Step Functions: Surface Support for Mocked Responses #12525
Conversation
MEPalma
added
the
semver: minor
MEPalma
requested review from
joe4dev,
gregfurman and
dominikschubert
as code owners
LocalStack Community integration with Pro 2 files 2 suites 1h 51m 54s ⏱️ Results for commit 376a73d. |
localstack-core/localstack/services/stepfunctions/mocking/mock_config.py
Show resolved
Hide resolved
| range_part_start = definition_parts[0] | ||
| range_part_end = definition_parts[1] |
There was a problem hiding this comment.
Should we be checking that range_part_start > range_part_end?
There was a problem hiding this comment.
We check when trying to construct the mock responses from these values localstack.services.stepfunctions.mocking.mock_config.StateMockedResponses.__init__
|
|
||
|
|
||
| def _parse_mocked_response_range(string_definition: str) -> tuple[int, int]: | ||
| definition_parts = string_definition.strip().split("-") |
There was a problem hiding this comment.
Do we need to validate these values at all? Like are there valid bounds?
There was a problem hiding this comment.
Yes indeed, we do this after parsing the values and constructing the StateMockedResponses object for the runtime later localstack.services.stepfunctions.mocking.mock_config.StateMockedResponses.__init__
localstack-core/localstack/services/stepfunctions/mocking/mock_config_file.py
Show resolved
Hide resolved
| Accepts any fields. | ||
| """ | ||
|
|
||
| model_config = {"extra": "allow", "frozen": True} |
There was a problem hiding this comment.
nit: Should we be using a pydantic.dataclass here instead? It contains a frozen attribute out-the-box so am wondering about the BaseModel approach
There was a problem hiding this comment.
Definitely a good point you raise! In this case we still need a few BaseModel features, I can see:
- extra="allow" on RawReturnResponse
- the root model on RawTestCase
- the model_validator in RawResponseModel
Pydantic’s dataclass doesn’t support any of those as far as I can see, and we'd also lose the model_dump() (although this may not be an issue by itself if we were using records)
| @lru_cache(maxsize=1) | ||
| def _read_sfn_raw_mock_config(file_path: str, modified_epoch: int) -> Optional[RawMockConfig]: # noqa |
There was a problem hiding this comment.
question: this will only cache a single call. I imagine we wont be reading in multiple mock files?
There was a problem hiding this comment.
correct, for now we'll be loading one file only
MEPalma
deleted the
MEP-SFN-mocked_responses_happy_path_and_preprocessing
branch
Motivation
These changes introduce base support for executing AWS Step Functions state machines in a mock-enabled mode, designed to streamline testing and simulation workflows. The primary motivation is to enable users to execute state machines using mock configuration files, allowing simulated task execution by sampling predefined responses from a mock oracle. This capability is particularly valuable for validating control flow logic and response handling in a deterministic manner without invoking actual downstream services AWS Docs
The changes in this PR surface support for all critical components necessary to enable this workflow. Future changes will refine and expand on each aspect introduced in this PR
Changes