|
| 1 | +/** |
| 2 | + * @vitest-environment node |
| 3 | + */ |
| 4 | +import { dbChainMockFns, resetDbChainMock } from '@sim/testing' |
| 5 | +import { beforeEach, describe, expect, it, vi } from 'vitest' |
| 6 | + |
| 7 | +const { |
| 8 | + mockCheckActorUsageLimits, |
| 9 | + mockGenerateEmbeddings, |
| 10 | + mockGetBoundWorkspaceFileSecretProvenanceByMetadata, |
| 11 | + mockGetFileMetadataByKeys, |
| 12 | + mockProcessDocument, |
| 13 | +} = vi.hoisted(() => ({ |
| 14 | + mockCheckActorUsageLimits: vi.fn(), |
| 15 | + mockGenerateEmbeddings: vi.fn(), |
| 16 | + mockGetBoundWorkspaceFileSecretProvenanceByMetadata: vi.fn(), |
| 17 | + mockGetFileMetadataByKeys: vi.fn(), |
| 18 | + mockProcessDocument: vi.fn(), |
| 19 | +})) |
| 20 | + |
| 21 | +vi.mock('@/lib/billing/calculations/usage-monitor', () => ({ |
| 22 | + checkActorUsageLimits: mockCheckActorUsageLimits, |
| 23 | +})) |
| 24 | + |
| 25 | +vi.mock('@/lib/knowledge/documents/document-processor', () => ({ |
| 26 | + processDocument: mockProcessDocument, |
| 27 | +})) |
| 28 | + |
| 29 | +vi.mock('@/lib/knowledge/embedding-models', () => ({ |
| 30 | + getEmbeddingModelInfo: vi.fn(() => ({ tokenizerProvider: 'openai' })), |
| 31 | +})) |
| 32 | + |
| 33 | +vi.mock('@/lib/knowledge/embeddings', () => ({ |
| 34 | + generateEmbeddings: mockGenerateEmbeddings, |
| 35 | +})) |
| 36 | + |
| 37 | +vi.mock('@/lib/uploads/contexts/workspace/workspace-file-secret-provenance', () => ({ |
| 38 | + getBoundWorkspaceFileSecretProvenanceByMetadata: |
| 39 | + mockGetBoundWorkspaceFileSecretProvenanceByMetadata, |
| 40 | +})) |
| 41 | + |
| 42 | +vi.mock('@/lib/uploads/core/storage-service', () => ({ |
| 43 | + deleteFile: vi.fn(), |
| 44 | +})) |
| 45 | + |
| 46 | +vi.mock('@/lib/uploads/server/metadata', () => ({ |
| 47 | + deleteFileMetadataByIdentity: vi.fn(), |
| 48 | + getFileMetadataByKeys: mockGetFileMetadataByKeys, |
| 49 | +})) |
| 50 | + |
| 51 | +import { processDocumentAsync } from '@/lib/knowledge/documents/service' |
| 52 | + |
| 53 | +const PERSISTED_KEY = 'workspace/workspace-1/persisted.pdf' |
| 54 | +const PERSISTED_URL = `/api/files/serve/${encodeURIComponent(PERSISTED_KEY)}?context=workspace` |
| 55 | +const CONTENT_UPDATED_AT = new Date('2026-08-05T12:00:00.000Z') |
| 56 | + |
| 57 | +const PERSISTED_CONTEXT = { |
| 58 | + workspaceId: null, |
| 59 | + knowledgeBaseUserId: 'knowledge-owner', |
| 60 | + chunkingConfig: null, |
| 61 | + embeddingModel: 'text-embedding-3-small', |
| 62 | + billedAccountUserId: null, |
| 63 | + uploadedBy: 'uploader-1', |
| 64 | + filename: 'persisted.pdf', |
| 65 | + fileUrl: PERSISTED_URL, |
| 66 | + fileSize: 512, |
| 67 | + mimeType: 'application/pdf', |
| 68 | + tag1: null, |
| 69 | + tag2: null, |
| 70 | + tag3: null, |
| 71 | + tag4: null, |
| 72 | + tag5: null, |
| 73 | + tag6: null, |
| 74 | + tag7: null, |
| 75 | + number1: null, |
| 76 | + number2: null, |
| 77 | + number3: null, |
| 78 | + number4: null, |
| 79 | + number5: null, |
| 80 | + date1: null, |
| 81 | + date2: null, |
| 82 | + boolean1: null, |
| 83 | + boolean2: null, |
| 84 | + boolean3: null, |
| 85 | +} |
| 86 | + |
| 87 | +const PERSISTED_PROVENANCE_ROW = { |
| 88 | + id: 'document-1', |
| 89 | + secretProvenanceVersion: null, |
| 90 | + filename: PERSISTED_CONTEXT.filename, |
| 91 | + fileUrl: PERSISTED_CONTEXT.fileUrl, |
| 92 | + contentHash: null, |
| 93 | + sourceUrl: null, |
| 94 | + tag1: null, |
| 95 | + tag2: null, |
| 96 | + tag3: null, |
| 97 | + tag4: null, |
| 98 | + tag5: null, |
| 99 | + tag6: null, |
| 100 | + tag7: null, |
| 101 | + number1: null, |
| 102 | + number2: null, |
| 103 | + number3: null, |
| 104 | + number4: null, |
| 105 | + number5: null, |
| 106 | + date1: null, |
| 107 | + date2: null, |
| 108 | + boolean1: null, |
| 109 | + boolean2: null, |
| 110 | + boolean3: null, |
| 111 | + provenanceSourceHash: null, |
| 112 | + status: null, |
| 113 | + entries: null, |
| 114 | +} |
| 115 | + |
| 116 | +const SOURCE_BINDING = { |
| 117 | + id: 'source-file-1', |
| 118 | + key: PERSISTED_KEY, |
| 119 | + userId: 'uploader-1', |
| 120 | + workspaceId: 'workspace-1', |
| 121 | + context: 'workspace', |
| 122 | + originalName: PERSISTED_CONTEXT.filename, |
| 123 | + displayName: PERSISTED_CONTEXT.filename, |
| 124 | + contentType: PERSISTED_CONTEXT.mimeType, |
| 125 | + size: PERSISTED_CONTEXT.fileSize, |
| 126 | + folderId: null, |
| 127 | + uploadedAt: CONTENT_UPDATED_AT, |
| 128 | + contentUpdatedAt: CONTENT_UPDATED_AT, |
| 129 | + deletedAt: null, |
| 130 | + secretProvenanceVersion: null, |
| 131 | +} |
| 132 | + |
| 133 | +describe('knowledge document processing source', () => { |
| 134 | + beforeEach(() => { |
| 135 | + vi.clearAllMocks() |
| 136 | + resetDbChainMock() |
| 137 | + dbChainMockFns.limit |
| 138 | + .mockResolvedValueOnce([PERSISTED_CONTEXT]) |
| 139 | + .mockResolvedValueOnce([PERSISTED_PROVENANCE_ROW]) |
| 140 | + .mockResolvedValueOnce([{ id: 'document-1' }]) |
| 141 | + mockCheckActorUsageLimits.mockResolvedValue({ isExceeded: false }) |
| 142 | + mockGetFileMetadataByKeys.mockImplementation(async (_keys: string[], context: string) => |
| 143 | + context === 'workspace' ? [SOURCE_BINDING] : [] |
| 144 | + ) |
| 145 | + mockGetBoundWorkspaceFileSecretProvenanceByMetadata.mockResolvedValue( |
| 146 | + new Map([[SOURCE_BINDING.id, { status: 'exact', entries: [] }]]) |
| 147 | + ) |
| 148 | + mockProcessDocument.mockResolvedValue({ |
| 149 | + chunks: [], |
| 150 | + metadata: { chunkCount: 0, tokenCount: 0, characterCount: 0 }, |
| 151 | + }) |
| 152 | + }) |
| 153 | + |
| 154 | + it('uses the persisted document source instead of stale queued source fields', async () => { |
| 155 | + await processDocumentAsync('knowledge-base-1', 'document-1', { |
| 156 | + filename: 'stale.pdf', |
| 157 | + fileUrl: 'https://example.com/stale.pdf', |
| 158 | + fileSize: 1, |
| 159 | + mimeType: 'text/plain', |
| 160 | + }) |
| 161 | + |
| 162 | + expect(mockGetFileMetadataByKeys).toHaveBeenCalledWith( |
| 163 | + [PERSISTED_KEY], |
| 164 | + 'workspace', |
| 165 | + expect.anything() |
| 166 | + ) |
| 167 | + expect(mockGetBoundWorkspaceFileSecretProvenanceByMetadata).toHaveBeenCalledWith( |
| 168 | + expect.anything(), |
| 169 | + [SOURCE_BINDING] |
| 170 | + ) |
| 171 | + expect(mockProcessDocument).toHaveBeenCalledWith( |
| 172 | + PERSISTED_CONTEXT.fileUrl, |
| 173 | + PERSISTED_CONTEXT.filename, |
| 174 | + PERSISTED_CONTEXT.mimeType, |
| 175 | + 1024, |
| 176 | + 200, |
| 177 | + 100, |
| 178 | + PERSISTED_CONTEXT.uploadedBy, |
| 179 | + null, |
| 180 | + undefined, |
| 181 | + undefined |
| 182 | + ) |
| 183 | + expect(mockGenerateEmbeddings).not.toHaveBeenCalled() |
| 184 | + }) |
| 185 | +}) |
0 commit comments