Feature/upgrade deps and modernize tests

README.md

@@ -1,3 +1,7 @@
+> **Note:** This is a temporary fork of [HumanCompatibleAI/overcooked_ai](https://github.com/HumanCompatibleAI/overcooked_ai), with full credit to [Micah Carroll](https://github.com/micahcarroll) and team for the original work. The purpose of this fork is to bring all Python dependencies up to date (Python 3.13, numpy 2.x, gymnasium, etc.) and modernize the test framework. We will close this fork if/when these changes can be merged into the original repository.
+
+---
+
 ![MDP python tests](https://github.com/HumanCompatibleAI/overcooked_ai/workflows/.github/workflows/pythontests.yml/badge.svg) ![overcooked-ai codecov](https://codecov.io/gh/HumanCompatibleAI/overcooked_ai/branch/master/graph/badge.svg) [![PyPI version](https://badge.fury.io/py/overcooked-ai.svg)](https://badge.fury.io/py/overcooked-ai) [!["Open Issues"](https://img.shields.io/github/issues-raw/HumanCompatibleAI/overcooked_ai.svg)](https://github.com/HumanCompatibleAI/minerl/overcooked_ai) [![GitHub issues by-label](https://img.shields.io/github/issues-raw/HumanCompatibleAI/overcooked_ai/bug.svg?color=red)](https://github.com/HumanCompatibleAI/overcooked_ai/issues?utf8=%E2%9C%93&q=is%3Aissue+is%3Aopen+label%3Abug) [![Downloads](https://pepy.tech/badge/overcooked-ai)](https://pepy.tech/project/overcooked-ai)
 [![arXiv](https://img.shields.io/badge/arXiv-1910.05789-bbbbbb.svg)](https://arxiv.org/abs/1910.05789)
 

pyproject.toml

@@ -1,38 +1,37 @@
 [build-system]
-requires = ["setuptools>=77.0.3"]
+requires = ["setuptools>=82.0.1"]
 build-backend = "setuptools.build_meta"
 
 [project]
 name = "overcooked_ai"
 version = "2.0.0"
 description = "Cooperative multi-agent environment based on Overcooked"
 readme = "README.md"
-requires-python = ">=3.10,<3.11"
+requires-python = ">=3.13"
 license = { file = "LICENSE" }
 authors = [{ name = "Micah Carroll", email = "mdc@berkeley.edu" }]
 urls = { homepage = "https://github.com/HumanCompatibleAI/overcooked_ai" }
 keywords = ["Overcooked", "AI", "Reinforcement Learning"]
 dependencies = [
-    "dill",
-    "gymnasium",
-    "ipywidgets",
-    "numpy<2",
-    "opencv-python",
-    "pygame",
-    "scipy",
-    "setuptools>=77.0.3",
-    "tqdm",
+    "dill>=0.4.1",
+    "gymnasium>=1.2.2",
+    "ipywidgets>=8.1.8",
+    "numpy>=2.4.4",
+    "opencv-python>=4.13.0.92",
+    "pygame>=2.6.1",
+    "scipy>=1.17.1",
+    "setuptools>=82.0.1",
+    "tqdm>=4.67.3",
 ]
 
 [project.optional-dependencies]
 harl = [
-    "ipykernel",
-    "ray[rllib,tune]==2.2",
-    "dm-tree",
-    "lz4",
-    "tensorflow[and-cuda]==2.19.0",
-    "wandb",
-    "gym",
+    "ipykernel>=7.2.0",
+    "ray[rllib,tune]>=2.54.1",
+    "dm-tree>=0.1.10",
+    "lz4>=4.4.5",
+    "tensorflow[and-cuda]>=2.21.0",
+    "wandb>=0.25.1",
 ]
 
 [project.scripts]

setup.py

@@ -38,32 +38,32 @@
         ],
     },
     install_requires=[
-        "dill",
-        "numpy<2.0.0",
-        "scipy",
-        "tqdm",
-        "gymnasium",
-        "ipython",
-        "pygame",
-        "ipywidgets",
-        "opencv-python",
-        "flask",
-        "flask-socketio",
+        "dill>=0.4.1",
+        "numpy>=2.4.4",
+        "scipy>=1.17.1",
+        "tqdm>=4.67.3",
+        "gymnasium>=1.2.2",
+        "ipython>=9.12.0",
+        "pygame>=2.6.1",
+        "ipywidgets>=8.1.8",
+        "opencv-python>=4.13.0.92",
+        "flask>=3.1.3",
+        "flask-socketio>=5.6.1",
     ],
     # removed overlapping dependencies
     extras_require={
         "harl": [
-            "wandb",
-            "GitPython",
-            "memory_profiler",
-            "sacred",
-            "pymongo",
-            "matplotlib",
-            "requests",
-            "seaborn==0.9.0",
-            "ray[rllib]>=2.5.0",
-            "protobuf",
-            "tensorflow>=2.14.0",
+            "wandb>=0.25.1",
+            "GitPython>=3.1.46",
+            "memory_profiler>=0.61.0",
+            "sacred>=0.8.7",
+            "pymongo>=4.16.0",
+            "matplotlib>=3.10.8",
+            "requests>=2.33.1",
+            "seaborn>=0.13.2",
+            "ray[rllib]>=2.54.1",
+            "protobuf>=7.34.1",
+            "tensorflow>=2.21.0",
         ]
     },
     entry_points={

src/human_aware_rl/human/tests.py

@@ -2,11 +2,9 @@
 import os
 import pickle
 import shutil
-import sys
-import unittest
 
 import numpy as np
-from numpy.testing._private.utils import assert_raises
+import pytest
 
 from human_aware_rl.human.process_dataframes import (
     csv_to_df_pickle,
@@ -29,7 +27,7 @@
 )
 
 
-class TestProcessDataFrames(unittest.TestCase):
+class TestProcessDataFrames:
     temp_data_dir = "this_is_a_temp"
     data_len_2019 = 3546
     data_len_2020 = 1189
@@ -51,61 +49,53 @@ class TestProcessDataFrames(unittest.TestCase):
         "layouts": ["cramped_room"],
     }
 
-    def setUp(self):
-        print(
-            "\nIn Class {}, in Method {}".format(
-                self.__class__.__name__, self._testMethodName
-            )
-        )
+    @pytest.fixture(autouse=True)
+    def _setup(self):
         if not os.path.exists(self.temp_data_dir):
             os.makedirs(self.temp_data_dir)
-
-    def tearDown(self):
+        yield
         shutil.rmtree(self.temp_data_dir)
 
     def test_csv_to_df_pickle_2019(self):
         params = copy.deepcopy(self.base_csv_to_df_params)
         params["csv_path"] = DUMMY_2019_RAW_HUMAN_DATA_PATH
         params["button_presses_threshold"] = 0.0
         data = csv_to_df_pickle(**params)
-        self.assertEqual(len(data), self.data_len_2019)
+        assert len(data) == self.data_len_2019
 
         params = copy.deepcopy(self.base_csv_to_df_params)
         params["csv_path"] = DUMMY_2019_RAW_HUMAN_DATA_PATH
         params["button_presses_threshold"] = 0.7
         data = csv_to_df_pickle(**params)
-        self.assertLess(len(data), self.data_len_2019)
+        assert len(data) < self.data_len_2019
 
     def test_csv_to_df_pickle_2020(self):
         params = copy.deepcopy(self.base_csv_to_df_params)
         params["button_presses_threshold"] = 0.0
         data = csv_to_df_pickle(**params)
-        self.assertEqual(len(data), self.data_len_2020)
+        assert len(data) == self.data_len_2020
 
         params = copy.deepcopy(self.base_csv_to_df_params)
         params["button_presses_threshold"] = 0.7
         data = csv_to_df_pickle(**params)
-        self.assertLess(len(data), self.data_len_2020)
+        assert len(data) < self.data_len_2020
 
     def test_csv_to_df_pickle(self):
-        # Try various button thresholds (hand-picked to lie between different values for dummy data games)
         button_thresholds = [0.2, 0.6, 0.7]
         lengths = []
         for threshold in button_thresholds:
-            # dummy dataset is too small to partion so we set train_test_split=False
             params = copy.deepcopy(self.base_csv_to_df_params)
             params["button_presses_threshold"] = threshold
             data = csv_to_df_pickle(**params)
             lengths.append(len(data))
 
-        # Filtered data size should be monotonically decreasing wrt button_threshold
         for i in range(len(lengths) - 1):
-            self.assertGreaterEqual(lengths[i], lengths[i + 1])
+            assert lengths[i] >= lengths[i + 1]
 
-        # Picking a threshold that's suficiently high discards all data, should result in value error
         params = copy.deepcopy(self.base_csv_to_df_params)
         params["button_presses_threshold"] = 0.8
-        self.assertRaises(ValueError, csv_to_df_pickle, **params)
+        with pytest.raises(ValueError):
+            csv_to_df_pickle(**params)
 
     def test_get_trajs_from_data_2019(self):
         params = copy.deepcopy(self.base_get_trajs_from_data_params)
@@ -117,22 +107,19 @@ def test_get_trajs_from_data_2019_featurize(self):
         trajectories, _ = get_trajs_from_data(**params)
 
     def test_get_trajs_from_data_2020(self):
-        # Ensure we can properly deserialize states with updated objects (i.e tomatoes)
         params = copy.deepcopy(self.base_get_trajs_from_data_params)
         params["layouts"] = ["inverse_marshmallow_experiment"]
         params["data_path"] = DUMMY_2020_CLEAN_HUMAN_DATA_PATH
         trajectories, _ = get_trajs_from_data(**params)
 
     def test_get_trajs_from_data_2020_featurize(self):
-        # Ensure we can properly featurize states with updated dynamics and updated objects (i.e tomatoes)
         params = copy.deepcopy(self.base_get_trajs_from_data_params)
         params["layouts"] = ["inverse_marshmallow_experiment"]
         params["data_path"] = DUMMY_2020_CLEAN_HUMAN_DATA_PATH
         params["featurize_states"] = True
         trajectories, _ = get_trajs_from_data(**params)
 
     def test_csv_to_df_to_trajs_integration(self):
-        # Ensure the output of 'csv_to_df_pickle' works as valid input to 'get_trajs_from_data'
         params = copy.deepcopy(self.base_csv_to_df_params)
         _ = csv_to_df_pickle(**params)
 
@@ -144,7 +131,7 @@ def test_csv_to_df_to_trajs_integration(self):
         _ = get_trajs_from_data(**params)
 
 
-class TestHumanDataConversion(unittest.TestCase):
+class TestHumanDataConversion:
     temp_dir = "this_is_also_a_temp"
     infile = DUMMY_2019_CLEAN_HUMAN_DATA_PATH
     horizon = 400
@@ -160,12 +147,8 @@ def _equal_pickle_and_env_state_dict(
             ["timestep", "all_orders", "bonus_orders"],
         )
 
-    def setUp(self):
-        print(
-            "\nIn Class {}, in Method {}".format(
-                self.__class__.__name__, self._testMethodName
-            )
-        )
+    @pytest.fixture(autouse=True)
+    def _setup(self):
         if not os.path.exists(self.temp_dir):
             os.makedirs(self.temp_dir)
 
@@ -191,7 +174,7 @@ def setUp(self):
         with open(outfile, "rb") as f:
             self.human_data = pickle.load(f)[self.layout_name]
 
-    def tearDown(self):
+        yield
         shutil.rmtree(self.temp_dir)
 
     def test_state(self):
@@ -200,17 +183,10 @@ def test_state(self):
             if state_dict.items() == self.starting_state_dict.items():
                 self.env.reset()
             else:
-                self.assertTrue(
-                    self._equal_pickle_and_env_state_dict(
-                        state_dict, self.env.state.to_dict()
-                    ),
-                    "Expected state:\t\n{}\n\nActual state:\t\n{}".format(
-                        self.env.state.to_dict(), state_dict
-                    ),
+                assert self._equal_pickle_and_env_state_dict(
+                    state_dict, self.env.state.to_dict()
+                ), "Expected state:\t\n{}\n\nActual state:\t\n{}".format(
+                    self.env.state.to_dict(), state_dict
                 )
             self.env.step(joint_action=joint_action)
             idx += 1
-
-
-if __name__ == "__main__":
-    unittest.main()