improvement(files): extend the OOXML size limits to the client previews (#6445)

Author: waleedlatif1

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

@@ -5,6 +5,7 @@ import { cn } from '@sim/emcn'
 import { createLogger } from '@sim/logger'
 import { toError } from '@sim/utils/errors'
 import { sanitizeRenderedHyperlinks, stripEmbeddedFrames } from '@/lib/core/security/url-safety'
+import { assertOoxmlPreviewWithinLimits } from '@/lib/file-parsers/ooxml-preview-guard'
 import type { WorkspaceFileRecord } from '@/lib/uploads/contexts/workspace'
 import { PREVIEW_LOADING_OVERLAY, PreviewError, resolvePreviewError } from './preview-shared'
 import { PreviewToolbar } from './preview-toolbar'
@@ -195,16 +196,18 @@ export const DocxPreview = memo(function DocxPreview({
   useEffect(() => {
     if (!containerRef.current || !fileData) return
 
+    const data = fileData
     let cancelled = false
 
     async function render() {
       try {
         setRendering(true)
+        await assertOoxmlPreviewWithinLimits(data)
         const { renderAsync } = await import('docx-preview')
         if (cancelled || !containerRef.current) return
         setRenderError(null)
         containerRef.current.innerHTML = ''
-        await renderAsync(fileData, containerRef.current, undefined, {
+        await renderAsync(data, containerRef.current, undefined, {
           inWrapper: true,
           ignoreWidth: false,
           ignoreHeight: false,

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

@@ -5,6 +5,7 @@ import { Chip } from '@sim/emcn'
 import { createLogger } from '@sim/logger'
 import { toError } from '@sim/utils/errors'
 import type { WorkBook } from 'xlsx'
+import { assertOoxmlPreviewWithinLimits } from '@/lib/file-parsers/ooxml-preview-guard'
 import type { WorkspaceFileRecord } from '@/lib/uploads/contexts/workspace'
 import { DataTable } from './data-table'
 import { PreviewError, PreviewLoadingFrame, resolvePreviewError } from './preview-shared'
@@ -46,6 +47,7 @@ export const XlsxPreview = memo(function XlsxPreview({
     async function parse() {
       try {
         setRenderError(null)
+        await assertOoxmlPreviewWithinLimits(data)
         const XLSX = await import('xlsx')
         const workbook = XLSX.read(new Uint8Array(data), { type: 'array' })
         if (!cancelled) {

apps/sim/lib/file-parsers/ooxml-limits.ts

@@ -0,0 +1,23 @@
+/**
+ * Shared size ceilings for OOXML (docx/xlsx/pptx) archives, applied both by the
+ * server-side parse guard ({@link ./zip-guard}) and the browser preview guard
+ * ({@link ./ooxml-preview-guard}). The downstream parsers build a full in-memory
+ * DOM/object graph many times the XML size, so a small archive that expands past
+ * these bounds can exhaust the shared server process or the viewer's browser tab.
+ *
+ * This module is dependency-free and browser-safe on purpose, so both the Node
+ * guard and the client previews resolve the same numbers from one source.
+ */
+
+/** Hard ceiling on the summed declared uncompressed size of all entries. */
+export const MAX_OOXML_TOTAL_UNCOMPRESSED_BYTES = 150 * 1024 * 1024
+
+/** Hard ceiling on any single entry's declared uncompressed size. */
+export const MAX_OOXML_ENTRY_UNCOMPRESSED_BYTES = 64 * 1024 * 1024
+
+export class ZipBombError extends Error {
+  constructor(message: string) {
+    super(message)
+    this.name = 'ZipBombError'
+  }
+}

apps/sim/lib/file-parsers/ooxml-preview-guard.test.ts

@@ -0,0 +1,57 @@
+/**
+ * @vitest-environment node
+ */
+import JSZip from 'jszip'
+import { describe, expect, it } from 'vitest'
+import { ZipBombError } from '@/lib/file-parsers/ooxml-limits'
+import { assertOoxmlPreviewWithinLimits } from '@/lib/file-parsers/ooxml-preview-guard'
+
+async function buildZip(entries: Record<string, string>): Promise<ArrayBuffer> {
+  const zip = new JSZip()
+  for (const [name, content] of Object.entries(entries)) {
+    zip.file(name, content)
+  }
+  const buffer = await zip.generateAsync({ type: 'nodebuffer', compression: 'DEFLATE' })
+  return buffer.buffer.slice(buffer.byteOffset, buffer.byteOffset + buffer.byteLength)
+}
+
+const TINY_LIMITS = {
+  maxTotalUncompressedBytes: 1024 * 1024,
+  maxEntryUncompressedBytes: 100_000,
+}
+
+describe('assertOoxmlPreviewWithinLimits', () => {
+  it('accepts an archive within the limits', async () => {
+    const data = await buildZip({ 'word/document.xml': '<w:document/>' })
+    await expect(assertOoxmlPreviewWithinLimits(data, TINY_LIMITS)).resolves.toBeUndefined()
+  })
+
+  it('rejects an archive whose single part exceeds the per-entry limit', async () => {
+    const data = await buildZip({ 'word/document.xml': 'A'.repeat(200_000) })
+    await expect(assertOoxmlPreviewWithinLimits(data, TINY_LIMITS)).rejects.toBeInstanceOf(
+      ZipBombError
+    )
+  })
+
+  it('rejects an archive whose summed parts exceed the total limit', async () => {
+    const data = await buildZip({
+      'a.xml': 'A'.repeat(60_000),
+      'b.xml': 'B'.repeat(60_000),
+      'c.xml': 'C'.repeat(60_000),
+    })
+    await expect(
+      assertOoxmlPreviewWithinLimits(data, {
+        maxTotalUncompressedBytes: 100_000,
+        maxEntryUncompressedBytes: 1024 * 1024,
+      })
+    ).rejects.toBeInstanceOf(ZipBombError)
+  })
+
+  it('accepts an ordinary document under the shared default limits', async () => {
+    const data = await buildZip({
+      '[Content_Types].xml': '<?xml version="1.0"?><Types/>',
+      'word/document.xml': `<w:document>${'text '.repeat(5000)}</w:document>`,
+    })
+    await expect(assertOoxmlPreviewWithinLimits(data)).resolves.toBeUndefined()
+  })
+})

apps/sim/lib/file-parsers/ooxml-preview-guard.ts

@@ -0,0 +1,53 @@
+import type { JSZipObject } from 'jszip'
+import {
+  MAX_OOXML_ENTRY_UNCOMPRESSED_BYTES,
+  MAX_OOXML_TOTAL_UNCOMPRESSED_BYTES,
+  ZipBombError,
+} from '@/lib/file-parsers/ooxml-limits'
+
+export interface OoxmlPreviewLimits {
+  maxTotalUncompressedBytes: number
+  maxEntryUncompressedBytes: number
+}
+
+/**
+ * The central directory records each entry's declared uncompressed size, so
+ * `JSZip.loadAsync` exposes it without inflating anything.
+ */
+function readDeclaredUncompressedSize(entry: JSZipObject): number | undefined {
+  const data = (entry as JSZipObject & { _data?: { uncompressedSize?: number } })._data
+  const size = data?.uncompressedSize
+  return typeof size === 'number' && Number.isFinite(size) ? size : undefined
+}
+
+/**
+ * Reject an OOXML archive whose declared uncompressed size exceeds the shared
+ * OOXML bounds, before a browser preview (docx-preview, SheetJS) inflates it and
+ * exhausts the tab. `JSZip.loadAsync` reads the central directory without
+ * inflating any entry, so this inspects the declared sizes cheaply.
+ *
+ * Declared sizes are attacker-controlled, so a lying archive can still slip past;
+ * the browser has no bounded synchronous inflate to verify them the way the
+ * server guard does. This catches the honest bombs, the realistic preview case.
+ */
+export async function assertOoxmlPreviewWithinLimits(
+  data: ArrayBuffer | Uint8Array,
+  {
+    maxTotalUncompressedBytes = MAX_OOXML_TOTAL_UNCOMPRESSED_BYTES,
+    maxEntryUncompressedBytes = MAX_OOXML_ENTRY_UNCOMPRESSED_BYTES,
+  }: Partial<OoxmlPreviewLimits> = {}
+): Promise<void> {
+  const JSZip = (await import('jszip')).default
+  const zip = await JSZip.loadAsync(data)
+
+  let total = 0
+  for (const entry of Object.values(zip.files)) {
+    if (entry.dir) continue
+    const size = readDeclaredUncompressedSize(entry)
+    if (size === undefined) continue
+    total += size
+    if (size > maxEntryUncompressedBytes || total > maxTotalUncompressedBytes) {
+      throw new ZipBombError('This file is too large to preview safely')
+    }
+  }
+}

apps/sim/lib/file-parsers/zip-guard.test.ts

@@ -3,11 +3,8 @@
  */
 import JSZip from 'jszip'
 import { describe, expect, it } from 'vitest'
-import {
-  assertOoxmlArchiveWithinLimits,
-  type OoxmlSizeLimits,
-  ZipBombError,
-} from '@/lib/file-parsers/zip-guard'
+import { ZipBombError } from '@/lib/file-parsers/ooxml-limits'
+import { assertOoxmlArchiveWithinLimits, type OoxmlSizeLimits } from '@/lib/file-parsers/zip-guard'
 
 const HIGH_LIMITS: OoxmlSizeLimits = {
   maxTotalUncompressedBytes: 1024 * 1024 * 1024,

apps/sim/lib/file-parsers/zip-guard.ts

@@ -1,5 +1,10 @@
 import { inflateRawSync } from 'zlib'
 import { createLogger } from '@sim/logger'
+import {
+  MAX_OOXML_ENTRY_UNCOMPRESSED_BYTES,
+  MAX_OOXML_TOTAL_UNCOMPRESSED_BYTES,
+  ZipBombError,
+} from '@/lib/file-parsers/ooxml-limits'
 
 const logger = createLogger('ZipBombGuard')
 
@@ -50,8 +55,6 @@ export interface OoxmlSizeLimits {
 }
 
 const ONE_HUNDRED_MEBIBYTES = 100 * 1024 * 1024
-const ONE_HUNDRED_FIFTY_MEBIBYTES = 150 * 1024 * 1024
-const SIXTY_FOUR_MEBIBYTES = 64 * 1024 * 1024
 
 /**
  * The downstream parsers (mammoth, SheetJS, officeparser) build a full in-memory
@@ -60,22 +63,17 @@ const SIXTY_FOUR_MEBIBYTES = 64 * 1024 * 1024
  * parses a second time for HTML. The old 1 GiB ceiling let a ~3.5 MB archive
  * expand past what the process could hold and OOM it. The total and per-entry
  * caps here keep a single parse's peak within a modest container's budget while
- * still admitting all but pathologically large documents.
+ * still admitting all but pathologically large documents. The size ceilings are
+ * shared with the browser preview guard via {@link ./ooxml-limits}; the ratio
+ * heuristic is server-only.
  */
 export const DEFAULT_OOXML_SIZE_LIMITS: OoxmlSizeLimits = {
-  maxTotalUncompressedBytes: ONE_HUNDRED_FIFTY_MEBIBYTES,
-  maxEntryUncompressedBytes: SIXTY_FOUR_MEBIBYTES,
+  maxTotalUncompressedBytes: MAX_OOXML_TOTAL_UNCOMPRESSED_BYTES,
+  maxEntryUncompressedBytes: MAX_OOXML_ENTRY_UNCOMPRESSED_BYTES,
   maxCompressionRatio: 150,
   ratioCheckFloorBytes: ONE_HUNDRED_MEBIBYTES,
 }
 
-export class ZipBombError extends Error {
-  constructor(message: string) {
-    super(message)
-    this.name = 'ZipBombError'
-  }
-}
-
 /**
  * Whether the buffer is shaped like a ZIP archive — i.e. begins with a local
  * file header (the leading signature of every non-empty ZIP, and thus every