fix(files): release purged history inside the file-row purge transaction

Author: waleedlatif1

apps/docs/openapi-v2-files-audit.json

@@ -4569,7 +4569,7 @@
           "source": {
             "type": "string",
             "enum": ["upload", "user", "api", "copilot", "workflow", "collab", "revert", "unknown"],
-            "description": "What wrote this version: `upload` (the original upload), `user` (a save in the Sim editor), `api` (an API, CLI, or MCP write), `copilot` (Sim, the agent), `workflow` (a workflow run), `collab` (collaborative editing), `revert` (a revert to an earlier version), or `unknown` (content written before version history existed)."
+            "description": "What wrote this version: `upload` (the original upload), `user` (a save in the Sim editor), `api` (an API, CLI, or MCP write), `copilot` (Sim, the agent), `workflow` (a workflow run), `collab` (collaborative editing), `revert` (a revert to an earlier version), or `unknown` (content written before version history existed, or by a writer with no source of its own)."
           },
           "authors": {
             "type": "array",

apps/sim/background/cleanup-file-versions.ts

@@ -9,10 +9,7 @@ import type { PlanCategory } from '@/lib/billing/plan-helpers'
 import { DEFAULT_DELETE_CHUNK_SIZE } from '@/lib/cleanup/batch-delete'
 import { retentionCleanupQueue } from '@/lib/cleanup/queue'
 import { enqueueWorkspaceFileStorageCleanups } from '@/lib/uploads/contexts/workspace/workspace-file-storage-cleanup-outbox'
-import {
-  FILE_VERSION_RETENTION_KEEP_LATEST,
-  MAX_SUPERSEDED_FILE_VERSIONS,
-} from '@/lib/uploads/contexts/workspace/workspace-file-versions'
+import { MAX_SUPERSEDED_FILE_VERSIONS } from '@/lib/uploads/contexts/workspace/workspace-file-versions'
 
 const logger = createLogger('CleanupFileVersions')
 
@@ -39,6 +36,9 @@ const MAX_SUPERSEDED_VERSIONS_BY_PLAN: Record<PlanCategory, number> = {
   enterprise: MAX_SUPERSEDED_FILE_VERSIONS,
 }
 
+/** Retention never prunes the newest versions of a file, whatever their age. */
+const FILE_VERSION_RETENTION_KEEP_LATEST = 10
+
 /** Newest superseded versions a file keeps whatever their age (the current version is the tenth). */
 const KEEP_SUPERSEDED = FILE_VERSION_RETENTION_KEEP_LATEST - 1
 

apps/sim/background/cleanup-soft-deletes.test.ts

@@ -82,11 +82,11 @@ vi.mock('@/lib/uploads', () => ({
 
 vi.mock('@/lib/uploads/server/metadata', () => ({ deleteFileMetadata: mockDeleteFileMetadata }))
 
-const { mockReleaseExpiredWorkspaceFileVersions } = vi.hoisted(() => ({
-  mockReleaseExpiredWorkspaceFileVersions: vi.fn(),
+const { mockReleaseWorkspaceFileVersionsForPurgeInTx } = vi.hoisted(() => ({
+  mockReleaseWorkspaceFileVersionsForPurgeInTx: vi.fn(),
 }))
 vi.mock('@/lib/uploads/contexts/workspace/workspace-file-versions', () => ({
-  releaseExpiredWorkspaceFileVersions: mockReleaseExpiredWorkspaceFileVersions,
+  releaseWorkspaceFileVersionsForPurgeInTx: mockReleaseWorkspaceFileVersionsForPurgeInTx,
 }))
 
 vi.mock('@/lib/workflows/utils', () => ({
@@ -129,7 +129,7 @@ describe('cleanup soft deletes', () => {
     })
   })
 
-  it('releases version history only for files whose current object was deleted', async () => {
+  it('releases version history inside the transaction that purges the file rows', async () => {
     mockSelectRowsByIdChunks
       .mockResolvedValueOnce([])
       .mockResolvedValueOnce([])
@@ -148,28 +148,26 @@ describe('cleanup soft deletes', () => {
           context: 'workspace',
           sizeBytes: 4,
         },
-        {
-          id: 'chat-file',
-          key: 'mothership/chat-file',
-          workspaceId: 'ws-1',
-          context: 'mothership',
-          sizeBytes: 3,
-        },
       ])
     mockDeleteFiles.mockResolvedValueOnce({
       deleted: 1,
       failed: [{ key: 'workspace/ws-1/file-failed', error: 'storage unavailable' }],
     })
+    dbChainMockFns.returning.mockResolvedValueOnce([{ id: 'file-purged', sizeBytes: 5 }])
 
     await runCleanupSoftDeletes(basePayload)
 
-    expect(mockReleaseExpiredWorkspaceFileVersions).toHaveBeenCalledWith(
-      expect.anything(),
+    expect(mockReleaseWorkspaceFileVersionsForPurgeInTx).toHaveBeenCalledOnce()
+    expect(mockReleaseWorkspaceFileVersionsForPurgeInTx).toHaveBeenCalledWith(
+      dbChainMock.db,
       ['file-purged'],
       expect.any(Date)
     )
-    expect(mockDeleteFiles.mock.invocationCallOrder[0]).toBeLessThan(
-      mockReleaseExpiredWorkspaceFileVersions.mock.invocationCallOrder[0]
+    expect(dbChainMockFns.transaction.mock.invocationCallOrder[0]).toBeLessThan(
+      mockReleaseWorkspaceFileVersionsForPurgeInTx.mock.invocationCallOrder[0]
+    )
+    expect(mockReleaseWorkspaceFileVersionsForPurgeInTx.mock.invocationCallOrder[0]).toBeLessThan(
+      dbChainMockFns.delete.mock.invocationCallOrder[0]
     )
   })
 

apps/sim/background/cleanup-soft-deletes.ts

@@ -44,7 +44,7 @@ import { hardDeleteDocuments } from '@/lib/knowledge/documents/service'
 import type { StorageContext } from '@/lib/uploads'
 import { isUsingCloudStorage, StorageService } from '@/lib/uploads'
 import { allocateUniqueWorkspaceFileName } from '@/lib/uploads/contexts/workspace/workspace-file-manager'
-import { releaseExpiredWorkspaceFileVersions } from '@/lib/uploads/contexts/workspace/workspace-file-versions'
+import { releaseWorkspaceFileVersionsForPurgeInTx } from '@/lib/uploads/contexts/workspace/workspace-file-versions'
 import { deleteFileMetadata } from '@/lib/uploads/server/metadata'
 import { getWorkspaceFileSize } from '@/lib/uploads/shared/types'
 import { deduplicateWorkflowName } from '@/lib/workflows/utils'
@@ -333,6 +333,11 @@ async function deleteExpiredBillableWorkspaceFileRows(
     for (const batch of chunkArray(workspaceRows, DEFAULT_DELETE_CHUNK_SIZE)) {
       try {
         const deletedCount = await db.transaction(async (tx) => {
+          await releaseWorkspaceFileVersionsForPurgeInTx(
+            tx,
+            batch.map(({ id }) => id),
+            retentionDate
+          )
           const deletedRows = await tx
             .delete(workspaceFiles)
             .where(
@@ -901,15 +906,6 @@ export async function runCleanupSoftDeletes(
   }
 
   const fileCleanup = await cleanupWorkspaceFileStorage(fileScope)
-  /**
-   * Only files whose current object is gone are committed to the purge, so a file whose object
-   * deletion failed keeps its history for the retry instead of losing it ahead of its own delete.
-   */
-  await releaseExpiredWorkspaceFileVersions(
-    cleanupDb,
-    fileCleanup.multiContextRows.filter((row) => row.context === 'workspace').map((row) => row.id),
-    retentionDate
-  )
   if (budgets && fileCleanup.filesFailed) throw new Error('File storage cleanup failed')
 
   let totalDeleted = 0

apps/sim/lib/api/contracts/v2/file-versions.ts

@@ -22,16 +22,16 @@ import {
 
 /**
  * v2 file version history contracts. Every content write that changes the bytes — an upload, an
- * API or editor save, a Sim edit, a workflow write, a revert — records a version; collaborative
- * edits and repeated workflow writes from one author fold into one version per ten-minute window. Renames and moves
- * are metadata changes and never create versions, so every version reads under the file's
- * current name.
+ * API or editor save, a Sim edit, a workflow write, a revert — records a version. Collaborative
+ * edits and repeated workflow writes from one author fold into the current version until it is ten
+ * minutes old or its writer has been idle for five. Renames and moves are metadata changes and never
+ * create versions, so every version reads under the file's current name.
  */
 
 export const v2FileVersionSourceSchema = z
   .enum(['upload', 'user', 'api', 'copilot', 'workflow', 'collab', 'revert', 'unknown'])
   .describe(
-    'What wrote this version: `upload` (the original upload), `user` (a save in the Sim editor), `api` (an API, CLI, or MCP write), `copilot` (Sim, the agent), `workflow` (a workflow run), `collab` (collaborative editing), `revert` (a revert to an earlier version), or `unknown` (content written before version history existed).'
+    'What wrote this version: `upload` (the original upload), `user` (a save in the Sim editor), `api` (an API, CLI, or MCP write), `copilot` (Sim, the agent), `workflow` (a workflow run), `collab` (collaborative editing), `revert` (a revert to an earlier version), or `unknown` (content written before version history existed, or by a writer with no source of its own).'
   )
 
 export const v2FileVersionAuthorSchema = z

apps/sim/lib/uploads/contexts/workspace/__integration__/file-versions.integration.ts

@@ -39,7 +39,7 @@ import { WORKSPACE_FILE_STORAGE_CLEANUP_OUTBOX_EVENT } from '@/lib/uploads/conte
 import {
   getCurrentWorkspaceFileVersion,
   queryWorkspaceFileVersions,
-  releaseExpiredWorkspaceFileVersions,
+  releaseWorkspaceFileVersionsForPurgeInTx,
 } from '@/lib/uploads/contexts/workspace/workspace-file-versions'
 import { revertWorkspaceFileVersion } from '@/lib/workspace-files/application/file-versions'
 import { runCleanupFileVersions } from '@/background/cleanup-file-versions'
@@ -402,7 +402,7 @@ describe('workspace file version history in PostgreSQL', () => {
     ).rejects.toMatchObject({ code: 'conflict' })
   })
 
-  it('releases an expired file history atomically and leaves a restored file untouched', async () => {
+  it('releases history only with the purge transaction and leaves a restored file untouched', async () => {
     const expired = await seedFile('one')
     const restored = await seedFile('one')
     for (const fixture of [expired, restored]) {
@@ -423,10 +423,17 @@ describe('workspace file version history in PostgreSQL', () => {
       .filter((row) => row.supersededAt !== null)
       .map((row) => row.key)
 
-    await releaseExpiredWorkspaceFileVersions(
-      db,
-      [expired.fileId, restored.fileId],
-      new Date(Date.now() - 30 * 24 * 60 * 60 * 1000)
+    const cutoff = new Date(Date.now() - 30 * 24 * 60 * 60 * 1000)
+    await expect(
+      db.transaction(async (tx) => {
+        await releaseWorkspaceFileVersionsForPurgeInTx(tx, [expired.fileId], cutoff)
+        throw new Error('purge failed')
+      })
+    ).rejects.toThrow('purge failed')
+    expect((await versionRows(expired.fileId)).map((row) => row.version)).toEqual([1, 2, 3])
+
+    await db.transaction((tx) =>
+      releaseWorkspaceFileVersionsForPurgeInTx(tx, [expired.fileId, restored.fileId], cutoff)
     )
 
     expect((await versionRows(expired.fileId)).map((row) => row.version)).toEqual([3])

apps/sim/lib/uploads/contexts/workspace/workspace-file-storage-accounting.test.ts

@@ -210,7 +210,7 @@ describe('workspace file metadata and storage accounting', () => {
     mockProcessWorkspaceFileLiveDocReconciliationNow.mockResolvedValue('completed')
     mockReplaceWorkspaceFileSecretProvenanceInTx.mockResolvedValue(undefined)
     mockSaveCollabDocStateInTx.mockResolvedValue(undefined)
-    mockRecordWorkspaceFileVersionInTx.mockResolvedValue({ releasedKeys: [] })
+    mockRecordWorkspaceFileVersionInTx.mockResolvedValue({ version: 2, releasedKeys: [] })
   })
 
   it('returns the canonical inserted record with the pre-resolved folder path', async () => {
@@ -785,7 +785,10 @@ describe('workspace file metadata and storage accounting', () => {
     dbChainMockFns.limit.mockResolvedValueOnce([FILE_ROW]).mockResolvedValueOnce([FILE_ROW])
     dbChainMockFns.returning.mockResolvedValueOnce([updatedFile])
     mockUploadFile.mockResolvedValueOnce({ key: replacementKey })
-    mockRecordWorkspaceFileVersionInTx.mockResolvedValueOnce({ releasedKeys: [FILE_ROW.key] })
+    mockRecordWorkspaceFileVersionInTx.mockResolvedValueOnce({
+      version: 2,
+      releasedKeys: [FILE_ROW.key],
+    })
 
     await updateWorkspaceFileContent(
       FILE_ROW.workspaceId,