market_screener/tests/news-pipeline.test.ts

import test from 'node:test';
import assert from 'node:assert/strict';
import { NewsPipeline } from '../server/domains/news/NewsPipeline.js';
import type { NewsRepository } from '../server/domains/news/NewsRepository.js';
import type { NormalizedStory } from '../server/domains/shared/types/index.js';

/** In-memory stub that records what the pipeline stores. */
class StubRepo {
  articles: Array<{ urlHash: string; tickers: string[]; catalyst: string | null }> = [];
  links: Array<{ ticker: string; day: string }> = [];
  seenTitles = new Set<string>();
  capCounts = new Map<string, number>(); // `${ticker}|${day}` → count

  insertArticle(a: { urlHash: string; tickers: string[]; catalyst: string | null }): boolean {
    if (this.articles.some((x) => x.urlHash === a.urlHash)) return false;
    this.articles.push(a);
    return true;
  }
  titleSeenSince(titleHash: string): boolean {
    return this.seenTitles.has(titleHash);
  }
  linkTicker(ticker: string, day: string): void {
    this.links.push({ ticker, day });
  }
  countTickerDay(ticker: string, day: string): number {
    return this.capCounts.get(`${ticker}|${day}`) ?? 0;
  }
  purgeBodiesBefore(): number {
    return 0;
  }
  deleteUnreferencedBefore(): number {
    return 0;
  }
}

const UNIVERSE = new Set(['AAPL', 'MSFT']);

function story(overrides: Partial<NormalizedStory> = {}): NormalizedStory {
  return {
    tickers: ['AAPL'],
    headline: 'Apple announces quarterly results beat estimates',
    source: 'prwire',
    url: `https://example.com/${Math.random()}`,
    publishedAt: '2026-06-09T14:00:00.000Z',
    ...overrides,
  };
}

function makePipeline(repo: StubRepo): NewsPipeline {
  return new NewsPipeline(repo as unknown as NewsRepository);
}

test('NewsPipeline', async (t) => {
  await t.test('stores universe stories and links tickers', () => {
    const repo = new StubRepo();
    const stats = makePipeline(repo).ingest([story()], UNIVERSE);
    assert.equal(stats.stored, 1);
    assert.equal(repo.links.length, 1);
    assert.equal(repo.links[0].ticker, 'AAPL');
    assert.equal(repo.links[0].day, '2026-06-09');
  });

  await t.test('drops stories with no universe ticker (§4.1)', () => {
    const repo = new StubRepo();
    const stats = makePipeline(repo).ingest([story({ tickers: ['ZZZZ'] })], UNIVERSE);
    assert.equal(stats.stored, 0);
    assert.equal(stats.droppedNoUniverseTicker, 1);
    assert.equal(repo.articles.length, 0);
  });

  await t.test('drops noise headlines, but never filings (§4.2)', () => {
    const repo = new StubRepo();
    const noise = story({ headline: '5 best stocks to buy now including Apple' });
    const filing = story({
      headline: '8-K filing: 5 best stocks edge case',
      source: 'edgar',
      catalystHint: 'regulatory',
    });
    const stats = makePipeline(repo).ingest([noise, filing], UNIVERSE);
    assert.equal(stats.droppedNoise, 1);
    assert.equal(stats.stored, 1);
    assert.equal(repo.articles[0].catalyst, 'regulatory');
  });

  await t.test('drops syndicated duplicates by normalized title (§4.3)', () => {
    const repo = new StubRepo();
    const pipeline = makePipeline(repo);
    // First copy stored; mark its normalized-title hash as seen
    pipeline.ingest([story({ headline: 'Apple Beats Q2 Estimates!' })], UNIVERSE);
    repo.seenTitles.add(sha256(NewsPipeline.normalizeTitle('Apple Beats Q2 Estimates!')));
    // Same story, different casing/punctuation/URL → syndicated copy
    const stats = pipeline.ingest(
      [story({ headline: 'APPLE BEATS Q2 ESTIMATES', url: 'https://other.com/copy' })],
      UNIVERSE,
    );
    assert.equal(stats.droppedDuplicate, 1);
  });

  await t.test('enforces per-ticker daily cap, filings exempt (§4.4)', () => {
    const repo = new StubRepo();
    repo.capCounts.set('AAPL|2026-06-09', 25); // at cap
    const wire = story();
    const filing = story({ source: 'edgar', catalystHint: 'ma', url: 'https://sec.gov/x' });
    const stats = makePipeline(repo).ingest([wire, filing], UNIVERSE);
    assert.equal(stats.droppedCapped, 1);
    assert.equal(stats.stored, 1); // the filing
  });

  await t.test('classifies catalysts with M&A taking priority', () => {
    assert.equal(NewsPipeline.classify('Acme to be acquired by MegaCorp in Q2 deal'), 'ma');
    assert.equal(NewsPipeline.classify('Acme reports record quarterly results'), 'earnings');
    assert.equal(NewsPipeline.classify('Acme raises full-year guidance'), 'guidance');
    assert.equal(NewsPipeline.classify('FDA approval granted for Acme drug'), 'regulatory');
    assert.equal(NewsPipeline.classify('Fed holds rates steady amid CPI data'), 'macro');
    assert.equal(NewsPipeline.classify('Acme appoints new CMO'), null);
  });

  await t.test('noise detector catches listicles and target reiterations', () => {
    assert.ok(NewsPipeline.isNoise('3 Top Stocks to Watch This Week'));
    assert.ok(NewsPipeline.isNoise('Analyst price target raised on momentum'));
    assert.ok(!NewsPipeline.isNoise('Apple announces $90B buyback'));
  });
});

// Helper mirroring NewsPipeline's title hashing for the dedupe test
import { createHash } from 'crypto';
function sha256(input: string): string {
  return createHash('sha256').update(input).digest('hex');
}