DynamoDB: Fix projection expression with binary attribute (#6816)

moto/dynamodb/models/dynamo_type.py

@@ -1,3 +1,4 @@
+import base64
 import copy
 import decimal
 
@@ -195,6 +196,10 @@ def size(self) -> int:
 
     def to_json(self) -> Dict[str, Any]:
         # Returns a regular JSON object where the value can still be/contain a DynamoType
+        if self.is_binary():
+            # Binary data cannot be represented in JSON
+            # AWS returns a base64-encoded value - the SDK's then convert that back
+            return {self.type: base64.b64encode(self.value).decode("utf-8")}
         return {self.type: self.value}
 
     def to_regular_json(self) -> Dict[str, Any]:
@@ -212,6 +217,8 @@ def to_regular_json(self) -> Dict[str, Any]:
                 val.to_regular_json() if isinstance(val, DynamoType) else val
                 for val in value
             ]
+        if self.is_binary():
+            value = base64.b64decode(value)
         return {self.type: value}
 
     def compare(self, range_comparison: str, range_objs: List[Any]) -> bool:
@@ -236,6 +243,9 @@ def is_list(self) -> bool:
     def is_map(self) -> bool:
         return self.type == DDBType.MAP
 
+    def is_binary(self) -> bool:
+        return self.type == DDBType.BINARY
+
     def same_type(self, other: "DynamoType") -> bool:
         return self.type == other.type
 

moto/dynamodb/models/utilities.py

@@ -14,7 +14,7 @@ def dynamo_json_dump(dynamo_object: Any) -> str:
 
 
 def bytesize(val: str) -> int:
-    return len(val.encode("utf-8"))
+    return len(val if isinstance(val, bytes) else val.encode("utf-8"))
 
 
 def find_nested_key(

tests/test_dynamodb/test_dynamodb.py

@@ -4,6 +4,7 @@
 
 import boto3
 from boto3.dynamodb.conditions import Attr, Key
+from boto3.dynamodb.types import Binary
 import re
 from moto import mock_dynamodb, settings
 from moto.core import DEFAULT_ACCOUNT_ID as ACCOUNT_ID
@@ -5727,6 +5728,30 @@ def test_projection_expression_execution_order():
     )
 
 
+@mock_dynamodb
+def test_projection_expression_with_binary_attr():
+    dynamo_resource = boto3.resource("dynamodb", region_name="us-east-1")
+    dynamo_resource.create_table(
+        TableName="test",
+        AttributeDefinitions=[
+            {"AttributeName": "pk", "AttributeType": "S"},
+            {"AttributeName": "sk", "AttributeType": "S"},
+        ],
+        KeySchema=[
+            {"AttributeName": "pk", "KeyType": "HASH"},
+            {"AttributeName": "sk", "KeyType": "RANGE"},
+        ],
+        BillingMode="PAY_PER_REQUEST",
+    )
+    table = dynamo_resource.Table("test")
+    table.put_item(Item={"pk": "pk", "sk": "sk", "key": b"value\xbf"})
+    assert table.get_item(
+        Key={"pk": "pk", "sk": "sk"},
+        ExpressionAttributeNames={"#key": "key"},
+        ProjectionExpression="#key",
+    )["Item"] == {"key": Binary(b"value\xbf")}
+
+
 @mock_dynamodb
 def test_invalid_projection_expressions():
     table_name = "test-projection-expressions-table"