-
Notifications
You must be signed in to change notification settings - Fork 4.3k
/
workflowmanager.py
218 lines (185 loc) · 8.22 KB
/
workflowmanager.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
import os
from typing import List, Optional
import autogen
from .datamodel import AgentConfig, AgentFlowSpec, AgentWorkFlowConfig, Message
from .utils import get_skills_from_prompt, clear_folder, sanitize_model
from datetime import datetime
class AutoGenWorkFlowManager:
"""
AutoGenWorkFlowManager class to load agents from a provided configuration and run a chat between them
"""
def __init__(
self,
config: AgentWorkFlowConfig,
history: Optional[List[Message]] = None,
work_dir: str = None,
clear_work_dir: bool = True,
) -> None:
"""
Initializes the AutoGenFlow with agents specified in the config and optional
message history.
Args:
config: The configuration settings for the sender and receiver agents.
history: An optional list of previous messages to populate the agents' history.
"""
self.work_dir = work_dir or "work_dir"
if clear_work_dir:
clear_folder(self.work_dir)
# given the config, return an AutoGen agent object
self.sender = self.load(config.sender)
# given the config, return an AutoGen agent object
self.receiver = self.load(config.receiver)
if config.receiver.type == "groupchat":
# append self.sender to the list of agents
self.receiver._groupchat.agents.append(self.sender)
print(self.receiver)
self.agent_history = []
if history:
self.populate_history(history)
def process_reply(self, recipient, messages, sender, config):
if "callback" in config and config["callback"] is not None:
callback = config["callback"]
callback(sender, recipient, messages[-1])
last_message = messages[-1]
sender = sender.name
recipient = recipient.name
if "name" in last_message:
sender = last_message["name"]
iteration = {
"recipient": recipient,
"sender": sender,
"message": last_message,
"timestamp": datetime.now().isoformat(),
}
self.agent_history.append(iteration)
return False, None
def _sanitize_history_message(self, message: str) -> str:
"""
Sanitizes the message e.g. remove references to execution completed
Args:
message: The message to be sanitized.
Returns:
The sanitized message.
"""
to_replace = ["execution succeeded", "exitcode"]
for replace in to_replace:
message = message.replace(replace, "")
return message
def populate_history(self, history: List[Message]) -> None:
"""
Populates the agent message history from the provided list of messages.
Args:
history: A list of messages to populate the agents' history.
"""
for msg in history:
if isinstance(msg, dict):
msg = Message(**msg)
if msg.role == "user":
self.sender.send(
msg.content,
self.receiver,
request_reply=False,
)
elif msg.role == "assistant":
self.receiver.send(
msg.content,
self.sender,
request_reply=False,
)
def sanitize_agent_spec(self, agent_spec: AgentFlowSpec) -> AgentFlowSpec:
"""
Sanitizes the agent spec by setting loading defaults
Args:
config: The agent configuration to be sanitized.
agent_type: The type of the agent.
Returns:
The sanitized agent configuration.
"""
agent_spec.config.is_termination_msg = agent_spec.config.is_termination_msg or (
lambda x: "TERMINATE" in x.get("content", "").rstrip()[-20:]
)
def get_default_system_message(agent_type: str) -> str:
if agent_type == "assistant":
return autogen.AssistantAgent.DEFAULT_SYSTEM_MESSAGE
else:
return "You are a helpful AI Assistant."
# sanitize llm_config if present
if agent_spec.config.llm_config is not False:
config_list = []
for llm in agent_spec.config.llm_config.config_list:
# check if api_key is present either in llm or env variable
if "api_key" not in llm and "OPENAI_API_KEY" not in os.environ:
error_message = f"api_key is not present in llm_config or OPENAI_API_KEY env variable for agent ** {agent_spec.config.name}**. Update your workflow to provide an api_key to use the LLM."
raise ValueError(error_message)
# only add key if value is not None
sanitized_llm = sanitize_model(llm)
config_list.append(sanitized_llm)
agent_spec.config.llm_config.config_list = config_list
if agent_spec.config.code_execution_config is not False:
code_execution_config = agent_spec.config.code_execution_config or {}
code_execution_config["work_dir"] = self.work_dir
# tbd check if docker is installed
code_execution_config["use_docker"] = False
agent_spec.config.code_execution_config = code_execution_config
if agent_spec.skills:
# get skill prompt, also write skills to a file named skills.py
skills_prompt = ""
skills_prompt = get_skills_from_prompt(agent_spec.skills, self.work_dir)
if agent_spec.config.system_message:
agent_spec.config.system_message = agent_spec.config.system_message + "\n\n" + skills_prompt
else:
agent_spec.config.system_message = get_default_system_message(agent_spec.type) + "\n\n" + skills_prompt
return agent_spec
def load(self, agent_spec: AgentFlowSpec) -> autogen.Agent:
"""
Loads an agent based on the provided agent specification.
Args:
agent_spec: The specification of the agent to be loaded.
Returns:
An instance of the loaded agent.
"""
agent_spec = self.sanitize_agent_spec(agent_spec)
if agent_spec.type == "groupchat":
agents = [
self.load(self.sanitize_agent_spec(agent_config)) for agent_config in agent_spec.groupchat_config.agents
]
group_chat_config = agent_spec.groupchat_config.dict()
group_chat_config["agents"] = agents
groupchat = autogen.GroupChat(**group_chat_config)
agent = autogen.GroupChatManager(groupchat=groupchat, **agent_spec.config.dict())
agent.register_reply([autogen.Agent, None], reply_func=self.process_reply, config={"callback": None})
return agent
else:
agent = self.load_agent_config(agent_spec.config, agent_spec.type)
return agent
def load_agent_config(self, agent_config: AgentConfig, agent_type: str) -> autogen.Agent:
"""
Loads an agent based on the provided agent configuration.
Args:
agent_config: The configuration of the agent to be loaded.
agent_type: The type of the agent to be loaded.
Returns:
An instance of the loaded agent.
"""
if agent_type == "assistant":
agent = autogen.AssistantAgent(**agent_config.dict())
elif agent_type == "userproxy":
agent = autogen.UserProxyAgent(**agent_config.dict())
else:
raise ValueError(f"Unknown agent type: {agent_type}")
agent.register_reply([autogen.Agent, None], reply_func=self.process_reply, config={"callback": None})
return agent
def run(self, message: str, clear_history: bool = False) -> None:
"""
Initiates a chat between the sender and receiver agents with an initial message
and an option to clear the history.
Args:
message: The initial message to start the chat.
clear_history: If set to True, clears the chat history before initiating.
"""
self.sender.initiate_chat(
self.receiver,
message=message,
clear_history=clear_history,
)
# pass