const test = require('node:test');
const assert = require('node:assert/strict');

const { makeMeasurementInstance } = require('../helpers/factories');

/**
 * Unit coverage for the three outlier detection strategies shipped by the
 * measurement node. Each test seeds the storedValues window first, then
 * probes the classifier directly. This keeps the assertions focused on the
 * detection logic rather than the full calculateInput pipeline.
 */

function makeDetector(method, threshold) {
  return makeMeasurementInstance({
    scaling: { enabled: false, inputMin: 0, inputMax: 1, absMin: -1000, absMax: 1000, offset: 0 },
    smoothing: { smoothWindow: 20, smoothMethod: 'none' },
    outlierDetection: { enabled: true, method, threshold },
  });
}

function seed(m, values) {
  // bypass calculateInput so we don't trigger outlier filtering while seeding
  m.storedValues = [...values];
}

test("zScore flags a value far above the mean as an outlier", () => {
  const m = makeDetector('zScore', 3);
  seed(m, [10, 11, 10, 9, 10, 11, 10, 11, 9, 10]);
  assert.equal(m.outlierDetection(100), true);
});

test("zScore does not flag a value inside the distribution", () => {
  const m = makeDetector('zScore', 3);
  seed(m, [10, 11, 10, 9, 10, 11, 10, 11, 9, 10]);
  assert.equal(m.outlierDetection(11), false);
});

test("iqr flags a value outside Q1/Q3 fences", () => {
  const m = makeDetector('iqr');
  seed(m, [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]);
  assert.equal(m.outlierDetection(100), true);
});

test("iqr does not flag a value inside Q1/Q3 fences", () => {
  const m = makeDetector('iqr');
  seed(m, [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]);
  assert.equal(m.outlierDetection(5), false);
});

test("modifiedZScore flags heavy-tailed outliers", () => {
  const m = makeDetector('modifiedZScore', 3.5);
  seed(m, [10, 11, 10, 9, 10, 11, 10, 11, 9, 10]);
  assert.equal(m.outlierDetection(1000), true);
});

test("modifiedZScore accepts normal data", () => {
  const m = makeDetector('modifiedZScore', 3.5);
  seed(m, [10, 11, 10, 9, 10, 11, 10, 11, 9, 10]);
  assert.equal(m.outlierDetection(11), false);
});

test("unknown outlier method falls back to schema default (zScore) and still runs", () => {
  // validateEnum replaces unknown values with the schema default. The
  // schema default is "zScore"; the dispatcher normalizes to lowercase
  // and routes to zScoreOutlierDetection. With a tight window, value=100
  // is a clear outlier -> returns true.
  const m = makeDetector('bogus', 3);
  seed(m, [1, 2, 3, 4, 5]);
  assert.equal(m.outlierDetection(100), true);
});

test("outlier detection returns false when window has < 2 samples", () => {
  const m = makeDetector('zScore', 3);
  m.storedValues = [];
  assert.equal(m.outlierDetection(500), false);
});

test("calculateInput ignores a value flagged as outlier", () => {
  const m = makeDetector('zScore', 3);
  // Build a tight baseline then throw a spike at it.
  [10, 10, 10, 10, 10].forEach((v) => m.calculateInput(v));
  const before = m.outputAbs;
  m.calculateInput(9999);
  // Output must not move to the spike (outlier rejected).
  assert.equal(m.outputAbs, before);
});

test("toggleOutlierDetection flips the flag without corrupting config", () => {
  const m = makeDetector('zScore', 3);
  const initial = m.config.outlierDetection.enabled;
  m.toggleOutlierDetection();
  assert.equal(m.config.outlierDetection.enabled, !initial);
  // Re-toggle restores
  m.toggleOutlierDetection();
  assert.equal(m.config.outlierDetection.enabled, initial);
  // Method is preserved (enum values are normalized to lowercase by validateEnum).
  assert.equal(m.config.outlierDetection.method.toLowerCase(), 'zscore');
});