import { verbatimOverlap, normWords } from '../overlap';

const SOURCE = `Saudi Arabia will apply excise tax to sweetened drinks from January 2026,
Gulf News reported, adding another category to a tax architecture that already
captures selected products across the Gulf region. For businesses outside the region,
the broader lesson is that excise tax in the GCC should be viewed as a shared framework
with local variations that change launch economics.`;

describe('normWords', () => {
  it('lowercases, strips punctuation, collapses whitespace', () => {
    expect(normWords('Hello,   WORLD! (2026)')).toEqual(['hello', 'world', '2026']);
    expect(normWords('')).toEqual([]);
    expect(normWords(null as unknown as string)).toEqual([]);
  });
});

describe('verbatimOverlap', () => {
  it('flags a long verbatim copy (whole sentence lifted from the source)', () => {
    const copied = 'Saudi Arabia will apply excise tax to sweetened drinks from January 2026, Gulf News reported.';
    const r = verbatimOverlap(copied, [SOURCE]);
    expect(r.longestRunWords).toBeGreaterThanOrEqual(12);
    expect(r.overlapRatio).toBeGreaterThan(0.5);
    expect(r.sample).toContain('excise tax to sweetened drinks');
  });

  it('passes a genuine paraphrase of the same facts', () => {
    const paraphrase =
      'From 2026 the Kingdom broadens its excise regime to cover sugary beverages, a reminder that Gulf tax rules differ market by market.';
    const r = verbatimOverlap(paraphrase, [SOURCE]);
    expect(r.longestRunWords).toBeLessThan(6);
    expect(r.overlapRatio).toBeLessThan(0.15);
  });

  it('ignores a short shared phrase (< minRun words)', () => {
    // "excise tax" is shared but only 2 words — below the 6-word shingle.
    const r = verbatimOverlap('Companies weighing excise tax exposure should plan ahead.', [SOURCE]);
    expect(r.longestRunWords).toBe(0);
  });

  it('is case- and punctuation-insensitive', () => {
    const copied = 'SHOULD BE VIEWED AS A SHARED FRAMEWORK — with local variations!!!';
    const r = verbatimOverlap(copied, [SOURCE]);
    expect(r.longestRunWords).toBeGreaterThanOrEqual(6);
    expect(r.sample).toContain('shared framework');
  });

  it('returns zeros for empty candidate or no sources', () => {
    expect(verbatimOverlap('', [SOURCE])).toEqual({ longestRunWords: 0, overlapRatio: 0, sample: '' });
    expect(verbatimOverlap('anything here', [])).toEqual({ longestRunWords: 0, overlapRatio: 0, sample: '' });
    expect(verbatimOverlap('anything here', [''])).toEqual({ longestRunWords: 0, overlapRatio: 0, sample: '' });
  });

  it('finds the copied span across multiple sources', () => {
    const copied = 'a tax architecture that already captures selected products';
    const r = verbatimOverlap(copied, ['unrelated text about camels', SOURCE]);
    expect(r.longestRunWords).toBeGreaterThanOrEqual(7);
  });
});