From 45e1847513fde2c1a680fd1b384c4efadc108e51 Mon Sep 17 00:00:00 2001
From: rene <41963722+renaynay@users.noreply.github.com>
Date: Thu, 4 Jan 2024 16:37:21 +0100
Subject: [PATCH] feat(nodebuilder/prune)!: Enable sampling window for light
 nodes (#2991)

This PR enforces a sampling window of 30 days' worth of seconds for
light nodes. The DASer will now skip over sampling headers that fall
outside this range.

I've labeled this PR as breaking as it contains a behavioural break
(sampling constrained to a 30-day window rather than all historical
headers).
---
 das/daser.go                |  2 ++
 das/daser_test.go           | 37 +++++++++++++++++++++++++++++++++++++
 nodebuilder/prune/module.go |  2 +-
 pruner/light/window.go      |  2 ++
 4 files changed, 42 insertions(+), 1 deletion(-)

diff --git a/das/daser.go b/das/daser.go
index 40eee3d316..7d569f7e0b 100644
--- a/das/daser.go
+++ b/das/daser.go
@@ -151,6 +151,8 @@ func (d *DASer) sample(ctx context.Context, h *header.ExtendedHeader) error {
 	// short-circuit if pruning is enabled and the header is outside the
 	// availability window
 	if !d.isWithinSamplingWindow(h) {
+		log.Debugw("skipping header outside sampling window", "height", h.Height(),
+			"time", h.Time())
 		return nil
 	}
 
diff --git a/das/daser_test.go b/das/daser_test.go
index fd1eb39f7d..9eec6392cc 100644
--- a/das/daser_test.go
+++ b/das/daser_test.go
@@ -2,6 +2,7 @@ package das
 
 import (
 	"context"
+	"strconv"
 	"testing"
 	"time"
 
@@ -244,6 +245,42 @@ func TestDASerSampleTimeout(t *testing.T) {
 	}
 }
 
+// TestDASer_SamplingWindow tests the sampling window determination
+// for headers.
+func TestDASer_SamplingWindow(t *testing.T) {
+	ds := ds_sync.MutexWrap(datastore.NewMapDatastore())
+	sub := new(headertest.Subscriber)
+	fserv := &fraudtest.DummyService[*header.ExtendedHeader]{}
+	getter := getterStub{}
+	avail := mocks.NewMockAvailability(gomock.NewController(t))
+
+	// create and start DASer
+	daser, err := NewDASer(avail, sub, getter, ds, fserv, newBroadcastMock(1),
+		WithSamplingWindow(time.Second))
+	require.NoError(t, err)
+
+	var tests = []struct {
+		timestamp    time.Time
+		withinWindow bool
+	}{
+		{timestamp: time.Now().Add(-(time.Second * 5)), withinWindow: false},
+		{timestamp: time.Now().Add(-(time.Millisecond * 800)), withinWindow: true},
+		{timestamp: time.Now().Add(-(time.Hour)), withinWindow: false},
+		{timestamp: time.Now().Add(-(time.Hour * 24 * 30)), withinWindow: false},
+		{timestamp: time.Now(), withinWindow: true},
+	}
+
+	for i, tt := range tests {
+		t.Run(strconv.Itoa(i), func(t *testing.T) {
+			eh := headertest.RandExtendedHeader(t)
+			eh.RawHeader.Time = tt.timestamp
+
+			assert.Equal(t, tt.withinWindow, daser.isWithinSamplingWindow(eh))
+		})
+	}
+
+}
+
 // createDASerSubcomponents takes numGetter (number of headers
 // to store in mockGetter) and numSub (number of headers to store
 // in the mock header.Subscriber), returning a newly instantiated
diff --git a/nodebuilder/prune/module.go b/nodebuilder/prune/module.go
index 330ef21cdc..2141b74bf1 100644
--- a/nodebuilder/prune/module.go
+++ b/nodebuilder/prune/module.go
@@ -39,7 +39,7 @@ func ConstructModule(tp node.Type) fx.Option {
 			fx.Provide(func() pruner.Pruner {
 				return light.NewPruner()
 			}),
-			fx.Supply(archival.Window), // TODO @renaynay: turn this into light.Window in following PR
+			fx.Supply(light.Window),
 		)
 	default:
 		panic("unknown node type")
diff --git a/pruner/light/window.go b/pruner/light/window.go
index 53bfe4a163..dc1a9e4444 100644
--- a/pruner/light/window.go
+++ b/pruner/light/window.go
@@ -6,4 +6,6 @@ import (
 	"github.com/celestiaorg/celestia-node/pruner"
 )
 
+// Window is the availability window for light nodes in the Celestia
+// network (30 days).
 const Window = pruner.AvailabilityWindow(time.Second * 86400 * 30)