chore(cleanup): remove dead uploads, tokenization, table, knowledge, and search exports

Author: waleedlatif1

apps/sim/app/api/files/delete/route.test.ts

@@ -54,7 +54,6 @@ describe('File Delete API Route', () => {
     filesAuthorizationMockFns.mockVerifyFileAccess.mockResolvedValue(true)
     storageServiceMockFns.mockDeleteFile.mockResolvedValue(undefined)
     storageServiceMockFns.mockHasCloudStorage.mockReturnValue(true)
-    uploadsMockFns.mockGetStorageProvider.mockReturnValue('s3')
     uploadsMockFns.mockIsUsingCloudStorage.mockReturnValue(true)
   })
 

apps/sim/lib/file-parsers/yaml-parser.ts

@@ -29,14 +29,6 @@ export class YamlComplexityError extends FileParserError {
   }
 }
 
-/**
- * Type guard for {@link YamlComplexityError}. Callers use this to fail closed on
- * a complexity-limit rejection instead of falling back to a generic parse.
- */
-export function isYamlComplexityError(error: unknown): error is YamlComplexityError {
-  return error instanceof YamlComplexityError
-}
-
 /**
  * Validate that a parsed YAML value stays within the file parser's expansion
  * limits, returning the document depth.

apps/sim/lib/internal/file/parser.test.ts

@@ -177,7 +177,7 @@ vi.mock('fs/promises', () => ({
   writeFile: mockFsWriteFile,
 }))
 
-const { mockGetStorageProvider, mockIsUsingCloudStorage } = uploadsMockFns
+const { mockIsUsingCloudStorage } = uploadsMockFns
 const { mockGetBoundWorkspaceFileSecretProvenance } = workspaceFileSecretProvenanceMockFns
 
 import { fileParseBodySchema } from '@/lib/api/contracts/storage-transfer'
@@ -237,14 +237,8 @@ async function POST(request: NextRequest): Promise<Response> {
   })
 }
 
-function setupFileApiMocks(
-  options: {
-    authenticated?: boolean
-    storageProvider?: 's3' | 'blob' | 'local'
-    cloudEnabled?: boolean
-  } = {}
-) {
-  const { authenticated = true, storageProvider = 's3', cloudEnabled = true } = options
+function setupFileApiMocks(options: { authenticated?: boolean; cloudEnabled?: boolean } = {}) {
+  const { authenticated = true, cloudEnabled = true } = options
 
   if (authenticated) {
     authMockFns.mockGetSession.mockResolvedValue({
@@ -272,7 +266,6 @@ function setupFileApiMocks(
     error: authenticated ? undefined : 'Unauthorized',
   })
 
-  mockGetStorageProvider.mockReturnValue(storageProvider)
   mockIsUsingCloudStorage.mockReturnValue(cloudEnabled)
 }
 
@@ -475,7 +468,6 @@ describe('file parser operation', () => {
   it('should keep known binary extensions as binary even when the bytes are valid UTF-8', async () => {
     setupFileApiMocks({
       cloudEnabled: true,
-      storageProvider: 's3',
       authenticated: true,
     })
     mockIsSupportedFileType.mockReturnValue(false)
@@ -496,7 +488,6 @@ describe('file parser operation', () => {
   it('should parse unknown extensions as text when the bytes look like UTF-8 text', async () => {
     setupFileApiMocks({
       cloudEnabled: true,
-      storageProvider: 's3',
       authenticated: true,
     })
     mockIsSupportedFileType.mockReturnValue(false)
@@ -521,7 +512,6 @@ describe('file parser operation', () => {
   it('reports degraded parser output as a failure instead of returning it as content', async () => {
     setupFileApiMocks({
       cloudEnabled: false,
-      storageProvider: 'local',
       authenticated: true,
     })
     mockParseBuffer.mockResolvedValue({
@@ -546,7 +536,6 @@ describe('file parser operation', () => {
   it('should reject parser complexity limits instead of returning raw text', async () => {
     setupFileApiMocks({
       cloudEnabled: true,
-      storageProvider: 's3',
       authenticated: true,
     })
     storageServiceMockFns.mockDownloadFile.mockResolvedValue(Buffer.from('{"value":true}'))
@@ -743,7 +732,6 @@ describe('file parser operation', () => {
   it('should reject oversized local files before materializing them', async () => {
     setupFileApiMocks({
       cloudEnabled: false,
-      storageProvider: 'local',
       authenticated: true,
     })
     mockFsStat.mockResolvedValue({ isFile: () => true, size: 104857601 })

apps/sim/lib/knowledge/application/search-integrations.test.ts

@@ -43,14 +43,6 @@ vi.mock('@/lib/sim-search/connectors', () => ({
     ['github', { name: 'GitHub' }],
     ['jira', { name: 'Jira' }],
   ],
-  searchMemberAccountProvider: (type: string) =>
-    type === 'google_drive'
-      ? 'google-drive'
-      : ['gmail', 'jira', 'github'].includes(type)
-        ? type === 'github'
-          ? 'github-repositories'
-          : type
-        : null,
 }))
 
 import { CredentialGroupProviderConfigurationError } from '@/lib/credential-groups/provider-adapter'

apps/sim/lib/knowledge/constants.ts

@@ -32,7 +32,6 @@ export const MAX_KNOWLEDGE_CONNECTOR_DOCUMENT_SEARCH_LENGTH = 200
 /** Bound viewer-specific source resolution and document counts to a single page. */
 export const SEARCH_SOURCE_PAGE_SIZE = 25
 export const SEARCH_SOURCE_CANDIDATE_PAGE_SIZE = 100
-export const MAX_SEARCH_SOURCE_PROVIDER_TYPES = 100
 
 /**
  * Chunking a knowledge base gets when its creator names no configuration.

apps/sim/lib/knowledge/documents/service.ts

@@ -145,7 +145,6 @@ import {
   enqueueKnowledgeStorageCleanup,
   getKnowledgeBaseStorageKey,
   isKnowledgeBaseOwnedStorageKey,
-  type KnowledgeStorageCleanupDocument,
 } from '@/lib/knowledge/documents/storage-cleanup'
 import { claimKnowledgeUploadForAttachment } from '@/lib/knowledge/documents/storage-upload'
 import {
@@ -4067,14 +4066,6 @@ export async function updateDocument(
   }
 }
 
-/** Persists standalone cleanup intents; document mutations supply their own transaction. */
-export async function deleteDocumentStorageFiles(
-  documentsToDelete: readonly KnowledgeStorageCleanupDocument[],
-  requestId: string
-): Promise<void> {
-  await enqueueKnowledgeStorageCleanup(db, documentsToDelete, requestId)
-}
-
 async function excludeConnectorDocuments(
   documentIds: string[],
   requestId: string

apps/sim/lib/knowledge/documents/workspace-source-provenance.test.ts

@@ -1,6 +1,6 @@
 import { dbChainMockFns, resetDbChainMock } from '@sim/testing'
 import { billingStorageMock, billingStorageMockFns } from '@sim/testing/mocks/billing-storage.mock'
-import { storageServiceMock, storageServiceMockFns } from '@sim/testing/mocks/storage-service.mock'
+import { storageServiceMock } from '@sim/testing/mocks/storage-service.mock'
 import {
   uploadsMetadataMock,
   uploadsMetadataMockFns,
@@ -22,11 +22,7 @@ vi.mock('@/lib/uploads/core/storage-service', () => storageServiceMock)
 
 vi.mock('@/lib/uploads/server/metadata', () => uploadsMetadataMock)
 
-import {
-  createDocumentRecords,
-  createSingleDocument,
-  deleteDocumentStorageFiles,
-} from '@/lib/knowledge/documents/service'
+import { createDocumentRecords, createSingleDocument } from '@/lib/knowledge/documents/service'
 
 const mockCheckStorageQuotaForBillingContext =
   billingStorageMockFns.mockCheckStorageQuotaForBillingContext
@@ -36,8 +32,6 @@ const mockMaybeNotifyStorageLimitForBillingContext =
   billingStorageMockFns.mockMaybeNotifyStorageLimitForBillingContext
 const mockResolveStorageBillingContext = billingStorageMockFns.mockResolveStorageBillingContext
 
-const mockDeleteFile = storageServiceMockFns.mockDeleteFile
-const mockDeleteFileMetadataByIdentity = uploadsMetadataMockFns.mockDeleteFileMetadataByIdentity
 const mockGetFileMetadataByKeys = uploadsMetadataMockFns.mockGetFileMetadataByKeys
 const mockGetBoundWorkspaceFileSecretProvenanceByMetadata =
   workspaceFileSecretProvenanceMockFns.mockGetBoundWorkspaceFileSecretProvenanceByMetadata
@@ -228,50 +222,4 @@ describe('knowledge workspace source provenance', () => {
       })
     }
   })
-
-  it('never deletes a referenced workspace source as knowledge-base storage', async () => {
-    await deleteDocumentStorageFiles(
-      [{ id: 'document-1', fileUrl: SOURCE_URL, workspaceId: WORKSPACE_ID }],
-      'request-1'
-    )
-
-    expect(mockGetFileMetadataByKeys).not.toHaveBeenCalled()
-    expect(mockDeleteFile).not.toHaveBeenCalled()
-    expect(mockDeleteFileMetadataByIdentity).not.toHaveBeenCalled()
-  })
-
-  it.each(['org-1', 'org-2', null])(
-    'only queues an organization cache for its exact owner: %s',
-    async (organizationId) => {
-      const storageKey = 'kb/org-source.pdf'
-      mockGetFileMetadataByKeys.mockResolvedValue([
-        {
-          ...SOURCE_BINDING,
-          key: storageKey,
-          context: 'knowledge-base',
-          workspaceId: null,
-          organizationId: 'org-1',
-        },
-      ])
-      const cleanup = deleteDocumentStorageFiles(
-        [
-          {
-            id: 'org-doc',
-            fileUrl: `/api/files/serve/${encodeURIComponent(storageKey)}`,
-            workspaceId: null,
-            organizationId,
-          },
-        ],
-        'request-1'
-      )
-      if (organizationId === 'org-1') {
-        await expect(cleanup).resolves.toBeUndefined()
-        expect(dbChainMockFns.values).toHaveBeenCalledOnce()
-      } else {
-        await expect(cleanup).rejects.toThrow()
-        expect(dbChainMockFns.values).not.toHaveBeenCalled()
-      }
-      expect(mockDeleteFile).not.toHaveBeenCalled()
-    }
-  )
 })

apps/sim/lib/knowledge/service.test.ts

@@ -1,6 +1,5 @@
 import {
   dbChainMockFns,
-  hasMockCondition,
   permissionsMock,
   permissionsMockFns,
   queueTableRows,
@@ -18,7 +17,6 @@ vi.mock('@/lib/billing/storage', () => billingStorageMock)
 
 import {
   attachKnowledgeBaseConnectors,
-  findActiveKnowledgeBasesByExactName,
   getWorkspaceKnowledgeBases,
   KnowledgeBasePermissionError,
   updateKnowledgeBase,
@@ -59,31 +57,6 @@ describe('getWorkspaceKnowledgeBases — paging', () => {
   })
 })
 
-/**
- * A VFS path names one knowledge base exactly. Resolving it by reading every base whose name
- * merely CONTAINS the term, then filtering in JS, makes a single-row lookup scale with the
- * workspace — the sibling `findActiveTablesByExactName` is the shape to match.
- */
-describe('findActiveKnowledgeBasesByExactName', () => {
-  beforeEach(() => {
-    resetDbChainMock()
-  })
-
-  it('matches the name exactly and reads at most two rows', async () => {
-    await findActiveKnowledgeBasesByExactName('ws-1', 'Docs')
-
-    const [condition] = dbChainMockFns.where.mock.calls[0] ?? []
-    expect(
-      hasMockCondition(
-        condition,
-        (node) =>
-          node.type === 'eq' && node.left === schemaMock.knowledgeBase.name && node.right === 'Docs'
-      )
-    ).toBe(true)
-    expect(dbChainMockFns.limit).toHaveBeenCalledWith(2)
-  })
-})
-
 /**
  * These tests guard the workspace mass-assignment fix:
  * a user with write/admin on the *source* workspace must not be able to move a

apps/sim/lib/knowledge/service.ts

@@ -424,22 +424,6 @@ export async function getWorkspaceKnowledgeBases(
   }
 }
 
-/** Loads at most two active exact-name matches so a caller can fail on corrupt ambiguity. */
-export async function findActiveKnowledgeBasesByExactName(
-  workspaceId: string,
-  name: string
-): Promise<ActiveKnowledgeBaseReference[]> {
-  return readKnowledgeBaseRows(
-    and(
-      eq(knowledgeBase.workspaceId, workspaceId),
-      eq(knowledgeBase.name, name),
-      isNull(knowledgeBase.deletedAt)
-    ),
-    listOrderBy(keysetColumns(KNOWLEDGE_BASE_SORTS.createdAt), 'asc'),
-    2
-  )
-}
-
 /**
  * Create a new knowledge base
  */

apps/sim/lib/sim-search/connectors.ts

@@ -1,9 +1,6 @@
 import type { ComponentType } from 'react'
 import type { IntegrationAvailabilityResponse } from '@/lib/api/contracts/common'
-import {
-  findCredentialGroupProviderFromProviderId,
-  isCredentialGroupStandardOAuthProvider,
-} from '@/lib/credential-groups/providers'
+import { findCredentialGroupProviderFromProviderId } from '@/lib/credential-groups/providers'
 import { getIntegrationsForCredentialProvider } from '@/lib/integrations/credential-display'
 import {
   getCanonicalScopesForProvider,
@@ -83,13 +80,6 @@ export const SEARCH_CONNECTORS: readonly SearchConnector[] = Object.entries(CONN
   })
   .sort((a, b) => a.meta.name.localeCompare(b.meta.name))
 
-/** Standard member sign-in configured by an explicit Search source addition. */
-export function searchMemberAccountProvider(connectorType: string) {
-  const connector = SEARCH_CONNECTORS.find((candidate) => candidate.type === connectorType)
-  const provider = connector && findCredentialGroupProviderFromProviderId(connector.providerId)
-  return provider && isCredentialGroupStandardOAuthProvider(provider) ? provider : null
-}
-
 /**
  * Every source an admin may set up for Sim Search, alphabetical by name: the
  * connectors that either mirror their source's permissions or connect per person.
@@ -125,11 +115,6 @@ export function personalSetupFields(meta: ConnectorMeta): ConnectorConfigField[]
   )
 }
 
-/** Personal sources use defaults even when they also support central indexing. Slack needs a custom app first. */
-export function canConnectWithDefaults(meta: ConnectorMeta): boolean {
-  return canConnectPersonally(meta) && meta.id !== 'slack' && personalSetupFields(meta).length === 0
-}
-
 /** The name a connector shows, from its registry entry. */
 export function connectorDisplayName(connectorType: string): string {
   return CONNECTOR_META_REGISTRY[connectorType]?.name ?? connectorType