data := new(data.Table)
+	for _, g := range groups {
+		data.AddColumn(g)
+	}
+	data.AddRow(counts...)
+	return chart.Draw(data)
+}
diff --git a/cmd/jp/line.go b/cmd/jp/line.go
index caca49f..8054f87 100644
--- a/cmd/jp/line.go
+++ b/cmd/jp/line.go
@@ -1,47 +1,48 @@
 package main
 
 import (
+	"log"
 	"reflect"
 
+	"github.com/sgreben/jp/pkg/data"
 	"github.com/sgreben/jp/pkg/draw"
 
 	"github.com/sgreben/jp/pkg/plot"
 )
 
-func linePlotData(xvv, yvv [][]reflect.Value) (x, y []float64) {
-	for _, xv := range xvv {
-		for i := range xv {
-			if xv[i].IsValid() && xv[i].CanInterface() {
-				xvi, ok := xv[i].Interface().(float64)
-				if ok {
-					x = append(x, xvi)
-				}
+func linePlotData(xv, yv []reflect.Value) (x, y []float64) {
+	for i := range xv {
+		if xv[i].IsValid() && xv[i].CanInterface() {
+			xvi, ok := xv[i].Interface().(float64)
+			if ok {
+				x = append(x, xvi)
 			}
 		}
 	}
-	for _, yv := range yvv {
-		for i := range yv {
-			if yv[i].IsValid() && yv[i].CanInterface() {
-				yvi, ok := yv[i].Interface().(float64)
-				if ok {
-					y = append(y, yvi)
-				}
+	for i := range yv {
+		if yv[i].IsValid() && yv[i].CanInterface() {
+			yvi, ok := yv[i].Interface().(float64)
+			if ok {
+				y = append(y, yvi)
 			}
 		}
 	}
 	return
 }
 
-func linePlot(xvv, yvv [][]reflect.Value, c draw.Canvas) string {
-	x, y := linePlotData(xvv, yvv)
+func linePlot(xv, yv []reflect.Value, c draw.Canvas) string {
+	x, y := linePlotData(xv, yv)
 	chart := plot.NewLineChart(c)
-	data := new(plot.DataTable)
+	data := new(data.Table)
 	data.AddColumn("x")
 	data.AddColumn("y")
 	n := len(x)
 	if len(y) > n {
 		n = len(y)
 	}
+	if len(y) == 0 {
+		log.Fatal("no valid y values given")
+	}
 	// If no valid xs are given, use the indices as x values.
 	if len(x) == 0 {
 		x = make([]float64, len(y))
diff --git a/cmd/jp/main.go b/cmd/jp/main.go
index 62f3db0..173ac27 100644
--- a/cmd/jp/main.go
+++ b/cmd/jp/main.go
@@ -23,6 +23,7 @@ type configuration struct {
 	PlotType   enumVar
 	CanvasType enumVar
 	InputType  enumVar
+	HistBins   uint
 }
 
 const (
@@ -51,6 +52,7 @@ var config = configuration{
 			plotTypeLine,
 			plotTypeBar,
 			plotTypeScatter,
+			plotTypeHist,
 		},
 	},
 	CanvasType: enumVar{
@@ -86,8 +88,10 @@ func init() {
 	flag.StringVar(&config.XY, "xy", "", "x,y value pairs (JSONPath expression). Overrides -x and -y if given.")
 	flag.IntVar(&config.Box.Width, "width", 0, "Plot width (default 0 (auto))")
 	flag.IntVar(&config.Box.Height, "height", 0, "Plot height (default 0 (auto))")
+	flag.UintVar(&config.HistBins, "bins", 0, "Number of histogram bins (default 0 (auto))")
 	flag.Parse()
 	log.SetOutput(os.Stderr)
+	log.SetFlags(log.LstdFlags | log.Lshortfile)
 
 	var err error
 	xPattern = jsonpath.New("x")
@@ -143,12 +147,12 @@ func main() {
 		}
 		in = parseRows(rows)
 	}
-	var x, y [][]reflect.Value
+	var x, y []reflect.Value
 	if xyPattern != nil {
 		x, y = split(match(in, xyPattern))
 	} else {
-		x = match(in, xPattern)
-		y = match(in, yPattern)
+		x = flatten(match(in, xPattern))
+		y = flatten(match(in, yPattern))
 	}
 	buffer := draw.NewBuffer(config.Box)
 	var p draw.Pixels
@@ -162,7 +166,6 @@ func main() {
 	}
 	p.Clear()
 	c := draw.Canvas{Pixels: p}
-	fmt.Println()
 	switch config.PlotType.Value {
 	case plotTypeLine:
 		fmt.Println(linePlot(x, y, c))
@@ -170,5 +173,7 @@ func main() {
 		fmt.Println(scatterPlot(x, y, c))
 	case plotTypeBar:
 		fmt.Println(barPlot(x, y, c))
+	case plotTypeHist:
+		fmt.Println(histogram(x, c, config.HistBins))
 	}
 }
diff --git a/cmd/jp/scatter.go b/cmd/jp/scatter.go
index 9df8b4a..2fdf0e6 100644
--- a/cmd/jp/scatter.go
+++ b/cmd/jp/scatter.go
@@ -3,39 +3,35 @@ package main
 import (
 	"reflect"
 
+	"github.com/sgreben/jp/pkg/data"
 	"github.com/sgreben/jp/pkg/draw"
-
 	"github.com/sgreben/jp/pkg/plot"
 )
 
-func scatterPlotData(xvv, yvv [][]reflect.Value) (x, y []float64) {
-	for _, xv := range xvv {
-		for i := range xv {
-			if xv[i].IsValid() && xv[i].CanInterface() {
-				xvi, ok := xv[i].Interface().(float64)
-				if ok {
-					x = append(x, xvi)
-				}
+func scatterPlotData(xv, yv []reflect.Value) (x, y []float64) {
+	for i := range xv {
+		if xv[i].IsValid() && xv[i].CanInterface() {
+			xvi, ok := xv[i].Interface().(float64)
+			if ok {
+				x = append(x, xvi)
 			}
 		}
 	}
-	for _, yv := range yvv {
-		for i := range yv {
-			if yv[i].IsValid() && yv[i].CanInterface() {
-				yvi, ok := yv[i].Interface().(float64)
-				if ok {
-					y = append(y, yvi)
-				}
+	for i := range yv {
+		if yv[i].IsValid() && yv[i].CanInterface() {
+			yvi, ok := yv[i].Interface().(float64)
+			if ok {
+				y = append(y, yvi)
 			}
 		}
 	}
 	return
 }
 
-func scatterPlot(xvv, yvv [][]reflect.Value, c draw.Canvas) string {
-	x, y := scatterPlotData(xvv, yvv)
+func scatterPlot(xv, yv []reflect.Value, c draw.Canvas) string {
+	x, y := scatterPlotData(xv, yv)
 	chart := plot.NewScatterChart(c)
-	data := new(plot.DataTable)
+	data := new(data.Table)
 	data.AddColumn("x")
 	data.AddColumn("y")
 	n := len(x)
diff --git a/cmd/jp/split.go b/cmd/jp/split.go
index 4f36825..efe59c9 100644
--- a/cmd/jp/split.go
+++ b/cmd/jp/split.go
@@ -11,10 +11,10 @@ func flatten(in [][]reflect.Value) (out []reflect.Value) {
 	return
 }
 
-func split(in [][]reflect.Value) (x, y [][]reflect.Value) {
+func split(in [][]reflect.Value) (x, y []reflect.Value) {
 	flat := flatten(in)
 	n := len(flat)
-	x = [][]reflect.Value{flat[:n/2]}
-	y = [][]reflect.Value{flat[n/2:]}
+	x = flat[:n/2]
+	y = flat[n/2:]
 	return
 }