breakpilot-compliance/admin-compliance/lib/sdk/drafting-engine/__tests__/intent-classifier.test.ts

import { IntentClassifier } from '../intent-classifier'

describe('IntentClassifier', () => {
  const classifier = new IntentClassifier()

  describe('classify - Draft mode', () => {
    it.each([
      ['Erstelle ein VVT fuer unseren Hauptprozess', 'draft'],
      ['Generiere eine TOM-Dokumentation', 'draft'],
      ['Schreibe eine Datenschutzerklaerung', 'draft'],
      ['Verfasse einen Entwurf fuer das Loeschkonzept', 'draft'],
      ['Create a DSFA document', 'draft'],
      ['Draft a privacy policy for us', 'draft'],
      ['Neues VVT anlegen', 'draft'],
    ])('"%s" should classify as %s', (input, expectedMode) => {
      const result = classifier.classify(input)
      expect(result.mode).toBe(expectedMode)
      expect(result.confidence).toBeGreaterThan(0.7)
    })
  })

  describe('classify - Validate mode', () => {
    it.each([
      ['Pruefe die Konsistenz meiner Dokumente', 'validate'],
      ['Ist mein VVT korrekt?', 'validate'],
      ['Validiere die TOM gegen das VVT', 'validate'],
      ['Check die Vollstaendigkeit', 'validate'],
      ['Stimmt das mit der DSFA ueberein?', 'validate'],
      ['Cross-Check VVT und TOM', 'validate'],
    ])('"%s" should classify as %s', (input, expectedMode) => {
      const result = classifier.classify(input)
      expect(result.mode).toBe(expectedMode)
      expect(result.confidence).toBeGreaterThan(0.7)
    })
  })

  describe('classify - Ask mode', () => {
    it.each([
      ['Was fehlt noch in meinem Profil?', 'ask'],
      ['Zeige mir die Luecken', 'ask'],
      ['Welche Dokumente fehlen noch?', 'ask'],
      ['Was ist der naechste Schritt?', 'ask'],
      ['Welche Informationen brauche ich noch?', 'ask'],
    ])('"%s" should classify as %s', (input, expectedMode) => {
      const result = classifier.classify(input)
      expect(result.mode).toBe(expectedMode)
      expect(result.confidence).toBeGreaterThan(0.6)
    })
  })

  describe('classify - Explain mode (fallback)', () => {
    it.each([
      ['Was ist DSGVO?', 'explain'],
      ['Erklaere mir Art. 30', 'explain'],
      ['Hallo', 'explain'],
      ['Danke fuer die Hilfe', 'explain'],
    ])('"%s" should classify as %s (fallback)', (input, expectedMode) => {
      const result = classifier.classify(input)
      expect(result.mode).toBe(expectedMode)
    })
  })

  describe('classify - confidence thresholds', () => {
    it('should have high confidence for clear draft intents', () => {
      const result = classifier.classify('Erstelle ein neues VVT')
      expect(result.confidence).toBeGreaterThanOrEqual(0.85)
    })

    it('should have lower confidence for ambiguous inputs', () => {
      const result = classifier.classify('Hallo')
      expect(result.confidence).toBeLessThan(0.6)
    })

    it('should boost confidence with document type detection', () => {
      const withDoc = classifier.classify('Erstelle VVT')
      const withoutDoc = classifier.classify('Erstelle etwas')
      expect(withDoc.confidence).toBeGreaterThanOrEqual(withoutDoc.confidence)
    })

    it('should boost confidence with multiple pattern matches', () => {
      const single = classifier.classify('Erstelle Dokument')
      const multi = classifier.classify('Erstelle und generiere ein neues Dokument')
      expect(multi.confidence).toBeGreaterThanOrEqual(single.confidence)
    })
  })

  describe('detectDocumentType', () => {
    it.each([
      ['VVT erstellen', 'vvt'],
      ['Verarbeitungsverzeichnis', 'vvt'],
      ['Art. 30 Dokumentation', 'vvt'],
      ['TOM definieren', 'tom'],
      ['technisch organisatorische Massnahmen', 'tom'],
      ['Art. 32 Massnahmen', 'tom'],
      ['DSFA durchfuehren', 'dsfa'],
      ['Datenschutz-Folgenabschaetzung', 'dsfa'],
      ['Art. 35 Pruefung', 'dsfa'],
      ['DPIA erstellen', 'dsfa'],
      ['Datenschutzerklaerung', 'dsi'],
      ['Privacy Policy', 'dsi'],
      ['Art. 13 Information', 'dsi'],
      ['Loeschfristen definieren', 'lf'],
      ['Loeschkonzept erstellen', 'lf'],
      ['Retention Policy', 'lf'],
      ['Auftragsverarbeitung', 'av_vertrag'],
      ['AVV erstellen', 'av_vertrag'],
      ['Art. 28 Vertrag', 'av_vertrag'],
      ['Einwilligung einholen', 'einwilligung'],
      ['Consent Management', 'einwilligung'],
      ['Cookie Banner', 'einwilligung'],
    ])('"%s" should detect document type %s', (input, expectedType) => {
      const result = classifier.detectDocumentType(input)
      expect(result).toBe(expectedType)
    })

    it('should return undefined for unrecognized types', () => {
      expect(classifier.detectDocumentType('Hallo Welt')).toBeUndefined()
      expect(classifier.detectDocumentType('Was kostet das?')).toBeUndefined()
    })
  })

  describe('classify - Umlaut handling', () => {
    it('should handle German umlauts correctly', () => {
      // With actual umlauts (ä, ö, ü)
      const result1 = classifier.classify('Prüfe die Vollständigkeit')
      expect(result1.mode).toBe('validate')

      // With ae/oe/ue substitution
      const result2 = classifier.classify('Pruefe die Vollstaendigkeit')
      expect(result2.mode).toBe('validate')
    })

    it('should handle ß correctly', () => {
      const result = classifier.classify('Schließe Lücken')
      // Should still detect via normalized patterns
      expect(result).toBeDefined()
    })
  })

  describe('classify - combined mode + document type', () => {
    it('should detect both mode and document type', () => {
      const result = classifier.classify('Erstelle ein VVT fuer unsere Firma')
      expect(result.mode).toBe('draft')
      expect(result.detectedDocumentType).toBe('vvt')
    })

    it('should detect validate + document type', () => {
      const result = classifier.classify('Pruefe mein TOM auf Konsistenz')
      expect(result.mode).toBe('validate')
      expect(result.detectedDocumentType).toBe('tom')
    })
  })
})