/**
 * earth-now.co
 * Author:  Simon-Pierre Boucher
 * Contact: contact@spboucher.ai
 * File:    packages/models/test/filters.test.ts
 * Purpose: Holt-Winters and Kalman tests — nowcasting recovery and observation/forecast blending behavior
 */

import { describe, expect, it } from "vitest";
import { blendWithForecast, fitHoltWinters, kalmanFilter } from "../src/index";

describe("fitHoltWinters", () => {
  it("recovers a clean level+trend+season series and forecasts it", () => {
    // 4 seasons of monthly data: level 100, trend +1/step, season = 10·sin.
    const m = 12;
    const series: number[] = [];
    for (let i = 0; i < 4 * m; i++) {
      series.push(100 + i + 10 * Math.sin((2 * Math.PI * (i % m)) / m));
    }
    const fit = fitHoltWinters(series, { alpha: 0.5, beta: 0.1, gamma: 0.3, seasonLength: m });
    const truthNext = (h: number) =>
      100 + (series.length + h - 1) + 10 * Math.sin((2 * Math.PI * ((series.length + h - 1) % m)) / m);
    // Exponential smoothing carries a small lag on trending series — accept < 1.5
    // on a signal of amplitude 10 with level ~150.
    for (const h of [1, 3, 6]) {
      expect(Math.abs(fit.forecast(h) - truthNext(h))).toBeLessThan(1.5);
    }
  });

  it("rejects too-short series and bad params", () => {
    expect(() => fitHoltWinters([1, 2, 3], { alpha: 0.5, beta: 0.1, gamma: 0.1, seasonLength: 12 })).toThrow();
    expect(() =>
      fitHoltWinters(new Array(30).fill(1), { alpha: 1.5, beta: 0.1, gamma: 0.1, seasonLength: 12 }),
    ).toThrow();
  });
});

describe("kalmanFilter + blendWithForecast", () => {
  it("converges to a constant signal", () => {
    const noisy = [10.4, 9.7, 10.1, 10.3, 9.9, 10.0, 10.2, 9.8, 10.0, 10.1];
    const steps = kalmanFilter(noisy, {
      processVariance: 0.001,
      measurementVariance: 0.25,
      initialValue: 0,
      initialVariance: 100,
    });
    const last = steps[steps.length - 1]!;
    expect(last.value).toBeCloseTo(10, 0);
    expect(last.variance).toBeLessThan(0.1);
  });

  it("blend tracks the observation near it and the forecast far out", () => {
    const lastState = { value: 100, variance: 1 };
    const forecast = { value: 110, variance: 4 };
    const near = blendWithForecast(lastState, 1, 0.01, forecast);
    const far = blendWithForecast(lastState, 10_000, 0.01, forecast);
    // Near: mostly the observation; far: mostly the forecast.
    expect(near.value).toBeLessThan(103);
    expect(far.value).toBeGreaterThan(106);
    expect(near.variance).toBeLessThan(far.variance);
  });
});