d3ro-voice/apps/mobile-rn/__tests__/knowledge-service.test.ts

const mockFrom = jest.fn()
const mockRealtimeOn = jest.fn()
const mockRealtimeSubscribe = jest.fn()
const mockRealtimeChannel: Record<string, unknown> = {
  on: mockRealtimeOn,
  subscribe: mockRealtimeSubscribe,
}
mockRealtimeOn.mockReturnValue(mockRealtimeChannel)
mockRealtimeSubscribe.mockReturnValue(mockRealtimeChannel)

jest.mock('../src/lib/supabase', () => ({
  supabase: {
    from: (...args: unknown[]) => mockFrom(...args),
    channel: jest.fn(() => mockRealtimeChannel),
    removeChannel: jest.fn(async () => 'ok'),
  },
}))

import type { KnowledgeDocument } from '@d3ro/api-client'
import {
  chunkKnowledgeText,
  createKnowledgeDocument,
  deleteKnowledgeDocument,
  indexKnowledgeDocument,
  KnowledgeServiceError,
  listKnowledgeDocuments,
  mergeKnowledgeDocuments,
  sanitizeKnowledgeSearch,
  searchKnowledge,
  subscribeToKnowledgeDocuments,
} from '../src/features/knowledge/knowledge-service'
import {
  classifyKnowledgeFile,
  validatePickedKnowledgeContent,
} from '../src/features/knowledge/knowledge-file-picker'

const USER_A = '11111111-1111-4111-8111-111111111111'
const USER_B = '22222222-2222-4222-8222-222222222222'

function document(overrides: Partial<KnowledgeDocument> = {}): KnowledgeDocument {
  return {
    id: 'aaaaaaaa-aaaa-4aaa-8aaa-aaaaaaaaaaaa',
    user_id: USER_A,
    team_id: null,
    title: 'Architecture',
    file_name: 'architecture.md',
    file_type: 'md',
    source_url: null,
    storage_key: null,
    chunk_count: 1,
    indexed: false,
    indexed_at: null,
    created_at: '2026-08-21T00:00:00.000Z',
    updated_at: '2026-08-21T00:00:00.000Z',
    ...overrides,
  }
}

function queryBuilder(result: Record<string, unknown>): Record<string, jest.Mock> {
  const builder: Record<string, jest.Mock> = {}
  for (const method of [
    'select',
    'eq',
    'ilike',
    'range',
    'order',
    'insert',
    'update',
    'delete',
    'not',
    'is',
  ]) {
    builder[method] = jest.fn(() => builder)
  }
  builder.single = jest.fn(async () => result)
  builder.maybeSingle = jest.fn(async () => result)
  builder.then = jest.fn((resolve, reject) => Promise.resolve(result).then(resolve, reject))
  return builder
}

beforeEach(() => {
  mockFrom.mockReset()
  mockRealtimeOn.mockClear()
  mockRealtimeSubscribe.mockClear()
  mockRealtimeOn.mockReturnValue(mockRealtimeChannel)
  mockRealtimeSubscribe.mockReturnValue(mockRealtimeChannel)
  jest.restoreAllMocks()
})

describe('knowledge input and file validation', () => {
  it('chunks at bounded newline boundaries without a source-text fallback', () => {
    const first = 'a'.repeat(650)
    const second = 'b'.repeat(500)
    const chunks = chunkKnowledgeText(`${first}\n${second}`)
    expect(chunks).toEqual([first, second])
    expect(chunks.every((chunk) => chunk.length <= 800)).toBe(true)
    expect(() => chunkKnowledgeText('binary\u0000payload')).toThrow(KnowledgeServiceError)
  })

  it('sanitizes PostgREST controls and only accepts UTF-8 TXT or Markdown metadata', () => {
    expect(sanitizeKnowledgeSearch("roadmap),user_id.eq.other,%_'2026"))
      .toBe("roadmap user id.eq.other '2026")
    expect(classifyKnowledgeFile('Roadmap.MD', 'text/markdown')).toEqual({
      fileName: 'Roadmap.MD',
      fileType: 'md',
      title: 'Roadmap',
    })
    expect(classifyKnowledgeFile('notes.txt', null).fileType).toBe('txt')
    expect(() => classifyKnowledgeFile('report.pdf', 'application/pdf'))
      .toThrow(KnowledgeServiceError)
    expect(validatePickedKnowledgeContent('\ufeff valid text ')).toBe('valid text')
    expect(() => validatePickedKnowledgeContent('\ufffd'.repeat(20)))
      .toThrow(KnowledgeServiceError)
  })
})

describe('knowledge document persistence', () => {
  it('lists RLS-visible personal and team documents with sanitized title search', async () => {
    const owned = document()
    const shared = document({
      id: 'bbbbbbbb-bbbb-4bbb-8bbb-bbbbbbbbbbbb',
      user_id: USER_B,
      team_id: 'cccccccc-cccc-4ccc-8ccc-cccccccccccc',
      title: 'Shared',
    })
    const listQuery = queryBuilder({ data: [owned, shared], error: null, count: 2 })
    mockFrom.mockReturnValueOnce(listQuery)

    await expect(listKnowledgeDocuments({
      userId: USER_A,
      search: 'roadmap),user_id.eq.other',
    })).resolves.toEqual(expect.objectContaining({
      documents: [owned, shared],
      total: 2,
      hasMore: false,
    }))
    expect(listQuery.ilike).toHaveBeenCalledWith('title', '%roadmap user id.eq.other%')
    expect(listQuery.eq).not.toHaveBeenCalledWith('user_id', USER_A)
  })

  it('removes an incomplete document when any chunk insert is not confirmed', async () => {
    const created = document({ chunk_count: 2 })
    const documentQuery = queryBuilder({ data: created, error: null })
    const chunkQuery = queryBuilder({ data: [{ id: 'one' }], error: null })
    const cleanupQuery = queryBuilder({ data: null, error: null })
    mockFrom
      .mockReturnValueOnce(documentQuery)
      .mockReturnValueOnce(chunkQuery)
      .mockReturnValueOnce(cleanupQuery)

    await expect(createKnowledgeDocument({
      userId: USER_A,
      title: 'Architecture',
      fileName: 'architecture.md',
      fileType: 'md',
      content: `${'a'.repeat(800)}\n${'b'.repeat(100)}`,
    })).rejects.toMatchObject({ code: 'invalid-response' })
    expect(cleanupQuery.delete).toHaveBeenCalledTimes(1)
    expect(cleanupQuery.eq).toHaveBeenCalledWith('id', created.id)
    expect(cleanupQuery.eq).toHaveBeenCalledWith('user_id', USER_A)
  })

  it('fails closed and resets pending state when indexed metadata lacks all embeddings', async () => {
    const pending = document()
    const ready = document({ indexed: true, indexed_at: '2026-08-21T00:01:00.000Z' })
    const documentQuery = queryBuilder({ data: ready, error: null })
    const countQuery = queryBuilder({ data: null, error: null, count: 0 })
    const resetQuery = queryBuilder({ data: null, error: null })
    mockFrom
      .mockReturnValueOnce(documentQuery)
      .mockReturnValueOnce(countQuery)
      .mockReturnValueOnce(resetQuery)
    jest.spyOn(global, 'fetch').mockResolvedValue(new Response(JSON.stringify({
      embedded: 1,
      total: 1,
      indexed: true,
    }), { status: 200 }))

    await expect(indexKnowledgeDocument({
      userId: USER_A,
      accessToken: 'access-token',
      document: pending,
    })).rejects.toMatchObject({ code: 'invalid-response' })
    expect(resetQuery.update).toHaveBeenCalledWith({ indexed: false, indexed_at: null })
  })

  it('deletes only an owner row guarded by its revision token', async () => {
    const current = document()
    const deleteQuery = queryBuilder({ data: { id: current.id }, error: null })
    mockFrom.mockReturnValueOnce(deleteQuery)
    await expect(deleteKnowledgeDocument(USER_A, current)).resolves.toBeUndefined()
    expect(deleteQuery.eq).toHaveBeenCalledWith('id', current.id)
    expect(deleteQuery.eq).toHaveBeenCalledWith('user_id', USER_A)
    expect(deleteQuery.eq).toHaveBeenCalledWith('updated_at', current.updated_at)

    await expect(deleteKnowledgeDocument(USER_A, document({ user_id: USER_B })))
      .rejects.toMatchObject({ code: 'forbidden' })
  })

  it('keeps the newest document revision when realtime and pagination overlap', () => {
    const newest = document({ updated_at: '2026-08-22T00:00:00.000Z', indexed: true })
    const stale = document({ updated_at: '2026-08-20T00:00:00.000Z', indexed: false })
    expect(mergeKnowledgeDocuments([newest], [stale])).toEqual([newest])
  })
})

describe('knowledge edge contracts', () => {
  it('maps an unconfigured embedding provider without fabricating search results', async () => {
    jest.spyOn(global, 'fetch').mockResolvedValue(new Response(JSON.stringify({
      error: 'embedding_provider_unavailable',
    }), { status: 503 }))
    await expect(searchKnowledge({
      accessToken: 'access-token',
      query: 'architecture',
    })).rejects.toMatchObject({ code: 'index-unavailable', status: 503 })
  })

  it('maps an upstream embedding failure separately from missing configuration', async () => {
    jest.spyOn(global, 'fetch').mockResolvedValue(new Response(JSON.stringify({
      error: 'embedding_upstream_failed',
    }), { status: 502 }))
    await expect(searchKnowledge({
      accessToken: 'access-token',
      query: 'architecture',
    })).rejects.toMatchObject({ code: 'index-failed', status: 502 })
  })

  it('rejects malformed semantic results instead of returning raw text', async () => {
    jest.spyOn(global, 'fetch').mockResolvedValue(new Response(JSON.stringify({
      results: [{ id: 'not-a-uuid', content: 'raw fallback', similarity: 0.9 }],
    }), { status: 200 }))
    await expect(searchKnowledge({
      accessToken: 'access-token',
      query: 'architecture',
    })).rejects.toMatchObject({ code: 'invalid-response' })
  })

  // redteam r3-20: 필터가 걸린 postgres_changes 는 DELETE 를 보내지 않고, 필터 없는 DELETE 구독은 남의 행 id 를 흘린다.
  // 삭제는 knowledge-realtime 의 다른 경로로 받는다 — knowledge-realtime-redteam-r3-20.test.ts 와 같은 계약.
  it('subscribes to document inserts and updates through RLS realtime, never to DELETE', () => {
    subscribeToKnowledgeDocuments(jest.fn(), jest.fn())
    const events = mockRealtimeOn.mock.calls.map((call) => (call[1] as { event: string }).event)
    expect(events).toEqual(['INSERT', 'UPDATE'])
  })
})