# api/main.py
import sentence_transformers
import pandas as pd
import os

from fastapi import FastAPI, HTTPException
from huggingface_hub import hf_hub_download, login
from src.processor import send_to_dataset,search_and_retrieve,generate_tech
from typing import List, Dict
from pydantic import BaseModel
from datasets import load_dataset
from dotenv import load_dotenv

load_dotenv()


login(token=os.getenv("HF_TOKEN"))

# This is the main application object that Uvicorn will run
app = FastAPI(
    title="My Standalone API",
    description="An API hosted on Hugging Face Spaces",
    version="1.0.0"
)


model = sentence_transformers.SentenceTransformer('sentence-transformers/all-MiniLM-L6-v2')
dataset = load_dataset("OrganizedProgrammers/Technologies", split="train")
dataset.add_faiss_index(column='embeddings')

class SearchInput(BaseModel):
    title: str

class SearchOutput(BaseModel):
    title: str
    purpose: str
    score: float
    top5: List[Dict]

class GenerateInput(BaseModel):
    title: str
    instructions: str
    force: bool = False

class GenerateOutput(BaseModel):
    name: str
    purpose: str
    problem_types_solved: str
    advantages: str
    limitations: str
    domain_tags: str

@app.post("/search-technologies", response_model=SearchOutput)
def post_search(payload: SearchInput):
    """
    Endpoint that returns a search result.
    """
    config = {"dataset": dataset, "model": model}
    res = search_and_retrieve(payload.title, config)
    return res

@app.post("/generate-technology", response_model=GenerateOutput)
def post_generate_and_push(payload: GenerateInput):
    """
    Endpoint to generate a technology and push it to the dataset
    """

    config = {"dataset": dataset, "model": model}
    res = search_and_retrieve(payload.title, config)
    if res["score"] >= 0.7 and not payload.force:
        raise HTTPException(status_code=500, detail=f"Cannot generate the technology a high score of {res['score']} have been found for the technology : {res['title']}")

    json_response = generate_tech(payload.title, payload.instructions)

    send_to_dataset(json_response, model)

    return json_response