From 80e52e03d5bd7ead65de9b4b5ec2e0d21b7dc3d7 Mon Sep 17 00:00:00 2001
From: Al Cutter <al.cutter@gmail.com>
Date: Fri, 10 Jan 2025 18:38:12 +0000
Subject: [PATCH] Pass in bundle hasher

---
 cmd/experimental/migrate/internal/migrate.go | 16 +++++++++++++++
 cmd/experimental/migrate/posix/main.go       |  2 +-
 storage/posix/files.go                       | 21 +++++++++++---------
 3 files changed, 29 insertions(+), 10 deletions(-)

diff --git a/cmd/experimental/migrate/internal/migrate.go b/cmd/experimental/migrate/internal/migrate.go
index 576d8226..fca11492 100644
--- a/cmd/experimental/migrate/internal/migrate.go
+++ b/cmd/experimental/migrate/internal/migrate.go
@@ -21,6 +21,8 @@ import (
 	"time"
 
 	"github.com/avast/retry-go/v4"
+	"github.com/transparency-dev/merkle/rfc6962"
+	"github.com/transparency-dev/trillian-tessera/api"
 	"github.com/transparency-dev/trillian-tessera/api/layout"
 	"github.com/transparency-dev/trillian-tessera/client"
 	"golang.org/x/sync/errgroup"
@@ -179,3 +181,17 @@ func (m *copier) migrateWorker(ctx context.Context) error {
 	}
 	return nil
 }
+
+// BundleHasher parses a C2SP tlog-tile bundle and returns the leaf hashes of each entry it contains.
+func BundleHasher(bundle []byte) ([][]byte, error) {
+	eb := &api.EntryBundle{}
+	if err := eb.UnmarshalText(bundle); err != nil {
+		return nil, fmt.Errorf("unmarshal: %v", err)
+	}
+	r := make([][]byte, 0, len(eb.Entries))
+	for _, e := range eb.Entries {
+		h := rfc6962.DefaultHasher.HashLeaf(e)
+		r = append(r, h[:])
+	}
+	return r, nil
+}
diff --git a/cmd/experimental/migrate/posix/main.go b/cmd/experimental/migrate/posix/main.go
index a3c8243e..90044e25 100644
--- a/cmd/experimental/migrate/posix/main.go
+++ b/cmd/experimental/migrate/posix/main.go
@@ -66,7 +66,7 @@ func main() {
 	}
 
 	// Create our Tessera storage backend:
-	st, err := posix.NewMigrationTarget(ctx, *storageDir, *initialise, tessera.WithCTLayout())
+	st, err := posix.NewMigrationTarget(ctx, *storageDir, *initialise, internal.BundleHasher, tessera.WithCTLayout())
 	if err != nil {
 		klog.Exitf("Failed to create new GCP storage: %v", err)
 	}
diff --git a/storage/posix/files.go b/storage/posix/files.go
index 5e5ce35a..08b74848 100644
--- a/storage/posix/files.go
+++ b/storage/posix/files.go
@@ -509,10 +509,14 @@ func createExclusive(f string, d []byte) error {
 	return nil
 }
 
+// BundleHasherFunc is the signature of a function which knows how to parse an entry bundle and calculate leaf hashes for its entries.
+type BundleHasherFunc func(entryBundle []byte) (LeafHashes [][]byte, err error)
+
 // NewMigrationTarget creates a new POSIX storage for the MigrationTarget lifecycle mode.
 // - path is a directory in which the log should be stored
 // - create must only be set when first creating the log, and will create the directory structure and an empty checkpoint
-func NewMigrationTarget(ctx context.Context, path string, create bool, opts ...func(*options.StorageOptions)) (*MigrationStorage, error) {
+// - bundleHasher knows how to parse the provided entry bundle, and returns a slice of leaf hashes for the entries it contains.
+func NewMigrationTarget(ctx context.Context, path string, create bool, bundleHasher BundleHasherFunc, opts ...func(*options.StorageOptions)) (*MigrationStorage, error) {
 	opt := storage.ResolveStorageOptions(opts...)
 
 	r := &MigrationStorage{
@@ -520,6 +524,7 @@ func NewMigrationTarget(ctx context.Context, path string, create bool, opts ...f
 			path:        path,
 			entriesPath: opt.EntriesPath,
 		},
+		bundleHasher: bundleHasher,
 	}
 	if err := r.s.initialise(create); err != nil {
 		return nil, err
@@ -528,7 +533,8 @@ func NewMigrationTarget(ctx context.Context, path string, create bool, opts ...f
 }
 
 type MigrationStorage struct {
-	s Storage
+	s            Storage
+	bundleHasher BundleHasherFunc
 }
 
 func (m *MigrationStorage) AwaitIntegration(ctx context.Context, sourceSize uint64, sourceRoot []byte) error {
@@ -626,14 +632,11 @@ func (m *MigrationStorage) fetchLeafHashes(ctx context.Context, from, to, source
 			return nil, fmt.Errorf("ReadEntryBundle(%d.%d): %v", idx, p, err)
 		}
 
-		eb := &api.EntryBundle{}
-		if err := eb.UnmarshalText(b); err != nil {
-			return nil, fmt.Errorf("unmarshal bundle index %d: %v", idx, err)
-		}
-		for _, e := range eb.Entries {
-			h := rfc6962.DefaultHasher.HashLeaf(e)
-			lh = append(lh, h[:])
+		bh, err := m.bundleHasher(b)
+		if err != nil {
+			return nil, fmt.Errorf("bundleHasherFunc for bundle index %d: %v", idx, err)
 		}
+		lh = append(lh, bh...)
 		n++
 		if n >= maxBundles {
 			break