Rename agent.py to langgraph_agent.py
Browse files- agent.py +0 -0
- langgraph_agent.py +115 -0
agent.py
DELETED
File without changes
|
langgraph_agent.py
ADDED
@@ -0,0 +1,115 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
from langgraph.graph import START, StateGraph, MessagesState
|
3 |
+
from langgraph.prebuilt import tools_condition, ToolNode
|
4 |
+
from langchain_openai import ChatOpenAI
|
5 |
+
from langchain_huggingface import ChatHuggingFace, HuggingFaceEndpoint
|
6 |
+
from langchain_community.tools.tavily_search import TavilySearchResults
|
7 |
+
from langchain_community.document_loaders import WikipediaLoader, ArxivLoader
|
8 |
+
from langchain_core.messages import SystemMessage, HumanMessage
|
9 |
+
from langchain_core.tools import tool
|
10 |
+
|
11 |
+
#tools
|
12 |
+
@tool
|
13 |
+
def multiply(a: int, b: int) -> int:
|
14 |
+
return a * b
|
15 |
+
|
16 |
+
@tool
|
17 |
+
def add(a: int, b: int) -> int:
|
18 |
+
return a + b
|
19 |
+
|
20 |
+
@tool
|
21 |
+
def subtract(a: int, b: int) -> int:
|
22 |
+
return a - b
|
23 |
+
|
24 |
+
@tool
|
25 |
+
def divide(a: int, b: int) -> float:
|
26 |
+
if b == 0:
|
27 |
+
raise ValueError("Cannot divide by zero.")
|
28 |
+
return a / b
|
29 |
+
|
30 |
+
@tool
|
31 |
+
def modulus(a: int, b: int) -> int:
|
32 |
+
return a % b
|
33 |
+
|
34 |
+
@tool
|
35 |
+
def wiki_search(query: str) -> dict:
|
36 |
+
docs = WikipediaLoader(query=query, load_max_docs=2).load()
|
37 |
+
formatted = "\n\n---\n\n".join(
|
38 |
+
f'<Document source="{d.metadata["source"]}"/>\n{d.page_content}'
|
39 |
+
for d in docs
|
40 |
+
)
|
41 |
+
return {"wiki_results": formatted}
|
42 |
+
|
43 |
+
@tool
|
44 |
+
def web_search(query: str) -> dict:
|
45 |
+
docs = TavilySearchResults(max_results=3).invoke(query=query)
|
46 |
+
formatted = "\n\n---\n\n".join(
|
47 |
+
f'<Document source="{d.metadata["source"]}"/>\n{d.page_content}'
|
48 |
+
for d in docs
|
49 |
+
)
|
50 |
+
return {"web_results": formatted}
|
51 |
+
|
52 |
+
@tool
|
53 |
+
def arvix_search(query: str) -> dict:
|
54 |
+
docs = ArxivLoader(query=query, load_max_docs=3).load()
|
55 |
+
formatted = "\n\n---\n\n".join(
|
56 |
+
f'<Document source="{d.metadata["source"]}"/>\n{d.page_content[:1000]}'
|
57 |
+
for d in docs
|
58 |
+
)
|
59 |
+
return {"arvix_results": formatted}
|
60 |
+
|
61 |
+
|
62 |
+
OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
|
63 |
+
HF_SPACE_TOKEN = os.getenv("HF_SPACE_TOKEN")
|
64 |
+
|
65 |
+
# βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
|
66 |
+
# 4) Assemble tool list
|
67 |
+
tools = [
|
68 |
+
multiply, add, subtract, divide, modulus,
|
69 |
+
wiki_search, web_search, arvix_search,
|
70 |
+
]
|
71 |
+
|
72 |
+
# βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
|
73 |
+
# 5) Load your system prompt
|
74 |
+
with open("system_prompt.txt", "r", encoding="utf-8") as f:
|
75 |
+
system_prompt = f.read()
|
76 |
+
sys_msg = SystemMessage(content=system_prompt)
|
77 |
+
|
78 |
+
# βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
|
79 |
+
def build_graph(provider: str = "openai"):
|
80 |
+
"""Build the LangGraph agent with chosen LLM (default: OpenAI)."""
|
81 |
+
if provider == "openai":
|
82 |
+
llm = ChatOpenAI(
|
83 |
+
model_name="o4-mini-2025-04-16",
|
84 |
+
openai_api_key=OPENAI_API_KEY,
|
85 |
+
# no temperature override here
|
86 |
+
)
|
87 |
+
elif provider == "huggingface":
|
88 |
+
llm = ChatHuggingFace(
|
89 |
+
llm=HuggingFaceEndpoint(
|
90 |
+
url="https://api-inference.huggingface.co/models/Meta-DeepLearning/llama-2-7b-chat-hf",
|
91 |
+
),
|
92 |
+
temperature=0,
|
93 |
+
)
|
94 |
+
else:
|
95 |
+
raise ValueError("Invalid provider. Choose 'openai' or 'huggingface'.")
|
96 |
+
|
97 |
+
llm_with_tools = llm.bind_tools(tools)
|
98 |
+
|
99 |
+
def assistant(state: MessagesState):
|
100 |
+
return {"messages": [llm_with_tools.invoke(state["messages"])]}
|
101 |
+
|
102 |
+
builder = StateGraph(MessagesState)
|
103 |
+
builder.add_node("assistant", assistant)
|
104 |
+
builder.add_node("tools", ToolNode(tools))
|
105 |
+
builder.add_edge(START, "assistant")
|
106 |
+
builder.add_conditional_edges("assistant", tools_condition)
|
107 |
+
builder.add_edge("tools", "assistant")
|
108 |
+
|
109 |
+
return builder.compile()
|
110 |
+
|
111 |
+
if __name__ == "__main__":
|
112 |
+
graph = build_graph()
|
113 |
+
msgs = graph.invoke({"messages":[ HumanMessage(content="Whatβs the capital of France?") ]})
|
114 |
+
for m in msgs["messages"]:
|
115 |
+
m.pretty_print()
|