fix(files): probe served bytes before handing off, keep generated documents in-page, version local objects by inode

Author: waleedlatif1

apps/sim/app/workspace/[workspaceId]/files/components/file-viewer/file-viewer.tsx

@@ -449,15 +449,19 @@ function StreamedMediaPlayer({
   const diagnosing = useRef(false)
 
   /**
-   * A media element reports only that it failed, not why, so the URL is asked directly. A 404 means
-   * the key was replaced; the record is re-resolved, and a moved key re-renders this preview at the
-   * new URL. Bytes that are there but would not play are the browser's format support.
+   * A media element reports only that it failed, not why, so the URL is asked directly for its
+   * first byte, which opens storage. A 404 means the key was replaced; the record is re-resolved,
+   * and a moved key re-renders this preview at the new URL. Bytes that are there but would not play
+   * are the browser's format support.
    */
   const handleError = useCallback(async () => {
     if (diagnosing.current) return
     diagnosing.current = true
-    // boundary-raw-fetch: HEAD probe of the binary serve path, which has no contract
-    const probe = await fetch(src, { method: 'HEAD', cache: 'no-store' }).catch(() => null)
+    // boundary-raw-fetch: ranged probe of the binary serve path, which has no contract
+    const probe = await fetch(src, { headers: { Range: 'bytes=0-0' }, cache: 'no-store' }).catch(
+      () => null
+    )
+    await probe?.body?.cancel()
     if (probe?.status === 404) await reresolveRecord()
     setStatus(probe?.ok ? 'unplayable' : 'unavailable')
   }, [src, reresolveRecord])

apps/sim/lib/uploads/client/download.test.ts

@@ -149,7 +149,7 @@ describe('large file downloads', () => {
     await triggerFileDownload(video)
 
     expect(fetchMock).toHaveBeenCalledOnce()
-    expect(fetchMock.mock.calls[0]?.[1]).toMatchObject({ method: 'HEAD' })
+    expect(fetchMock.mock.calls[0]?.[1]).toMatchObject({ headers: { Range: 'bytes=0-0' } })
     expect(createObjectURL).not.toHaveBeenCalled()
     expect(downloadedHref).toMatch(/^\/api\/files\/serve\/workspace%2Fworkspace-1%2Fclip\.mp4\?/)
     expect(downloadedName).toBe('clip.mp4')
@@ -165,7 +165,15 @@ describe('large file downloads', () => {
   it('keeps fetching a file small enough to hold, so a rejection still surfaces as an error', async () => {
     await triggerFileDownload({ ...video, size: 1024 })
 
-    expect(fetchMock.mock.calls[0]?.[1]).not.toMatchObject({ method: 'HEAD' })
+    expect(fetchMock.mock.calls[0]?.[1]).not.toHaveProperty('headers')
+    expect(createObjectURL).toHaveBeenCalledOnce()
+  })
+
+  it('keeps fetching a large generated document, which must compile before it can be served', async () => {
+    await triggerFileDownload({ ...video, name: 'report.pdf', type: 'text/x-pdflibjs' })
+
+    expect(fetchMock).toHaveBeenCalledOnce()
+    expect(fetchMock.mock.calls[0]?.[1]).not.toHaveProperty('headers')
     expect(createObjectURL).toHaveBeenCalledOnce()
   })
 })

apps/sim/lib/uploads/client/download.ts

@@ -3,7 +3,7 @@ import { requestRaw } from '@/lib/api/client/request'
 import { downloadWorkspaceFileItemsContract } from '@/lib/api/contracts/workspace-file-folders'
 import { exportWorkspaceFileSnapshotContract } from '@/lib/api/contracts/workspace-files'
 import type { WorkspaceFileRecord } from '@/lib/uploads/contexts/workspace'
-import { isMarkdownFile } from '@/lib/uploads/utils/file'
+import { isMarkdownFile, needsRenderedArtifact } from '@/lib/uploads/utils/file'
 import { normalizeMimeType } from '@/lib/uploads/utils/mime'
 
 /** Action-time content from the mounted viewer, scoped so another file cannot consume it. */
@@ -14,11 +14,11 @@ export interface FileDownloadSource {
 }
 
 /**
- * Above this a file is handed to the browser's download manager instead of being read into the
- * page. Holding it as a `Blob` costs its whole size in tab memory and shows no progress until the
- * last byte lands; the download manager streams it to disk. Below it, fetching first keeps a
- * rejection — a document still compiling — on screen as an error rather than as a failed entry
- * in the browser's downloads list.
+ * Above this a stored file is handed to the browser's download manager instead of being read into
+ * the page. Holding it as a `Blob` costs its whole size in tab memory and shows no progress until
+ * the last byte lands; the download manager streams it to disk. Below it, and for a generated
+ * document, which must compile first, fetching keeps a rejection on screen as an error rather than
+ * as a failed entry in the browser's downloads list.
  */
 const IN_PAGE_DOWNLOAD_MAX_BYTES = 100 * 1024 * 1024
 
@@ -97,9 +97,16 @@ export async function triggerFileDownload(
     ? `/api/files/export/${encodeURIComponent(record.id)}`
     : `/api/files/serve/${encodeURIComponent(record.key)}?context=${record.storageContext ?? 'workspace'}&t=${Date.now()}`
 
-  if (!exportsMarkdown && record.size > IN_PAGE_DOWNLOAD_MAX_BYTES) {
-    // boundary-raw-fetch: HEAD probe of a binary path with no contract, so a refusal still surfaces here
-    const probe = await fetch(url, { method: 'HEAD', cache: 'no-store' })
+  if (
+    !exportsMarkdown &&
+    !needsRenderedArtifact(record.type, record.name) &&
+    record.size > IN_PAGE_DOWNLOAD_MAX_BYTES
+  ) {
+    // A first-byte read opens storage itself, so a missing object is refused here, in place,
+    // rather than as a failed entry in the browser's downloads list.
+    // boundary-raw-fetch: ranged probe of a binary path with no contract
+    const probe = await fetch(url, { headers: { Range: 'bytes=0-0' }, cache: 'no-store' })
+    await probe.body?.cancel()
     if (!probe.ok) throw new Error(`Failed to download "${record.name}"`)
     clickDownloadLink(url, record.name)
     return

apps/sim/lib/uploads/core/storage-service.ts

@@ -808,8 +808,8 @@ export async function headObject(
   const { UPLOAD_DIR_SERVER } = await getLocalStorageSetup()
   try {
     const file = await stat(join(UPLOAD_DIR_SERVER, sanitizeFileKey(key)))
-    // A local write replaces the file in place, so its modification time is what changes with it.
-    return { size: file.size, version: `${file.mtimeMs}` }
+    const { localObjectVersion } = await import('@/lib/uploads/upload-session/provider')
+    return { size: file.size, version: localObjectVersion(file) }
   } catch (error) {
     const code = (error as NodeJS.ErrnoException).code
     if (code === 'ENOENT') return null

apps/sim/lib/uploads/upload-session/provider.ts

@@ -790,7 +790,7 @@ async function headLocalObject(key: string): Promise<UploadObjectHead | null> {
     size: file.size,
     contentType: metadata.contentType,
     uploadId: metadata.uploadId,
-    version: localVersion(file),
+    version: localObjectVersion(file),
   }
 }
 
@@ -803,7 +803,7 @@ async function deleteLocalObjectVersion(key: string, version: string): Promise<v
     if ((error as NodeJS.ErrnoException).code === 'ENOENT') return
     throw error
   }
-  if (localVersion(current) !== version) return
+  if (localObjectVersion(current) !== version) return
   await unlink(path)
   await rm(localMetadataPath(key), { force: true })
   await rmdir(dirname(path)).catch((error: NodeJS.ErrnoException) => {
@@ -841,7 +841,11 @@ async function readLocalMetadata(path: string): Promise<LocalUploadMetadata> {
   }
 }
 
-function localVersion(file: Awaited<ReturnType<typeof stat>>): string {
+/**
+ * Version of a local upload object: what changes whenever its bytes do, including a same-size
+ * replacement within one timestamp tick.
+ */
+export function localObjectVersion(file: Awaited<ReturnType<typeof stat>>): string {
   return `${file.dev}:${file.ino}:${file.size}:${file.mtimeMs}`
 }