multi-agent-rlenv 1.0.0__tar.gz

multi_agent_rlenv-1.0.0/.github/workflows/ci.yaml

@@ -0,0 +1,95 @@
+name: Build & Test
+
+on:
+  push:
+    branches:
+      - main
+      - master
+      - dev
+    tags:
+      - '*'
+  pull_request:
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  test:
+    name: 🧪 Test
+    strategy:
+      matrix:
+        os:
+          - ubuntu-latest
+          - windows-latest
+          - macOS-latest
+        target:
+          - x86_64
+          - aarch64
+        python-version: 
+          - '3.10'
+          - '3.11'
+          - '3.12' # Dependency ale-py does not yet support Python 3.12
+    runs-on: ${{ matrix.os }}
+    steps:
+      # Checkout the repository
+      - name: Checkout
+        uses: actions/checkout@v4
+      - name: Python setup
+        uses: actions/setup-python@v5
+        with:
+          python-version: ${{ matrix.python-version }}
+      - name: Install uv
+        uses: yezz123/setup-uv@v4
+        with:
+          uv-version: 0.3.0
+      - name: Install dependencies and run pytest
+        run: |
+          uv sync
+          uv run pytest -n 4
+
+  build:
+    name: 📦 Build package
+    if: startsWith(github.ref, 'refs/tags/')  # only publish to PyPI on tag pushes
+    needs: test
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@v4
+    - name: Set up Python
+      uses: actions/setup-python@v5
+      with:
+        python-version: 3.12
+    - name: Install UV
+      uses: yezz123/setup-uv@v4
+      with:
+          uv-version: 0.3.0
+    - name: Build wheels
+      run: |
+        uv venv
+        uv pip install pip build
+        uv run python -m build
+    
+    - name: 📡 Upload package distributions
+      uses: actions/upload-artifact@v3
+      with:
+        name: wheels
+        path: dist/
+
+
+  publish-to-pypi:
+    name: 📤 Publish to PyPI
+    needs: build
+    runs-on: ubuntu-latest
+    environment:
+      name: pypi
+      url: https://pypi.org/p/multi-agent-rlenv 
+    permissions:
+      id-token: write  # IMPORTANT: this permission is mandatory for trusted publishing
+    steps:
+    - name: 📡 Download package distributions
+      uses: actions/download-artifact@v3
+      with:
+        name: wheels
+        path: dist/
+    - name: 📤 Publish 📦 to PyPI
+      uses: pypa/gh-action-pypi-publish@release/v1

multi_agent_rlenv-1.0.0/.gitignore

@@ -0,0 +1,187 @@
+
+# Created by https://www.toptal.com/developers/gitignore/api/python,visualstudiocode
+# Edit at https://www.toptal.com/developers/gitignore?templates=python,visualstudiocode
+
+### Python ###
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+#poetry.lock
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+./env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#  and can be added to the global gitignore or merged into this file.  For a more nuclear
+#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
+#.idea/
+
+### VisualStudioCode ###
+.vscode
+
+
+# Local History for Visual Studio Code
+.history/
+
+# Built Visual Studio Code Extensions
+*.vsix
+
+### VisualStudioCode Patch ###
+# Ignore all local history of files
+.history
+.ionide
+
+# Support for Project snippet scope
+
+# End of https://www.toptal.com/developers/gitignore/api/python,visualstudiocode
+
+# Do not push nohup logs
+nohup.out
+# Do not push Docker configuration
+.devcontainer
+# Do not push Tensorboard reports
+runs
+logs
+data/game_images/*.png
+poetry.lock
+uv.lock

multi_agent_rlenv-1.0.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2023 Yannick Molinghen
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

multi_agent_rlenv-1.0.0/PKG-INFO

@@ -0,0 +1,50 @@
+Metadata-Version: 2.3
+Name: multi-agent-rlenv
+Version: 1.0.0
+Summary: A strongly typed Multi-Agent Reinforcement Learning framework
+Project-URL: repository, https://github.com/yamoling/ma-rlenv
+Author-email: Yannick Molinghen <yannick.molinghen@ulb.be>
+License-File: LICENSE
+Classifier: Operating System :: OS Independent
+Classifier: Programming Language :: Python :: 3
+Requires-Python: <4,>=3.10
+Requires-Dist: gymnasium>=0.29.1
+Requires-Dist: numpy>=2.0.0
+Requires-Dist: opencv-python>=4.10.0.84
+Description-Content-Type: text/markdown
+
+# RLEnv: yet another RL framework
+This framework aims at high level abstractions of RL models, allowing to build algorithms on top of it.
+
+## Designing an environment
+To create an environment that is compatible with RLEnv, you should inherit from the `RLEnv` class.
+
+## Instanciating an environment
+### Simple environments
+```python
+import marlenv as menv
+print(menv.__version__)
+
+# From Gym
+env = menv.make("CartPole-v1")
+
+# From pettingzoo
+from pettingzoo.sisl import pursuit_v4
+env = menv.make(pursuit_v4.parallel_env())
+```
+
+### Adding extra information to the observations
+```python
+import marlenv as menv
+# Building the environment with additional information
+from pettingzoo.sisl import pursuit_v4
+env = menv.Builder(pursuit_v4.parallel_env())\
+    .with_agent_id()\
+    .with_last_action()\
+    .build()
+# 8 agents  + 5 actions = 13 extras
+assert env.extra_feature_shape == (13, )
+```
+
+# Related projects
+- MARL: multi-agent reinforcement framework [https://github.com/yamoling/marl](https://github.com/yamoling/marl)

multi_agent_rlenv-1.0.0/README.md

@@ -0,0 +1,35 @@
+# RLEnv: yet another RL framework
+This framework aims at high level abstractions of RL models, allowing to build algorithms on top of it.
+
+## Designing an environment
+To create an environment that is compatible with RLEnv, you should inherit from the `RLEnv` class.
+
+## Instanciating an environment
+### Simple environments
+```python
+import marlenv as menv
+print(menv.__version__)
+
+# From Gym
+env = menv.make("CartPole-v1")
+
+# From pettingzoo
+from pettingzoo.sisl import pursuit_v4
+env = menv.make(pursuit_v4.parallel_env())
+```
+
+### Adding extra information to the observations
+```python
+import marlenv as menv
+# Building the environment with additional information
+from pettingzoo.sisl import pursuit_v4
+env = menv.Builder(pursuit_v4.parallel_env())\
+    .with_agent_id()\
+    .with_last_action()\
+    .build()
+# 8 agents  + 5 actions = 13 extras
+assert env.extra_feature_shape == (13, )
+```
+
+# Related projects
+- MARL: multi-agent reinforcement framework [https://github.com/yamoling/marl](https://github.com/yamoling/marl)

multi_agent_rlenv-1.0.0/pyproject.toml

@@ -0,0 +1,39 @@
+[project]
+name = "multi-agent-rlenv"
+version = "1.0.0"
+description = "A strongly typed Multi-Agent Reinforcement Learning framework"
+authors = [
+    { "name" = "Yannick Molinghen", "email" = "yannick.molinghen@ulb.be" },
+]
+readme = "README.md"
+requires-python = ">=3.10, <4"
+dependencies = ["numpy>=2.0.0", "opencv-python>=4.10.0.84", "gymnasium>=0.29.1"]
+urls = { "repository" = "https://github.com/yamoling/ma-rlenv" }
+classifiers = [
+    "Programming Language :: Python :: 3",
+    "Operating System :: OS Independent",
+]
+
+
+[build-system]
+requires = ["hatchling"]
+build-backend = "hatchling.build"
+
+[tool.ruff]
+line-length = 140
+
+[tool.hatch.build.targets.wheel]
+packages = ["src/marlenv"]
+
+[tool.hatch.metadata]
+allow-direct-references = true
+
+[tool.uv]
+dev-dependencies = ["pytest>=8.3.2", "pytest-xdist>=3.6.1"]
+
+
+[tool.pytest.ini_options]
+# Run tests on the maximum number of available CPUs
+addopts = "-n auto"
+testpaths = ["tests"]
+pythonpath = "src"

multi_agent_rlenv-1.0.0/src/marlenv/__init__.py

@@ -0,0 +1,52 @@
+"""
+RLEnv is a strongly typed library for multi-agent and multi-objective reinforcement learning.
+
+RLEnv
+    - provides a simple and consistent interface for reinforcement learning environments
+    - provides fundamental models such as `Observation`s, `Episode`s, `Transition`s, ...
+    - works with gymnasium, pettingzoo and SMAC out of the box
+    - provides helpful wrappers to add intrinsic rewards, agent ids, record videos, ...
+"""
+
+__version__ = "1.0.4"
+
+from . import models
+from . import wrappers
+from . import adapters
+from .models import spaces
+
+
+from .env_builder import make, Builder
+from .models import (
+    RLEnv,
+    Observation,
+    Episode,
+    EpisodeBuilder,
+    Transition,
+    DiscreteSpace,
+    ContinuousSpace,
+    ActionSpace,
+    DiscreteActionSpace,
+    ContinuousActionSpace,
+)
+from .mock_env import MockEnv
+
+__all__ = [
+    "models",
+    "wrappers",
+    "adapters",
+    "spaces",
+    "make",
+    "Builder",
+    "RLEnv",
+    "Observation",
+    "Episode",
+    "EpisodeBuilder",
+    "Transition",
+    "ActionSpace",
+    "DiscreteSpace",
+    "ContinuousSpace",
+    "DiscreteActionSpace",
+    "ContinuousActionSpace",
+    "MockEnv",
+]

multi_agent_rlenv-1.0.0/src/marlenv/adapters/__init__.py

@@ -0,0 +1,24 @@
+from .pymarl_adapter import PymarlAdapter
+from typing import Any
+
+__all__ = ["PymarlAdapter"]
+try:
+    from .gym_adapter import Gym
+
+    __all__.append("Gym")
+except ImportError:
+    Gym = Any
+
+try:
+    from .pettingzoo_adapter import PettingZoo
+
+    __all__.append("PettingZoo")
+except ImportError:
+    PettingZoo = Any
+
+try:
+    from .smac_adapter import SMAC
+
+    __all__.append("SMAC")
+except ImportError:
+    SMAC = Any

multi_agent_rlenv-1.0.0/src/marlenv/adapters/gym_adapter.py

@@ -0,0 +1,64 @@
+from gymnasium import Env, spaces
+import numpy as np
+
+from marlenv.models import (
+    RLEnv,
+    Observation,
+    ActionSpace,
+    DiscreteSpace,
+    ContinuousSpace,
+)
+
+
+class Gym(RLEnv[ActionSpace]):
+    """Wraps a gym envronment in an RLEnv"""
+
+    def __init__(self, env: Env):
+        if env.observation_space.shape is None:
+            raise NotImplementedError("Observation space must have a shape")
+        match env.action_space:
+            case spaces.Discrete() as s:
+                space = ActionSpace(1, DiscreteSpace(int(s.n)))
+            case spaces.Box() as s:
+                low = s.low.astype(np.float32)
+                high = s.high.astype(np.float32)
+                if not isinstance(low, np.ndarray):
+                    low = np.full(s.shape, s.low, dtype=np.float32)
+                if not isinstance(high, np.ndarray):
+                    high = np.full(s.shape, s.high, dtype=np.float32)
+                space = ActionSpace(1, ContinuousSpace(low=low, high=high))
+            case other:
+                raise NotImplementedError(f"Action space {other} not supported")
+        super().__init__(space, env.observation_space.shape, (1,))
+        self.env = env
+        if self.env.unwrapped.spec is not None:
+            self.name = self.env.unwrapped.spec.id
+        else:
+            self.name = "gym-no-id"
+
+    def step(self, actions):
+        obs_, reward, done, truncated, info = self.env.step(list(actions)[0])
+        obs_ = Observation(
+            np.array([obs_], dtype=np.float32),
+            self.available_actions(),
+            self.get_state(),
+        )
+        return obs_, np.array([reward], dtype=np.float32), done, truncated, info
+
+    def get_state(self):
+        return np.zeros(1, dtype=np.float32)
+
+    def reset(self):
+        obs_data, _info = self.env.reset()
+        obs = Observation(
+            np.array([obs_data], dtype=np.float32),
+            self.available_actions(),
+            self.get_state(),
+        )
+        return obs
+
+    def render(self, mode: str = "human"):
+        return self.env.render()
+
+    def seed(self, seed_value: int):
+        self.env.reset(seed=seed_value)

multi_agent_rlenv-1.0.0/src/marlenv/adapters/pettingzoo_adapter.py

@@ -0,0 +1,58 @@
+from pettingzoo import ParallelEnv
+from gymnasium import spaces  # pettingzoo uses gymnasium spaces
+from marlenv.models import RLEnv, Observation, ActionSpace, DiscreteActionSpace, ContinuousActionSpace, ContinuousSpace
+import numpy as np
+import numpy.typing as npt
+
+
+class PettingZoo(RLEnv[ActionSpace]):
+    def __init__(self, env: ParallelEnv):
+        env.reset()
+        aspace = env.action_space(env.possible_agents[0])
+
+        match aspace:
+            case spaces.Discrete() as s:
+                space = DiscreteActionSpace(env.num_agents, int(s.n))
+
+            case spaces.Box() as s:
+                low = s.low.astype(np.float32)
+                high = s.high.astype(np.float32)
+                if not isinstance(low, np.ndarray):
+                    low = np.full(s.shape, s.low, dtype=np.float32)
+                if not isinstance(high, np.ndarray):
+                    high = np.full(s.shape, s.high, dtype=np.float32)
+                space = ContinuousActionSpace(env.num_agents, low, high=high)
+            case other:
+                raise NotImplementedError(f"Action space {other} not supported")
+
+        obs_space = env.observation_space(env.possible_agents[0])
+        if obs_space.shape is None:
+            raise NotImplementedError("Only discrete observation spaces are supported")
+        self._env = env
+        super().__init__(space, obs_space.shape, self.get_state().shape)
+        self.agents = env.possible_agents
+
+    def get_state(self):
+        try:
+            return self._env.state()
+        except NotImplementedError:
+            return np.array([0])
+
+    def step(self, actions: npt.NDArray[np.int64]):
+        action_dict = dict(zip(self.agents, actions))
+        obs, reward, term, trunc, info = self._env.step(action_dict)
+        obs_data = np.array([v for v in obs.values()])
+        reward = np.sum([r for r in reward.values()], keepdims=True)
+        observation = Observation(obs_data, self.available_actions(), self.get_state())
+        return observation, reward, any(term.values()), any(trunc.values()), info
+
+    def reset(self) -> Observation:
+        obs = self._env.reset()[0]
+        obs_data = np.array([v for v in obs.values()])
+        return Observation(obs_data, self.available_actions(), self.get_state())
+
+    def seed(self, seed_value: int):
+        self._env.reset(seed=seed_value)
+
+    def render(self, *_):
+        return self._env.render()