feat: solution for 'Практическое задание: AI-агент на LangChain'

.gitignore

@@ -0,0 +1,5 @@
+node_modules/
+.env
+dist/
+build/
+*.log

README.md

@@ -0,0 +1,53 @@
+# Shopping List AI Agent
+
+This project demonstrates a simple hierarchical AI agent built with LangChain that helps plan a shopping list, fetches prices for each item using a sub‑agent, and calculates the total cost.
+
+## Prerequisites
+
+- Python 3.10+
+- A local LLM server (e.g., LM Studio) running at `http://localhost:1234/v1`.  
+  The server must expose an OpenAI‑compatible API.
+
+## Installation
+
+bash
+# Clone the repository
+git clone https://github.com/your-username/shopping-list-agent.git
+cd shopping-list-agent
+
+# Create a virtual environment (optional but recommended)
+python -m venv venv
+source venv/bin/activate   # On Windows use `venv\\Scripts\\activate`
+
+# Install dependencies
+pip install -r requirements.txt
+
+
+## Running the Agent
+
+bash
+python src/main.py
+
+
+The script will ask the agent to plan a shopping list for the following request:
+
+
+Помоги составить список покупок: молоко, хлеб, яблоки. Я нахожусь в Казани.
+
+
+The agent will:
+
+1. Parse the list of products and the city.
+2. For each product, invoke the `get_price` tool, which internally creates a sub‑agent that generates a realistic price table.
+3. Sum the prices and output the final total.
+
+All intermediate tool calls and the final answer are printed to the console.
+
+## Customization
+
+- **Model**: Change the `model` parameter in `src/main.py` to match the model name exposed by your LM Studio instance.
+- **Temperature**: Adjust the `temperature` argument to control the randomness of the responses.
+
+## License
+
+MIT License

main.py

@@ -0,0 +1,70 @@
+import json
+from langchain_openai import ChatOpenAI
+from pydantic import SecretStr
+from langchain.tools import tool
+from langchain.agents import create_agent
+
+# Configure the local LLM
+llm = ChatOpenAI(
+    model="gpt-4o-mini",  # Replace with your LM Studio model name
+    base_url="http://localhost:1234/v1",
+    api_key=SecretStr("fake"),
+    temperature=0.7,
+)
+
+@tool
+def get_price(product: str, city: str) -> str:
+    """Get realistic price for a product in a city. Returns a markdown table."""
+    # Create a sub-agent that generates a price table
+    sub_agent = create_agent(
+        model=llm,
+        tools=[],
+        system_prompt=f"You are a price estimator. Provide a realistic price for {product} in {city}. Return a markdown table with columns: Product, Price (rub.), Store.",
+    )
+    # Invoke the sub-agent
+    result = sub_agent.invoke(
+        {
+            "messages": [
+                {"role": "user", "content": f"Provide price for {product} in {city}."}
+            ]
+        }
+    )
+    # Extract the assistant message content
+    messages = result.get("messages", [])
+    for msg in messages:
+        if msg.get("role") == "assistant" and msg.get("content"):
+            return msg["content"]
+    return "No price data available."
+
+def main():
+    # Main agent that uses the get_price tool
+    agent = create_agent(
+        model=llm,
+        tools=[get_price],
+        system_prompt="You are a shopping list planner. Use the get_price tool to find prices for items.",
+    )
+    # Sample user query
+    user_query = "Помоги составить список покупок: молоко, хлеб, яблоки. Я нахожусь в Казани."
+    # Invoke the agent
+    result = agent.invoke(
+        {
+            "messages": [
+                {"role": "human", "content": user_query}
+            ]
+        }
+    )
+    # Print all messages, including tool calls and final answer
+    for msg in result.get("messages", []):
+        role = msg.get("role")
+        content = msg.get("content")
+        tool_calls = msg.get("tool_calls")
+        if content:
+            print(f"{role}: {content}")
+        elif tool_calls:
+            for call in tool_calls:
+                print(f"{role} calls {call.get('name')} with args {call.get('arguments')}")
+        else:
+            print(f"{role}: (no content)")
+
+if __name__ == "__main__":
+    main()

requirements.txt

@@ -0,0 +1,2 @@
+langchain==1.2.10
+langchain-openai==1.1.9

src/main.py

@@ -0,0 +1,50 @@
+import os
+from langchain_openai import ChatOpenAI
+from langchain.tools import tool
+from langchain.agents import create_agent
+from pydantic import SecretStr
+
+# Initialize the LLM
+llm = ChatOpenAI(
+    model="gpt-4o-mini",
+    base_url="http://localhost:1234/v1",
+    api_key=SecretStr("fake"),
+    temperature=0.7,
+)
+
+@tool
+def get_price(product: str, city: str) -> str:
+    """Get price for a product in a city. Returns a markdown table with columns: Продукт, Цена (руб.), Магазин."""
+    # Create a sub-agent to estimate the price
+    sub_agent = create_agent(
+        model=llm,
+        tools=[],
+        system_prompt=f"""You are a price estimator for product '{product}' in city '{city}'. Provide a realistic price based on historical data. Output a markdown table with columns: Продукт, Цена (руб.), Магазин."""
+    )
+    # Invoke the sub-agent
+    sub_response = sub_agent.invoke({"messages": [{"role": "human", "content": ""}]})
+    # Return the content of the last message
+    return sub_response["messages"][-1]["content"]
+
+# Main agent
+main_agent = create_agent(
+    model=llm,
+    tools=[get_price],
+    system_prompt="Ты помощник по планированию покупок. Принимаешь список продуктов и город. Для каждого продукта вызывай инструмент get_price, затем суммируй цены и выдавай итоговую стоимость."
+)
+
+def main():
+    question = "Помоги составить список покупок: молоко, хлеб, яблоки. Я нахожусь в Казани."
+    response = main_agent.invoke({"messages": [{"role": "human", "content": question}]})
+    # Print all messages
+    for msg in response["messages"]:
+        role = msg.get("role", "")
+        if role == "assistant":
+            print(msg["content"])
+        elif role == "tool":
+            print(f"Tool call: {msg['name']} with args {msg.get('arguments')}")
+        else:
+            print(msg["content"])
+
+if __name__ == "__main__":
+    main()