Overview

Wrappers offer a flexible approach to enhancing or modifying an existing environment without changing its core implementation. By leveraging wrappers, you can reduce redundant code, promote modularity, and streamline your workflow. A significant advantage of wrappers is their ability to be stacked, allowing you to combine multiple enhancements seamlessly.

To apply a wrapper, you first need to initialize the base environment. Afterward, you can instantiate the wrapper by passing the environment and any required parameters to its constructor.

1# Example usage:
2import textarena as ta
3
4env = ta.make(env_id="UltimateTicTacToe-v0")
5env = ta.wrappers.LLMObservationWrapper(env=env)
6env = ta.wrappers.PrettyRenderWrapper(
7    env=env, 
8    record_video=True,
9    video_path="ultimatetictactoe_game.mp4"
10)

There are three common things you might want a wrapper to do:

• Transform actions before applying them to the base environment
• Transform observations that are returned by the base environment
• Apply a render to base environment for friendly visualisations

Base Wrapper Class

The base class for environment wrappers. It intercepts all calls to the wrapped environment and forwards them unless explicitly overridden.

Methods

Executes a single step in the environment based on the action provided by the agent.

Resets the environment and initializes it with a given seed. If no seed is provided, a random one is used.

The reset method is critical for initializing the environment to a consistent state, ensuring reproducibility when a seed is provided.

Retrieves the current player ID and their observations, providing insights into the current state of the environment.

Closes the environment and performs cleanup operations. This is essential for releasing resources or saving final states.