refactored event store

.gitignore

@@ -4,5 +4,10 @@
 # ignore all .env
 **/*.env
 
+# ignore all sqlite files
+**/*.sqlite
+**/*-shm
+**/*-wal
+
 # ignore the debugs
 .vscode

cmd/config.go

@@ -47,7 +47,7 @@ type Config struct {
 	Firehose pipe.FirehoseConfig
 	Fetcher  pipe.FetcherConfig
 	Arbiter  pipe.ArbiterConfig
-	Processor pipe.ProcessorConfig
+	Engine   pipe.EngineConfig
 }
 
 // NewConfig returns a config with default parameters
@@ -57,7 +57,7 @@ func NewConfig() *Config {
 		Firehose:     pipe.NewFirehoseConfig(),
 		Fetcher:      pipe.NewFetcherConfig(),
 		Arbiter:      pipe.NewArbiterConfig(),
-		Processor:    pipe.NewProcessorConfig(),
+		Engine:       pipe.NewEngineConfig(),
 	}
 }
 
@@ -66,7 +66,7 @@ func (c *Config) Print() {
 	c.Firehose.Print()
 	c.Fetcher.Print()
 	c.Arbiter.Print()
-	c.Processor.Print()
+	c.Engine.Print()
 }
 
 // LoadConfig reads the enviroment variables and parses them into a [Config] struct
@@ -174,14 +174,26 @@ func LoadConfig() (*Config, error) {
 			}
 			config.Arbiter.PromotionWait = time.Duration(wait) * time.Second
 
-		case "PROCESSOR_CACHE_CAPACITY":
-			config.Processor.CacheCapacity, err = strconv.Atoi(val)
+		case "ENGINE_PRINT_EVERY":
+			config.Engine.PrintEvery, err = strconv.Atoi(val)
 			if err != nil {
 				return nil, fmt.Errorf("error parsing %v: %v", keyVal, err)
 			}
 
-		case "PROCESSOR_PRINT_EVERY":
-			config.Processor.PrintEvery, err = strconv.Atoi(val)
+		case "ENGINE_UPDATER_CAPACITY":
+			config.Engine.UpdaterCapacity, err = strconv.Atoi(val)
+			if err != nil {
+				return nil, fmt.Errorf("error parsing %v: %v", keyVal, err)
+			}
+
+		case "ENGINE_CACHE_CAPACITY":
+			config.Engine.CacheCapacity, err = strconv.Atoi(val)
+			if err != nil {
+				return nil, fmt.Errorf("error parsing %v: %v", keyVal, err)
+			}
+
+		case "ENGINE_ARCHIVE_CAPACITY":
+			config.Engine.ArchiverCapacity, err = strconv.Atoi(val)
 			if err != nil {
 				return nil, fmt.Errorf("error parsing %v: %v", keyVal, err)
 			}

cmd/crawler.go

@@ -17,6 +17,7 @@ import (
 
 	"github.com/nbd-wtf/go-nostr"
 	"github.com/redis/go-redis/v9"
+	"github.com/vertex-lab/relay/pkg/eventstore"
 )
 
 func main() {
@@ -32,6 +33,11 @@ func main() {
 	events := make(chan *nostr.Event, config.EventsCapacity)
 	pubkeys := make(chan string, config.PubkeysCapacity)
 
+	store, err := eventstore.New(config.SQLiteURL)
+	if err != nil {
+		panic(err)
+	}
+
 	db := redb.New(&redis.Options{Addr: config.RedisAddress})
 	count, err := db.NodeCount(ctx)
 	if err != nil {
@@ -63,30 +69,36 @@ func main() {
 		log.Printf("correctly added %d init pubkeys", len(config.InitPubkeys))
 	}
 
-	var wg sync.WaitGroup
-	wg.Add(4)
+	var producers sync.WaitGroup
+	var consumers sync.WaitGroup
 
+	producers.Add(3)
 	go func() {
-		defer wg.Done()
+		defer producers.Done()
 		pipe.Firehose(ctx, config.Firehose, db, enqueue(events))
 	}()
 
 	go func() {
-		defer wg.Done()
+		defer producers.Done()
 		pipe.Fetcher(ctx, config.Fetcher, pubkeys, enqueue(events))
 	}()
 
 	go func() {
-		defer wg.Done()
+		defer producers.Done()
 		pipe.Arbiter(ctx, config.Arbiter, db, enqueue(pubkeys))
+		close(pubkeys) // Arbiter is the only pubkey sender
 	}()
 
+	consumers.Add(1)
 	go func() {
-		defer wg.Done()
-		pipe.Processor(ctx, config.Processor, db, events)
+		defer consumers.Done()
+		pipe.Engine(ctx, config.Engine, store, db, events)
 	}()
 
-	wg.Wait()
+	producers.Wait()
+	close(events)
+
+	consumers.Wait()
 }
 
 // handleSignals listens for OS signals and triggers context cancellation.

go.mod

@@ -8,6 +8,7 @@ require (
 	github.com/joho/godotenv v1.5.1
 	github.com/nbd-wtf/go-nostr v0.51.12
 	github.com/redis/go-redis/v9 v9.8.0
+	github.com/vertex-lab/relay v0.4.5
 )
 
 require (
@@ -26,6 +27,7 @@ require (
 	github.com/json-iterator/go v1.1.12 // indirect
 	github.com/klauspost/cpuid/v2 v2.2.10 // indirect
 	github.com/mailru/easyjson v0.9.0 // indirect
+	github.com/mattn/go-sqlite3 v1.14.24 // indirect
 	github.com/modern-go/concurrent v0.0.0-20180306012644-bacd9c7ef1dd // indirect
 	github.com/modern-go/reflect2 v1.0.2 // indirect
 	github.com/puzpuzpuz/xsync/v3 v3.5.1 // indirect

go.sum

@@ -43,6 +43,8 @@ github.com/klauspost/cpuid/v2 v2.2.10/go.mod h1:hqwkgyIinND0mEev00jJYCxPNVRVXFQe
 github.com/knz/go-libedit v1.10.1/go.mod h1:MZTVkCWyz0oBc7JOWP3wNAzd002ZbM/5hgShxwh4x8M=
 github.com/mailru/easyjson v0.9.0 h1:PrnmzHw7262yW8sTBwxi1PdJA3Iw/EKBa8psRf7d9a4=
 github.com/mailru/easyjson v0.9.0/go.mod h1:1+xMtQp2MRNVL/V1bOzuP3aP8VNwRW55fQUto+XFtTU=
+github.com/mattn/go-sqlite3 v1.14.24 h1:tpSp2G2KyMnnQu99ngJ47EIkWVmliIizyZBfPrBWDRM=
+github.com/mattn/go-sqlite3 v1.14.24/go.mod h1:Uh1q+B4BYcTPb+yiD3kU8Ct7aC0hY9fxUwlHK0RXw+Y=
 github.com/modern-go/concurrent v0.0.0-20180228061459-e0a39a4cb421/go.mod h1:6dJC0mAP4ikYIbvyc7fijjWJddQyLn8Ig3JB5CqoB9Q=
 github.com/modern-go/concurrent v0.0.0-20180306012644-bacd9c7ef1dd h1:TRLaZ9cD/w8PVh93nsPXa1VrQ6jlwL5oN8l14QlcNfg=
 github.com/modern-go/concurrent v0.0.0-20180306012644-bacd9c7ef1dd/go.mod h1:6dJC0mAP4ikYIbvyc7fijjWJddQyLn8Ig3JB5CqoB9Q=
@@ -76,6 +78,8 @@ github.com/tidwall/pretty v1.2.1 h1:qjsOFOWWQl+N3RsoF5/ssm1pHmJJwhjlSbZ51I6wMl4=
 github.com/tidwall/pretty v1.2.1/go.mod h1:ITEVvHYasfjBbM0u2Pg8T2nJnzm8xPwvNhhsoaGGjNU=
 github.com/twitchyliquid64/golang-asm v0.15.1 h1:SU5vSMR7hnwNxj24w34ZyCi/FmDZTkS4MhqMhdFk5YI=
 github.com/twitchyliquid64/golang-asm v0.15.1/go.mod h1:a1lVb/DtPvCB8fslRZhAngC2+aY1QWCk3Cedj/Gdt08=
+github.com/vertex-lab/relay v0.4.5 h1:6elykMvZFV+y/JYNhaNo3+pxbK66QnrmAIv3/3CN+VU=
+github.com/vertex-lab/relay v0.4.5/go.mod h1:K3Utw2y0FhhDigT86A2gc/ZqK6MrF06mw6p0CihD4I4=
 golang.org/x/arch v0.15.0 h1:QtOrQd0bTUnhNVNndMpLHNWrDmYzZ2KDqSrEymqInZw=
 golang.org/x/arch v0.15.0/go.mod h1:JmwW7aLIoRUKgaTzhkiEFxvcEiQGyOg9BMonBJUS7EE=
 golang.org/x/exp v0.0.0-20250305212735-054e65f0b394 h1:nDVHiLt8aIbd/VzvPWN6kSOPE7+F/fNFDSXLVYkE/Iw=

pkg/pipe/intake.go

@@ -101,10 +101,11 @@ func (b *buffer) Contains(ID string) bool {
 // Firehose connects to a list of relays and pulls [relevantKinds] events that are newer than [FirehoseConfig.Since].
 // It discards events from unknown pubkeys as an anti-spam mechanism.
 func Firehose(ctx context.Context, config FirehoseConfig, check PubkeyChecker, send func(*nostr.Event) error) {
-	pool := nostr.NewSimplePool(ctx)
-	defer close(pool)
 	defer log.Println("Firehose: shutting down...")
 
+	pool := nostr.NewSimplePool(ctx)
+	defer shutdown(pool)
+
 	filter := nostr.Filter{
 		Kinds: relevantKinds,
 		Since: config.Since(),
@@ -160,19 +161,24 @@ func (c FetcherConfig) Print() {
 // - when the batch is bigger than config.Batch
 // - after config.Interval since the last query.
 func Fetcher(ctx context.Context, config FetcherConfig, pubkeys <-chan string, send func(*nostr.Event) error) {
+	defer log.Println("Fetcher: shutting down...")
+
 	batch := make([]string, 0, config.Batch)
 	timer := time.After(config.Interval)
 
 	pool := nostr.NewSimplePool(ctx)
-	defer close(pool)
+	defer shutdown(pool)
 
 	for {
 		select {
 		case <-ctx.Done():
-			log.Println("Fetcher: shutting down...")
 			return
 
-		case pubkey := <-pubkeys:
+		case pubkey, ok := <-pubkeys:
+			if !ok {
+				return
+			}
+
 			batch = append(batch, pubkey)
 			if len(batch) < config.Batch {
 				continue
@@ -244,8 +250,8 @@ func fetch(ctx context.Context, pool *nostr.SimplePool, relays, pubkeys []string
 	return events, nil
 }
 
-// Close iterates over the relays in the pool and closes all connections.
-func close(pool *nostr.SimplePool) {
+// Shutdown iterates over the relays in the pool and closes all connections.
+func shutdown(pool *nostr.SimplePool) {
 	pool.Relays.Range(func(_ string, relay *nostr.Relay) bool {
 		relay.Close()
 		return true

pkg/pipe/processor.go

@@ -10,39 +10,140 @@ import (
 	"github/pippellia-btc/crawler/pkg/walks"
 	"log"
 	"slices"
+	"sync/atomic"
 	"time"
 
 	"github.com/nbd-wtf/go-nostr"
+	"github.com/vertex-lab/relay/pkg/eventstore"
 )
 
+// EventTracker tracks the number of events processed
+var EventTracker atomic.Int32
+
 var ErrUnsupportedKind = errors.New("unsupported event kind")
 
-type ProcessorConfig struct {
-	CacheCapacity int
+type EngineConfig struct {
 	PrintEvery int
+
+	// for the GraphUpdater
+	UpdaterCapacity int
+	CacheCapacity   int
+
+	// for the archiveEngine
+	ArchiverCapacity int
 }
 
-func NewProcessorConfig() ProcessorConfig {
-	return ProcessorConfig{
-		CacheCapacity: 10000,
-		PrintEvery:    5000}
+func NewEngineConfig() EngineConfig {
+	return EngineConfig{
+		PrintEvery:       5000,
+		UpdaterCapacity:  1000,
+		CacheCapacity:    100_000,
+		ArchiverCapacity: 1000,
+	}
 }
 
-func (c ProcessorConfig) Print() {
-	fmt.Printf("Processor\n")
-	fmt.Printf("  CacheCapacity: %d\n", c.CacheCapacity)
+func (c EngineConfig) Print() {
+	fmt.Printf("Engine\n")
 	fmt.Printf("  PrintEvery: %d\n", c.PrintEvery)
+	fmt.Printf("  UpdaterCapacity: %d\n", c.UpdaterCapacity)
+	fmt.Printf("  CacheCapacity: %d\n", c.CacheCapacity)
+	fmt.Printf("  ArchiveCapacity: %d\n", c.ArchiverCapacity)
 }
 
-func Processor(
+// Engine is responsible for dispacting the correct events to the [Archiver] or [GraphUpdater].
+func Engine(
 	ctx context.Context,
-	config ProcessorConfig,
+	config EngineConfig,
+	store *eventstore.Store,
 	db redb.RedisDB,
-	//store *eventstore.Store,
 	events chan *nostr.Event) {
 
-	var err error
-	var processed int
+	defer log.Println("Engine: shutting down...")
+
+	graphEvents := make(chan *nostr.Event, config.UpdaterCapacity)
+	archiveEvents := make(chan *nostr.Event, config.ArchiverCapacity)
+	defer close(graphEvents)
+	defer close(archiveEvents)
+
+	go GraphUpdater(ctx, config, store, db, graphEvents)
+	go Archiver(ctx, config, store, archiveEvents)
+
+	log.Println("Engine: ready to process events")
+
+	for {
+		select {
+		case <-ctx.Done():
+			return
+
+		case event, ok := <-events:
+			if !ok {
+				return
+			}
+
+			switch event.Kind {
+			case nostr.KindFollowList:
+				graphEvents <- event
+
+			case nostr.KindProfileMetadata:
+				archiveEvents <- event
+
+			default:
+				logEvent(event, ErrUnsupportedKind)
+			}
+		}
+	}
+}
+
+// Archiver consumes events that are not graph-related and stores them.
+func Archiver(
+	ctx context.Context,
+	config EngineConfig,
+	store *eventstore.Store,
+	events chan *nostr.Event) {
+
+	for {
+		select {
+		case <-ctx.Done():
+			return
+
+		case event, ok := <-events:
+			if !ok {
+				return
+			}
+
+			err := func() error {
+				opctx, cancel := context.WithTimeout(context.Background(), 2*time.Second)
+				defer cancel()
+
+				switch {
+				case nostr.IsRegularKind(event.Kind):
+					return store.Save(opctx, event)
+
+				case nostr.IsReplaceableKind(event.Kind):
+					_, err := store.Replace(opctx, event)
+					return err
+
+				default:
+					return nil
+				}
+			}()
+
+			if err != nil {
+				logEvent(event, err)
+			}
+
+			EventTracker.Add(1)
+		}
+	}
+}
+
+// GraphUpdater consumes events to update the graph and random walks.
+func GraphUpdater(
+	ctx context.Context,
+	config EngineConfig,
+	store *eventstore.Store,
+	db redb.RedisDB,
+	events chan *nostr.Event) {
 
 	cache := walks.NewWalker(
 		walks.WithCapacity(config.CacheCapacity),
@@ -50,34 +151,36 @@ func Processor(
 		walks.WithLogFile("cache.log"),
 	)
 
-	log.Println("Processor: ready to process events")
-
 	for {
 		select {
 		case <-ctx.Done():
-			log.Println("Processor: shutting down...")
 			return
 
-		case event := <-events:
-			switch event.Kind {
-			case nostr.KindFollowList:
-				err = processFollowList(cache, db, event)
-
-			case nostr.KindProfileMetadata:
-				err = nil
-
-			default:
-				err = ErrUnsupportedKind
+		case event, ok := <-events:
+			if !ok {
+				return
 			}
 
+			err := func() error {
+				opctx, cancel := context.WithTimeout(context.Background(), 15*time.Second)
+				defer cancel()
+
+				replaced, err := store.Replace(opctx, event)
+				if err != nil {
+					return err
+				}
+
+				if replaced {
+					return processFollowList(opctx, db, cache, event)
+				}
+				return nil
+			}()
+
 			if err != nil {
-				log.Printf("Processor: event ID %s, kind %d by %s: %v", event.ID, event.Kind, event.PubKey, err)
+				logEvent(event, err)
 			}
 
-			processed++
-			if processed%config.PrintEvery == 0 {
-				log.Printf("Processor: processed %d events", processed)
-			}
+			EventTracker.Add(1)
 		}
 	}
 }
@@ -85,10 +188,7 @@ func Processor(
 // processFollowList parses the pubkeys listed in the event, and uses them to:
 // - update the follows of the author (db and cache)
 // - update the author's random walks and signal the number to the [WalksTracker]
-func processFollowList(cache *walks.CachedWalker, db redb.RedisDB, event *nostr.Event) error {
-	ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
-	defer cancel()
-
+func processFollowList(ctx context.Context, db redb.RedisDB, cache *walks.CachedWalker, event *nostr.Event) error {
 	author, err := db.NodeByKey(ctx, event.PubKey)
 	if err != nil {
 		return err
@@ -173,6 +273,11 @@ func parsePubkeys(event *nostr.Event) []string {
 	return unique(pubkeys)
 }
 
+func logEvent(e *nostr.Event, extra any) {
+	msg := fmt.Sprintf("Engine: event ID %s, kind %d by %s: ", e.ID, e.Kind, e.PubKey)
+	log.Printf(msg+"%v", extra)
+}
+
 // Unique returns a slice of unique elements of the input slice.
 func unique[E cmp.Ordered](slice []E) []E {
 	if len(slice) == 0 {

pkg/redb/graph.go

@@ -447,9 +447,13 @@ func (db RedisDB) Pubkeys(ctx context.Context, nodes ...graph.ID) ([]string, err
 
 type MissingHandler func(ctx context.Context, db RedisDB, pubkey string) (graph.ID, error)
 
+// Ignore pubkeys that are not found
 func Ignore(context.Context, RedisDB, string) (graph.ID, error) { return "", nil }
+
+// Return a sentinel value ("-1") as the node ID of pubkeys not found
 func Sentinel(context.Context, RedisDB, string) (graph.ID, error) { return "-1", nil }
 
+// AddValid pubkeys to the database if they were not already present
 func AddValid(ctx context.Context, db RedisDB, pubkey string) (graph.ID, error) {
 	if !nostr.IsValidPublicKey(pubkey) {
 		return "", nil