app.py

import os
import json
import faiss
import torch
import numpy as np

from fastapi import FastAPI
from contextlib import asynccontextmanager
from pydantic import BaseModel
from huggingface_hub import snapshot_download
from sentence_transformers import SentenceTransformer
from transformers import AutoTokenizer, AutoModelForCausalLM

# ─────────────────────────────
# CONFIG
# ─────────────────────────────
MODEL_REPO = "Qwen/Qwen2.5-0.5B-Instruct"
RAG_REPO = "Rady10/Agriculture-Rag-Data-Index"
EMBED_MODEL = "sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2"

DEVICE = "cpu"
MAX_TOKENS = 256

os.environ["TOKENIZERS_PARALLELISM"] = "false"

# ─────────────────────────────
# GLOBALS
# ─────────────────────────────
tokenizer = None
model = None
embedder = None
faiss_index = None
rag_chunks = None

# ─────────────────────────────
# SYSTEM PROMPT
# ─────────────────────────────
SYSTEM_PROMPT = """
You are an agriculture assistant.
Answer clearly and concisely in English or Arabic.
Focus on plant diseases, pests, irrigation, and farming advice.
"""

# ─────────────────────────────
# FASTAPI LIFESPAN (IMPORTANT)
# ─────────────────────────────
@asynccontextmanager
async def lifespan(app: FastAPI):

    global tokenizer, model, embedder, faiss_index, rag_chunks

    print("Loading RAG...")

    rag_dir = snapshot_download(
        repo_id=RAG_REPO,
        repo_type="dataset",
        local_dir="./rag"
    )

    faiss_index = faiss.read_index(
        os.path.join(rag_dir, "agro.index")
    )

    with open(os.path.join(rag_dir, "chunks.json"), "r", encoding="utf-8") as f:
        rag_chunks = json.load(f)

    print("Loading embedder...")
    embedder = SentenceTransformer(EMBED_MODEL)

    print("Loading tokenizer...")
    tokenizer = AutoTokenizer.from_pretrained(
        MODEL_REPO,
        trust_remote_code=True
    )

    print("Loading model...")
    model = AutoModelForCausalLM.from_pretrained(
        MODEL_REPO,
        device_map="cpu",
        torch_dtype=torch.float32,
        trust_remote_code=True
    )

    model.eval()

    print("ALL LOADED")

    yield

app = FastAPI(lifespan=lifespan)

# ─────────────────────────────
# REQUEST MODEL
# ─────────────────────────────
class ChatRequest(BaseModel):
    message: str

# ─────────────────────────────
# RAG
# ─────────────────────────────
def retrieve(query, k=3):

    if not query:
        return ""

    emb = embedder.encode([query], normalize_embeddings=True).astype(np.float32)

    scores, idxs = faiss_index.search(emb, k)

    results = []

    for score, idx in zip(scores[0], idxs[0]):
        if idx != -1 and score > 0.3:
            results.append(rag_chunks[idx]["text"])

    return "\n\n".join(results)

# ─────────────────────────────
# GENERATION
# ─────────────────────────────
def generate(text):

    context = retrieve(text)

    prompt = SYSTEM_PROMPT
    if context:
        prompt += "\n\nKnowledge:\n" + context

    messages = [
        {"role": "system", "content": prompt},
        {"role": "user", "content": text}
    ]

    input_text = tokenizer.apply_chat_template(
        messages,
        tokenize=False,
        add_generation_prompt=True
    )

    inputs = tokenizer(input_text, return_tensors="pt").to(model.device)

    with torch.no_grad():
        output = model.generate(
            **inputs,
            max_new_tokens=MAX_TOKENS,
            temperature=0.7,
            top_p=0.9
        )

    return tokenizer.decode(
        output[0][inputs.input_ids.shape[1]:],
        skip_special_tokens=True
    )

# ─────────────────────────────
# API ROUTES
# ─────────────────────────────
@app.get("/")
def home():
    return {"status": "running"}

@app.post("/chat")
def chat(req: ChatRequest):

    response = generate(req.message)

    return {"response": response}