From b2e14f01b32855928ef6fd1b8cd5148622e3662d Mon Sep 17 00:00:00 2001
From: "Masih H. Derkani" <m@derkani.org>
Date: Wed, 6 Jul 2022 16:27:40 +0100
Subject: [PATCH] Empty identity CID should be indexed when options are set

There is an edge-case where if the storing identity CIDs are enabled in
a CARv2, one could technically store an empty identity CID which ends up
with a `singlewidthindex` width of 8.

Such CAR files indeed exist out there and the validation changes
introduced in `2.4.0` means such CAR file indices are no longer readable
, even if regenerated.

The question is should this be considered a valid index/readable index?
---
 v2/blockstore/readwrite_test.go | 50 +++++++++++++++++++++++++++++++++
 v2/index/indexsorted.go         |  4 +--
 2 files changed, 52 insertions(+), 2 deletions(-)

diff --git a/v2/blockstore/readwrite_test.go b/v2/blockstore/readwrite_test.go
index 11cc99a2..22525e78 100644
--- a/v2/blockstore/readwrite_test.go
+++ b/v2/blockstore/readwrite_test.go
@@ -8,6 +8,7 @@ import (
 	"io/ioutil"
 	"math/rand"
 	"os"
+	"path"
 	"path/filepath"
 	"sync"
 	"testing"
@@ -943,3 +944,52 @@ func TestReadWrite_ReWritingCARv1WithIdentityCidIsIdenticalToOriginalWithOptions
 	require.Equal(t, wantWritten, gotWritten)
 	require.Equal(t, wantSum, gotSum)
 }
+
+func TestBlockstore_IdentityCidWithEmptyDataIsIndexed(t *testing.T) {
+	p := path.Join(t.TempDir(), "car-id-cid-empty.carv2")
+	var noData []byte
+
+	mh, err := multihash.Sum(noData, multihash.IDENTITY, -1)
+	require.NoError(t, err)
+	w, err := blockstore.OpenReadWrite(p, nil, carv2.StoreIdentityCIDs(true))
+	require.NoError(t, err)
+
+	blk, err := blocks.NewBlockWithCid(noData, cid.NewCidV1(cid.Raw, mh))
+	require.NoError(t, err)
+
+	err = w.Put(context.TODO(), blk)
+	require.NoError(t, err)
+	require.NoError(t, w.Finalize())
+
+	r, err := carv2.OpenReader(p)
+	require.NoError(t, err)
+	defer func() { require.NoError(t, r.Close()) }()
+
+	dr, err := r.DataReader()
+	require.NoError(t, err)
+	header, err := carv1.ReadHeader(dr, carv1.DefaultMaxAllowedHeaderSize)
+	require.NoError(t, err)
+	wantOffset, err := carv1.HeaderSize(header)
+	require.NoError(t, err)
+
+	ir, err := r.IndexReader()
+	require.NoError(t, err)
+	idx, err := index.ReadFrom(ir)
+	require.NoError(t, err)
+
+	itidx, ok := idx.(index.IterableIndex)
+	require.True(t, ok)
+	var count int
+	err = itidx.ForEach(func(m multihash.Multihash, u uint64) error {
+		dm, err := multihash.Decode(m)
+		require.NoError(t, err)
+		require.Equal(t, multicodec.Identity, multicodec.Code(dm.Code))
+		require.Equal(t, 0, dm.Length)
+		require.Empty(t, dm.Digest)
+		require.Equal(t, wantOffset, u)
+		count++
+		return nil
+	})
+	require.NoError(t, err)
+	require.Equal(t, 1, count)
+}
diff --git a/v2/index/indexsorted.go b/v2/index/indexsorted.go
index aeed4c11..ed94ed8f 100644
--- a/v2/index/indexsorted.go
+++ b/v2/index/indexsorted.go
@@ -91,8 +91,8 @@ func (s *singleWidthIndex) Unmarshal(r io.Reader) error {
 }
 
 func (s *singleWidthIndex) checkUnmarshalLengths(width uint32, dataLen, extra uint64) error {
-	if width <= 8 {
-		return errors.New("malformed index; width must be bigger than 8")
+	if width < 8 {
+		return errors.New("malformed index; width must be at least 8")
 	}
 	const maxWidth = 32 << 20 // 32MiB, to ~match the go-cid maximum
 	if width > maxWidth {