Agents with create_agent

from standins import ScriptedModel, call
from langchain.tools import tool
from langchain.agents import create_agent
from langgraph.checkpoint.memory import InMemorySaver
from langchain.messages import AIMessage
from pydantic import BaseModel

@tool
def get_weather(city: str) -> str:
    """Get the current weather of an Indian city."""
    return {"pune": "28°C, cloudy", "goa": "30°C, sunny", "manali": "6°C, snow"}.get(city.lower(), "25°C, clear")

@tool
def convert_currency(amount: float, from_currency: str, to_currency: str) -> str:
    """Convert money between INR, USD and EUR."""
    rate = {"INR": 1, "USD": 84, "EUR": 91}
    return f"{amount} {from_currency} = {round(amount * rate[from_currency.upper()] / rate[to_currency.upper()], 2)} {to_currency}"

tools = [get_weather, convert_currency]
llm = ScriptedModel([
    AIMessage("", tool_calls=[{"name": "convert_currency", "args": {"amount": 500, "from_currency": "USD", "to_currency": "INR"}, "id": "a"},
                              {"name": "get_weather", "args": {"city": "Goa"}, "id": "b"}]),
    "500 USD is about Rs 42,000. Goa is 30°C and sunny - pack sunscreen! 🏖️"])
agent = create_agent(model=llm, tools=tools, system_prompt="You are a helpful travel assistant. Use tools when needed. Keep answers short.")
result = agent.invoke({"messages": [{"role": "user", "content": "I am going to Goa with 500 USD. How much is that in INR and how is the weather?"}]})
for m in result["messages"]:
    print(f"{m.type:5s} | " + (", ".join(f"{c['name']}({c['args']})" for c in m.tool_calls) if getattr(m, "tool_calls", None) else m.content))

print("\n--- stream_mode='updates' ---")
llm2 = ScriptedModel([call("get_weather", {"city": "Manali"}), "Manali is 6°C with snow."])
for chunk in create_agent(llm2, tools).stream({"messages": [{"role": "user", "content": "Weather in Manali?"}]}, stream_mode="updates"):
    for step, data in chunk.items():
        last = data["messages"][-1]
        print(f"[{step}]", last.tool_calls[0]["name"] if getattr(last, "tool_calls", None) else last.content)

print("\n--- memory: checkpointer + thread_id ---")
mem_llm = ScriptedModel(["Noted: budget 800 EUR, trip to Pune.", "Your budget is 800 EUR.", "I don't know your budget yet - please tell me."])
memory_agent = create_agent(mem_llm, tools, checkpointer=InMemorySaver())
cfg = {"configurable": {"thread_id": "ajeet-1"}}
memory_agent.invoke({"messages": [{"role": "user", "content": "My budget is 800 EUR and I am going to Pune."}]}, cfg)
r = memory_agent.invoke({"messages": [{"role": "user", "content": "What is my budget?"}]}, cfg)
print("thread ajeet-1 :", len(r["messages"]), "messages saved ->", r["messages"][-1].content)
r = memory_agent.invoke({"messages": [{"role": "user", "content": "What is my budget?"}]}, {"configurable": {"thread_id": "new-user"}})
print("thread new-user:", len(r["messages"]), "messages saved ->", r["messages"][-1].content)

print("\n--- structured response ---")
class TripPlan(BaseModel):
    city: str; weather: str; budget_inr: float; packing_tips: list[str]
plan_llm = ScriptedModel([call("TripPlan", {"city": "Manali", "weather": "6°C, snow", "budget_inr": 25200, "packing_tips": ["thermals", "gloves", "snow boots"]})])
r = create_agent(plan_llm, tools, response_format=TripPlan).invoke({"messages": [{"role": "user", "content": "Plan for Manali with 300 USD"}]})
print(repr(r["structured_response"]))
print("\n" + agent.get_graph().draw_ascii())