125 lines
4.2 KiB
Python
125 lines
4.2 KiB
Python
import os
|
||
import asyncio
|
||
from pathlib import Path
|
||
from langchain_openai import ChatOpenAI, OpenAIEmbeddings
|
||
from langchain_chroma import Chroma
|
||
from langchain_core.documents import Document
|
||
from langchain.tools import tool
|
||
from deepagents import create_deep_agent
|
||
from deepagents.backends import FilesystemBackend, LocalShellBackend, CompositeBackend
|
||
|
||
# ----------------- Configuration -----------------
|
||
# Load API key from .env or environment variable
|
||
os.environ.setdefault("OPENAI_API_KEY", os.getenv("OPENAI_API_KEY", ""))
|
||
|
||
# LLM via OpenRouter
|
||
llm = ChatOpenAI(
|
||
model="openai/gpt-oss-20b:free",
|
||
base_url="https://openrouter.ai/api/v1",
|
||
api_key=os.getenv("OPENAI_API_KEY"),
|
||
temperature=0.0,
|
||
)
|
||
|
||
# Embeddings for Chroma
|
||
embeddings = OpenAIEmbeddings(
|
||
model="text-embedding-3-small",
|
||
base_url="https://openrouter.ai/api/v1",
|
||
api_key=os.getenv("OPENAI_API_KEY"),
|
||
)
|
||
|
||
# ----------------- Chroma DB -----------------
|
||
CHROMA_PATH = Path("./chroma_faq")
|
||
CHROMA_COLLECTION = "faq_collection"
|
||
vector_store = Chroma(
|
||
collection_name=CHROMA_COLLECTION,
|
||
embedding_function=embeddings,
|
||
persist_directory=str(CHROMA_PATH),
|
||
)
|
||
|
||
# Load markdown files into Chroma (idempotent)
|
||
DATA_DIR = Path("./data")
|
||
if not CHROMA_PATH.exists() or not list(CHROMA_PATH.iterdir()):
|
||
docs = []
|
||
for md_file in DATA_DIR.glob("*.md"):
|
||
text = md_file.read_text(encoding="utf-8")
|
||
docs.append(Document(page_content=text, metadata={"source": md_file.name}))
|
||
vector_store.add_documents(docs)
|
||
vector_store.persist()
|
||
|
||
# ----------------- Tools -----------------
|
||
@tool
|
||
def search_course_docs(query: str) -> str:
|
||
"""Search the local FAQ collection for relevant passages."""
|
||
results = vector_store.similarity_search(query, k=3)
|
||
if not results:
|
||
return "No relevant information found in the course materials."
|
||
return "\n---\n".join(f"{doc.metadata.get('source', 'unknown')}\n{doc.page_content}" for doc in results)
|
||
|
||
# Mock MCP tool – static JSON data
|
||
COURSE_META = {
|
||
"schedule": "Monday 10:00-12:00, Wednesday 14:00-16:00",
|
||
"instructor": "Dr. Ivanov",
|
||
"location": "Room 101",
|
||
}
|
||
|
||
@tool
|
||
def fetch_course_meta(query: str) -> str:
|
||
"""Return course metadata matching the query keyword.
|
||
For example, query="schedule" returns the schedule string.
|
||
"""
|
||
key = query.lower().strip()
|
||
return COURSE_META.get(key, f"No metadata found for '{query}'.")
|
||
|
||
# ----------------- Backend -----------------
|
||
backend = CompositeBackend([
|
||
LocalShellBackend(workspace_dir="./workspace"),
|
||
FilesystemBackend(),
|
||
])
|
||
|
||
# ----------------- Agent -----------------
|
||
agent = create_deep_agent(
|
||
model=llm,
|
||
tools=[search_course_docs, fetch_course_meta],
|
||
backend=backend,
|
||
system_prompt=(
|
||
"You are a helpful FAQ assistant for the course.\n"
|
||
"When answering a question, first decide whether the answer comes from the course materials (use search_course_docs)\n"
|
||
"or from course metadata (use fetch_course_meta).\n"
|
||
"Do not call both tools unless absolutely necessary.\n"
|
||
"In your final answer, prepend 'source: chroma' or 'source: mcp_meta' to indicate the origin."
|
||
),
|
||
)
|
||
|
||
# ----------------- CLI -----------------
|
||
PRESET_QUESTIONS = [
|
||
"What topics are covered in the first lecture?", # should hit chroma
|
||
"Who is the instructor for this course?", # should hit mcp_meta
|
||
"When is the next class?", # should hit chroma (or meta if schedule)
|
||
]
|
||
|
||
async def run_question(question: str):
|
||
result = await agent.ainvoke(
|
||
{"messages": ["HumanMessage(content=\"{}\")".format(question)]},
|
||
{"configurable": {"thread_id": "session-1"}},
|
||
)
|
||
# The agent returns a dict with 'messages'; extract last content
|
||
content = result["messages"][-1].content
|
||
print(f"\nQ: {question}\nA: {content}\n")
|
||
|
||
async def interactive():
|
||
print("Enter a question (or 'exit' to quit):")
|
||
while True:
|
||
q = input("> ")
|
||
if q.lower() in {"exit", "quit"}:
|
||
break
|
||
await run_question(q)
|
||
|
||
async def main():
|
||
print("Running preset questions...")
|
||
for q in PRESET_QUESTIONS:
|
||
await run_question(q)
|
||
await interactive()
|
||
|
||
if __name__ == "__main__":
|
||
asyncio.run(main())
|