fix(declarative): keep components code out of AirbyteCodeTamperedError and match standard-test checksums

airbyte_cdk/sources/declarative/parsers/custom_code_compiler.py

@@ -89,7 +89,6 @@ def validate_python_code(
                     {
                         "expected_checksum": checksum,
                         "actual_checksum": calculated_checksum,
-                        "code_text": code_text,
                     }
                 ),
             )

airbyte_cdk/test/standard_tests/declarative_sources.py

@@ -16,12 +16,11 @@
 
 
 def md5_checksum(file_path: Path) -> str:
-    """Helper function to calculate the MD5 checksum of a file.
+    """Calculate the MD5 checksum of the file text, as injected by `create_connector`.
 
     This is used to calculate the checksum of the `components.py` file, if it exists.
     """
-    with open(file_path, "rb") as file:
-        return md5(file.read()).hexdigest()
+    return md5(file_path.read_text().encode()).hexdigest()
 
 
 class DeclarativeSourceTestSuite(SourceTestSuiteBase):

unit_tests/source_declarative_manifest/test_source_declarative_w_custom_components.py

@@ -158,17 +158,23 @@ def test_invalid_checksum_fails_to_run(
         datetime.datetime.now() - datetime.timedelta(days=2)
     ).strftime("%Y-%m-%d")
 
+    actual_checksum = py_components_config_dict["__injected_components_py_checksums"][hash_type]
     py_components_config_dict["__injected_components_py_checksums"][hash_type] = "invalid_checksum"
 
     with NamedTemporaryFile(delete=False, suffix=".json") as temp_config_file:
         json_str = json.dumps(py_components_config_dict)
         Path(temp_config_file.name).write_text(json_str)
         temp_config_file.flush()
-        with pytest.raises(AirbyteCodeTamperedError):
+        with pytest.raises(AirbyteCodeTamperedError) as exc_info:
             source = create_declarative_source(
                 ["check", "--config", temp_config_file.name],
             )
 
+    # No components code in the message: it lands in failure reasons and logs
+    assert str(exc_info.value) == f"{hash_type} checksum does not match." + str(
+        {"expected_checksum": "invalid_checksum", "actual_checksum": actual_checksum}
+    )
+
 
 @pytest.mark.parametrize(
     "env_value, should_raise",

unit_tests/test/test_standard_tests.py

@@ -9,9 +9,11 @@
 from airbyte_cdk.sources.declarative.concurrent_declarative_source import (
     ConcurrentDeclarativeSource,
 )
+from airbyte_cdk.sources.declarative.parsers.custom_code_compiler import validate_python_code
 from airbyte_cdk.sources.source import Source
 from airbyte_cdk.test.models.scenario import ConnectorTestScenario
 from airbyte_cdk.test.standard_tests._job_runner import IConnector
+from airbyte_cdk.test.standard_tests.declarative_sources import md5_checksum
 from airbyte_cdk.test.standard_tests.docker_base import DockerConnectorTestSuite
 from airbyte_cdk.test.standard_tests.pytest_hooks import _scenario_test_ids
 
@@ -108,3 +110,14 @@ def test_dedup_scenarios_conflicting_statuses_raise() -> None:
 def test_scenario_test_ids(config_paths: list[Path], expected_ids: list[str]) -> None:
     scenarios = [ConnectorTestScenario(config_path=path) for path in config_paths]
     assert _scenario_test_ids(scenarios) == expected_ids
+
+
+def test_md5_checksum_matches_injected_text_with_crlf(tmp_path: Path) -> None:
+    """Assert the checksum covers `read_text()`, which `create_connector` injects, not raw bytes."""
+    components_py_path = tmp_path / "components.py"
+    components_py_path.write_bytes(b"def sample_function():\r\n    return 1\r\n")
+
+    validate_python_code(
+        code_text=components_py_path.read_text(),
+        checksums={"md5": md5_checksum(components_py_path)},
+    )