main.py

from fastapi import FastAPI, Request, Form
from fastapi.responses import HTMLResponse
from fastapi.staticfiles import StaticFiles
from fastapi.templating import Jinja2Templates
from langchain_community.vectorstores import FAISS
from langchain_huggingface import HuggingFaceEmbeddings
from langchain_core.prompts import PromptTemplate
from langchain_core.runnables import RunnablePassthrough, RunnableLambda
import os
import requests
from dotenv import load_dotenv

load_dotenv()

app = FastAPI()

# Mount static files
app.mount("/static", StaticFiles(directory="static"), name="static")
templates = Jinja2Templates(directory="templates")

# Load FAISS Index
FAISS_PATH = "faiss_index"
embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")

if os.path.exists(FAISS_PATH):
    vector_db = FAISS.load_local(FAISS_PATH, embeddings, allow_dangerous_deserialization=True)
    retriever = vector_db.as_retriever(search_kwargs={"k": 3})
    print("DEBUG: FAISS index loaded.")
else:
    print("WARNING: FAISS index not found. Run ingest.py first.")
    retriever = None

# Custom DeepSeek V3.2 Wrapper
def call_deepseek_v3(prompt_input):
    if hasattr(prompt_input, "to_string"):
        prompt_text = prompt_input.to_string()
    else:
        prompt_text = str(prompt_input)

    api_url = "https://router.huggingface.co/v1/chat/completions"
    token = os.getenv("HUGGINGFACEHUB_API_TOKEN")
    
    if not token:
        return "Error: No API Token found."

    headers = {
        "Authorization": f"Bearer {token}",
        "Content-Type": "application/json",
    }
    
    payload = {
        "model": "deepseek-ai/DeepSeek-V3.2",
        "messages": [
            {
                "role": "system",
                "content": "You are a Stranger Things expert. Answer clearly."
            },
            {
                "role": "user",
                "content": prompt_text
            }
        ],
        "temperature": 0.3,
        "max_tokens": 700,
        "stream": False 
    }

    try:
        response = requests.post(api_url, headers=headers, json=payload)
        if response.status_code != 200:
             return f"DeepSeek Error ({response.status_code}): {response.text}"
        return response.json()["choices"][0]["message"]["content"]
    except Exception as e:
        return f"Connection Error: {e}"

llm = RunnableLambda(call_deepseek_v3)

# Updated Prompt Template (User Request)
template = """
You are a Stranger Things expert assistant. Answer the user's question using ONLY the provided context.

Important rules:
- Do NOT copy sentences directly from the context. Rewrite in your own words.
- If the context does NOT contain the answer, say: "I don’t have enough information in the provided context to answer that fully."
- Keep the tone natural, friendly, and engaging.
- Write at least 4–6 sentences unless the question is very simple.

If the question is about a CHARACTER, you MUST include:
1) Full name + who they are in the story
2) Key relationships (friends, family, love interest, major connections)
3) Role / occupation / passion (student, sheriff, journalist, Dungeon Master, etc.)
4) Iconic traits (personality, behavior, famous moments or skills)

If the question is about an EVENT / LOCATION / OBJECT, you MUST include:
1) What it is
2) Why it matters in the story
3) Who is involved
4) Any major consequences or impact

Context:
{context}

User Question:
{question}

Answer (detailed and structured):
"""

prompt = PromptTemplate.from_template(template)

def format_docs(docs):
    return "\n\n".join(doc.page_content for doc in docs)

if retriever:
    rag_chain = (
        {"context": retriever | format_docs, "question": RunnablePassthrough()}
        | prompt
        | llm
    )
else:
    rag_chain = None

@app.get("/", response_class=HTMLResponse)
async def read_root(request: Request):
    return templates.TemplateResponse("index.html", {"request": request})

from pydantic import BaseModel

class QueryRequest(BaseModel):
    query: str

@app.post("/query")
async def query_endpoint(request: QueryRequest):
    if not rag_chain:
        return {"answer": "System Error: RAG chain not initialized."}
    
    answer = rag_chain.invoke(request.query)
    return {"answer": answer}

@app.post("/get_response")
async def get_response(request: Request, query: str = Form(...)):
    if not rag_chain:
        return templates.TemplateResponse("index.html", {"request": request, "response": "System Error: RAG chain not initialized."})
    
    result = rag_chain.invoke(query)
    return templates.TemplateResponse("index.html", {"request": request, "response": result})

if __name__ == "__main__":
    import uvicorn
    uvicorn.run(app, host="0.0.0.0", port=8001)