Merge branch 'dev' of https://github.com/geekan/MetaGPT into geekan/dev

.github/workflows/build-package.yaml

@@ -0,0 +1,34 @@
+name: Build and upload python package
+
+on:
+  release:
+    types: [created]
+
+jobs:
+  deploy:
+    runs-on: ubuntu-latest
+
+    steps:
+    - uses: actions/checkout@v4
+    - name: Set up Python
+      uses: actions/setup-python@v4
+      with:
+        python-version: '3.9'
+        cache: 'pip'
+    - name: Install dependencies
+      run: |
+        python -m pip install --upgrade pip
+        pip install -r requirements.txt
+        pip install -e.
+        pip install setuptools wheel twine
+    - name: Set package version
+      run: |
+        export VERSION="${GITHUB_REF#refs/tags/v}"
+        sed -i "s/version=.*/version=\"${VERSION}\",/" setup.py
+    - name: Build and publish
+      env:
+        TWINE_USERNAME: __token__
+        TWINE_PASSWORD: ${{ secrets.PYPI_API_TOKEN }}
+      run: |
+        python setup.py bdist_wheel sdist
+        twine upload dist/*

.github/workflows/unittest.yaml

@@ -1,13 +1,10 @@
-name: Python application test
+name: Unit Tests
 
 on:
   workflow_dispatch:
   pull_request_target:
   push: 
-    branches: 
-      - 'main'
-      - 'dev'
-      - '*-release'
+    branches:
       - '*-debugger'
 
 jobs:
@@ -79,3 +76,4 @@ jobs:
       uses: codecov/codecov-action@v3
       env:
         CODECOV_TOKEN: ${{ secrets.CODECOV_TOKEN }}
+      if: ${{ always() }}

docs/ROADMAP.md

@@ -9,24 +9,22 @@ ### Short-term Objective
 
 1. Become the multi-agent framework with the highest ROI.
 2. Support fully automatic implementation of medium-sized projects (around 2000 lines of code).
-3. Implement most identified tasks, reaching version 0.5.
+3. Implement most identified tasks, reaching version 1.0.
 
 ### Tasks
 
-To reach version v0.5, approximately 70% of the following tasks need to be completed.
-
 1. Usability
    1. ~~Release v0.01 pip package to try to solve issues like npm installation (though not necessarily successfully)~~ (v0.3.0)
-   2. Support for overall save and recovery of software companies
+   2. ~~Support for overall save and recovery of software companies~~ (v0.6.0)
    3. ~~Support human confirmation and modification during the process~~ (v0.3.0) New: Support human confirmation and modification with fewer constrainsts and a more user-friendly interface
    4. Support process caching: Consider carefully whether to add server caching mechanism
    5. ~~Resolve occasional failure to follow instruction under current prompts, causing code parsing errors, through stricter system prompts~~ (v0.4.0, with function call)
    6. Write documentation, describing the current features and usage at all levels (ongoing, continuously adding contents to [documentation site](https://docs.deepwisdom.ai/main/en/guide/get_started/introduction.html))
    7. ~~Support Docker~~
 2. Features
-   1. Support a more standard and stable parser (need to analyze the format that the current LLM is better at)
-   2. ~~Establish a separate output queue, differentiated from the message queue~~
-   3. Attempt to atomize all role work, but this may significantly increase token overhead
+   1. ~~Support a more standard and stable parser (need to analyze the format that the current LLM is better at)~~ (v0.5.0)
+   2. ~~Establish a separate output queue, differentiated from the message queue~~ (v0.5.0)
+   3. ~~Attempt to atomize all role work, but this may significantly increase token overhead~~ (v0.5.0)
    4. Complete the design and implementation of module breakdown
    5. Support various modes of memory: clearly distinguish between long-term and short-term memory
    6. Perfect the test role, and carry out necessary interactions with humans
@@ -43,10 +41,10 @@ ### Tasks
 4. Actions
    1. ~~Implementation: Search~~ (v0.2.1)
    2. Implementation: Knowledge search, supporting 10+ data formats
-   3. Implementation: Data EDA (expected v0.6.0)
-   4. Implementation: Review
-   5. ~~Implementation~~: Add Document (v0.5.0)
-   6. ~~Implementation~~: Delete Document (v0.5.0)
+   3. Implementation: Data EDA (expected v0.7.0)
+   4. Implementation: Review & Revise (expected v0.7.0)
+   5. ~~Implementation: Add Document~~ (v0.5.0)
+   6. ~~Implementation: Delete Document~~ (v0.5.0)
    7. Implementation: Self-training
    8. ~~Implementation: DebugError~~ (v0.2.1)
    9. Implementation: Generate reliable unit tests based on YAPI
@@ -64,23 +62,23 @@ ### Tasks
    3. ~~Support Playwright apis~~
 7. Roles
    1. Perfect the action pool/skill pool for each role
-   2. Red Book blogger
-   3. E-commerce seller
-   4. Data analyst (expected v0.6.0)
-   5. News observer
-   6. ~~Institutional researcher~~ (v0.2.1)
+   2. E-commerce seller
+   3. Data analyst (expected v0.7.0)
+   4. News observer
+   5. ~~Institutional researcher~~ (v0.2.1)
 8. Evaluation
    1. Support an evaluation on a game dataset (experimentation done with game agents)
    2. Reproduce papers, implement full skill acquisition for a single game role, achieving SOTA results (experimentation done with game agents)
-   3. Support an evaluation on a math dataset (expected v0.6.0)
+   3. Support an evaluation on a math dataset (expected v0.7.0)
    4. Reproduce papers, achieving SOTA results for current mathematical problem solving process
 9. LLM
    1. Support Claude underlying API
    2. ~~Support Azure asynchronous API~~
    3. Support streaming version of all APIs
    4. ~~Make gpt-3.5-turbo available (HARD)~~
+   5. Support 
 10. Other
-    1. Clean up existing unused code
+    1. ~~Clean up existing unused code~~
     2. Unify all code styles and establish contribution standards
-    3. Multi-language support
-    4. Multi-programming-language support
+    3. ~~Multi-language support~~
+    4. ~~Multi-programming-language support~~

examples/debate_simple.py

@@ -13,7 +13,9 @@ from metagpt.roles import Role
 from metagpt.team import Team
 
 action1 = Action(name="AlexSay", instruction="Express your opinion with emotion and don't repeat it")
+action1.llm.model = "gpt-4-1106-preview"
 action2 = Action(name="BobSay", instruction="Express your opinion with emotion and don't repeat it")
+action2.llm.model = "gpt-3.5-turbo-1106"
 alex = Role(name="Alex", profile="Democratic candidate", goal="Win the election", actions=[action1], watch=[action2])
 bob = Role(name="Bob", profile="Republican candidate", goal="Win the election", actions=[action2], watch=[action1])
 env = Environment(desc="US election live broadcast")

metagpt/actions/action.py

@@ -25,7 +25,7 @@ from metagpt.utils.project_repo import ProjectRepo
 
 
 class Action(SerializationMixin, ContextMixin, BaseModel):
-    model_config = ConfigDict(arbitrary_types_allowed=True, exclude=["llm"])
+    model_config = ConfigDict(arbitrary_types_allowed=True)
 
     name: str = ""
     i_context: Union[dict, CodingContext, CodeSummarizeContext, TestingContext, RunCodeContext, str, None] = ""

metagpt/config2.py

@@ -18,7 +18,7 @@ from metagpt.configs.redis_config import RedisConfig
 from metagpt.configs.s3_config import S3Config
 from metagpt.configs.search_config import SearchConfig
 from metagpt.configs.workspace_config import WorkspaceConfig
-from metagpt.const import METAGPT_ROOT
+from metagpt.const import CONFIG_ROOT, METAGPT_ROOT
 from metagpt.utils.yaml_model import YamlModel
 
 
@@ -81,6 +81,11 @@ class Config(CLIParams, YamlModel):
     AZURE_TTS_REGION: str = ""
     mermaid_engine: str = "nodejs"
 
+    @classmethod
+    def from_home(cls, path):
+        """Load config from ~/.metagpt/config.yaml"""
+        return Config.model_validate_yaml(CONFIG_ROOT / path)
+
     @classmethod
     def default(cls):
         """Load default config

metagpt/const.py

@@ -47,7 +47,7 @@ def get_metagpt_root():
 
 
 # METAGPT PROJECT ROOT AND VARS
-
+CONFIG_ROOT = Path.home() / ".metagpt"
 METAGPT_ROOT = get_metagpt_root()  # Dependent on METAGPT_PROJECT_ROOT
 DEFAULT_WORKSPACE_ROOT = METAGPT_ROOT / "workspace"
 

metagpt/context_mixin.py

@@ -19,6 +19,11 @@ class ContextMixin(BaseModel):
 
     model_config = ConfigDict(arbitrary_types_allowed=True)
 
+    # Pydantic has bug on _private_attr when using inheritance, so we use private_* instead
+    # - https://github.com/pydantic/pydantic/issues/7142
+    # - https://github.com/pydantic/pydantic/issues/7083
+    # - https://github.com/pydantic/pydantic/issues/7091
+
     # Env/Role/Action will use this context as private context, or use self.context as public context
     private_context: Optional[Context] = Field(default=None, exclude=True)
     # Env/Role/Action will use this config as private config, or use self.context.config as public config
@@ -52,6 +57,8 @@ class ContextMixin(BaseModel):
     def set_config(self, config: Config, override=False):
         """Set config"""
         self.set("private_config", config, override)
+        if config is not None:
+            _ = self.llm  # init llm
 
     def set_llm(self, llm: BaseLLM, override=False):
         """Set llm"""

metagpt/llm.py

@@ -5,13 +5,15 @@
 @Author  : alexanderwu
 @File    : llm.py
 """
+from typing import Optional
 
-
+from metagpt.configs.llm_config import LLMConfig
 from metagpt.context import CONTEXT
 from metagpt.provider.base_llm import BaseLLM
 
 
-def LLM() -> BaseLLM:
+def LLM(llm_config: Optional[LLMConfig] = None) -> BaseLLM:
     """get the default llm provider if name is None"""
-    # context.use_llm(name=name, provider=provider)
+    if llm_config is not None:
+        CONTEXT.llm_with_cost_manager_from_llm_config(llm_config)
     return CONTEXT.llm()

metagpt/provider/__init__.py

@@ -15,6 +15,7 @@ from metagpt.provider.zhipuai_api import ZhiPuAILLM
 from metagpt.provider.azure_openai_api import AzureOpenAILLM
 from metagpt.provider.metagpt_api import MetaGPTLLM
 from metagpt.provider.human_provider import HumanProvider
+from metagpt.provider.spark_api import SparkLLM
 
 __all__ = [
     "FireworksLLM",
@@ -26,4 +27,5 @@ __all__ = [
     "MetaGPTLLM",
     "OllamaLLM",
     "HumanProvider",
+    "SparkLLM",
 ]

metagpt/provider/base_llm.py

@@ -59,7 +59,9 @@ class BaseLLM(ABC):
         if system_msgs:
             message = self._system_msgs(system_msgs)
         else:
-            message = [self._default_system_msg()] if self.use_system_prompt else []
+            message = [self._default_system_msg()]
+        if not self.use_system_prompt:
+            message = []
         if format_msgs:
             message.extend(format_msgs)
         message.append(self._user_msg(msg))

metagpt/provider/openai_api.py

@@ -220,10 +220,12 @@ class OpenAILLM(BaseLLM):
 
     @handle_exception
     def _update_costs(self, usage: CompletionUsage):
-        if self.config.calc_usage and usage:
+        if self.config.calc_usage and usage and self.cost_manager:
             self.cost_manager.update_cost(usage.prompt_tokens, usage.completion_tokens, self.model)
 
     def get_costs(self) -> Costs:
+        if not self.cost_manager:
+            return Costs()
         return self.cost_manager.get_costs()
 
     def _get_max_tokens(self, messages: list[dict]):

metagpt/provider/spark_api.py

@@ -26,14 +26,14 @@ from metagpt.provider.llm_provider_registry import register_provider
 class SparkLLM(BaseLLM):
     def __init__(self, config: LLMConfig):
         self.config = config
-        logger.warning("当前方法无法支持异步运行。当你使用acompletion时，并不能并行访问。")
+        logger.warning("SparkLLM：当前方法无法支持异步运行。当你使用acompletion时，并不能并行访问。")
 
     def get_choice_text(self, rsp: dict) -> str:
         return rsp["payload"]["choices"]["text"][-1]["content"]
 
     async def acompletion_text(self, messages: list[dict], stream=False, timeout: int = 3) -> str:
         # 不支持
-        logger.warning("当前方法无法支持异步运行。当你使用acompletion时，并不能并行访问。")
+        # logger.warning("当前方法无法支持异步运行。当你使用acompletion时，并不能并行访问。")
         w = GetMessageFromWeb(messages, self.config)
         return w.run()
 

metagpt/roles/engineer.py

@@ -104,7 +104,7 @@ class Engineer(Role):
             # Code review
             if review:
                 action = WriteCodeReview(i_context=coding_context, context=self.context, llm=self.llm)
-                self._init_action_system_message(action)
+                self._init_action(action)
                 coding_context = await action.run()
             await self.project_repo.srcs.save(
                 filename=coding_context.filename,

metagpt/roles/role.py

@@ -132,6 +132,13 @@ class Role(SerializationMixin, ContextMixin, BaseModel):
 
     role_id: str = ""
     states: list[str] = []
+
+    # scenarios to set action system_prompt:
+    #   1. `__init__` while using Role(actions=[...])
+    #   2. add action to role while using `role.set_action(action)`
+    #   3. set_todo while using `role.set_todo(action)`
+    #   4. when role.system_prompt is being updated (e.g. by `role.system_prompt = "..."`)
+    # Additional, if llm is not set, we will use role's llm
     actions: list[SerializeAsAny[Action]] = Field(default=[], validate_default=True)
     rc: RoleContext = Field(default_factory=RoleContext)
     addresses: set[str] = set()
@@ -146,6 +153,10 @@ class Role(SerializationMixin, ContextMixin, BaseModel):
         self.pydantic_rebuild_model()
         super().__init__(**data)
 
+        if self.is_human:
+            self.llm = HumanProvider(None)
+
+        self._check_actions()
         self.llm.system_prompt = self._get_prefix()
         self._watch(data.get("watch") or [UserRequirement])
 
@@ -225,7 +236,16 @@ class Role(SerializationMixin, ContextMixin, BaseModel):
     def _setting(self):
         return f"{self.name}({self.profile})"
 
-    def _init_action_system_message(self, action: Action):
+    def _check_actions(self):
+        """Check actions and set llm and prefix for each action."""
+        self.set_actions(self.actions)
+        return self
+
+    def _init_action(self, action: Action):
+        if not action.private_config:
+            action.set_llm(self.llm, override=True)
+        else:
+            action.set_llm(self.llm, override=False)
         action.set_prefix(self._get_prefix())
 
     def set_action(self, action: Action):
@@ -241,7 +261,7 @@ class Role(SerializationMixin, ContextMixin, BaseModel):
         self._reset()
         for action in actions:
             if not isinstance(action, Action):
-                i = action(name="", llm=self.llm)
+                i = action()
             else:
                 if self.is_human and not isinstance(action.llm, HumanProvider):
                     logger.warning(
@@ -250,7 +270,7 @@ class Role(SerializationMixin, ContextMixin, BaseModel):
                         f"try passing in Action classes instead of initialized instances"
                     )
                 i = action
-            self._init_action_system_message(i)
+            self._init_action(i)
             self.actions.append(i)
             self.states.append(f"{len(self.actions)}. {action}")
 
@@ -308,6 +328,7 @@ class Role(SerializationMixin, ContextMixin, BaseModel):
         if env:
             env.set_addresses(self, self.addresses)
             self.llm.system_prompt = self._get_prefix()
+            self.set_actions(self.actions)  # reset actions to update llm and prefix
 
     def _get_prefix(self):
         """Get the role prefix"""
@@ -320,7 +341,8 @@ class Role(SerializationMixin, ContextMixin, BaseModel):
             prefix += CONSTRAINT_TEMPLATE.format(**{"constraints": self.constraints})
 
         if self.rc.env and self.rc.env.desc:
-            other_role_names = ", ".join(self.rc.env.role_names())
+            all_roles = self.rc.env.role_names()
+            other_role_names = ", ".join([r for r in all_roles if r != self.name])
             env_desc = f"You are in {self.rc.env.desc} with roles({other_role_names})."
             prefix += env_desc
         return prefix
@@ -480,7 +502,6 @@ class Role(SerializationMixin, ContextMixin, BaseModel):
             if not msg.cause_by:
                 msg.cause_by = UserRequirement
             self.put_message(msg)
-
         if not await self._observe():
             # If there is no new information, suspend and wait
             logger.debug(f"{self._setting}: no news. waiting.")

metagpt/startup.py

@@ -7,7 +7,7 @@ from pathlib import Path
 import typer
 
 from metagpt.config2 import config
-from metagpt.const import METAGPT_ROOT
+from metagpt.const import CONFIG_ROOT, METAGPT_ROOT
 
 app = typer.Typer(add_completion=False, pretty_exceptions_show_locals=False)
 
@@ -118,7 +118,7 @@ def startup(
 
 def copy_config_to(config_path=METAGPT_ROOT / "config" / "config2.yaml"):
     """Initialize the configuration file for MetaGPT."""
-    target_path = Path.home() / ".metagpt" / "config2.yaml"
+    target_path = CONFIG_ROOT / "config2.yaml"
 
     # 创建目标目录（如果不存在）
     target_path.parent.mkdir(parents=True, exist_ok=True)

tests/metagpt/actions/test_action_node.py

@@ -23,14 +23,12 @@ from metagpt.team import Team
 async def test_debate_two_roles():
     action1 = Action(name="AlexSay", instruction="Express your opinion with emotion and don't repeat it")
     action2 = Action(name="BobSay", instruction="Express your opinion with emotion and don't repeat it")
-    biden = Role(
+    alex = Role(
         name="Alex", profile="Democratic candidate", goal="Win the election", actions=[action1], watch=[action2]
     )
-    trump = Role(
-        name="Bob", profile="Republican candidate", goal="Win the election", actions=[action2], watch=[action1]
-    )
+    bob = Role(name="Bob", profile="Republican candidate", goal="Win the election", actions=[action2], watch=[action1])
     env = Environment(desc="US election live broadcast")
-    team = Team(investment=10.0, env=env, roles=[biden, trump])
+    team = Team(investment=10.0, env=env, roles=[alex, bob])
 
     history = await team.run(idea="Topic: climate change. Under 80 words per message.", send_to="Alex", n_round=3)
     assert "Alex" in history
@@ -39,9 +37,9 @@ async def test_debate_two_roles():
 @pytest.mark.asyncio
 async def test_debate_one_role_in_env():
     action = Action(name="Debate", instruction="Express your opinion with emotion and don't repeat it")
-    biden = Role(name="Alex", profile="Democratic candidate", goal="Win the election", actions=[action])
+    alex = Role(name="Alex", profile="Democratic candidate", goal="Win the election", actions=[action])
     env = Environment(desc="US election live broadcast")
-    team = Team(investment=10.0, env=env, roles=[biden])
+    team = Team(investment=10.0, env=env, roles=[alex])
     history = await team.run(idea="Topic: climate change. Under 80 words per message.", send_to="Alex", n_round=3)
     assert "Alex" in history
 
@@ -49,8 +47,8 @@ async def test_debate_one_role_in_env():
 @pytest.mark.asyncio
 async def test_debate_one_role():
     action = Action(name="Debate", instruction="Express your opinion with emotion and don't repeat it")
-    biden = Role(name="Alex", profile="Democratic candidate", goal="Win the election", actions=[action])
-    msg: Message = await biden.run("Topic: climate change. Under 80 words per message.")
+    alex = Role(name="Alex", profile="Democratic candidate", goal="Win the election", actions=[action])
+    msg: Message = await alex.run("Topic: climate change. Under 80 words per message.")
 
     assert len(msg.content) > 10
     assert msg.sent_from == "metagpt.roles.role.Role"

tests/metagpt/provider/mock_llm_config.py

@@ -39,5 +39,6 @@ mock_llm_config_zhipu = LLMConfig(
     llm_type="zhipu",
     api_key="mock_api_key.zhipu",
     base_url="mock_base_url",
+    model="mock_zhipu_model",
     proxy="http://localhost:8080",
 )

tests/metagpt/provider/test_spark_api.py

@@ -4,6 +4,7 @@
 
 import pytest
 
+from metagpt.config2 import Config
 from metagpt.provider.spark_api import GetMessageFromWeb, SparkLLM
 from tests.metagpt.provider.mock_llm_config import mock_llm_config
 
@@ -33,6 +34,14 @@ def mock_spark_get_msg_from_web_run(self) -> str:
     return resp_content
 
 
+@pytest.mark.asyncio
+async def test_spark_aask():
+    llm = SparkLLM(Config.from_home("spark.yaml").llm)
+
+    resp = await llm.aask("Hello!")
+    print(resp)
+
+
 @pytest.mark.asyncio
 async def test_spark_acompletion(mocker):
     mocker.patch("metagpt.provider.spark_api.GetMessageFromWeb.run", mock_spark_get_msg_from_web_run)

tests/metagpt/roles/test_role.py

@@ -3,6 +3,7 @@
 # @Desc   : unittest of Role
 import pytest
 
+from metagpt.llm import HumanProvider
 from metagpt.roles.role import Role
 
 
@@ -12,5 +13,10 @@ def test_role_desc():
     assert role.desc == "Best Seller"
 
 
+def test_role_human():
+    role = Role(is_human=True)
+    assert isinstance(role.llm, HumanProvider)
+
+
 if __name__ == "__main__":
     pytest.main([__file__, "-s"])

tests/metagpt/test_config.py

@@ -5,15 +5,10 @@
 @Author  : alexanderwu
 @File    : test_config.py
 """
-from pydantic import BaseModel
 
 from metagpt.config2 import Config
 from metagpt.configs.llm_config import LLMType
-from metagpt.context_mixin import ContextMixin
-from tests.metagpt.provider.mock_llm_config import (
-    mock_llm_config,
-    mock_llm_config_proxy,
-)
+from tests.metagpt.provider.mock_llm_config import mock_llm_config
 
 
 def test_config_1():
@@ -27,57 +22,3 @@ def test_config_from_dict():
     cfg = Config(llm=mock_llm_config)
     assert cfg
     assert cfg.llm.api_key == "mock_api_key"
-
-
-class ModelX(ContextMixin, BaseModel):
-    a: str = "a"
-    b: str = "b"
-
-
-class WTFMixin(BaseModel):
-    c: str = "c"
-    d: str = "d"
-
-
-class ModelY(WTFMixin, ModelX):
-    pass
-
-
-def test_config_mixin_1():
-    new_model = ModelX()
-    assert new_model.a == "a"
-    assert new_model.b == "b"
-
-
-def test_config_mixin_2():
-    i = Config(llm=mock_llm_config)
-    j = Config(llm=mock_llm_config_proxy)
-    obj = ModelX(config=i)
-    assert obj.config == i
-    assert obj.config.llm == mock_llm_config
-
-    obj.set_config(j)
-    # obj already has a config, so it will not be set
-    assert obj.config == i
-
-
-def test_config_mixin_3():
-    """Test config mixin with multiple inheritance"""
-    i = Config(llm=mock_llm_config)
-    j = Config(llm=mock_llm_config_proxy)
-    obj = ModelY(config=i)
-    assert obj.private_config == i
-    assert obj.private_config.llm == mock_llm_config
-
-    obj.set_config(j)
-    # obj already has a config, so it will not be set
-    assert obj.private_config == i
-    assert obj.private_config.llm == mock_llm_config
-
-    assert obj.a == "a"
-    assert obj.b == "b"
-    assert obj.c == "c"
-    assert obj.d == "d"
-
-    print(obj.__dict__.keys())
-    assert "_config" in obj.__dict__.keys()

tests/metagpt/test_context_mixin.py

@@ -0,0 +1,128 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+"""
+@Time    : 2024/1/11 19:24
+@Author  : alexanderwu
+@File    : test_context_mixin.py
+"""
+import pytest
+from pydantic import BaseModel
+
+from metagpt.actions import Action
+from metagpt.config2 import Config
+from metagpt.context_mixin import ContextMixin
+from metagpt.environment import Environment
+from metagpt.roles import Role
+from metagpt.team import Team
+from tests.metagpt.provider.mock_llm_config import (
+    mock_llm_config,
+    mock_llm_config_proxy,
+    mock_llm_config_zhipu,
+)
+
+
+class ModelX(ContextMixin, BaseModel):
+    a: str = "a"
+    b: str = "b"
+
+
+class WTFMixin(BaseModel):
+    c: str = "c"
+    d: str = "d"
+
+
+class ModelY(WTFMixin, ModelX):
+    pass
+
+
+def test_config_mixin_1():
+    new_model = ModelX()
+    assert new_model.a == "a"
+    assert new_model.b == "b"
+
+
+def test_config_mixin_2():
+    i = Config(llm=mock_llm_config)
+    j = Config(llm=mock_llm_config_proxy)
+    obj = ModelX(config=i)
+    assert obj.config == i
+    assert obj.config.llm == mock_llm_config
+
+    obj.set_config(j)
+    # obj already has a config, so it will not be set
+    assert obj.config == i
+
+
+def test_config_mixin_3_multi_inheritance_not_override_config():
+    """Test config mixin with multiple inheritance"""
+    i = Config(llm=mock_llm_config)
+    j = Config(llm=mock_llm_config_proxy)
+    obj = ModelY(config=i)
+    assert obj.config == i
+    assert obj.config.llm == mock_llm_config
+
+    obj.set_config(j)
+    # obj already has a config, so it will not be set
+    assert obj.config == i
+    assert obj.config.llm == mock_llm_config
+
+    assert obj.a == "a"
+    assert obj.b == "b"
+    assert obj.c == "c"
+    assert obj.d == "d"
+
+    print(obj.__dict__.keys())
+    assert "private_config" in obj.__dict__.keys()
+
+
+def test_config_mixin_4_multi_inheritance_override_config():
+    """Test config mixin with multiple inheritance"""
+    i = Config(llm=mock_llm_config)
+    j = Config(llm=mock_llm_config_zhipu)
+    obj = ModelY(config=i)
+    assert obj.config == i
+    assert obj.config.llm == mock_llm_config
+
+    obj.set_config(j, override=True)
+    # override obj.config
+    assert obj.config == j
+    assert obj.config.llm == mock_llm_config_zhipu
+
+    assert obj.a == "a"
+    assert obj.b == "b"
+    assert obj.c == "c"
+    assert obj.d == "d"
+
+    print(obj.__dict__.keys())
+    assert "private_config" in obj.__dict__.keys()
+    assert obj.llm.model == "mock_zhipu_model"
+
+
+@pytest.mark.asyncio
+async def test_config_priority():
+    """If action's config is set, then its llm will be set, otherwise, it will use the role's llm"""
+    gpt4t = Config.from_home("gpt-4-1106-preview.yaml")
+    gpt35 = Config.default()
+    gpt4 = Config.default()
+    gpt4.llm.model = "gpt-4-0613"
+
+    a1 = Action(config=gpt4t, name="Say", instruction="Say your opinion with emotion and don't repeat it")
+    a2 = Action(name="Say", instruction="Say your opinion with emotion and don't repeat it")
+    a3 = Action(name="Vote", instruction="Vote for the candidate, and say why you vote for him/her")
+
+    # it will not work for a1 because the config is already set
+    A = Role(name="A", profile="Democratic candidate", goal="Win the election", actions=[a1], watch=[a2], config=gpt4)
+    # it will work for a2 because the config is not set
+    B = Role(name="B", profile="Republican candidate", goal="Win the election", actions=[a2], watch=[a1], config=gpt4)
+    # ditto
+    C = Role(name="C", profile="Voter", goal="Vote for the candidate", actions=[a3], watch=[a1, a2], config=gpt35)
+
+    env = Environment(desc="US election live broadcast")
+    Team(investment=10.0, env=env, roles=[A, B, C])
+
+    assert a1.llm.model == "gpt-4-1106-preview"
+    assert a2.llm.model == "gpt-4-0613"
+    assert a3.llm.model == "gpt-3.5-turbo-1106"
+
+    # history = await team.run(idea="Topic: climate change. Under 80 words per message.", send_to="a1", n_round=3)
+    # assert "Alex" in history

tests/metagpt/utils/test_redis.py

@@ -5,25 +5,26 @@
 @Author  : mashenquan
 @File    : test_redis.py
 """
+from unittest.mock import AsyncMock
 
-import mock
 import pytest
+from pytest_mock import mocker
 
 from metagpt.config2 import Config
 from metagpt.utils.redis import Redis
 
 
 async def async_mock_from_url(*args, **kwargs):
-    mock_client = mock.AsyncMock()
+    mock_client = AsyncMock()
     mock_client.set.return_value = None
     mock_client.get.side_effect = [b"test", b""]
     return mock_client
 
 
 @pytest.mark.asyncio
-@mock.patch("aioredis.from_url", return_value=async_mock_from_url())
 async def test_redis(i):
     redis = Config.default().redis
+    mocker.patch("aioredis.from_url", return_value=async_mock_from_url())
 
     conn = Redis(redis)
     await conn.set("test", "test", timeout_sec=0)

tests/mock/mock_llm.py

@@ -42,7 +42,9 @@ class MockLLM(OpenAILLM):
         if system_msgs:
             message = self._system_msgs(system_msgs)
         else:
-            message = [self._default_system_msg()] if self.use_system_prompt else []
+            message = [self._default_system_msg()]
+        if not self.use_system_prompt:
+            message = []
         if format_msgs:
             message.extend(format_msgs)
         message.append(self._user_msg(msg))