update app.py

[github-actions[bot]]

(🤖 AI Generated)

Summary

Refactored app.py to enhance modularity, update command-line argument parsing, and introduce new agent configurations. Moved AppConfig and GraphExecutor out of app.py, adjusted default behaviors, and added new command-line options for improved flexibility.

Files Changed

• code/app.py

Code Changes

1. Removed Local AppConfig Class

• Deleted the AppConfig dataclass from app.py.
• Imported AppConfig from fabric_agent_action.config.

- from dataclasses import dataclass
- from typing import Optional, TextIO
- from typing_extensions import Literal
+ from typing import TextIO
+ from fabric_agent_action.config import AppConfig

2. Modified Argument Parsing

• Made --input-file Required

  - parser.add_argument(
  -     "-i",
  -     "--input-file",
  -     type=argparse.FileType("r"),
  -     default=sys.stdin,
  -     help="Input file (default: stdin)",
  + parser.add_argument(
  +     "-i",
  +     "--input-file",
  +     type=argparse.FileType("r"),
  +     required=True,
  +     help="Input file",
  )

• Updated --agent-type Choices and Defaults

  - agent_group.add_argument(
  -     "--agent-type",
  -     type=str,
  -     choices=["single_command", "react"],
  -     default="single_command",
  -     help="Type of agent (default: single_command)",
  + agent_group.add_argument(
  +     "--agent-type",
  +     type=str,
  +     choices=["router", "react", "react_issue", "react_pr"],
  +     default="router",
  +     help="Type of agent (default: router)",
  )

• Added New Agent Configuration Options

  agent_group.add_argument(
    "--agent-preamble-enabled",
    action="store_true",
    help="Enable preamble in output",
  )
  agent_group.add_argument(
    "--agent-preamble",
    type=str,
    default="##### (🤖 AI Generated)",
    help="Preamble added to the beginning of output (default: ##### (🤖 AI Generated)",
  )

• Included Defaults for Fabric Patterns and Max Turns

  fabric_group.add_argument(
    "--fabric-patterns-included",
    type=str,
    default="",
    help="Comma separated list of fabric patterns to include in agent",
  )
  fabric_group.add_argument(
    "--fabric-patterns-excluded",
    type=str,
    default="",
    help="Comma separated list of fabric patterns to exclude in agent",
  )
  fabric_group.add_argument(
    "--fabric-max-num-turns",
    type=int,
    default=10,
    help="Maximum number of turns to LLM when running fabric patterns (default: 10)",
  )

3. Refactored Execution Flow

• Moved Execution Logic to app Function

  - def main() -> None:
  -     # Existing code...
  -     input_str = read_input(config.input_file)
  -     # Agent initialization and execution
  -     executor.execute(graph, input_str)
  + def app(config: AppConfig) -> None:
  +     input_str = read_input(config.input_file)
  +     # Agent initialization and execution
  +     executor.execute(graph, input_str)

• Simplified main Function

  def main() -> None:
    try:
        config = parse_arguments()
        setup_logging(config.verbose, config.debug)
        logger.info("Starting Fabric Agent Action")
        app(config)
        logger.info("Fabric Agent Action completed successfully")
    except Exception as e:
        logger.error(f"An error occurred: {str(e)}")
        sys.exit(1)
    finally:
        if config.output_file is not sys.stdout:
            config.output_file.close()

4. Removed GraphExecutor Class

• Deleted the GraphExecutor class from app.py.

• Introduced GraphExecutorFactory from fabric_agent_action.graphs.

  - class GraphExecutor:
  -     """Handles graph execution and output generation"""
  -     # Implementation details...
  + from fabric_agent_action.graphs import GraphExecutorFactory

• Updated execution to use the factory:

  executor = GraphExecutorFactory.create(config)
  executor.execute(graph, input_str)

Reason for Changes

• Modularity: Moving AppConfig and GraphExecutor to their respective modules improves code organization and reusability.
• Clarity: Separating the application logic into an app function makes app.py cleaner and the execution flow more understandable.
• Flexibility: Updating command-line arguments and defaults allows users to have more control over agent configurations.
• Maintainability: Modular code is easier to maintain and test, facilitating future enhancements.

Impact of Changes

• User Experience: Users must now specify an input file explicitly; the application no longer defaults to reading from stdin.
• Agent Configuration: Additional agent types and configurations provide users with more options to tailor the agent's behavior.
• Codebase: The application is better structured, making it easier for new contributors to understand and for developers to maintain.

Test Plan

• Unit Tests: Run existing unit tests to ensure no regressions have been introduced.
• Command-Line Testing: Verify that the application correctly parses new command-line arguments and handles required inputs.
  • Test with various --agent-type options.
  • Test enabling and customizing --agent-preamble.
  • Test the new fabric pattern options and --fabric-max-num-turns.
• Integration Tests: Ensure that the application integrates correctly with the modularized AppConfig and GraphExecutorFactory.

Additional Notes

• Documentation: Update the README and any relevant documentation to reflect changes in command-line arguments and defaults.
• Backward Compatibility: Users relying on default stdin input or previous default agent types will need to adjust their usage.
• Dependency Management: Ensure that fabric_agent_action.config and fabric_agent_action.graphs modules are correctly installed and accessible.

Shortlist of Changes for Reviewers

• Confirm the removal of the local AppConfig and its import from the config module.
• Review the new command-line arguments and defaults for correctness and usability.
• Check the refactored execution flow for any potential issues with application initialization.
• Verify that the deletion of GraphExecutor and the use of GraphExecutorFactory maintains existing functionality.

End of PR Description