Make CLI pip-installable (#8772)

Co-authored-by: openhands <openhands@all-hands.dev>

.github/workflows/ghcr-build.yml

@@ -293,7 +293,7 @@ jobs:
       - name: Install poetry via pipx
         run: pipx install poetry
       - name: Install Python dependencies using Poetry
-        run: make install-python-dependencies POETRY_GROUP=main,test,runtime INSTALL_PLAYWRIGHT=0
+        run: make install-python-dependencies INSTALL_PLAYWRIGHT=0
       - name: Run docker runtime tests
         run: |
           # We install pytest-xdist in order to run tests across CPUs

.github/workflows/integration-runner.yml

@@ -54,7 +54,7 @@ jobs:
             Hi! I started running the integration tests on your PR. You will receive a comment with the results shortly.
 
       - name: Install Python dependencies using Poetry
-        run: poetry install --without evaluation
+        run: poetry install --with dev,test,runtime
 
       - name: Configure config.toml for testing with Haiku
         env:

.github/workflows/py-unit-tests.yml

@@ -44,7 +44,7 @@ jobs:
           python-version: ${{ matrix.python-version }}
           cache: 'poetry'
       - name: Install Python dependencies using Poetry
-        run: poetry install --without evaluation
+        run: poetry install --with dev,test,runtime
       - name: Build Environment
         run: make build
       - name: Run Unit Tests
@@ -71,7 +71,7 @@ jobs:
           python-version: ${{ matrix.python-version }}
           cache: 'poetry'
       - name: Install Python dependencies using Poetry
-        run: poetry install --without evaluation
+        run: poetry install --with dev,test,runtime
       - name: Run Windows unit tests
         run: poetry run pytest -svv tests/unit/test_windows_bash.py
       - name: Run Windows runtime tests with LocalRuntime

Makefile

@@ -151,7 +151,7 @@ install-python-dependencies:
 		echo "Installing only POETRY_GROUP=${POETRY_GROUP}"; \
 		poetry install --only $${POETRY_GROUP}; \
 	else \
-		poetry install; \
+		poetry install --with dev,test,runtime; \
 	fi
 	@if [ "${INSTALL_PLAYWRIGHT}" != "false" ] && [ "${INSTALL_PLAYWRIGHT}" != "0" ]; then \
 		if [ -f "/etc/manjaro-release" ]; then \

containers/app/Dockerfile

@@ -26,7 +26,7 @@ RUN apt-get update -y \
 
 COPY ./pyproject.toml ./poetry.lock ./
 RUN touch README.md
-RUN export POETRY_CACHE_DIR && poetry install --without evaluation --no-root && rm -rf $POETRY_CACHE_DIR
+RUN export POETRY_CACHE_DIR && poetry install --no-root && rm -rf $POETRY_CACHE_DIR
 
 FROM python:3.12.3-slim AS openhands-app
 

openhands/cli/main.py

@@ -325,7 +325,7 @@ async def run_session(
     return new_session_requested
 
 
-async def main(loop: asyncio.AbstractEventLoop) -> None:
+async def main_with_loop(loop: asyncio.AbstractEventLoop) -> None:
     """Runs the agent in CLI mode."""
     args = parse_arguments()
 
@@ -417,11 +417,11 @@ async def main(loop: asyncio.AbstractEventLoop) -> None:
         )
 
 
-if __name__ == '__main__':
+def main():
     loop = asyncio.new_event_loop()
     asyncio.set_event_loop(loop)
     try:
-        loop.run_until_complete(main(loop))
+        loop.run_until_complete(main_with_loop(loop))
     except KeyboardInterrupt:
         print('Received keyboard interrupt, shutting down...')
     except ConnectionRefusedError as e:
@@ -443,3 +443,7 @@ if __name__ == '__main__':
         except Exception as e:
             print(f'Error during cleanup: {e}')
             sys.exit(1)
+
+
+if __name__ == '__main__':
+    main()

pyproject.toml

@@ -20,47 +20,36 @@ packages = [
 
 [tool.poetry.dependencies]
 python = "^3.12,<3.14"
-litellm = "^1.60.0, !=1.64.4, !=1.67.*"              # avoid 1.64.4 (known bug) & 1.67.* (known bug #10272)
+litellm = "^1.60.0, !=1.64.4, !=1.67.*"           # avoid 1.64.4 (known bug) & 1.67.* (known bug #10272)
-aiohttp = ">=3.9.0,!=3.11.13"                        # Pin to avoid yanked version 3.11.13
+aiohttp = ">=3.9.0,!=3.11.13"                     # Pin to avoid yanked version 3.11.13
-google-generativeai = "*"                            # To use litellm with Gemini Pro API
+google-generativeai = "*"                         # To use litellm with Gemini Pro API
-google-api-python-client = "^2.164.0"                # For Google Sheets API
+google-api-python-client = "^2.164.0"             # For Google Sheets API
-google-auth-httplib2 = "*"                           # For Google Sheets authentication
+google-auth-httplib2 = "*"                        # For Google Sheets authentication
-google-auth-oauthlib = "*"                           # For Google Sheets OAuth
+google-auth-oauthlib = "*"                        # For Google Sheets OAuth
 termcolor = "*"
 docker = "*"
 fastapi = "*"
 toml = "*"
-uvicorn = "*"
 types-toml = "*"
+uvicorn = "*"
 numpy = "*"
 json-repair = "*"
-browsergym-core = "0.13.3"                           # integrate browsergym-core as the browsing interface
+browsergym-core = "0.13.3"                        # integrate browsergym-core as the browsing interface
 html2text = "*"
 e2b = ">=1.0.5,<1.4.0"
 pexpect = "*"
 jinja2 = "^3.1.3"
 python-multipart = "*"
-boto3 = "*"
-minio = "^7.2.8"
 tenacity = ">=8.5,<10.0"
 zope-interface = "7.2"
 pathspec = "^0.12.1"
-google-cloud-aiplatform = "*"
-anthropic = { extras = [ "vertex" ], version = "*" }
-tree-sitter = "^0.24.0"
-bashlex = "^0.18"
 pyjwt = "^2.9.0"
 dirhash = "*"
-python-frontmatter = "^1.1.0"
-python-docx = "*"
-PyPDF2 = "*"
-python-pptx = "*"
-pylatexenc = "*"
 tornado = "*"
 python-dotenv = "*"
 rapidfuzz = "^3.9.0"
 whatthepatch = "^1.0.6"
-protobuf = "^4.21.6,<5.0.0"                          # chromadb currently fails on 5.0+
+protobuf = "^4.21.6,<5.0.0"                       # chromadb currently fails on 5.0+
 opentelemetry-api = "1.25.0"
 opentelemetry-exporter-otlp-proto-grpc = "1.25.0"
 modal = ">=0.66.26,<0.78.0"
@@ -70,14 +59,8 @@ pygithub = "^2.5.0"
 joblib = "*"
 openhands-aci = "0.3.0"
 python-socketio = "^5.11.4"
-redis = ">=5.2,<7.0"
 sse-starlette = "^2.1.3"
 psutil = "*"
-stripe = ">=11.5,<13.0"
-ipywidgets = "^8.1.5"
-qtconsole = "^5.6.1"
-memory-profiler = "^0.61.0"
-daytona-sdk = "0.18.1"
 python-json-logger = "^3.2.1"
 prompt-toolkit = "^3.0.50"
 poetry = "^2.1.2"
@@ -85,6 +68,27 @@ anyio = "4.9.0"
 pythonnet = "*"
 fastmcp = "^2.5.2"
 mcpm = "1.12.0"
+python-frontmatter = "^1.1.0"
+# TODO: Should these go into the runtime group?
+ipywidgets = "^8.1.5"
+qtconsole = "^5.6.1"
+PyPDF2 = "*"
+python-pptx = "*"
+pylatexenc = "*"
+python-docx = "*"
+bashlex = "^0.18"
+
+# TODO: These are integrations that should probably be optional
+redis = ">=5.2,<7.0"
+minio = "^7.2.8"
+daytona-sdk = "0.18.1"
+stripe = ">=11.5,<13.0"
+google-cloud-aiplatform = "*"
+anthropic = { extras = [ "vertex" ], version = "*" }
+boto3 = "*"
+
+[tool.poetry.group.dev]
+optional = true
 
 [tool.poetry.group.dev.dependencies]
 ruff = "0.11.11"
@@ -93,6 +97,9 @@ pre-commit = "4.2.0"
 build = "*"
 types-setuptools = "*"
 
+[tool.poetry.group.test]
+optional = true
+
 [tool.poetry.group.test.dependencies]
 pytest = "*"
 pytest-cov = "*"
@@ -104,11 +111,18 @@ pandas = "*"
 reportlab = "*"
 gevent = ">=24.2.1,<26.0.0"
 
+[tool.poetry.group.runtime]
+optional = true
+
 [tool.poetry.group.runtime.dependencies]
 jupyterlab = "*"
 notebook = "*"
 jupyter_kernel_gateway = "*"
 flake8 = "*"
+memory-profiler = "^0.61.0"
+
+[tool.poetry.group.evaluation]
+optional = true
 
 [tool.poetry.group.evaluation.dependencies]
 streamlit = "*"
@@ -132,6 +146,7 @@ browsergym-visualwebarena = "0.13.3"
 boto3-stubs = { extras = [ "s3" ], version = "^1.37.19" }
 pyarrow = "20.0.0"                                                             # transitive dependency, pinned here to avoid conflicts
 datasets = "*"
+joblib = "*"
 
 [tool.poetry.scripts]
 openhands = "openhands.cli.main:main"

tests/unit/test_agent_controller.py

@@ -1381,6 +1381,7 @@ async def test_first_user_message_with_identical_content(test_event_stream, mock
     await controller.close()
 
 
+@pytest.mark.asyncio
 async def test_agent_controller_processes_null_observation_with_cause():
     """Test that AgentController processes NullObservation events with a cause value.
 
@@ -1395,6 +1396,9 @@ async def test_agent_controller_processes_null_observation_with_cause():
 
     # Create a mock agent with necessary attributes
     mock_agent = MagicMock(spec=Agent)
+    mock_agent.get_system_message = MagicMock(
+        return_value=None,
+    )
     mock_agent.llm = MagicMock(spec=LLM)
     mock_agent.llm.metrics = Metrics()
     mock_agent.llm.config = OpenHandsConfig().get_llm_config()
@@ -1408,14 +1412,14 @@ async def test_agent_controller_processes_null_observation_with_cause():
     )
 
     # Patch the controller's step method to track calls
-    with patch.object(controller, 'step') as mock_step:
+    with patch.object(controller, '_step') as mock_step:
         # Create and add the first user message (will have ID 0)
         user_message = MessageAction(content='First user message')
         user_message._source = EventSource.USER  # type: ignore[attr-defined]
         event_stream.add_event(user_message, EventSource.USER)
 
         # Give it a little time to process
-        await asyncio.sleep(0.3)
+        await asyncio.sleep(1)
 
         # Get all events from the stream
         events = list(event_stream.get_events())

tests/unit/test_cli.py

@@ -381,7 +381,7 @@ async def test_main_without_task(
     mock_run_session.return_value = False
 
     # Run the function
-    await cli.main(loop)
+    await cli.main_with_loop(loop)
 
     # Assertions
     mock_parse_args.assert_called_once()
@@ -458,7 +458,7 @@ async def test_main_with_task(
     mock_run_session.side_effect = [True, False]
 
     # Run the function
-    await cli.main(loop)
+    await cli.main_with_loop(loop)
 
     # Assertions
     mock_parse_args.assert_called_once()
@@ -553,7 +553,7 @@ async def test_main_with_session_name_passes_name_to_run_session(
     mock_run_session.return_value = False
 
     # Run the function
-    await cli.main(loop)
+    await cli.main_with_loop(loop)
 
     # Assertions
     mock_parse_args.assert_called_once()
@@ -713,7 +713,7 @@ async def test_main_security_check_fails(
     mock_check_security.return_value = False
 
     # Run the function
-    await cli.main(loop)
+    await cli.main_with_loop(loop)
 
     # Assertions
     mock_parse_args.assert_called_once()
@@ -796,7 +796,7 @@ async def test_config_loading_order(
     mock_run_session.return_value = False  # No new session requested
 
     # Run the function
-    await cli.main(loop)
+    await cli.main_with_loop(loop)
 
     # Assertions for argument parsing and config setup
     mock_parse_args.assert_called_once()