Fix output type loss in AgentOperator's human-in-the-loop review

[ColtenOuO]

Summary

AgentOperator's human-in-the-loop review path (enable_hitl_review=True) loses the original output type for any output_type that is neither str nor a Pydantic BaseModel (e.g. list[str], bool, dict).

The output is round-tripped through a string while it's shown to a human reviewer: HITLReviewMixin._to_string serializes it before review, and AgentOperator.execute() deserializes it back into output_type after approval. The serialization side used str(output), which produces a Python repr (e.g. "['tag-a', 'tag-b']"), not valid JSON.

The deserialization side then tried json.loads() on that repr, which always raised, and silently fell back to returning the raw repr string instead of the original list/bool/dict — a type change downstream tasks don't expect.

regenerate_with_feedback() (used when a reviewer requests changes) had the identical str(output) bug.

This is the same defect class already fixed for the require_approval=True path in #70075, which switched to TypeAdapter(...).dump_json(...). That fix wasn't applied to the separate enable_hitl_review=True path, so it regressed here.

Changes

• HITLReviewMixin._to_string and AgentOperator.regenerate_with_feedback
  now serialize non-str/non-BaseModel output with
  TypeAdapter(type(output)).dump_json(output), matching the pattern
  already used by LLMApprovalMixin.defer_for_approval.
• AgentOperator.execute()'s HITL-review branch now reuses the existing
  rehydrate_pydantic_output helper (already used by the approval path)
  instead of a bespoke json.loads/except fallback, so both review paths
  behave consistently.

---

Was generative AI tooling used to co-author this PR?

• Yes — Claude Code (Sonnet 5)

Generated-by: Claude Code (Sonnet 5) following the guidelines

[kaxil]

Dropping the json.loads fallback here regresses the exact case this PR targets. rehydrate_pydantic_output returns raw unchanged when output_type isn't a BaseModel subclass (its docstring notes the caller is expected to apply its own json.loads), so with output_type=list[str] this returns the JSON string '["tag-a","tag-b"]' instead of the list. The new test_execute_with_hitl_rehydrates_non_base_model_output fails on exactly this (assert '["tag-a", "tag-b"]' == ['tag-a', 'tag-b']). Keeping the previous json.loads/except fallback after the rehydrate call fixes it while preserving the BaseModel handling.

[ColtenOuO]

In PR #70075 (merged), rehydrate_pydantic_output was updated to handle any non-str output_type, not just BaseModel subclasses:

if output_type is str:
    return raw
try:
    rehydrated = TypeAdapter(output_type).validate_json(raw)
except (ValidationError, ValueError, TypeError):
    return raw

uv run --project providers/common/ai python -c "tic_output
from airflow.providers.common.ai.utils.output_type import rehydrate_pydantic_outputtput=False)
result = rehydrate_pydantic_output(list[str], '[\"tag-a\",\"tag-b\"]', serialize_output=False)
print(repr(result), type(result))
"

Output:

['tag-a', 'tag-b'] <class 'list'>

So I don't think the json.loads fallback needs to come back (?

Let me know what you think, or if I might have missed anything!

Thanks for your review!