[aryanpatil-07]

Link to Issue or Description of Change

1. Link to an existing issue (if applicable):

• Closes: ToolConfirmation.from_response_dict crashes with TypeError on parsed dicts #7681

2. Or, if no issue exists, describe the change:

Problem:
ToolConfirmation.from_response_dict assumed response["response"] was always a JSON-encoded string and passed it directly to json.loads(). When receiving an already-deserialized dictionary (e.g., from FastAPI or test runners), json.loads() raises TypeError. In _workflow_hitl_utils.py, only ValueError is caught, causing HITL workflow resumption to crash.

Solution:
Inspect response["response"] first: if it is a str, bytes, or bytearray, deserialize it with json.loads(); otherwise, pass the pre-parsed dictionary directly to model_validate().

Testing Plan

Unit Tests:

• I have added or updated unit tests for my change.
• All unit tests pass locally.

Summary of test results:

tests/unittests/tools/test_tool_confirmation.py ............... [100%]
============================= 15 passed in 1.29s ==============================

Manual End-to-End (E2E) Tests:

Verified that ToolConfirmation.from_response_dict({"response": {"confirmed": True}}) returns a valid ToolConfirmation(confirmed=True) instance without raising a TypeError.

Checklist

• I have read the CONTRIBUTING.md document.
• I have performed a self-review of my own code.
• I have commented my code, particularly in hard-to-understand areas.
• I have added tests that prove my fix is effective or that my feature works.
• New and existing unit tests pass locally with my changes.
• I have manually tested my changes end-to-end.
• Any dependent changes have been merged and published in downstream modules.

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

[BichengWang]

Reviewed 3f190e9acde7788548eceefab1a03d0dc2bf59bc against the requested coverage in #7681. No correctness blocker found: pre-parsed dictionaries reach validation, existing JSON/bytes handling is preserved, and invalid inner values become WorkflowDataError in the HITL parser.

Offline validation on macOS arm64 / Python 3.13.16: 118 passed across test_tool_confirmation.py, test_workflow_hitl_utils.py, and test_workflow.py. Restoring only the original decoder from base facd7ddafe0e23715e079e50971ad27b251e2e98 in-process makes four added cases fail with TypeError; the bytes control still passes. Sixteen additional probes cover approval/rejection with dict/string/bytes/bytearray payloads, nested payload preservation, unchanged inputs, and invalid values through the resume/parser helpers.

Small nonblocking cleanup: isort --check-only flags the imports in test_workflow_hitl_utils.py; please sort them. Production-file Ruff and diff whitespace checks pass. Pyink could not be installed from the offline cache; I did not run the full tox matrix or a live model/HTTP workflow.

[aryanpatil-07]

Hi @BichengWang,

Thanks for the thorough review and testing!

I've sorted the imports in tests/unittests/workflow/utils/test_workflow_hitl_utils.py per isort in commit acd2d4e and verified that isort --check-only now passes cleanly.