Reduce archivedb key lengths by 1 byte

x/archivedb/batch.go

@@ -22,7 +22,7 @@ type batch struct {
 func (c *batch) Write() error {
 	batch := c.db.db.NewBatch()
 	for _, op := range c.Ops {
-		key, _ := newDBKey(op.Key, c.height)
+		key, _ := newDBKeyFromUser(op.Key, c.height)
 		var value []byte
 		if !op.Delete {
 			value = newDBValue(op.Value)

x/archivedb/db_test.go

@@ -163,7 +163,7 @@ func TestDBKeySpace(t *testing.T) {
 
 	var (
 		key1    = []byte("key1")
-		key2, _ = newDBKey([]byte("key1"), 2)
+		key2, _ = newDBKeyFromUser([]byte("key1"), 2)
 		key3    = []byte("key3")
 		value1  = []byte("value1@1")
 		value2  = []byte("value2@2")

x/archivedb/key.go

@@ -11,76 +11,85 @@ import (
 )
 
 var (
-	ErrEmptyKey           = errors.New("empty key")
 	ErrParsingKeyLength   = errors.New("failed reading key length")
 	ErrIncorrectKeyLength = errors.New("incorrect key length")
 
-	heightKey = []byte{1}
+	heightKey = newDBKeyFromMetadata([]byte{})
 )
 
-// newDBKey converts a user formatted key and a height into a database formatted
-// key.
-//
-// A database key contains additional information alongside the given user key.
-//
 // The requirements of a database key are:
 //
 // 1. A given user key must have a unique database key prefix. This guarantees
 // that user keys can not overlap on disk.
 // 2. Inside of a database key prefix, the database keys must be sorted by
 // decreasing height.
 // 3. User keys must never overlap with any metadata keys.
+
+// newDBKeyFromUser converts a user key and height into a database formatted
+// key.
 //
-// To meet these requirements, a database key prefix is defined by concatinating
-// the zero byte, the length of the user key, and the user key. The suffix of
-// the database key is the negation of the big endian encoded height. This
-// suffix guarantees the keys are sorted correctly.
+// To meet the requirements of a database key, the prefix is defined by
+// concatenating the length of the user key and the user key. The suffix of the
+// database key is the negation of the big endian encoded height. This suffix
+// guarantees the keys are sorted correctly.
 //
 //	Example (Asumming heights are 1 byte):
-//	 |  User given  |  Stored as  |
-//	 |--------------|-------------|
-//	 |    foo:10    |  3:foo:245  |
-//	 |    foo:20    |  3:foo:235  |
+//	 |  User key  |  Stored as  |
+//	 |------------|-------------|
+//	 |   foo:10   |  3:foo:245  |
+//	 |   foo:20   |  3:foo:235  |
 //
 // Returns:
 // - The database key
 // - The database key prefix, which is independent of the height
-func newDBKey(key []byte, height uint64) ([]byte, []byte) {
+func newDBKeyFromUser(key []byte, height uint64) ([]byte, []byte) {
 	keyLen := len(key)
-	dbKeyMaxSize := 1 + binary.MaxVarintLen64 + keyLen + wrappers.LongLen
+	dbKeyMaxSize := binary.MaxVarintLen64 + keyLen + wrappers.LongLen
 	dbKey := make([]byte, dbKeyMaxSize)
-	offset := 1
-	offset += binary.PutUvarint(dbKey[offset:], uint64(keyLen))
+	offset := binary.PutUvarint(dbKey, uint64(keyLen))
 	offset += copy(dbKey[offset:], key)
 	prefixOffset := offset
 	binary.BigEndian.PutUint64(dbKey[offset:], ^height)
 	offset += wrappers.LongLen
 	return dbKey[:offset], dbKey[:prefixOffset]
 }
 
-// parseDBKey takes a database formatted key and returns the user formatted key
+// parseDBKeyFromUser takes a database formatted key and returns the user key
 // along with its height.
 //
 // Note: An error should only be returned from this function if the database has
 // been corrupted.
-func parseDBKey(dbKey []byte) ([]byte, uint64, error) {
-	dbKeyLen := uint64(len(dbKey))
-	if dbKeyLen == 0 {
-		return nil, 0, ErrEmptyKey
-	}
-
-	keyLen, offset := binary.Uvarint(dbKey[1:])
+func parseDBKeyFromUser(dbKey []byte) ([]byte, uint64, error) {
+	keyLen, offset := binary.Uvarint(dbKey)
 	if offset <= 0 {
 		return nil, 0, ErrParsingKeyLength
 	}
 
-	keyIndex := 1 + uint64(offset)
-	heightIndex := keyIndex + keyLen
-	if dbKeyLen != heightIndex+wrappers.LongLen {
+	heightIndex := uint64(offset) + keyLen
+	if uint64(len(dbKey)) != heightIndex+wrappers.LongLen {
 		return nil, 0, ErrIncorrectKeyLength
 	}
 
-	key := dbKey[keyIndex:heightIndex]
+	key := dbKey[offset:heightIndex]
 	height := ^binary.BigEndian.Uint64(dbKey[heightIndex:])
 	return key, height, nil
 }
+
+// newDBKeyFromMetadata converts a metadata key into a database formatted key.
+//
+// To meet the requirements of a database key, the key is defined by
+// concatenating the length of the metadata key + 1 and the metadata key.
+//
+//	Example:
+//	 |  Metadata key  |  Stored as  |
+//	 |----------------|-------------|
+//	 |       foo      |    4:foo    |
+//	 |       fo       |    3:fo     |
+func newDBKeyFromMetadata(key []byte) []byte {
+	keyLen := len(key)
+	dbKeyMaxSize := binary.MaxVarintLen64 + keyLen
+	dbKey := make([]byte, dbKeyMaxSize)
+	offset := binary.PutUvarint(dbKey, uint64(keyLen)+1)
+	offset += copy(dbKey[offset:], key)
+	return dbKey[:offset]
+}

x/archivedb/key_test.go

@@ -13,10 +13,10 @@ import (
 )
 
 func TestNaturalDescSortingForSameKey(t *testing.T) {
-	key0, _ := newDBKey(make([]byte, 0), 0)
-	key1, _ := newDBKey(make([]byte, 0), 1)
-	key2, _ := newDBKey(make([]byte, 0), 2)
-	key3, _ := newDBKey(make([]byte, 0), 3)
+	key0, _ := newDBKeyFromUser(make([]byte, 0), 0)
+	key1, _ := newDBKeyFromUser(make([]byte, 0), 1)
+	key2, _ := newDBKeyFromUser(make([]byte, 0), 2)
+	key3, _ := newDBKeyFromUser(make([]byte, 0), 3)
 
 	entry := [][]byte{key0, key1, key2, key3}
 	expected := [][]byte{key3, key2, key1, key0}
@@ -29,10 +29,10 @@ func TestNaturalDescSortingForSameKey(t *testing.T) {
 }
 
 func TestSortingDifferentPrefix(t *testing.T) {
-	key0, _ := newDBKey([]byte{0}, 0)
-	key1, _ := newDBKey([]byte{0}, 1)
-	key2, _ := newDBKey([]byte{1}, 0)
-	key3, _ := newDBKey([]byte{1}, 1)
+	key0, _ := newDBKeyFromUser([]byte{0}, 0)
+	key1, _ := newDBKeyFromUser([]byte{0}, 1)
+	key2, _ := newDBKeyFromUser([]byte{1}, 0)
+	key3, _ := newDBKeyFromUser([]byte{1}, 1)
 
 	entry := [][]byte{key0, key1, key2, key3}
 	expected := [][]byte{key1, key0, key3, key2}
@@ -49,10 +49,20 @@ func TestParseDBKey(t *testing.T) {
 
 	key := []byte{0, 1, 2, 3, 4, 5}
 	height := uint64(102310)
-	dbKey, _ := newDBKey(key, height)
+	dbKey, _ := newDBKeyFromUser(key, height)
 
-	parsedKey, parsedHeight, err := parseDBKey(dbKey)
+	parsedKey, parsedHeight, err := parseDBKeyFromUser(dbKey)
 	require.NoError(err)
 	require.Equal(key, parsedKey)
 	require.Equal(height, parsedHeight)
 }
+
+func FuzzMetadataKeyInvariant(f *testing.F) {
+	f.Fuzz(func(t *testing.T, userKey []byte, metadataKey []byte) {
+		// The prefix is independent of the height, so its value doesn't matter
+		// for this test.
+		_, dbKeyPrefix := newDBKeyFromUser(userKey, 0)
+		dbKey := newDBKeyFromMetadata(metadataKey)
+		require.False(t, bytes.HasPrefix(dbKey, dbKeyPrefix))
+	})
+}

x/archivedb/prefix_test.go

@@ -52,7 +52,7 @@ func TestDBEfficientLookups(t *testing.T) {
 
 	var (
 		key             = []byte("key")
-		maliciousKey, _ = newDBKey(key, 2)
+		maliciousKey, _ = newDBKeyFromUser(key, 2)
 	)
 
 	db := New(&limitIterationDB{Database: memdb.New()})

x/archivedb/reader.go

@@ -35,7 +35,7 @@ func (r *Reader) Get(key []byte) ([]byte, error) {
 // modified at, and a boolean to indicate if the last modification was an
 // insertion. If the key has never been modified, ErrNotFound will be returned.
 func (r *Reader) GetEntry(key []byte) ([]byte, uint64, bool, error) {
-	it := r.db.db.NewIteratorWithStartAndPrefix(newDBKey(key, r.height))
+	it := r.db.db.NewIteratorWithStartAndPrefix(newDBKeyFromUser(key, r.height))
 	defer it.Release()
 
 	next := it.Next()
@@ -48,7 +48,7 @@ func (r *Reader) GetEntry(key []byte) ([]byte, uint64, bool, error) {
 		return nil, 0, false, database.ErrNotFound
 	}
 
-	_, height, err := parseDBKey(it.Key())
+	_, height, err := parseDBKeyFromUser(it.Key())
 	if err != nil {
 		return nil, 0, false, err
 	}