Adding pre-commit and CI for ruff and mypy (#69)

* don't modify directories

* oops typo

* dev_config/python

* add config to CI

* bump CI python to 3.10

* 3.11?

* del actions/

* add suggestions

* delete unused code

* missed some

* oops missed another one

* remove a file

.github/workflows/lint.yml

@@ -3,7 +3,7 @@ name: Lint
 on: [push, pull_request]
 
 jobs:
-  lint:
+  lint-frontend:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v2
@@ -15,4 +15,20 @@ jobs:
           npm ci --legacy-peer-deps
       - run: |
           cd frontend
-          npm run lint
+          npm run lint
+
+  lint-python:
+    name: Lint python
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v2
+      - name: Set up python
+        uses: actions/setup-python@v2
+        with:
+          python-version: 3.11
+      - name: Install dependencies
+        run: pip install ruff mypy types-requests
+      - name: Run ruff
+        run: ruff check --config dev_config/python/ruff.toml opendevin/ server/ agenthub/
+      - name: Run mypy
+        run: mypy --config-file dev_config/python/mypy.ini opendevin/ server/ agenthub/

agenthub/__init__.py

@@ -1,2 +1,4 @@
 from . import langchains_agent
 from . import codeact_agent
+
+__all__ = ['langchains_agent', 'codeact_agent']

agenthub/codeact_agent/__init__.py

@@ -1,11 +1,12 @@
 import os
 import re
-import argparse
 from litellm import completion
 from termcolor import colored
 from typing import List, Dict
 
 from opendevin.agent import Agent, Message, Role
+from opendevin.lib.event import Event
+from opendevin.lib.command_manager import CommandManager
 from opendevin.sandbox.sandbox import DockerInteractive
 
 assert (
@@ -93,12 +94,13 @@ class CodeActAgent(Agent):
             command = re.search(r"<execute>(.*)</execute>", action, re.DOTALL)
             if command is not None:
                 # a command was found
-                command = command.group(1)
-                if command.strip() == "exit":
+                command_group = command.group(1)
+                if command_group.strip() == "exit":
                     print(colored("Exit received. Exiting...", "red"))
                     break
                 # execute the code
-                observation = self.env.execute(command)
+                # TODO: does exit_code get loaded into Message?
+                exit_code, observation = self.env.execute(command_group)
                 self._history.append(Message(Role.ASSISTANT, observation))
                 print(colored("===ENV OBSERVATION:===\n" + observation, "blue"))
             else:
@@ -120,5 +122,15 @@ class CodeActAgent(Agent):
         """
         raise NotImplementedError
 
+    # TODO: implement these abstract methods
+    def add_event(self, event: Event) -> None:
+        raise NotImplementedError("Implement this abstract method")
+
+    def step(self, cmd_mgr: CommandManager) -> Event:
+        raise NotImplementedError("Implement this abstract method")
+
+    def search_memory(self, query: str) -> List[str]:
+        raise NotImplementedError("Implement this abstract method")
+
 
 Agent.register("CodeActAgent", CodeActAgent)

agenthub/langchains_agent/__init__.py

@@ -1,8 +1,6 @@
-import os
-import argparse
-from typing import List, Dict, Type
+from typing import List
 
-from opendevin.agent import Agent, Message
+from opendevin.agent import Agent
 
 from agenthub.langchains_agent.utils.agent import Agent as LangchainsAgentImpl
 from opendevin.lib.event import Event

agenthub/langchains_agent/utils/llm.py

@@ -7,7 +7,7 @@ if os.getenv("DEBUG"):
     set_debug(True)
 
 from typing import List
-from langchain_core.pydantic_v1 import BaseModel, Field
+from langchain_core.pydantic_v1 import BaseModel
 
 from langchain.chains import LLMChain
 from langchain.prompts import PromptTemplate

agenthub/langchains_agent/utils/memory.py

@@ -1,14 +1,10 @@
-import os
 from . import json
 
 import chromadb
 
 from llama_index.core import Document
 from llama_index.core.retrievers import VectorIndexRetriever
-from llama_index.core import VectorStoreIndex, StorageContext, load_index_from_storage
-from llama_index.core.storage.docstore import SimpleDocumentStore
-from llama_index.core.vector_stores import SimpleVectorStore
-
+from llama_index.core import VectorStoreIndex
 from llama_index.vector_stores.chroma import ChromaVectorStore
 
 class LongTermMemory:
@@ -16,7 +12,6 @@ class LongTermMemory:
         db = chromadb.Client()
         self.collection = db.create_collection(name="memories")
         vector_store = ChromaVectorStore(chroma_collection=self.collection)
-        storage_context = StorageContext.from_defaults(vector_store=vector_store)
         self.index = VectorStoreIndex.from_vector_store(vector_store)
         self.thought_idx = 0
 

dev_config/python/.pre-commit-config.yaml

@@ -0,0 +1,17 @@
+repos:
+  - repo: https://github.com/astral-sh/ruff-pre-commit
+    rev: v0.3.3
+    hooks:
+      - id: ruff
+        entry: ruff check --config dev_config/python/ruff.toml opendevin/ server/ agenthub/
+        always_run: true
+        pass_filenames: false
+
+  - repo: https://github.com/pre-commit/mirrors-mypy
+    rev: v1.9.0
+    hooks:
+      - id: mypy
+        additional_dependencies: [types-requests, types-setuptools]
+        entry: mypy --config-file dev_config/python/mypy.ini opendevin/ server/ agenthub/
+        always_run: true
+        pass_filenames: false

dev_config/python/mypy.ini

@@ -0,0 +1,11 @@
+[mypy]
+warn_unused_configs = True
+ignore_missing_imports = True
+check_untyped_defs = True
+explicit_package_bases = True
+warn_unreachable = True
+warn_redundant_casts = True
+no_implicit_optional = True
+strict_optional = True
+
+exclude = agenthub/langchains_agent/regression

dev_config/python/ruff.toml

@@ -0,0 +1,3 @@
+exclude = [
+    "agenthub/langchains_agent/regression/",
+]

opendevin/agent.py

@@ -5,7 +5,6 @@ from enum import Enum
 
 from .lib.event import Event
 from .lib.command_manager import CommandManager
-from .controller import AgentController
 
 class Role(Enum):
     SYSTEM = "system"  # system message for LLM
@@ -79,7 +78,7 @@ class Agent(ABC):
         return self._complete
 
     @property
-    def history(self) -> List[str]:
+    def history(self) -> List[Message]:
         """
         Provides the history of interactions or state changes since the instruction was initiated.
 
@@ -125,7 +124,7 @@ class Agent(ABC):
         to prepare the agent for restarting the instruction or cleaning up before destruction.
 
         """
-        self.instruction = None
+        self.instruction = ''
         self._complete = False
         self._history = []
 

opendevin/controller.py

@@ -27,7 +27,6 @@ class AgentController:
         return out_event
 
     def start_loop(self):
-        output = None
         for i in range(self.max_iterations):
             print("STEP", i, flush=True)
             log_events = self.command_manager.get_background_events()

opendevin/lib/actions/__init__.py

@@ -2,3 +2,4 @@ from .browse import browse
 from .write import write
 from .read import read
 
+__all__ = ['run', 'kill', 'browse', 'write', 'read']

opendevin/lib/command_manager.py

@@ -1,5 +1,3 @@
-import subprocess
-import select
 from typing import List
 
 from opendevin.lib.event import Event
@@ -41,7 +39,7 @@ class CommandManager:
         self.background_commands[bg_cmd.id] = bg_cmd
         return "Background command started. To stop it, send a `kill` action with id " + str(bg_cmd.id)
 
-    def kill_command(self, id: int) -> str:
+    def kill_command(self, id: int):
         # TODO: get log events before killing
         self.background_commands[id].shell.close()
         del self.background_commands[id]

opendevin/lib/event.py

@@ -1,5 +1,3 @@
-import os
-import json
 import opendevin.lib.actions as actions
 
 ACTION_TYPES = ['run', 'kill', 'browse', 'read', 'write', 'recall', 'think', 'summarize', 'output', 'error', 'finish']

opendevin/main.py

@@ -1,6 +1,6 @@
+from typing import Type
 import argparse
 
-import agenthub  # for the agent registry
 from opendevin.agent import Agent
 from opendevin.controller import AgentController
 
@@ -14,7 +14,7 @@ if __name__ == "__main__":
 
     print(f"Running agent {args.agent_cls} (model: {args.model_name}, directory: {args.directory}) with task: \"{args.task}\"")
 
-    AgentCls: Agent = Agent.get_cls(args.agent_cls)
+    AgentCls: Type[Agent] = Agent.get_cls(args.agent_cls)
     agent = AgentCls(
         instruction=args.task,
         workspace_dir=args.directory,

opendevin/sandbox/sandbox.py

@@ -1,19 +1,15 @@
 import os
-import pty
 import sys
 import uuid
 import time
-import shlex
 import select
-import subprocess
 import docker
-import time
-from typing import List, Tuple
+from typing import Tuple
 from collections import namedtuple
 import atexit
 
-InputType = namedtuple("InputDtype", ["content"])
-OutputType = namedtuple("OutputDtype", ["content"])
+InputType = namedtuple("InputType", ["content"])
+OutputType = namedtuple("OutputType", ["content"])
 
 CONTAINER_IMAGE = os.getenv("SANDBOX_CONTAINER_IMAGE", "opendevin/sandbox:latest")
 
@@ -21,15 +17,15 @@ class DockerInteractive:
 
     def __init__(
         self,
-        workspace_dir: str = None,
-        container_image: str = None,
+        workspace_dir: str | None = None,
+        container_image: str | None = None,
         timeout: int = 120,
-        id: str = None
+        id: str | None = None
     ):
         if id is not None:
-            self.instance_id: str = id
+            self.instance_id = id
         else:
-            self.instance_id: str = uuid.uuid4()
+            self.instance_id = str(uuid.uuid4())
         if workspace_dir is not None:
             assert os.path.exists(workspace_dir), f"Directory {workspace_dir} does not exist."
             # expand to absolute path
@@ -52,7 +48,6 @@ class DockerInteractive:
         self.container_name = f"sandbox-{self.instance_id}"
 
         self.restart_docker_container()
-        uid = os.getuid()
         self.execute('useradd --shell /bin/bash -u {uid} -o -c \"\" -m devin && su devin')
         # regester container cleanup function
         atexit.register(self.cleanup)
@@ -62,9 +57,9 @@ class DockerInteractive:
             return ""
         logs = ""
         while True:
-            ready_to_read, _, _ = select.select([self.log_generator], [], [], .1)
+            ready_to_read, _, _ = select.select([self.log_generator], [], [], .1) # type: ignore[has-type]
             if ready_to_read:
-                data = self.log_generator.read(4096)
+                data = self.log_generator.read(4096) # type: ignore[has-type]
                 if not data:
                     break
                 # FIXME: we're occasionally seeing some escape characters like `\x02` and `\x00` in the logs...
@@ -171,7 +166,7 @@ if __name__ == "__main__":
                 print("\nExiting...")
                 break
             if user_input.lower() == "exit":
-                print(f"Exiting...")
+                print("Exiting...")
                 break
             exit_code, output = docker_interactive.execute(user_input)
             print("exit code:", exit_code)

server/server.py

@@ -1,4 +1,3 @@
-import asyncio
 import json
 import os
 from time import sleep
@@ -56,7 +55,6 @@ async def websocket_endpoint(websocket: WebSocket):
                     await send_message_to_client(get_error_payload("Failed to start container: " + str(e)))
                     continue
 
-                agent_listener = asyncio.create_task(listen_for_agent_messages())
             if action == "terminal":
                 msg = {
                     "action": "terminal",
@@ -67,7 +65,6 @@ async def websocket_endpoint(websocket: WebSocket):
                 if agent_websocket is None:
                     await send_message_to_client(get_error_payload("Agent not connected"))
                     continue
-                await send_message_to_agent(data)
 
     except WebSocketDisconnect:
         print("Client websocket disconnected")