from fastapi import FastAPI
from pydantic import BaseModel
import pickle
import numpy as np
from fastapi.middleware.cors import CORSMiddleware
from fastapi.responses import RedirectResponse

# Cargar el modelo desde el archivo .pkl
with open("miarbolcancer.pkl", "rb") as f:
    model = pickle.load(f)

# Definir el modelo de datos con Pydantic (sin ca_cervix como entrada)
class PredictionInput(BaseModel):
    behavior_sexualRisk: float
    behavior_eating: float
    behavior_personalHygine: float
    intention_aggregation: float
    intention_commitment: float
    attitude_consistency: float
    attitude_spontaneity: float
    norm_significantPerson: float
    norm_fulfillment: float
    perception_vulnerability: float
    perception_severity: float
    motivation_strength: float
    motivation_willingness: float
    socialSupport_emotionality: float
    socialSupport_appreciation: float
    socialSupport_instrumental: float
    empowerment_knowledge: float
    empowerment_abilities: float
    empowerment_desires: float

# Crear la aplicación FastAPI
app = FastAPI()
# CORS
app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

# Redirigir de "/" a "/docs"
@app.get("/")
def redirect_to_docs():
    return RedirectResponse(url="/docs")
    
# Definir el endpoint de predicción
@app.post("/predict/")
def predict(input_data: PredictionInput):
    # Convertir los datos de entrada en un array numpy
    input_array = np.array([[input_data.behavior_sexualRisk, input_data.behavior_eating, input_data.behavior_personalHygine,
                             input_data.intention_aggregation, input_data.intention_commitment, input_data.attitude_consistency,
                             input_data.attitude_spontaneity, input_data.norm_significantPerson, input_data.norm_fulfillment,
                             input_data.perception_vulnerability, input_data.perception_severity, input_data.motivation_strength,
                             input_data.motivation_willingness, input_data.socialSupport_emotionality, input_data.socialSupport_appreciation,
                             input_data.socialSupport_instrumental, input_data.empowerment_knowledge, input_data.empowerment_abilities,
                             input_data.empowerment_desires]])
    
    # Realizar la predicción (el modelo debe predecir ca_cervix)
    prediction = model.predict(input_array)
    
    # Convertir la predicción a tipo nativo Python (int o float)
    prediction_value = prediction[0] if isinstance(prediction[0], (int, float)) else prediction[0].item()

    # Retornar la predicción (ca_cervix)
    return {"ca_cervix_prediction": prediction_value}