import yfinance as yf
import pandas as pd
from fastapi import FastAPI
import uvicorn

app = FastAPI(title="Vertex Coders Risk API")

class VertexRiskEngine:
    def __init__(self, ticker):
        self.ticker_str = ticker
        self.company = yf.Ticker(ticker)
        self.scores = {}

    def get_value(self, df, possible_keys):
        """Busca un valor en el DataFrame intentando varias llaves posibles."""
        for key in possible_keys:
            if key in df.index:
                return df.loc[key]
        raise ValueError(f"Ninguna de las llaves {possible_keys} encontrada.")

    def run_audit(self):
        try:
            balance = self.company.balance_sheet
            financials = self.company.financials
            
            # --- ALTMAN Z-SCORE ---
            # Usamos iloc[0] para el año más reciente
            working_capital = self.get_value(balance, ['Working Capital']).iloc[0]
            total_assets = self.get_value(balance, ['Total Assets']).iloc[0]
            retained_earnings = self.get_value(balance, ['Retained Earnings']).iloc[0]
            ebit = self.get_value(financials, ['EBIT']).iloc[0]
            total_liabilities = self.get_value(balance, ['Total Liabilities Net Minority Interest', 'Total Liabilities']).iloc[0]
            sales = self.get_value(financials, ['Total Revenue']).iloc[0]
            market_cap = self.company.info.get('marketCap', 1)

            z = (1.2 * (working_capital/total_assets) + 
                 1.4 * (retained_earnings/total_assets) + 
                 3.3 * (ebit/total_assets) + 
                 0.6 * (market_cap/total_liabilities) + 
                 1.0 * (sales/total_assets))
            
            self.scores['altman_z'] = round(z, 2)

            # --- BENEISH M-SCORE (DSRI) ---
            # Comparamos Año T (iloc[0]) vs Año T-1 (iloc[1])
            sales_t = sales
            sales_t1 = self.get_value(financials, ['Total Revenue']).iloc[1]
            
            receivables_keys = ['Net Receivables', 'Accounts Receivable', 'Receivables']
            rec_t = self.get_value(balance, receivables_keys).iloc[0]
            rec_t1 = self.get_value(balance, receivables_keys).iloc[1]
            
            dsri = (rec_t / sales_t) / (rec_t1 / sales_t1)
            self.scores['m_score_dsri'] = round(dsri, 2)

            return self.scores
        except Exception as e:
            return {"error": str(e)}

@app.get("/audit/{ticker}")
def audit_company(ticker: str):
    engine = VertexRiskEngine(ticker.upper())
    result = engine.run_audit()
    
    # 1. Validar si el motor devolvió un error de ejecución
    if "error" in result:
        return {
            "ticker": ticker.upper(),
            "status": "ERROR_TECNICO",
            "analysis": None,
            "msg": f"No se pudo completar la auditoría: {result['error']}"
        }

    # 2. Extraer valores con seguridad (Type Casting)
    # Forzamos a float para evitar errores de comparación con tipos desconocidos
    try:
        z_score = float(result.get('altman_z', 0))
        m_dsri = float(result.get('m_score_dsri', 0))
    except (ValueError, TypeError):
        z_score = 0.0
        m_dsri = 0.0
    
    # 3. Lógica de Semáforo Robusta
    # Prioridad: Rojo (Peligro inminente o posible fraude)
    if z_score < 1.8 or m_dsri > 1.4:
        status = "ROJO"
        msg = "Riesgo crítico detectado: Posible quiebra o manipulación."
    # Zona Gris
    elif 1.8 <= z_score < 3.0:
        status = "AMARILLO"
        msg = "Empresa en zona gris. Requiere supervisión manual."
    # Zona Segura
    else:
        status = "VERDE"
        msg = "Fundamentos financieros sólidos según modelos automáticos."

    return {
        "ticker": ticker.upper(),
        "status": status,
        "analysis": result,
        "msg": msg,
        "engine_version": "1.0-Némesis"
    }
if __name__ == "__main__":
    uvicorn.run(app, host="0.0.0.0", port=8010)