fix(files): share one served-bytes predicate between the serve route and large downloads

apps/sim/app/api/files/serve/[...path]/route.test.ts

@@ -571,17 +571,53 @@ describe('File Serve API Route', () => {
     describe('outside the workspace-file model', () => {
       const key = 'workspace/ws/1790000000000-0123456789abcdef-clip.mp4'
 
-      function serveAttachment(headers: Record<string, string> = {}) {
+      function serveAttachment(headers: Record<string, string> = {}, objectKey = key) {
         mockIsUsingCloudStorage.mockReturnValue(true)
         storageServiceMockFns.mockDownloadFileStream.mockImplementation(async () =>
           Readable.from([Buffer.from('attachment')])
         )
         return GET(
-          new NextRequest(`http://localhost:3000/api/files/serve/${key}`, { headers }),
-          createRouteContext({ path: key.split('/') })
+          new NextRequest(`http://localhost:3000/api/files/serve/${objectKey}`, { headers }),
+          createRouteContext({ path: objectKey.split('/') })
         )
       }
 
+      describe('a document name', () => {
+        const documentKey = 'workspace/ws/1790000000000-0123456789abcdef-scan.pdf'
+
+        beforeEach(() => {
+          storageServiceMockFns.mockHeadObject.mockResolvedValue({ size: 10, version: '"v1"' })
+          storageServiceMockFns.mockDownloadFile.mockResolvedValue(Buffer.from('rendered'))
+        })
+
+        it('streams when its file record says it is a finished upload', async () => {
+          uploadsMetadataMockFns.mockGetFileMetadataByKey.mockResolvedValue({
+            key: documentKey,
+            contentType: 'application/pdf',
+          })
+
+          const response = await serveAttachment({}, documentKey)
+
+          expect(response.status).toBe(200)
+          expect(storageServiceMockFns.mockDownloadFileStream).toHaveBeenCalled()
+          expect(storageServiceMockFns.mockDownloadFile).not.toHaveBeenCalled()
+        })
+
+        it.each([
+          ['its record says it is a generation source', { contentType: 'text/x-pdflibjs' }],
+          ['it has no file record', null],
+        ])('stays buffered when %s', async (_label, record) => {
+          uploadsMetadataMockFns.mockGetFileMetadataByKey.mockResolvedValue(
+            record && { key: documentKey, ...record }
+          )
+
+          await serveAttachment({}, documentKey)
+
+          expect(storageServiceMockFns.mockDownloadFileStream).not.toHaveBeenCalled()
+          expect(storageServiceMockFns.mockDownloadFile).toHaveBeenCalled()
+        })
+      })
+
       it('streams a range of an attachment sized by storage, never buffering it', async () => {
         storageServiceMockFns.mockHeadObject.mockResolvedValue({
           size: 500_000_000,

apps/sim/app/api/files/serve/[...path]/route.ts

@@ -28,20 +28,22 @@ import {
   workspaceFileCacheControl,
 } from '@/lib/uploads/server/delivery'
 import { resolveServableImageBytes } from '@/lib/uploads/server/image-derivative'
-import { resolveStoredFileContext } from '@/lib/uploads/server/metadata'
+import { getFileMetadataByKey, resolveStoredFileContext } from '@/lib/uploads/server/metadata'
 import { MAX_BUFFERED_TRANSFER_BYTES } from '@/lib/uploads/shared/types'
 import {
   inferContextFromKey,
-  needsRenderedArtifact,
+  isRenderableDocumentName,
   resolveEffectiveMimeType,
 } from '@/lib/uploads/utils/file'
 import { internalWorkspaceFileServeAuth } from '@/lib/workspace-files/api'
 import { fetchAuthorizedServableWorkspaceFileBuffer } from '@/lib/workspace-files/application/fetch-servable-workspace-file-buffer'
 import { finishFileDelivery } from '@/lib/workspace-files/application/finish-file-delivery'
 import { readWorkspaceFileRecordByKey } from '@/lib/workspace-files/application/read-workspace-file-record-by-key'
 import { streamAuthorizedWorkspaceFile } from '@/lib/workspace-files/application/stream-authorized-workspace-file'
-import { isSimPageSource, SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-compile'
+import { isSimPageSource } from '@/lib/workspace-files/page-compile'
+import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-content-type'
 import { renderSimPageDocumentWithAssets } from '@/lib/workspace-files/page-document.server'
+import { servesStoredBytes } from '@/lib/workspace-files/stored-bytes'
 import { type KnowledgeFileAccess, verifyFileAccess } from '@/app/api/files/authorization'
 import {
   createConditionalFileResponse,
@@ -203,10 +205,10 @@ async function resolveTransformedBytes(params: {
  * Whether a request is answered with the stored object byte-for-byte, so its bytes can stream from
  * storage instead of being read whole and capped at the transfer ceiling.
  *
- * Every other answer is produced from the bytes and stays buffered: a Sim page renders, a generated
- * document compiles, and a `preview` may be a HEIC transcode — sniffed from the content, so no name
- * or type rules it out. What streams is exactly what {@link resolveServableBytes} and the
- * workspace resolver would have passed through unchanged.
+ * Every other answer is produced from the bytes and stays buffered: see {@link servesStoredBytes},
+ * plus a `preview`, which may be a HEIC transcode — sniffed from the content, so no name or type
+ * rules it out. What streams is exactly what {@link resolveServableBytes} and the workspace
+ * resolver would have passed through unchanged.
  */
 function servesStoredObject(
   filename: string,
@@ -215,12 +217,7 @@ function servesStoredObject(
 ): boolean {
   if (options.raw) return true
   if (options.preview) return false
-  if (recordedType === SIM_PAGE_CONTENT_TYPE || filename.toLowerCase().endsWith('.html')) {
-    return false
-  }
-  // A recorded type tells a finished upload from a generation source; without one, every
-  // document name may be a source.
-  return !needsRenderedArtifact(recordedType, filename)
+  return servesStoredBytes(recordedType, filename)
 }
 
 /**
@@ -231,11 +228,26 @@ function storedObjectContentType(filename: string, options: ServeOptions): strin
   return options.raw ? getContentType(filename) : resolveEffectiveMimeType(undefined, filename)
 }
 
+/**
+ * The type recorded for an object outside the workspace-file model, read only where it decides the
+ * answer: a document name may be a finished upload or a generation source, and the object's file
+ * record, where it has one, says which. An object with no record is judged by its name.
+ */
+async function recordedDocumentType(
+  key: string,
+  context: StorageContext,
+  filename: string,
+  options: ServeOptions
+): Promise<string | undefined> {
+  if (options.raw || options.preview || !isRenderableDocumentName(filename)) return undefined
+  return (await getFileMetadataByKey(key, context))?.contentType
+}
+
 /**
  * Streams a stored object outside the workspace-file model (execution outputs, chat and copilot
  * attachments, knowledge documents) when the request is answered with it as-is, or returns null
- * for the buffered path. The object is its only record, so storage is asked for its length and
- * version, and the version is what proves a revalidation still holds.
+ * for the buffered path. Storage is asked for the object's length and version, and the version is
+ * what proves a revalidation still holds.
  */
 async function streamStoredObject(params: {
   key: string
@@ -245,7 +257,8 @@ async function streamStoredObject(params: {
   request: NextRequest
 }): Promise<NextResponse | null> {
   const { key, context, filename, options } = params
-  if (!servesStoredObject(filename, undefined, options)) return null
+  const recordedType = await recordedDocumentType(key, context, filename, options)
+  if (!servesStoredObject(filename, recordedType, options)) return null
   const stored = await headObject(key, context)
   if (!stored) throw new FileNotFoundError(`File not found: ${key}`)
 

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

@@ -27,7 +27,8 @@ import {
 import type { FileDownloadSource } from '@/lib/uploads/client/download'
 import type { WorkspaceFileRecord } from '@/lib/uploads/contexts/workspace'
 import { getFileExtension } from '@/lib/uploads/utils/file'
-import { isSimPageSource, SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-compile'
+import { isSimPageSource } from '@/lib/workspace-files/page-compile'
+import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-content-type'
 import { EditorContextMenu } from '@/app/workspace/[workspaceId]/files/components/file-viewer/editor-context-menu'
 import { FileSaveConflict } from '@/app/workspace/[workspaceId]/files/components/file-viewer/file-save-conflict'
 import type { PreviewMode } from '@/app/workspace/[workspaceId]/files/components/file-viewer/file-viewer'

apps/sim/app/workspace/[workspaceId]/files/files.tsx

@@ -48,7 +48,7 @@ import {
   isFileBrowserSize,
   isFileBrowserType,
 } from '@/lib/workspace-files/browser'
-import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-compile'
+import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-content-type'
 import { FindBar } from '@/app/workspace/[workspaceId]/components/find-bar/find-bar'
 import { useFindShortcut } from '@/app/workspace/[workspaceId]/components/find-bar/use-find-shortcut'
 import type { MoveOptionNode } from '@/app/workspace/[workspaceId]/components/folders'

apps/sim/app/workspace/[workspaceId]/home/components/mothership-view/mothership-view.tsx

@@ -11,7 +11,7 @@ import {
 } from '@/lib/mothership/resources/types'
 import type { FileDownloadSource } from '@/lib/uploads/client/download'
 import { getFileExtension } from '@/lib/uploads/utils/file'
-import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-compile'
+import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-content-type'
 import type { PreviewMode } from '@/app/workspace/[workspaceId]/files/components/file-viewer'
 import {
   isCsvStreamOnly,

apps/sim/lib/mothership/tools/server/files/edit-content.ts

@@ -21,8 +21,8 @@ import {
   HAND_WRITTEN_PAGE_MESSAGE,
   isHandWrittenCompiledPage,
   isSimPageSource,
-  SIM_PAGE_CONTENT_TYPE,
 } from '@/lib/workspace-files/page-compile'
+import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-content-type'
 import { buildEmbeddedImageRefWarning } from './embedded-image-refs'
 import { waitForLatestFileIntent } from './intent-store'
 import { compileDocForWrite, getDocumentFormatInfo, inferContentType } from './workspace-file'

apps/sim/lib/projects/files/application/artifacts.ts

@@ -56,11 +56,8 @@ import {
   textExtractionUnsupportedError,
 } from '@/lib/workspace-files/application/read-workspace-file-text'
 import { fileOwnerCondition } from '@/lib/workspace-files/ownership-query'
-import {
-  isSimPageSource,
-  ProjectPageForeignEmbedError,
-  SIM_PAGE_CONTENT_TYPE,
-} from '@/lib/workspace-files/page-compile'
+import { isSimPageSource, ProjectPageForeignEmbedError } from '@/lib/workspace-files/page-compile'
+import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-content-type'
 import { renderSimPageDocument } from '@/lib/workspace-files/page-document'
 import {
   collectSimPageFileReferences,

apps/sim/lib/projects/files/application/content.ts

@@ -43,7 +43,7 @@ import {
 import { parseWorkspaceFileRevision } from '@/lib/workspace-files/application/file-revision'
 import { resolveWorkspaceFileVersionWrite } from '@/lib/workspace-files/application/file-version-write'
 import { MAX_WORKSPACE_FILE_CONTENT_BYTES } from '@/lib/workspace-files/orchestration'
-import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-compile'
+import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-content-type'
 import { restoreSimPageSourceBuffer } from '@/lib/workspace-files/page-source-embed'
 
 interface ContentUseCaseArgs<I, P = undefined> {

apps/sim/lib/projects/files/application/downloads.ts

@@ -59,7 +59,7 @@ import {
 } from '@/lib/workspace-files/download-selection'
 import { MAX_ZIP_DOWNLOAD_BYTES, MAX_ZIP_DOWNLOAD_FILES } from '@/lib/workspace-files/limits'
 import { fileOwnerCondition } from '@/lib/workspace-files/ownership-query'
-import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-compile'
+import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-content-type'
 import {
   createFileReadReceipt,
   type FileReadReceipt,

apps/sim/lib/projects/files/application/uploads.ts

@@ -56,7 +56,7 @@ import {
   type UploadSessionRecord,
 } from '@/lib/uploads/upload-session/service'
 import type { WorkspaceFileUploadSource } from '@/lib/uploads/upload-session/workspace-file-provenance'
-import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-compile'
+import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-content-type'
 import {
   MAX_SIM_PAGE_UPLOAD_SNIFF_BYTES,
   restoreSimPageSourceBuffer,

apps/sim/lib/public-shares/application/read.ts

@@ -37,11 +37,8 @@ import { resolveEffectiveMimeType } from '@/lib/uploads/utils/file'
 import { sniffImageContentType } from '@/lib/uploads/utils/validation'
 import { reportWorkspaceFileDelivery } from '@/lib/workspace-files/application/file-delivery-observer'
 import { fileOwnerCondition } from '@/lib/workspace-files/ownership-query'
-import {
-  isSimPageSource,
-  ProjectPageForeignEmbedError,
-  SIM_PAGE_CONTENT_TYPE,
-} from '@/lib/workspace-files/page-compile'
+import { isSimPageSource, ProjectPageForeignEmbedError } from '@/lib/workspace-files/page-compile'
+import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-content-type'
 import { renderSimPageDocument } from '@/lib/workspace-files/page-document'
 import {
   collectSimPageFileReferences,

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

@@ -169,11 +169,18 @@ describe('large file downloads', () => {
     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()
-  })
+  it.each([
+    ['a generated document', { name: 'report.pdf', type: 'text/x-pdflibjs' }],
+    ['a page recorded as one', { name: 'guide', type: 'text/x-sim-page' }],
+    ['an .html page', { name: 'guide.html', type: 'text/html' }],
+  ])(
+    'keeps fetching %s, which the server renders before it can be served',
+    async (_label, kind) => {
+      await triggerFileDownload({ ...video, ...kind })
+
+      expect(fetchMock).toHaveBeenCalledOnce()
+      expect(fetchMock.mock.calls[0]?.[1]).not.toHaveProperty('headers')
+      expect(createObjectURL).toHaveBeenCalledOnce()
+    }
+  )
 })

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

@@ -3,8 +3,9 @@ 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, needsRenderedArtifact } from '@/lib/uploads/utils/file'
+import { isMarkdownFile } from '@/lib/uploads/utils/file'
 import { normalizeMimeType } from '@/lib/uploads/utils/mime'
+import { servesStoredBytes } from '@/lib/workspace-files/stored-bytes'
 
 /** Action-time content from the mounted viewer, scoped so another file cannot consume it. */
 export interface FileDownloadSource {
@@ -16,9 +17,9 @@ export interface FileDownloadSource {
 /**
  * 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.
+ * the last byte lands; the download manager streams it to disk. Below it, and for a file the
+ * server renders first (a page, a generated document), fetching keeps a rejection on screen
+ * rather than as a failed entry in the browser's downloads list.
  */
 const IN_PAGE_DOWNLOAD_MAX_BYTES = 100 * 1024 * 1024
 
@@ -99,7 +100,7 @@ export async function triggerFileDownload(
 
   if (
     !exportsMarkdown &&
-    !needsRenderedArtifact(record.type, record.name) &&
+    servesStoredBytes(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,

apps/sim/lib/uploads/contexts/workspace/file-manager-page.test.ts

@@ -20,7 +20,8 @@ vi.mock('@/lib/uploads/contexts/workspace/file-folder-manager', () => workspaceF
 
 import { createWorkspaceFileBodySchema } from '@/lib/api/contracts/workspace-files'
 import { uploadWorkspaceFile } from '@/lib/uploads/contexts/workspace/file-manager'
-import { compileSimPage, SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-compile'
+import { compileSimPage } from '@/lib/workspace-files/page-compile'
+import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-content-type'
 
 workspaceFileFoldersMockFns.mockBuildWorkspaceFileFolderPathMap.mockImplementation(() => new Map())
 workspaceFileFoldersMockFns.mockNormalizeWorkspaceFileItemName.mockImplementation(

apps/sim/lib/uploads/contexts/workspace/file-manager.ts

@@ -161,7 +161,7 @@ import {
   resolveFileOwner,
 } from '@/lib/workspace-files/ownership'
 import { fileFolderOwnerCondition, fileOwnerCondition } from '@/lib/workspace-files/ownership-query'
-import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-compile'
+import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-content-type'
 import {
   MAX_SIM_PAGE_UPLOAD_SNIFF_BYTES,
   restoreSimPageSourceBuffer,

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

@@ -35,7 +35,8 @@ import {
   type RawFileInput,
   resolveTrustedFileContext,
 } from '@/lib/uploads/utils/file'
-import { isSimPageSource, SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-compile'
+import { isSimPageSource } from '@/lib/workspace-files/page-compile'
+import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-content-type'
 import { renderSimPageDocumentWithContributors } from '@/lib/workspace-files/page-document.server'
 import { type KnowledgeFileAccess, verifyFileAccess } from '@/app/api/files/authorization'
 import type { UserFile } from '@/executor/types'

apps/sim/lib/workspace-files/application/stream-authorized-workspace-file.ts

@@ -11,6 +11,10 @@ import { createFileReadReceipt } from '@/lib/workspace-files/read-receipt'
  * Opens an already-authorized workspace file's stored bytes for delivery, whole or as one inclusive
  * span, without reading them into memory.
  *
+ * Callers must already have authorized and admitted the file (`'opaque'`) through their own use
+ * case: storage is opened before this function's own recheck, which only decides whether the
+ * opened stream is handed out.
+ *
  * Access is rechecked once storage has opened and before the stream is handed out, as every
  * streamed read does: a revocation, a content write, or a provenance change landing in between
  * disposes the stream instead of publishing it. `authorize` is the caller's own use-case check, so

apps/sim/lib/workspace-files/archive-extraction.ts

@@ -11,7 +11,7 @@ import {
 import type { WorkspaceFileSecretProvenance } from '@/lib/uploads/contexts/workspace/file-secret-provenance'
 import { getFileExtension, getMimeTypeFromExtension } from '@/lib/uploads/utils/file'
 import type { EditableFileOwner } from '@/lib/workspace-files/ownership'
-import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-compile'
+import { SIM_PAGE_CONTENT_TYPE } from '@/lib/workspace-files/page-content-type'
 import { restoreSimPageSourceBuffer } from '@/lib/workspace-files/page-source-embed'
 
 interface StagedArchiveEntry {

apps/sim/lib/workspace-files/page-compile.ts

@@ -36,15 +36,6 @@ import type { EditableFileOwner } from '@/lib/workspace-files/ownership'
  */
 export const SIM_PAGE_MARKER = '<!--sim-page-->'
 
-/**
- * The INTERNAL content type stamped onto a page file's record when the write
- * path detects page source. The file stays `.html` to the user (serving and
- * downloads emit text/html); the record type is how surfaces know the file's
- * nature before any content loads — force the rendered view, hide the code
- * toggle, expect source.
- */
-export const SIM_PAGE_CONTENT_TYPE = 'text/x-sim-page'
-
 const frontmatterSchema = z.object({
   title: z.string().min(1),
   /** Tolerated for old sources; no longer rendered — the docs have no eyebrow. */

apps/sim/lib/workspace-files/page-content-type.ts

@@ -0,0 +1,11 @@
+/**
+ * The INTERNAL content type stamped onto a page file's record when the write
+ * path detects page source. The file stays `.html` to the user (serving and
+ * downloads emit text/html); the record type is how surfaces know the file's
+ * nature before any content loads — force the rendered view, hide the code
+ * toggle, expect source.
+ *
+ * Kept free of the page compiler so client surfaces that only need to recognize a page
+ * do not load the compiler's parsers.
+ */
+export const SIM_PAGE_CONTENT_TYPE = 'text/x-sim-page'