fix(files): address PDF renderer edge cases

Author: Bill Leoutsakos

apps/sim/app/api/files/export/[id]/markdown-pdf.test.ts

@@ -120,6 +120,17 @@ ${repeatedParagraphs}`
     expect(tablePages.every((page) => page.includes('Name') && page.includes('Value'))).toBe(true)
   })
 
+  it('renders a table that contains only a header', async () => {
+    const buffer = await renderMarkdownPdf({
+      markdown: '| Name | Value |\n| --- | --- |',
+      title: 'Header-only table',
+    })
+
+    const text = (await pdfPagesText(buffer)).join(' ')
+    expect(text).toContain('Name')
+    expect(text).toContain('Value')
+  })
+
   it('falls back instead of decoding an image above the pixel ceiling', async () => {
     const oversizedSvg = Buffer.from(
       '<svg xmlns="http://www.w3.org/2000/svg" width="20000" height="20000"><rect width="100%" height="100%" fill="red"/></svg>'

apps/sim/app/api/files/export/[id]/markdown-pdf.tsx

@@ -267,19 +267,31 @@ function fontRuns(value: string): FontRun[] {
     }
   })
 
-  for (const [index, character] of characters.entries()) {
-    if (character.family !== 'Geist' || !character.neutral) continue
-
-    let previous = index - 1
-    while (previous >= 0 && characters[previous].neutral) previous -= 1
-    let next = index + 1
-    while (next < characters.length && characters[next].neutral) next += 1
-    const previousFamily = characters[previous]?.family
-    const nextFamily = characters[next]?.family
-    if (previousFamily && previousFamily !== 'Geist' && previousFamily === nextFamily) {
-      character.family = previousFamily
-    } else if (previousFamily && previousFamily !== 'Geist' && next >= characters.length) {
-      character.family = previousFamily
+  let index = 0
+  while (index < characters.length) {
+    if (!characters[index].neutral) {
+      index += 1
+      continue
+    }
+
+    const start = index
+    while (index < characters.length && characters[index].neutral) index += 1
+
+    const previousFamily = characters[start - 1]?.family
+    const nextFamily = characters[index]?.family
+    const inheritedFamily =
+      previousFamily &&
+      previousFamily !== 'Geist' &&
+      (previousFamily === nextFamily || index === characters.length)
+        ? previousFamily
+        : undefined
+
+    if (inheritedFamily) {
+      for (let neutralIndex = start; neutralIndex < index; neutralIndex += 1) {
+        if (characters[neutralIndex].family === 'Geist') {
+          characters[neutralIndex].family = inheritedFamily
+        }
+      }
     }
   }
 
@@ -489,6 +501,10 @@ function estimateTableRowHeight(row: JSONContent, columnCount: number): number {
 function chunkTableRows(header: JSONContent, rows: JSONContent[]): TableChunk[] {
   const columnCount = header.content?.length ?? rows[0]?.content?.length ?? 1
   const headerHeight = estimateTableRowHeight(header, columnCount)
+  if (rows.length === 0) {
+    return [{ rows: [], unbreakable: headerHeight <= MAX_UNBREAKABLE_TABLE_HEIGHT }]
+  }
+
   const chunks: TableChunk[] = []
   let current: JSONContent[] = []
   let currentHeight = headerHeight

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

@@ -25,7 +25,7 @@ import {
   type ResolvedEmbeddedFileRef,
   replaceEmbeddedFileRefs,
 } from '@/lib/uploads/utils/embedded-image-ref'
-import { formatFileSize } from '@/lib/uploads/utils/file-utils'
+import { formatFileSize, isMarkdownFile } from '@/lib/uploads/utils/file-utils'
 import { verifyFileAccess } from '@/app/api/files/authorization'
 import { encodeFilenameForHeader } from '@/app/api/files/utils'
 
@@ -54,15 +54,6 @@ const PDF_EXPORT_RATE_LIMIT: TokenBucketConfig = {
   refillIntervalMs: 60_000,
 }
 
-const MARKDOWN_MIME_TYPES = new Set(['text/markdown', 'text/x-markdown'])
-const MARKDOWN_EXTENSIONS = new Set(['md', 'markdown'])
-
-function isMarkdown(originalName: string, contentType: string): boolean {
-  if (MARKDOWN_MIME_TYPES.has(contentType)) return true
-  const ext = originalName.split('.').pop()?.toLowerCase() ?? ''
-  return MARKDOWN_EXTENSIONS.has(ext)
-}
-
 function safeFilename(name: string): string {
   return path
     .basename(name)
@@ -141,7 +132,7 @@ export const GET = withRouteHandler(
       )
     }
 
-    if (!isMarkdown(record.originalName, record.contentType)) {
+    if (!isMarkdownFile({ name: record.originalName, type: record.contentType })) {
       if (format === 'pdf') {
         return NextResponse.json(
           { error: 'PDF export is only available for Markdown files.' },

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

@@ -2,6 +2,7 @@ import { requestRaw } from '@/lib/api/client/request'
 import { fileExportContract } from '@/lib/api/contracts/storage-transfer'
 import { downloadWorkspaceFileItemsContract } from '@/lib/api/contracts/workspace-file-folders'
 import type { WorkspaceFileRecord } from '@/lib/uploads/contexts/workspace'
+import { isMarkdownFile } from '@/lib/uploads/utils/file-utils'
 
 export function saveBlob(blob: Blob, fileName: string): void {
   const objectUrl = URL.createObjectURL(blob)
@@ -35,10 +36,7 @@ export async function triggerFileDownload(
   record: WorkspaceFileRecord,
   options?: { format?: 'pdf' }
 ): Promise<void> {
-  const isMarkdown =
-    record.type === 'text/markdown' ||
-    record.type === 'text/x-markdown' ||
-    /\.(?:md|markdown)$/i.test(record.name)
+  const isMarkdown = isMarkdownFile(record)
 
   if (options?.format === 'pdf' && !isMarkdown) {
     throw new Error('PDF export is only available for Markdown files')

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

@@ -27,9 +27,10 @@ describe('isMarkdownFile', () => {
     expect(isMarkdownFile({ name: 'doc.markdown' })).toBe(true)
   })
 
-  it('is true for a text/markdown MIME even without a .md name', () => {
+  it('is true for Markdown MIME types even without a .md name', () => {
     expect(isMarkdownFile({ type: 'text/markdown', name: 'notes' })).toBe(true)
     expect(isMarkdownFile({ type: 'text/markdown', name: 'doc.txt' })).toBe(true)
+    expect(isMarkdownFile({ type: 'text/x-markdown', name: 'legacy' })).toBe(true)
   })
 
   it('is false for non-markdown files', () => {

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

@@ -213,12 +213,12 @@ export function getFileExtension(filename: string): string {
 /**
  * Whether a file renders in the collaborative rich markdown editor. Server-safe counterpart to the
  * client's `isMarkdownFile` (which uses `resolvePreviewType`): the editor treats a file as markdown by
- * its `text/markdown` MIME *or* a `.md`/`.markdown` extension — MIME first, matching the client — so a
- * `text/markdown` file with a non-`.md` name still counts. Used to gate server work (e.g. the live-doc
- * merge) to exactly the files that can be open in that editor.
+ * its Markdown MIME *or* a `.md`/`.markdown` extension — MIME first, matching the client — so a
+ * Markdown file with a non-`.md` name still counts. Used to gate server work (e.g. the live-doc merge)
+ * to exactly the files that can be open in that editor.
  */
 export function isMarkdownFile(file: { type?: string | null; name: string }): boolean {
-  if (file.type === 'text/markdown') return true
+  if (file.type === 'text/markdown' || file.type === 'text/x-markdown') return true
   const ext = getFileExtension(file.name)
   return ext === 'md' || ext === 'markdown'
 }