import { describe, expect, it } from 'vitest';
import type { RegulatoryControlDocument } from '../../src/domain/regulatory-rag';
import { chunkRegulatoryControl, regulatoryChunkEmbeddingText, validateRegulatoryControlDocument } from '../../server/regulatory/regulatory-chunker';

const document: RegulatoryControlDocument = {
  schemaVersion: 1,
  corpusId: 'iso-27001',
  corpusVersion: '2022-de-1',
  documentId: 'iso-27001-2022-a.5.1-de',
  framework: 'ISO/IEC 27001',
  frameworkVersion: '2022',
  controlId: 'A.5.1',
  language: 'de',
  title: 'Informationssicherheitsrichtlinien',
  text: `${'Erster kontrollierter Absatz. '.repeat(8)}\r\n\r\n${'Second controlled paragraph. '.repeat(8)}`,
  sourceUri: 'https://example.invalid/iso-27001/a-5-1'
};

describe('regulatory chunker', () => {
  it('creates deterministic bounded chunks with complete metadata', () => {
    const first = chunkRegulatoryControl(document, { maximumCharacters: 220 });
    const second = chunkRegulatoryControl(document, { maximumCharacters: 220 });

    expect(first).toEqual(second);
    expect(first.length).toBeGreaterThan(1);
    expect(first.every(chunk => chunk.text.length <= 220)).toBe(true);
    expect(first.map(chunk => chunk.chunkIndex)).toEqual(first.map((_, index) => index));
    expect(first.every(chunk => chunk.framework === 'ISO/IEC 27001' && chunk.language === 'de')).toBe(true);
    expect(first.every(chunk => /^[a-f0-9]{64}$/.test(chunk.textSha256))).toBe(true);
    expect(regulatoryChunkEmbeddingText(first[0])).toContain('ISO/IEC 27001 2022 | A.5.1');
  });

  it('validates the version, required values and language before indexing', () => {
    expect(() => validateRegulatoryControlDocument({ ...document, schemaVersion: 2 })).toThrow(/schemaVersion 1/);
    expect(() => validateRegulatoryControlDocument({ ...document, controlId: ' ' })).toThrow(/controlId/);
    expect(() => validateRegulatoryControlDocument({ ...document, language: 'not_a_tag' })).toThrow(/BCP 47/);
  });
});