fix: first block num

Author: marino39

common.go

@@ -9,6 +9,7 @@ import (
 	"errors"
 	"fmt"
 	"io"
+	"math"
 	"os"
 	"path"
 	"sort"
@@ -22,6 +23,8 @@ import (
 	"github.com/c2h5oh/datasize"
 )
 
+const NoBlockNum = uint64(math.MaxUint64)
+
 type Dataset struct {
 	Name      string
 	Version   string

filter.go

@@ -43,8 +43,6 @@ func (o FilterBuilderOptions[T]) WithDefaults() FilterBuilderOptions[T] {
 }
 
 type filterBuilder[T any] struct {
-	ctx context.Context
-
 	indexes map[IndexName]Index[T]
 	fs      storage.FS
 }

filter_test.go

@@ -58,7 +58,7 @@ func generateMixedIntBlocks() []Block[[]int] {
 	// 45-49 generate 5 blocks with no data
 	// 50-69 generate 20 blocks with random but repeating huge numbers
 
-	for i := 1; i <= 20; i++ {
+	for i := 0; i <= 20; i++ {
 		blocks = append(blocks, Block[[]int]{
 			Hash:   common.BytesToHash([]byte{byte(i)}),
 			Number: uint64(i),
@@ -288,7 +288,7 @@ func TestIntMixFiltering(t *testing.T) {
 	}
 
 	onlyEvenResults := onlyEvenFilter.IndexIterator(context.Background())
-	assert.Len(t, onlyEvenResults.Bitmap().ToArray(), 20)
+	assert.Len(t, onlyEvenResults.Bitmap().ToArray(), 21)
 	for _, block := range onlyEvenResults.Bitmap().ToArray() {
 		assert.True(t, block <= 20)
 	}
@@ -373,7 +373,7 @@ func TestLowestIndexedBlockNum(t *testing.T) {
 	})
 	assert.NoError(t, err)
 	lowestBlockIndexed = indexer.BlockNum()
-	assert.Equal(t, uint64(0), lowestBlockIndexed)
+	assert.Equal(t, NoBlockNum, lowestBlockIndexed)
 	blocks := generateIntBlocks()
 	for _, block := range blocks[:50] {
 		err = indexer.Index(context.Background(), block)

index.go

@@ -50,7 +50,7 @@ func (u *IndexUpdate) Merge(update *IndexUpdate) {
 		u.BlockBitmap[indexValue].Or(bm)
 	}
 
-	if u.LastBlockNum < update.LastBlockNum {
+	if u.LastBlockNum == NoBlockNum || u.LastBlockNum < update.LastBlockNum {
 		u.LastBlockNum = update.LastBlockNum
 	}
 }
@@ -97,7 +97,7 @@ func (i *Index[T]) IndexBlock(ctx context.Context, fs storage.FS, block Block[T]
 			return nil, fmt.Errorf("unexpected: failed to get number of blocks indexed: %w", err)
 		}
 
-		if block.Number <= numBlocksIndexed {
+		if numBlocksIndexed != NoBlockNum && block.Number <= numBlocksIndexed {
 			return nil, nil
 		}
 	}
@@ -148,7 +148,7 @@ func (i *Index[T]) Store(ctx context.Context, fs storage.FS, indexUpdate *IndexU
 	if err != nil {
 		return fmt.Errorf("failed to get number of blocks indexed: %w", err)
 	}
-	if lastBlockNumIndexed >= indexUpdate.LastBlockNum {
+	if lastBlockNumIndexed != NoBlockNum && lastBlockNumIndexed >= indexUpdate.LastBlockNum {
 		return nil
 	}
 
@@ -169,12 +169,14 @@ func (i *Index[T]) Store(ctx context.Context, fs storage.FS, indexUpdate *IndexU
 
 		bmap.Or(bmUpdate)
 
+		fmt.Println("---- writing bm", i.name, indexValue, bmap.GetCardinality())
 		err = file.Write(ctx, bmap)
 		if err != nil {
 			return err
 		}
 	}
 
+	fmt.Println("---- writing last block num", i.name, indexUpdate.LastBlockNum)
 	err = i.storeLastBlockNumIndexed(ctx, fs, indexUpdate.LastBlockNum)
 	if err != nil {
 		return fmt.Errorf("failed to index number of blocks indexed: %w", err)
@@ -191,19 +193,19 @@ func (i *Index[T]) LastBlockNumIndexed(ctx context.Context, fs storage.FS) (uint
 	file, err := fs.Open(ctx, indexedBlockNumFilePath(string(i.name)), nil)
 	if err != nil {
 		// file doesn't exist
-		return 0, nil
+		return NoBlockNum, nil
 	}
 	defer file.Close()
 
 	buf, err := io.ReadAll(file)
 	if err != nil {
-		return 0, fmt.Errorf("failed to read IndexBlock file: %w", err)
+		return NoBlockNum, fmt.Errorf("failed to read IndexBlock file: %w", err)
 	}
 
 	var numBlocksIndexed uint64
 	err = binary.Read(bytes.NewReader(buf), binary.BigEndian, &numBlocksIndexed)
 	if err != nil {
-		return 0, fmt.Errorf("failed to unmarshal bitmap: %w", err)
+		return NoBlockNum, fmt.Errorf("failed to unmarshal bitmap: %w", err)
 	}
 
 	i.numBlocksIndexed = &atomic.Uint64{}
@@ -219,7 +221,7 @@ func (i *Index[T]) storeLastBlockNumIndexed(ctx context.Context, fs storage.FS,
 		prevBlockIndexed = blocksIndexed
 	}
 
-	if prevBlockIndexed >= numBlocksIndexed {
+	if prevBlockIndexed != NoBlockNum && prevBlockIndexed >= numBlocksIndexed {
 		return nil
 	}
 

indexer.go

@@ -134,16 +134,21 @@ func (i *Indexer[T]) BlockNum() uint64 {
 	i.mu.Lock()
 	defer i.mu.Unlock()
 
+	// initialize lowestBlockNum to math.MaxUint64
 	var lowestBlockNum uint64 = math.MaxUint64
 	for _, indexUpdate := range i.indexUpdates {
+		// if no blocks have been indexed, return NoBlockNum
+		if indexUpdate.LastBlockNum == NoBlockNum {
+			return NoBlockNum
+		}
+
+		// update lowestBlockNum if the current indexUpdate has a lower last block number
 		if indexUpdate.LastBlockNum < lowestBlockNum {
 			lowestBlockNum = indexUpdate.LastBlockNum
 		}
 	}
 
-	if lowestBlockNum == math.MaxUint64 {
-		return 0
-	}
+	// return the lowest block number indexed by all indexes
 	return lowestBlockNum
 }
 

reader.go

@@ -5,6 +5,7 @@ import (
 	"errors"
 	"fmt"
 	"io"
+	"math"
 	"os"
 	"sync"
 	"time"
@@ -29,10 +30,9 @@ type Reader[T any] interface {
 }
 
 type reader[T any] struct {
-	options        Options
-	path           string
-	fs             storage.FS
-	useCompression bool
+	options Options
+	path    string
+	fs      storage.FS
 
 	closer io.Closer
 
@@ -140,8 +140,11 @@ func (r *reader[T]) Read(ctx context.Context) (Block[T], error) {
 		}
 	}
 
-	var block Block[T]
-	for structs.IsZero(block) || block.Number <= r.lastBlockNum {
+	var (
+		block                    Block[T]
+		hasSuccessfullyReadBlock bool
+	)
+	for !hasSuccessfullyReadBlock || block.Number <= r.lastBlockNum && r.lastBlockNum != math.MaxUint64 {
 		select {
 		case <-ctx.Done():
 			return Block[T]{}, ctx.Err()
@@ -168,6 +171,8 @@ func (r *reader[T]) Read(ctx context.Context) (Block[T], error) {
 					r.lastBlockNum = block.Number
 				}
 
+				hasSuccessfullyReadBlock = true
+
 				if !r.isBlockWithin(block) {
 					currentFile := r.fileIndex.At(r.currFileIndex)
 					return Block[T]{}, fmt.Errorf("block number %d is out of file block %d-%d range",
@@ -181,6 +186,8 @@ func (r *reader[T]) Read(ctx context.Context) (Block[T], error) {
 			return Block[T]{}, fmt.Errorf("failed to decode file data: %w", err)
 		}
 
+		hasSuccessfullyReadBlock = true
+
 		if !r.isBlockWithin(block) {
 			currentFile := r.fileIndex.At(r.currFileIndex)
 			return Block[T]{}, fmt.Errorf("block number %d is out of file block %d-%d range",

writer.go

@@ -5,6 +5,7 @@ import (
 	"context"
 	"fmt"
 	"io"
+	"math"
 	"os"
 	"sync"
 
@@ -77,18 +78,26 @@ func NewWriter[T any](opt Options) (Writer[T], error) {
 		return nil, fmt.Errorf("failed to load file index: %w", err)
 	}
 
+	var firstBlockNum uint64
 	var lastBlockNum uint64
 	var fileIndexFileList = fileIndex.Files()
 	if len(fileIndexFileList) > 0 {
 		lastBlockNum = fileIndexFileList[len(fileIndexFileList)-1].LastBlockNum
 	}
 
+	if lastBlockNum == 0 {
+		firstBlockNum = NoBlockNum
+		lastBlockNum = NoBlockNum
+	} else {
+		firstBlockNum = lastBlockNum + 1
+	}
+
 	// create new writer
 	return &writer[T]{
 		options:       opt,
 		path:          datasetPath,
 		fs:            fs,
-		firstBlockNum: lastBlockNum + 1,
+		firstBlockNum: firstBlockNum,
 		lastBlockNum:  lastBlockNum,
 		fileIndex:     fileIndex,
 		buffer:        bytes.NewBuffer(make([]byte, 0, defaultFileSize)),
@@ -103,7 +112,7 @@ func (w *writer[T]) Write(ctx context.Context, b Block[T]) error {
 	w.mu.Lock()
 	defer w.mu.Unlock()
 
-	if w.lastBlockNum >= b.Number {
+	if w.lastBlockNum != math.MaxUint64 && w.lastBlockNum >= b.Number {
 		return nil
 	}
 
@@ -118,6 +127,10 @@ func (w *writer[T]) Write(ctx context.Context, b Block[T]) error {
 		return fmt.Errorf("failed to encode file data: %w", err)
 	}
 
+	if w.firstBlockNum == NoBlockNum {
+		w.firstBlockNum = b.Number
+	}
+
 	w.lastBlockNum = b.Number
 	w.options.FileRollPolicy.onBlockProcessed(w.lastBlockNum)
 	return nil

writer_no_gap.go

@@ -2,6 +2,7 @@ package ethwal
 
 import (
 	"context"
+	"fmt"
 
 	"github.com/0xsequence/ethwal/storage"
 )
@@ -13,7 +14,8 @@ type noGapWriter[T any] struct {
 }
 
 func NewWriterNoGap[T any](w Writer[T]) Writer[T] {
-	return &noGapWriter[T]{w: w}
+	fmt.Println("NewWriterNoGap", w.BlockNum())
+	return &noGapWriter[T]{w: w, lastBlockNum: w.BlockNum()}
 }
 
 func (n *noGapWriter[T]) FileSystem() storage.FS {
@@ -24,7 +26,7 @@ func (n *noGapWriter[T]) Write(ctx context.Context, b Block[T]) error {
 	defer func() { n.lastBlockNum = b.Number }()
 
 	// skip if block number is less than or equal to last block number
-	if b.Number <= n.lastBlockNum {
+	if n.lastBlockNum != NoBlockNum && b.Number <= n.lastBlockNum {
 		return nil
 	}
 
@@ -40,6 +42,7 @@ func (n *noGapWriter[T]) Write(ctx context.Context, b Block[T]) error {
 			return err
 		}
 	}
+
 	return n.w.Write(ctx, b)
 }