fix: remove double-base64 encoding in upload_file/download_file (#458)

Author: Hweinstock

src/bedrock_agentcore/tools/code_interpreter_client.py

@@ -4,7 +4,6 @@
 applications to start, stop, and invoke code execution in a managed sandbox environment.
 """
 
-import base64
 import logging
 import re
 import uuid
@@ -482,7 +481,7 @@ def upload_file(
                 'scripts/analysis.py'). Must be relative to the working directory.
                 Absolute paths starting with '/' are not allowed.
             content: File content as string (text files) or bytes (binary files).
-                    Binary content will be base64 encoded automatically.
+                    Binary content will be encoded automatically by botocore.
             description: Optional semantic description of the file contents.
                         This is stored as metadata and can help LLMs understand
                         the data structure (e.g., "CSV with columns: date, revenue, product_id").
@@ -514,7 +513,7 @@ def upload_file(
 
         # Handle binary content
         if isinstance(content, bytes):
-            file_content = {"path": path, "blob": base64.b64encode(content).decode("utf-8")}
+            file_content = {"path": path, "blob": content}
         else:
             file_content = {"path": path, "text": content}
 
@@ -564,7 +563,7 @@ def upload_files(
                 raise ValueError(f"Path must be relative, not absolute. Got: {path}")
 
             if isinstance(content, bytes):
-                file_contents.append({"path": path, "blob": base64.b64encode(content).decode("utf-8")})
+                file_contents.append({"path": path, "blob": content})
             else:
                 file_contents.append({"path": path, "text": content})
 
@@ -649,7 +648,7 @@ def download_file(
                             if "text" in resource:
                                 return resource["text"]
                             elif "blob" in resource:
-                                raw = base64.b64decode(resource["blob"])
+                                raw = resource["blob"]
                                 try:
                                     return raw.decode("utf-8")
                                 except (UnicodeDecodeError, ValueError):
@@ -690,7 +689,7 @@ def download_files(
                             if "text" in resource:
                                 files[file_path] = resource["text"]
                             elif "blob" in resource:
-                                raw = base64.b64decode(resource["blob"])
+                                raw = resource["blob"]
                                 try:
                                     files[file_path] = raw.decode("utf-8")
                                 except (UnicodeDecodeError, ValueError):

tests/bedrock_agentcore/tools/test_code_interpreter_client.py

@@ -1,4 +1,3 @@
-import base64
 import datetime
 from unittest.mock import ANY, MagicMock, patch
 
@@ -657,7 +656,6 @@ def test_upload_file_binary_content(self, mock_boto3):
         client.data_plane_client.invoke_code_interpreter.return_value = mock_response
 
         binary_content = b"\x89PNG\r\n\x1a\n"  # PNG header bytes
-        expected_b64 = base64.b64encode(binary_content).decode("utf-8")
 
         # Act
         result = client.upload_file(path="image.png", content=binary_content)
@@ -667,7 +665,7 @@ def test_upload_file_binary_content(self, mock_boto3):
             codeInterpreterIdentifier="test.identifier",
             sessionId="test-session-id",
             name="writeFiles",
-            arguments={"content": [{"path": "image.png", "blob": expected_b64}]},
+            arguments={"content": [{"path": "image.png", "blob": binary_content}]},
         )
         assert result == mock_response
 
@@ -758,7 +756,6 @@ def test_upload_files_mixed_content(self, mock_boto3):
         client.data_plane_client.invoke_code_interpreter.return_value = mock_response
 
         binary_content = b"\x00\x01\x02\x03"
-        expected_b64 = base64.b64encode(binary_content).decode("utf-8")
 
         files = [
             {"path": "text.txt", "content": "hello world"},
@@ -776,7 +773,7 @@ def test_upload_files_mixed_content(self, mock_boto3):
             arguments={
                 "content": [
                     {"path": "text.txt", "text": "hello world"},
-                    {"path": "binary.bin", "blob": expected_b64},
+                    {"path": "binary.bin", "blob": binary_content},
                 ]
             },
         )
@@ -963,14 +960,13 @@ def test_download_file_binary(self, mock_boto3):
         client.session_id = "test-session-id"
 
         binary_content = b"\x89PNG\r\n\x1a\n"  # PNG header bytes
-        encoded_content = base64.b64encode(binary_content).decode("utf-8")
 
         mock_response = {
             "stream": [
                 {
                     "result": {
                         "content": [
-                            {"type": "resource", "resource": {"uri": "file://image.png", "blob": encoded_content}}
+                            {"type": "resource", "resource": {"uri": "file://image.png", "blob": binary_content}}
                         ]
                     }
                 }
@@ -997,14 +993,16 @@ def test_download_file_blob_utf8_returns_str(self, mock_boto3):
         client.session_id = "test-session-id"
 
         text_content = "hello world"
-        encoded_content = base64.b64encode(text_content.encode("utf-8")).decode("utf-8")
 
         mock_response = {
             "stream": [
                 {
                     "result": {
                         "content": [
-                            {"type": "resource", "resource": {"uri": "file://data.bin", "blob": encoded_content}}
+                            {
+                                "type": "resource",
+                                "resource": {"uri": "file://data.bin", "blob": text_content.encode("utf-8")},
+                            }
                         ]
                     }
                 }
@@ -1102,7 +1100,6 @@ def test_download_files_binary(self, mock_boto3):
         client.session_id = "test-session-id"
 
         binary_content = b"\x89PNG\r\n\x1a\n"  # PNG header bytes
-        encoded_binary = base64.b64encode(binary_content).decode("utf-8")
 
         mock_response = {
             "stream": [
@@ -1120,7 +1117,7 @@ def test_download_files_binary(self, mock_boto3):
                                 "type": "resource",
                                 "resource": {
                                     "uri": "file:///opt/amazon/genesis1p-tools/var/chart.png",
-                                    "blob": encoded_binary,
+                                    "blob": binary_content,
                                 },
                             },
                         ]
@@ -1150,9 +1147,7 @@ def test_download_files_blob_utf8_returns_str(self, mock_boto3):
         client.session_id = "test-session-id"
 
         text_content = "some utf-8 blob content"
-        encoded_text = base64.b64encode(text_content.encode("utf-8")).decode("utf-8")
         binary_content = b"\x89PNG\r\n\x1a\n"
-        encoded_binary = base64.b64encode(binary_content).decode("utf-8")
 
         mock_response = {
             "stream": [
@@ -1163,14 +1158,14 @@ def test_download_files_blob_utf8_returns_str(self, mock_boto3):
                                 "type": "resource",
                                 "resource": {
                                     "uri": "file:///opt/amazon/genesis1p-tools/var/data.bin",
-                                    "blob": encoded_text,
+                                    "blob": text_content.encode("utf-8"),
                                 },
                             },
                             {
                                 "type": "resource",
                                 "resource": {
                                     "uri": "file:///opt/amazon/genesis1p-tools/var/chart.png",
-                                    "blob": encoded_binary,
+                                    "blob": binary_content,
                                 },
                             },
                         ]

tests_integ/tools/test_code_interpreter_client.py

@@ -0,0 +1,75 @@
+"""Integration tests for code interpreter binary file upload/download.
+
+Verifies that binary content survives a roundtrip through upload_file/download_file
+without double-base64 encoding (issue #458).
+
+Run with:
+    pytest tests_integ/tools/test_code_interpreter_client.py -xvs
+"""
+
+import os
+
+import pytest
+
+from bedrock_agentcore.tools.code_interpreter_client import CodeInterpreter
+
+# 67 bytes of binary data; base64 encoding would produce 92 bytes
+PAYLOAD = b"\x89PNG\r\n\x1a\n" + bytes(range(59))
+EXPECTED_SIZE = len(PAYLOAD)  # 67
+
+
+def _extract_stdout(stream):
+    """Extract stdout content from an execute_code stream response.
+
+    Returns the stdout string or raises AssertionError if not found.
+    """
+    for event in stream:
+        r = event.get("result", {})
+        stdout = r.get("structuredContent", {}).get("stdout", "")
+        if stdout:
+            return stdout
+        content = r.get("content", "")
+        if content:
+            return str(content)
+    raise AssertionError("stdout not found in stream response")
+
+
+@pytest.mark.integration
+class TestCodeInterpreterClient:
+    """Integration tests for CodeInterpreter client."""
+
+    @classmethod
+    def setup_class(cls):
+        cls.region = os.environ.get("BEDROCK_TEST_REGION", "us-east-1")
+        cls.client = CodeInterpreter(cls.region)
+        cls.client.start()
+        cls.client.upload_file(path="test.bin", content=PAYLOAD)
+
+    @classmethod
+    def teardown_class(cls):
+        cls.client.stop()
+
+    def test_upload_file_writes_correct_size(self):
+        """upload_file with binary bytes writes the correct size to disk (not double-base64 encoded)."""
+        result = self.client.execute_code("import os\nprint(os.path.getsize('test.bin'))")
+
+        stdout = _extract_stdout(result["stream"])
+        disk_size = None
+        for line in stdout.splitlines():
+            if line.strip().isdigit():
+                disk_size = int(line.strip())
+
+        assert disk_size is not None, "Could not parse disk size from stdout"
+        assert disk_size == EXPECTED_SIZE, (
+            f"Expected {EXPECTED_SIZE} bytes on disk, got {disk_size} (92 would indicate double-base64 encoding)"
+        )
+
+    def test_download_file_returns_original_bytes(self):
+        """download_file returns the exact original bytes that were uploaded."""
+        downloaded = self.client.download_file("test.bin")
+
+        assert isinstance(downloaded, bytes), f"Expected bytes, got {type(downloaded).__name__}"
+        assert downloaded == PAYLOAD, (
+            f"Downloaded content does not match original payload. "
+            f"Got {len(downloaded)} bytes, expected {EXPECTED_SIZE} bytes."
+        )