feat: use *array.Boolean to store page views

columns/columns.go

@@ -11,3 +11,4 @@ var Session = camel.Case(v1.Filters_Session.String())
 var Bounce = camel.Case(v1.Filters_Bounce.String())
 var Duration = camel.Case(v1.Filters_Duration.String())
 var Timestamp = camel.Case(v1.Filters_Timestamp.String())
+var View = camel.Case(v1.Filters_View.String())

proto/staples/v1/scan.proto

@@ -33,6 +33,7 @@ message Filters {
     Bounce = 2;
     Session = 3;
     Duration = 4;
+    View = 26;
 
     Browser = 5;
     BrowserVersion = 6;

staples/event.go

@@ -32,6 +32,7 @@ type Event struct {
 	// stayed and browsed the website.
 	Bounce   *bool
 	Session  bool
+	View     bool
 	Duration float64
 
 	Browser        string
@@ -100,10 +101,16 @@ func (e *Event) Release() {
 
 func (e *Event) TS() int64 { return e.Timestamp }
 
+const pageView = "pageview"
+
 func (e *Event) Hit() {
 	e.EntryPage = e.Path
 	e.Bounce = True
 	e.Session = true
+	if e.Event == pageView {
+		e.Event = ""
+		e.View = true
+	}
 }
 
 func (s *Event) Update(e *Event) {

staples/index.go

@@ -7,10 +7,9 @@ import (
 	"github.com/RoaringBitmap/roaring"
 	"github.com/apache/arrow/go/v15/arrow"
 	"github.com/apache/arrow/go/v15/arrow/array"
-	"github.com/vinceanalytics/vince/camel"
+	"github.com/vinceanalytics/vince/columns"
 	"github.com/vinceanalytics/vince/db"
 	"github.com/vinceanalytics/vince/filters"
-	v1 "github.com/vinceanalytics/vince/gen/go/staples/v1"
 	"github.com/vinceanalytics/vince/index"
 	"github.com/vinceanalytics/vince/logger"
 )
@@ -24,11 +23,12 @@ func NewIndex() *Index {
 var _ index.Index = (*Index)(nil)
 
 var skip = map[string]bool{
-	camel.Case(v1.Filters_Timestamp.String()): true,
-	camel.Case(v1.Filters_ID.String()):        true,
-	camel.Case(v1.Filters_Session.String()):   true,
-	camel.Case(v1.Filters_Bounce.String()):    true,
-	camel.Case(v1.Filters_Duration.String()):  true,
+	columns.Timestamp: true,
+	columns.ID:        true,
+	columns.Session:   true,
+	columns.Bounce:    true,
+	columns.Duration:  true,
+	columns.View:      true,
 }
 
 func (idx *Index) Index(r arrow.Record) (index.Full, error) {

stats/compute.go

@@ -63,6 +63,7 @@ func (c *Compute) Metric(ctx context.Context, m v1.Metric) (float64, error) {
 func (c *Compute) Events() float64 {
 	return float64(c.mapping[columns.Event].Len())
 }
+
 func (c *Compute) ViewsPerVisits() float64 {
 	views := c.PageView()
 	visits := c.Visits()
@@ -121,27 +122,11 @@ func (c *Compute) PageView() float64 {
 	if c.view != nil {
 		return *c.view
 	}
-	view := calcPageViews(c.mapping[columns.Event])
+	view := float64(countSetBits(c.mapping[columns.View].(*array.Boolean)))
 	c.view = &view
 	return view
 }
 
-const viewStr = "pageview"
-
-func calcPageViews(a arrow.Array) (n float64) {
-	d := a.(*array.Dictionary)
-	x := d.Dictionary().(*array.String)
-	for i := 0; i < d.Len(); i++ {
-		if d.IsNull(i) {
-			continue
-		}
-		if x.Value(d.GetValueIndex(i)) == viewStr {
-			n++
-		}
-	}
-	return
-}
-
 func metricsToProjection(f *v1.Filters, me []v1.Metric, props ...v1.Property) []string {
 	m := make(map[v1.Filters_Projection]struct{})
 	m[v1.Filters_Timestamp] = struct{}{}
@@ -151,7 +136,7 @@ func metricsToProjection(f *v1.Filters, me []v1.Metric, props ...v1.Property) []
 	for _, v := range me {
 		switch v {
 		case v1.Metric_pageviews:
-			m[v1.Filters_Event] = struct{}{}
+			m[v1.Filters_View] = struct{}{}
 		case v1.Metric_visitors:
 			m[v1.Filters_ID] = struct{}{}
 		case v1.Metric_visits:
@@ -160,10 +145,13 @@ func metricsToProjection(f *v1.Filters, me []v1.Metric, props ...v1.Property) []
 			m[v1.Filters_Session] = struct{}{}
 			m[v1.Filters_Bounce] = struct{}{}
 		case v1.Metric_visit_duration:
+			m[v1.Filters_Session] = struct{}{}
 			m[v1.Filters_Duration] = struct{}{}
 		case v1.Metric_views_per_visit:
-			m[v1.Filters_Event] = struct{}{}
+			m[v1.Filters_View] = struct{}{}
 			m[v1.Filters_Duration] = struct{}{}
+		case v1.Metric_events:
+			m[v1.Filters_Event] = struct{}{}
 		}
 	}
 	cols := make([]string, 0, len(m))