Record each snapshot's new chunk size when its backup runs (closes #311)

snapshot list worked out NEW CHUNK SIZE from the snapshot's file rows. A
file's row is shared by every snapshot that lists it and rewritten when
the file changes, so a snapshot whose change was a modified file showed
0 B, and older snapshots' figures moved after later backups.

The scanner now totals the size of the chunks it stores that were not
stored before (the chunks chunk_count counts), the backup writes that
total to a new snapshots column, new_chunk_size, and snapshot list reads
it.

Judgement call: the figure is fixed when the backup runs, so removing an
earlier snapshot no longer raises a later snapshot's figure.

Model: opus-5-5

internal/database/schema/001.sql

@@ -86,6 +86,7 @@ CREATE TABLE IF NOT EXISTS snapshots (
     completed_at INTEGER,
     file_count INTEGER NOT NULL DEFAULT 0,
     chunk_count INTEGER NOT NULL DEFAULT 0,
+    new_chunk_size INTEGER NOT NULL DEFAULT 0,
     blob_count INTEGER NOT NULL DEFAULT 0,
     total_size INTEGER NOT NULL DEFAULT 0,
     blob_size INTEGER NOT NULL DEFAULT 0,

internal/database/snapshots.go

@@ -129,6 +129,7 @@ func (r *SnapshotRepository) UpdateExtendedStats(
 	compressionLevel int,
 	uploadBytes int64,
 	uploadDurationMs int64,
+	newChunkSize int64,
 ) error {
 	compressionRatio, err := r.extendedCompressionRatio(
 		ctx, tx, snapshotID, blobUncompressedSize,
@@ -143,18 +144,19 @@ func (r *SnapshotRepository) UpdateExtendedStats(
 		    compression_ratio = ?,
 		    compression_level = ?,
 		    upload_bytes = ?,
-		    upload_duration_ms = ?
+		    upload_duration_ms = ?,
+		    new_chunk_size = ?
 		WHERE id = ?
 	`
 
 	if tx != nil {
 		_, err = tx.ExecContext(ctx, query,
 			blobUncompressedSize, compressionRatio, compressionLevel,
-			uploadBytes, uploadDurationMs, snapshotID)
+			uploadBytes, uploadDurationMs, newChunkSize, snapshotID)
 	} else {
 		_, err = r.db.ExecWithLog(ctx, query,
 			blobUncompressedSize, compressionRatio, compressionLevel,
-			uploadBytes, uploadDurationMs, snapshotID)
+			uploadBytes, uploadDurationMs, newChunkSize, snapshotID)
 	}
 
 	if err != nil {
@@ -594,45 +596,24 @@ func (r *SnapshotRepository) GetSnapshotUncompressedChunkSize(
 	return totalSize, nil
 }
 
-// GetSnapshotNewChunkSize returns the sum of plaintext sizes of chunks that are
-// referenced by this snapshot but not by any earlier completed snapshot known to
-// the local database. The result is the marginal uncompressed data this snapshot
-// added to the dedup pool — i.e., the delta from prior snapshots.
+// GetSnapshotNewChunkSize returns the total plaintext size of the chunks the
+// snapshot's backup stored that were not stored before, as the backup
+// recorded it. It is not derived from the snapshot's files: a file's row is
+// shared by every snapshot that lists it and is rewritten when the file
+// changes, so it shows only the file's latest chunks.
 func (r *SnapshotRepository) GetSnapshotNewChunkSize(
 	ctx context.Context, snapshotID string,
 ) (int64, error) {
-	query := `
-		WITH this_snap_chunks AS (
-			SELECT DISTINCT fc.chunk_hash
-			FROM snapshot_files sf
-			JOIN file_chunks fc ON sf.file_id = fc.file_id
-			WHERE sf.snapshot_id = ?
-		),
-		prior_chunks AS (
-			SELECT DISTINCT fc.chunk_hash
-			FROM snapshots s
-			JOIN snapshot_files sf ON sf.snapshot_id = s.id
-			JOIN file_chunks fc ON fc.file_id = sf.file_id
-			WHERE s.completed_at IS NOT NULL
-			  AND s.id != ?
-			  AND s.started_at < (SELECT started_at FROM snapshots WHERE id = ?)
-		)
-		SELECT COALESCE(SUM(c.size), 0)
-		FROM chunks c
-		JOIN this_snap_chunks t ON c.chunk_hash = t.chunk_hash
-		WHERE c.chunk_hash NOT IN (SELECT chunk_hash FROM prior_chunks)
-	`
+	query := `SELECT new_chunk_size FROM snapshots WHERE id = ?`
 
-	var totalSize int64
+	var newChunkSize int64
 
-	err := r.db.conn.QueryRowContext(
-		ctx, query, snapshotID, snapshotID, snapshotID,
-	).Scan(&totalSize)
+	err := r.db.conn.QueryRowContext(ctx, query, snapshotID).Scan(&newChunkSize)
 	if err != nil {
 		return 0, fmt.Errorf("querying new chunk size: %w", err)
 	}
 
-	return totalSize, nil
+	return newChunkSize, nil
 }
 
 // GetIncompleteSnapshots returns all snapshots that haven't been completed

internal/snapshot/scanner.go

@@ -135,7 +135,9 @@ type ScannerConfig struct {
 // are counted per file: BytesScanned is the size of the new and changed
 // files, BytesSkipped that of the unchanged ones. FilesFailed counts the
 // new and changed files that phase 2 could not store; FilesScanned
-// includes them and BytesScanned does not.
+// includes them and BytesScanned does not. ChunksCreated counts the
+// chunks that were not already stored, and NewChunkSize is their total
+// size.
 type ScanResult struct {
 	FilesScanned   int
 	FilesSkipped   int
@@ -145,6 +147,7 @@ type ScanResult struct {
 	BytesSkipped   int64
 	BytesDeleted   int64
 	ChunksCreated  int
+	NewChunkSize   int64
 	BlobsCreated   int
 	BlobsUploaded  int
 	BytesUploaded  int64
@@ -1866,11 +1869,12 @@ func (s *Scanner) processFileStreaming(
 	return nil
 }
 
-// updateChunkStats counts a chunk that was not already stored. The scan
-// result's file counts, BytesScanned and BytesSkipped are not touched
-// here: the scan phase counts each file once.
+// updateChunkStats counts a chunk that was not already stored, and its
+// size. The scan result's file counts, BytesScanned and BytesSkipped are
+// not touched here: the scan phase counts each file once.
 func (s *Scanner) updateChunkStats(chunkSize int64, result *ScanResult) {
 	result.ChunksCreated++
+	result.NewChunkSize += chunkSize
 
 	if s.progress != nil {
 		s.progress.GetStats().ChunksCreated.Add(1)

internal/snapshot/snapshot.go

@@ -199,6 +199,7 @@ func (sm *SnapshotManager) UpdateSnapshotStatsExtended(
 			stats.CompressionLevel,
 			stats.BytesUploaded,
 			stats.UploadDurationMs,
+			stats.NewChunkSize,
 		)
 	})
 }
@@ -906,6 +907,7 @@ type ExtendedBackupStats struct {
 	BlobUncompressedSize int64 // Total uncompressed size of all referenced blobs
 	CompressionLevel     int   // Compression level used for this snapshot
 	UploadDurationMs     int64 // Total milliseconds spent uploading to S3
+	NewChunkSize         int64 // Total size of the chunks counted in ChunksCreated
 }
 
 // deleteOtherSnapshots deletes all snapshots except the current one

internal/vaultik/new_chunk_size_test.go

@@ -0,0 +1,133 @@
+package vaultik_test
+
+import (
+	"bytes"
+	"context"
+	"maps"
+	"path/filepath"
+	"testing"
+
+	"github.com/spf13/afero"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"sneak.berlin/go/vaultik/internal/chunker"
+	"sneak.berlin/go/vaultik/internal/database"
+	"sneak.berlin/go/vaultik/internal/log"
+	"sneak.berlin/go/vaultik/internal/storage"
+	"sneak.berlin/go/vaultik/internal/vaultik"
+)
+
+// TestNewChunkSizeOfChangedFileStaysFixed covers
+// https://git.eeqj.de/sneak/vaultik/issues/311. A backup whose only change
+// is a modified file lists the size of the chunks that change added as its
+// NEW CHUNK SIZE, and a later backup that changes the file again leaves
+// every earlier snapshot's figure as it was. The file's row in the local
+// index is shared by all three snapshots and rewritten by each backup.
+func TestNewChunkSizeOfChangedFileStaysFixed(t *testing.T) {
+	log.Initialize(log.Config{})
+	t.Parallel()
+
+	fs := afero.NewOsFs()
+	tempDir := t.TempDir()
+	dataDir := filepath.Join(tempDir, "src")
+	dbPath := filepath.Join(tempDir, "index.sqlite")
+	changedPath := filepath.Join(dataDir, "changed.bin")
+
+	ctx := context.Background()
+	files := writeFaultSourceTree(t, fs, dataDir)
+	cfg := changedFileConfig(dataDir, dbPath)
+
+	appendRandomBytes(t, fs, files, changedPath, appendedFileSize)
+
+	store, err := storage.NewFileStorer(filepath.Join(tempDir, "remote"))
+	require.NoError(t, err)
+
+	db, err := database.New(ctx, dbPath)
+	require.NoError(t, err)
+	t.Cleanup(func() { _ = db.Close() })
+
+	repos := database.NewRepositories(db)
+	v := newBackupVaultik(ctx, cfg, store, repos, db, fs)
+
+	require.NoError(t, backUp(v, "first"))
+
+	firstFiles := maps.Clone(files)
+	firstID := localSnapshotID(ctx, t, repos, "first")
+
+	appendRandomBytes(t, fs, files, changedPath, faultChunkSize)
+	require.NoError(t, backUp(v, "second"))
+
+	secondID := localSnapshotID(ctx, t, repos, "second")
+	secondAdded := sizeOfNewChunks(t, files, firstFiles)
+	require.Positive(t, secondAdded)
+	assert.Equal(t, secondAdded, listedNewChunkSizes(t, v)[secondID],
+		"the second snapshot must count the chunks the change added")
+
+	appendRandomBytes(t, fs, files, changedPath, faultChunkSize)
+	require.NoError(t, backUp(v, "third"))
+
+	sizes := listedNewChunkSizes(t, v)
+	assert.Equal(t, sizeOfNewChunks(t, firstFiles, nil), sizes[firstID],
+		"a later backup must not change the first snapshot's figure")
+	assert.Equal(t, secondAdded, sizes[secondID],
+		"a later backup must not change the second snapshot's figure")
+}
+
+// sizeOfNewChunks returns the total size of the distinct chunks of the
+// files in after that no file in before has, cut the way the backup cuts
+// them.
+func sizeOfNewChunks(t *testing.T, after, before map[string][]byte) int64 {
+	t.Helper()
+
+	known := make(map[string]bool)
+	for _, chunk := range chunksOf(t, before) {
+		known[chunk.Hash] = true
+	}
+
+	var total int64
+
+	for _, chunk := range chunksOf(t, after) {
+		if !known[chunk.Hash] {
+			known[chunk.Hash] = true
+			total += chunk.Size
+		}
+	}
+
+	return total
+}
+
+// chunksOf returns the chunks of every file in files.
+func chunksOf(t *testing.T, files map[string][]byte) []chunker.Chunk {
+	t.Helper()
+
+	c := chunker.NewChunker(faultChunkSize)
+
+	var all []chunker.Chunk
+
+	for _, content := range files {
+		chunks, err := c.ChunkReader(bytes.NewReader(content))
+		require.NoError(t, err)
+
+		all = append(all, chunks...)
+	}
+
+	return all
+}
+
+// listedNewChunkSizes runs `snapshot list --json` and returns each listed
+// snapshot's new_chunk_size by snapshot ID.
+func listedNewChunkSizes(t *testing.T, v *vaultik.Vaultik) map[string]int64 {
+	t.Helper()
+
+	var out bytes.Buffer
+
+	v.Stdout = &out
+	require.NoError(t, v.ListSnapshots(true))
+
+	sizes := make(map[string]int64)
+	for _, row := range decodeListJSON(t, out.String()) {
+		sizes[row.ID] = row.NewChunkSize
+	}
+
+	return sizes
+}

internal/vaultik/snapshot.go

@@ -209,6 +209,7 @@ type snapshotStats struct {
 	totalFiles         int
 	totalBytes         int64
 	totalChunks        int
+	newChunkSize       int64
 	totalBlobs         int
 	totalBytesSkipped  int64
 	totalFilesSkipped  int
@@ -341,6 +342,7 @@ func (v *Vaultik) scanAllDirectories(
 		stats.totalFiles += result.FilesScanned
 		stats.totalBytes += result.BytesScanned
 		stats.totalChunks += result.ChunksCreated
+		stats.newChunkSize += result.NewChunkSize
 		stats.totalBlobs += result.BlobsCreated
 		stats.totalFilesSkipped += result.FilesSkipped
 		stats.totalFilesFailed += result.FilesFailed
@@ -403,6 +405,7 @@ func (v *Vaultik) finalizeSnapshotMetadata(
 		BlobUncompressedSize: stats.blobUncompressedSize,
 		CompressionLevel:     v.Config.CompressionLevel,
 		UploadDurationMs:     stats.uploadDuration.Milliseconds(),
+		NewChunkSize:         stats.newChunkSize,
 	}
 
 	err = v.SnapshotManager.UpdateSnapshotStatsExtended(v.ctx, snapshotID, extStats)

internal/vaultik/snapshot_list_test.go

@@ -445,6 +445,7 @@ type listJSONRow struct {
 	RemoteKey      string `json:"remote_key"`
 	Timestamp      string `json:"timestamp"`
 	CompressedSize int64  `json:"compressed_size"`
+	NewChunkSize   int64  `json:"new_chunk_size"`
 	LocallyTracked bool   `json:"locally_tracked"`
 	RemotePresent  *bool  `json:"remote_present"`
 }