🔨 Modularizing Event Loop Logic

Commit: 8cb8422
Author: Fede Kamelhar
Date: May 25, 2025
Message: Modularizing Event Loop

⸻

✨ What’s Included

1. Extracted Tool Execution Logic

Moved all logic related to handling tool_use into a dedicated private function: _handle_tool_execution().

Benefit:
Improves readability, simplifies event_loop_cycle(), and separates concerns for better maintainability and testing.

New function:

def _handle_tool_execution(...) -> Tuple[StopReason, Message, EventLoopMetrics, Dict[str, Any]]:

if stop_reason == "tool_use":
    return _handle_tool_execution(...)

2. Centralized Retry Configuration

Introduced constants for retry settings, replacing hardcoded values:

MAX_ATTEMPTS = 6
INITIAL_DELAY = 4
MAX_DELAY = 240  # 4 minutes

Benefit:
Easier tuning of retry behavior and removal of magic numbers.

3. Type Hint Improvements

Refined function signatures for stronger typing:
• initialize_state() now accepts a Dict[str, Any] instead of **kwargs.
• callback_handler is now typed as Callable[..., Any].

Benefit:
Improves IDE support, readability, and static analysis.

⸻

4. Small Refactors & Cleanup
   • Replaced inline retry configuration with named constants.
   • Removed redundant state handling and improved control flow.
   • Added a detailed docstring to _handle_tool_execution().

5. Updated Tests

Updated test_initialize_state() to use the new function signature:

kwargs = strands.event_loop.event_loop.initialize_state(kwargs)

Benefit:
Keeps tests aligned with refactored API for better coverage and accuracy.