This commit is contained in:
cjw
2026-02-17 02:31:39 +08:00
parent 2f0e25c400
commit 27a1b8a7e9
14 changed files with 877 additions and 1 deletions
+76
View File
@@ -0,0 +1,76 @@
from typing import Any, Dict, List, Optional
from langchain_core.messages import BaseMessage, HumanMessage
from langchain_openai import ChatOpenAI
from langgraph.graph import StateGraph, END
from config import Config
class AgentState:
"""State definition for the agent workflow"""
messages: List[BaseMessage]
current_step: str
context: Dict[str, Any]
def __init__(self, messages: List[BaseMessage] = None, current_step: str = "start", context: Dict[str, Any] = None):
self.messages = messages or []
self.current_step = current_step
self.context = context or {}
class BaseAgent:
"""Base agent class with common functionality"""
def __init__(self, model_name: str = Config.DEFAULT_MODEL):
self.model = ChatOpenAI(
model=model_name,
api_key=Config.OPENAI_API_KEY,
temperature=0.1,
max_retries=Config.MAX_RETRIES,
timeout=Config.TIMEOUT
)
self.graph = self._build_graph()
def _build_graph(self) -> StateGraph:
"""Build the state graph for the agent"""
workflow = StateGraph(AgentState)
# Add nodes and edges
workflow.add_node("process_input", self._process_input)
workflow.add_node("generate_response", self._generate_response)
# Define edges
workflow.add_edge("process_input", "generate_response")
workflow.add_edge("generate_response", END)
# Set entry point
workflow.set_entry_point("process_input")
return workflow.compile()
def _process_input(self, state: AgentState) -> AgentState:
"""Process user input"""
# This is a base implementation - subclasses should override
state.current_step = "processed"
return state
def _generate_response(self, state: AgentState) -> AgentState:
"""Generate response using the LLM"""
if state.messages:
response = self.model.invoke(state.messages)
state.messages.append(response)
return state
def run(self, user_input: str, **kwargs) -> Dict[str, Any]:
"""Run the agent with user input"""
initial_state = AgentState(
messages=[HumanMessage(content=user_input)],
context=kwargs
)
result = self.graph.invoke(initial_state)
return {
"messages": result.messages,
"context": result.context,
"final_step": result.current_step
}
+94
View File
@@ -0,0 +1,94 @@
from typing import Dict, Any, List
from langchain_core.messages import BaseMessage, HumanMessage, AIMessage
from langgraph.graph import StateGraph, END
from .base_agent import BaseAgent, AgentState
class ConversationAgent(BaseAgent):
"""Agent for handling multi-turn conversations"""
def __init__(self, model_name: str = None):
super().__init__(model_name)
self.conversation_history: List[BaseMessage] = []
def _build_graph(self) -> StateGraph:
"""Build conversation-specific graph"""
workflow = StateGraph(AgentState)
# Add nodes
workflow.add_node("analyze_intent", self._analyze_intent)
workflow.add_node("generate_response", self._generate_response)
workflow.add_node("update_context", self._update_context)
# Define edges
workflow.add_edge("analyze_intent", "generate_response")
workflow.add_edge("generate_response", "update_context")
workflow.add_edge("update_context", END)
# Set entry point
workflow.set_entry_point("analyze_intent")
return workflow.compile()
def _analyze_intent(self, state: AgentState) -> AgentState:
"""Analyze user intent and conversation context"""
# Simple intent analysis - can be enhanced with more sophisticated logic
user_message = state.messages[-1] if state.messages else None
if user_message and isinstance(user_message, HumanMessage):
content = user_message.content.lower()
# Basic intent detection
if any(word in content for word in ["hello", "hi", "hey", "greetings"]):
state.context["intent"] = "greeting"
elif any(word in content for word in ["help", "assist", "support"]):
state.context["intent"] = "help"
elif "?" in content:
state.context["intent"] = "question"
else:
state.context["intent"] = "general"
state.current_step = "intent_analyzed"
return state
def _generate_response(self, state: AgentState) -> AgentState:
"""Generate response considering conversation history"""
# Combine conversation history with current message
all_messages = self.conversation_history + state.messages
if all_messages:
response = self.model.invoke(all_messages)
state.messages.append(response)
state.current_step = "response_generated"
return state
def _update_context(self, state: AgentState) -> AgentState:
"""Update conversation context and history"""
# Add the conversation to history (excluding system messages)
for message in state.messages:
if isinstance(message, (HumanMessage, AIMessage)):
self.conversation_history.append(message)
# Limit conversation history to avoid token limits
if len(self.conversation_history) > 10:
self.conversation_history = self.conversation_history[-10:]
state.current_step = "context_updated"
return state
def run(self, user_input: str, **kwargs) -> Dict[str, Any]:
"""Run conversation with history management"""
initial_state = AgentState(
messages=[HumanMessage(content=user_input)],
context=kwargs
)
result = self.graph.invoke(initial_state)
return {
"messages": result.messages,
"context": result.context,
"conversation_history": self.conversation_history,
"final_step": result.current_step
}
+86
View File
@@ -0,0 +1,86 @@
from typing import Dict, Any, List, Optional
from langchain_core.messages import BaseMessage, HumanMessage, AIMessage, ToolMessage
from langchain_core.tools import BaseTool
from langgraph.graph import StateGraph, END
from langgraph.prebuilt import ToolNode
from .base_agent import BaseAgent, AgentState
from tools.calculator import CalculatorTool
from tools.web_search import WebSearchTool
class ToolAgent(BaseAgent):
"""Agent that can use tools to accomplish tasks"""
def __init__(self, model_name: str = None, tools: List[BaseTool] = None):
# Initialize with default tools if none provided
if tools is None:
tools = [CalculatorTool(), WebSearchTool()]
self.tools = tools
self.tool_node = ToolNode(tools)
super().__init__(model_name)
def _build_graph(self) -> StateGraph:
"""Build tool-using graph"""
workflow = StateGraph(AgentState)
# Add nodes
workflow.add_node("agent", self._agent_node)
workflow.add_node("tools", self.tool_node)
# Define edges
workflow.add_edge("tools", "agent")
# Conditional routing
workflow.add_conditional_edges(
"agent",
self._should_use_tools,
{
"tools": "tools",
"end": END,
}
)
# Set entry point
workflow.set_entry_point("agent")
return workflow.compile()
def _agent_node(self, state: AgentState) -> AgentState:
"""Agent node that decides whether to use tools"""
# Bind tools to the model
model_with_tools = self.model.bind_tools(self.tools)
# Get the last message
if state.messages:
response = model_with_tools.invoke(state.messages)
state.messages.append(response)
return state
def _should_use_tools(self, state: AgentState) -> str:
"""Determine if tools should be used"""
last_message = state.messages[-1]
# If the last message has tool calls, route to tools
if hasattr(last_message, 'tool_calls') and last_message.tool_calls:
return "tools"
# Otherwise, end the workflow
return "end"
def run(self, user_input: str, **kwargs) -> Dict[str, Any]:
"""Run the tool-using agent"""
initial_state = AgentState(
messages=[HumanMessage(content=user_input)],
context=kwargs
)
result = self.graph.invoke(initial_state)
return {
"messages": result.messages,
"context": result.context,
"tools_used": [tool.name for tool in self.tools],
"final_step": result.current_step
}