chore: Reset to stable state

.github/workflows/code-embedder.yml

@@ -2,15 +2,20 @@ name: Code Embedder
 
 on: pull_request
 
+permissions:
+  contents: write
+
 jobs:
   code_embedder:
     name: "Code embedder"
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
+        with:
+          ref: ${{ github.event.pull_request.head.ref }}
 
       - name: Run code embedder
-        uses: kvankova/code-embedder@v0.4.0
+        uses: kvankova/code-embedder@v1.1.1
         env:
           GITHUB_TOKEN: ${{ secrets.CODE_EMBEDDER }}

.pre-commit-config.yaml

@@ -30,14 +30,6 @@ repos:
         types:
           - python
 
-      - id: mypy
-        name: mypy
-        entry: poetry run mypy .
-        pass_filenames: false
-        language: system
-        types:
-          - python
-
       - id: run-pytest
         name: Run tests
         entry: make test

README.md

@@ -46,29 +46,37 @@ pip install -e .
 ## 🌱 Getting Started
 Creating an agent is very simple. Start by subclassing an `Agent` class just like
 [`RandomAgent`](./generals/agents/random_agent.py) or [`ExpanderAgent`](./generals/agents/expander_agent.py).
-You can specify your agent `id` (name) and `color` and the only thing remaining is to implement the `act` function,
+You can specify your agent `id` (name) and the only thing remaining is to implement the `act` function,
 that has the signature explained in sections down below.
 
 
-### Usage Example (🤸 Gymnasium)
+### Usage Example (🦁 PettingZoo)
 The example loop for running the game looks like this
-```python:examples/gymnasium_example.py
-import gymnasium as gym
-
+```python:examples/pettingzoo_example.py
 from generals.agents import RandomAgent, ExpanderAgent
+from generals.envs import PettingZooGenerals
 
 # Initialize agents
-agent = RandomAgent()
-npc = ExpanderAgent()
+random = RandomAgent(id="Random")
+expander = ExpanderAgent(id="Expander")
+
+# Store agents in a dictionary
+agents = {
+    random.id: random,
+    expander.id: expander
+}
 
 # Create environment
-env = gym.make("gym-generals-v0", agent=agent, npc=npc, render_mode="human")
+env = PettingZooGenerals(agent_ids=[random.id, expander.id], to_render=True)
+observations, info = env.reset()
 
-observation, info = env.reset()
 terminated = truncated = False
 while not (terminated or truncated):
-    action = agent.act(observation)
-    observation, reward, terminated, truncated, info = env.step(action)
+    actions = {}
+    for agent in env.agent_ids:
+        actions[agent] = agents[agent].act(observations[agent])
+    # All agents perform their actions
+    observations, rewards, terminated, truncated, info = env.step(actions)
     env.render()
 
 ```
@@ -80,7 +88,7 @@ while not (terminated or truncated):
 Grids on which the game is played on are generated via `GridFactory`. You can instantiate the class with desired grid properties, and it will generate
 grid with these properties for each run.
 ```python
-import gymnasium as gym
+from generals.envs import PettingZooGenerals
 from generals import GridFactory
 
 grid_factory = GridFactory(
@@ -89,21 +97,20 @@ grid_factory = GridFactory(
     mountain_density=0.2,                  # Probability of a mountain in a cell
     city_density=0.05,                     # Probability of a city in a cell
     general_positions=[(0,3),(5,7)],       # Positions of generals (i, j)
-    padding=True                           # Whether to pad grids to max_grid_dims with mountains (default=True)
 )
 
 # Create environment
-env = gym.make(
-    "gym-generals-v0",
+env = PettingZooGenerals(
     grid_factory=grid_factory,
     ...
 )
 ```
 You can also specify grids manually, as a string via `options` dict:
 ```python
-import gymnasium as gym
+from generals.envs import PettingZooGenerals
+
+env = PettingZooGenerals(agent_ids=[agent1.id, agent2.id])
 
-env = gym.make("gym-generals-v0", ...)
 grid = """
 .3.#
 #..A
@@ -123,13 +130,14 @@ Grids are created using a string format where:
 - numbers `0-9` and `x`, where `x=10`, represent cities, where the number specifies amount of neutral army in the city,
   which is calculated as `40 + number`. The reason for `x=10` is that the official game has cities in range `[40, 50]`
 
+> [!TIP]
+> Check out [complete example](./examples/complete_example.py) for concrete example in the wild!
+
 ## 🔬 Interactive Replays
 We can store replays and then analyze them in an interactive fashion. `Replay` class handles replay related functionality.
 ### Storing a replay
 ```python
-import gymnasium as gym
-
-env = gym.make("gym-generals-v0", ...)
+env = ...
 
 options = {"replay_file": "my_replay"}
 env.reset(options=options) # The next game will be encoded in my_replay.pkl
@@ -220,15 +228,19 @@ or create your own connection workflow. Our implementations expect that your age
 implemented the required methods.
 ```python:examples/client_example.py
 from generals.remote import autopilot
+from generals.agents import ExpanderAgent
+
 import argparse
+
 parser = argparse.ArgumentParser()
 parser.add_argument("--user_id", type=str, default=...) # Register yourself at generals.io and use this id
-parser.add_argument("--lobby_id", type=str, default=...) # The last part of the lobby url
-parser.add_argument("--agent_id", type=str, default="Expander") # agent_id should be "registered" in AgentFactory
+parser.add_argument("--lobby_id", type=str, default="psyo") # After you create a private lobby, copy last part of the url
 
 if __name__ == "__main__":
     args = parser.parse_args()
-    autopilot(args.agent_id, args.user_id, args.lobby_id)
+    agent = ExpanderAgent()
+    autopilot(agent, args.user_id, args.lobby_id)
+
 ```
 This script will run `ExpanderAgent` in the specified lobby.
 ## 🙌 Contributing

examples/gymnasium_example.py

@@ -1,17 +1,28 @@
+#  type: ignore
 import gymnasium as gym
+import numpy as np
 
-from generals.agents import RandomAgent, ExpanderAgent
+from generals.envs import GymnasiumGenerals
 
-# Initialize agents
-agent = RandomAgent()
-npc = ExpanderAgent()
+agent_names = ["007", "Generalissimo"]
 
-# Create environment
-env = gym.make("gym-generals-v0", agent=agent, npc=npc, render_mode="human")
+n_envs = 12
+envs = gym.vector.AsyncVectorEnv(
+    [lambda: GymnasiumGenerals(agents=agent_names, truncation=500) for _ in range(n_envs)],
+)
 
-observation, info = env.reset()
-terminated = truncated = False
-while not (terminated or truncated):
-    action = agent.act(observation)
-    observation, reward, terminated, truncated, info = env.step(action)
-    env.render()
+
+observations, infos = envs.reset()
+terminated = [False] * len(observations)
+truncated = [False] * len(observations)
+
+while True:
+    agent_actions = [envs.single_action_space.sample() for _ in range(n_envs)]
+    npc_actions = [envs.single_action_space.sample() for _ in range(n_envs)]
+
+    # Stack actions together
+    actions = np.stack([agent_actions, npc_actions], axis=1)
+    observations, rewards, terminated, truncated, infos = envs.step(actions)
+    masks = [np.stack([info[-1] for info in infos[agent_name]]) for agent_name in agent_names]
+    if any(terminated) or any(truncated):
+        break

generals/__init__.py

@@ -1,39 +1,18 @@
-from gymnasium.envs.registration import register
-
 from generals.agents.agent import Agent
 from generals.core.game import Action
 from generals.core.grid import Grid, GridFactory
 from generals.core.observation import Observation
 from generals.core.replay import Replay
+from generals.envs.gymnasium_generals import GymnasiumGenerals
 from generals.envs.pettingzoo_generals import PettingZooGenerals
 
 __all__ = [
     "Action",
     "Agent",
     "GridFactory",
     "PettingZooGenerals",
+    "GymnasiumGenerals",
     "Grid",
     "Replay",
     "Observation",
-    "GeneralsIOClientError",
 ]
-
-
-def _register_gym_generals_envs():
-    register(
-        id="gym-generals-v0",
-        entry_point="generals.envs.gymnasium_generals:GymnasiumGenerals",
-    )
-
-    register(
-        id="gym-generals-image-v0",
-        entry_point="generals.envs.initializers:gym_image_observations",
-    )
-
-    register(
-        id="gym-generals-rllib-v0",
-        entry_point="generals.envs.initializers:gym_rllib",
-    )
-
-
-_register_gym_generals_envs()

generals/envs/__init__.py

@@ -1,10 +1,8 @@
 #  type: ignore
 from generals.envs.gymnasium_generals import GymnasiumGenerals
-from generals.envs.multiagent_gymnasium_generals import MultiAgentGymnasiumGenerals
 from generals.envs.pettingzoo_generals import PettingZooGenerals
 
 __all__ = [
     "PettingZooGenerals",
     "GymnasiumGenerals",
-    "MultiAgentGymnasiumGenerals",
 ]