fix(files): resolve every generated-document source type, not four of five

The archive keyed off a locally enumerated set of source content types that
omitted text/x-pdflibjs — the isolated-vm PDF generator. Those files failed the
check and streamed their generator source under a .pdf name, which is the exact
corruption this change exists to fix.

A canonical five-member set already existed in the file viewer; enumerating a
fourth copy was the mistake, not the missing string. The set moves to file-utils
beside the other file-type predicates, the viewer imports it, and the route asks
isGeneratedDocumentSourceType rather than carrying its own list. A parameterized
test pins all five, so adding a generator without extending the set fails.

Author: waleedlatif1

apps/sim/app/api/workspaces/[id]/files/download/route.test.ts

@@ -248,6 +248,27 @@ describe('workspace files download route', () => {
     expect(body.error).not.toContain('second.docx')
   })
 
+  it.each([
+    ['text/x-docxjs', 'report.docx'],
+    ['text/x-pptxgenjs', 'deck.pptx'],
+    ['text/x-pdflibjs', 'isolated.pdf'],
+    ['text/x-python-pdf', 'sandboxed.pdf'],
+    ['text/x-python-xlsx', 'sheet.xlsx'],
+  ])('resolves %s rather than streaming its source', async (type, name) => {
+    // Both PDF generators must be covered: the isolated-vm path stores pdf-lib JS and
+    // the E2B path stores Python, and either one streamed raw is the corruption bug.
+    mockListWorkspaceFiles.mockResolvedValue([{ ...workspaceFile('f1', name), type }])
+    mockFetchServableWorkspaceFileBuffer.mockResolvedValue({
+      buffer: Buffer.from('PKrendered'),
+      contentType: 'application/octet-stream',
+    })
+
+    await zipFrom(await GET(requestFor('fileIds=f1'), context))
+
+    expect(mockFetchServableWorkspaceFileBuffer).toHaveBeenCalledTimes(1)
+    expect(mockDownloadFileStream).not.toHaveBeenCalled()
+  })
+
   it('streams an uploaded office file rather than resolving it', async () => {
     // A real upload serves exactly its stored bytes, so it must not take the buffered
     // path — otherwise a selection of large decks is held in memory for nothing.

apps/sim/app/api/workspaces/[id]/files/download/route.ts

@@ -6,12 +6,6 @@ import { type NextRequest, NextResponse } from 'next/server'
 import { downloadWorkspaceFileItemsContract } from '@/lib/api/contracts/workspace-file-folders'
 import { parseRequest } from '@/lib/api/server'
 import { getSession } from '@/lib/auth'
-import {
-  DOCXJS_SOURCE_MIME,
-  PPTXGENJS_SOURCE_MIME,
-  PYTHON_PDF_SOURCE_MIME,
-  PYTHON_XLSX_SOURCE_MIME,
-} from '@/lib/copilot/tools/server/files/doc-compile'
 import { nodeReadableToWebStream } from '@/lib/core/utils/node-stream'
 import { PayloadSizeLimitError } from '@/lib/core/utils/stream-limits'
 import { withRouteHandler } from '@/lib/core/utils/with-route-handler'
@@ -26,6 +20,7 @@ import {
 import { downloadFileStream } from '@/lib/uploads/core/storage-service'
 import {
   formatFileSize,
+  isGeneratedDocumentSourceType,
   isRenderableDocumentName,
   MAX_RENDERED_DOCUMENT_BYTES,
 } from '@/lib/uploads/utils/file-utils'
@@ -37,14 +32,6 @@ const logger = createLogger('WorkspaceFilesDownloadAPI')
 const MAX_ZIP_DOWNLOAD_FILES = 100
 const MAX_ZIP_DOWNLOAD_BYTES = 250 * 1024 * 1024
 
-/** Content types under which a generated document's *source* is stored. */
-const GENERATED_SOURCE_TYPES = new Set<string>([
-  DOCXJS_SOURCE_MIME,
-  PPTXGENJS_SOURCE_MIME,
-  PYTHON_PDF_SOURCE_MIME,
-  PYTHON_XLSX_SOURCE_MIME,
-])
-
 /**
  * Whether this entry's stored bytes are a generation source that has to be resolved
  * before it can go in the archive. An ordinary uploaded `.docx` serves exactly what is
@@ -55,7 +42,7 @@ const GENERATED_SOURCE_TYPES = new Set<string>([
  * a document name.
  */
 function needsRendering(file: WorkspaceFileRecord): boolean {
-  return file.type ? GENERATED_SOURCE_TYPES.has(file.type) : isRenderableDocumentName(file.name)
+  return file.type ? isGeneratedDocumentSourceType(file.type) : isRenderableDocumentName(file.name)
 }
 
 /**

apps/sim/app/workspace/[workspaceId]/files/components/file-viewer/use-editable-file-content.ts

@@ -3,6 +3,7 @@
 import { useCallback, useEffect, useMemo, useReducer, useRef } from 'react'
 import { toast } from '@sim/emcn'
 import type { WorkspaceFileRecord } from '@/lib/uploads/contexts/workspace'
+import { GENERATED_DOCUMENT_SOURCE_TYPES } from '@/lib/uploads/utils/file-utils'
 import {
   useUpdateWorkspaceFileContent,
   useWorkspaceFileContent,
@@ -20,13 +21,7 @@ import {
  * editable text is the source program, not the compiled artifact. The serve route
  * returns that source only when asked for the raw representation.
  */
-const GENERATED_SOURCE_FILE_TYPES = new Set([
-  'text/x-pptxgenjs',
-  'text/x-docxjs',
-  'text/x-pdflibjs',
-  'text/x-python-pdf',
-  'text/x-python-xlsx',
-])
+const GENERATED_SOURCE_FILE_TYPES = GENERATED_DOCUMENT_SOURCE_TYPES
 
 /**
  * Poll cadence for the content query while the post-stream reconcile waits for a fetch showing the

apps/sim/lib/uploads/utils/file-utils.ts

@@ -217,6 +217,25 @@ export function getFileExtension(filename: string): string {
  */
 const RENDERABLE_DOCUMENT_EXTENSIONS = new Set(['pdf', 'docx', 'pptx', 'xlsx'])
 
+/**
+ * Content types under which a generated document's *generation source* is stored. A
+ * file carrying one of these renders to something other than its stored bytes, so any
+ * surface that hands out the file itself has to resolve it first. Both PDF generators
+ * are here: the E2B path stores Python, the isolated-vm path stores pdf-lib JS.
+ */
+export const GENERATED_DOCUMENT_SOURCE_TYPES = new Set<string>([
+  'text/x-docxjs',
+  'text/x-pptxgenjs',
+  'text/x-pdflibjs',
+  'text/x-python-pdf',
+  'text/x-python-xlsx',
+])
+
+/** True when the stored bytes for `contentType` are a generation source. */
+export function isGeneratedDocumentSourceType(contentType: string | undefined | null): boolean {
+  return contentType ? GENERATED_DOCUMENT_SOURCE_TYPES.has(contentType) : false
+}
+
 /**
  * Ceiling on a single rendered generated document. A generator source is text and is
  * orders of magnitude smaller than the document it produces, so the declared size is no