Fix: preserve non-ASCII filenames in document downloads (#9702)

src-ui/src/app/utils/http.spec.ts

@@ -0,0 +1,31 @@
+import { getFilenameFromContentDisposition } from './http'
+
+describe('getFilenameFromContentDisposition', () => {
+  it('should extract filename from Content-Disposition header with filename*', () => {
+    const header = "attachment; filename*=UTF-8''example%20file.txt"
+    expect(getFilenameFromContentDisposition(header)).toBe('example file.txt')
+  })
+
+  it('should extract filename from Content-Disposition header with filename=', () => {
+    const header = 'attachment; filename="example-file.txt"'
+    expect(getFilenameFromContentDisposition(header)).toBe('example-file.txt')
+  })
+
+  it('should prioritize filename* over filename if both are present', () => {
+    const header =
+      'attachment; filename="fallback.txt"; filename*=UTF-8\'\'preferred%20file.txt'
+    const result = getFilenameFromContentDisposition(header)
+    expect(result).toBe('preferred file.txt')
+  })
+
+  it('should gracefully fall back to null', () => {
+    // invalid UTF-8 sequence
+    expect(
+      getFilenameFromContentDisposition("attachment; filename*=UTF-8''%E0%A4%A")
+    ).toBeNull()
+    // missing filename
+    expect(getFilenameFromContentDisposition('attachment;')).toBeNull()
+    // empty header
+    expect(getFilenameFromContentDisposition(null)).toBeNull()
+  })
+})

src-ui/src/app/utils/http.ts

@@ -0,0 +1,23 @@
+export function getFilenameFromContentDisposition(header: string): string {
+  if (!header) {
+    return null
+  }
+
+  // Try filename* (RFC 5987)
+  const filenameStar = header.match(/filename\*=(?:UTF-\d['']*)?([^;]+)/i)
+  if (filenameStar?.[1]) {
+    try {
+      return decodeURIComponent(filenameStar[1])
+    } catch (e) {
+      // Ignore decoding errors and fall through
+    }
+  }
+
+  // Fallback to filename=
+  const filenameMatch = header.match(/filename="?([^"]+)"?/)
+  if (filenameMatch?.[1]) {
+    return filenameMatch[1]
+  }
+
+  return null
+}