NousResearch/hermes-agent · warning · Error

Invalid PDF data URL type

Error message

Invalid PDF data URL type

What it means

Second guard in dataUrlToBlob: the metadata segment between 'data:' and the comma must declare media type 'application/pdf' and include a 'base64' parameter (case-insensitive after trimming). This error means the data URL is structurally valid but is not a base64-encoded PDF — e.g. an image/png data URL, a text/plain payload, or a PDF encoded without the base64 marker.

Source

Thrown at apps/desktop/src/app/chat/right-rail/preview-file.tsx:237

  return suspicious / Math.min(bytes.length, 4096) > 0.12
}

function dataUrlToBlob(dataUrl: string) {
  const comma = dataUrl.indexOf(',')

  if (comma < 0 || !dataUrl.startsWith('data:')) {
    throw new Error('Invalid PDF data URL')
  }

  const metadata = dataUrl
    .slice(5, comma)
    .split(';')
    .map(part => part.trim().toLowerCase())

  const payload = dataUrl.slice(comma + 1)

  if (metadata[0] !== 'application/pdf' || !metadata.slice(1).includes('base64')) {
    throw new Error('Invalid PDF data URL type')
  }

  let binary: string

  try {
    binary = atob(decodeURIComponent(payload))
  } catch {
    throw new Error('Invalid PDF data URL payload')
  }

  if (!binary.startsWith('%PDF-')) {
    throw new Error('Invalid PDF file header')
  }

  const bytes = new Uint8Array(binary.length)

  for (let index = 0; index < binary.length; index += 1) {
    bytes[index] = binary.charCodeAt(index)

View on GitHub (pinned to c896c09c42)

Solutions

  1. Detect the MIME from the data URL itself and dispatch to the matching preview (image viewer for image/*, text for text/*) instead of assuming PDF.
  2. Fix the payload producer to emit `data:application/pdf;base64,...` exactly.
  3. If the PDF arrives without base64 (raw percent-encoded), add an encoding branch instead of loosening this check.
  4. Validate attachment metadata against the actual data URL header before opening the preview.

Example fix

// before
if (metadata[0] !== 'application/pdf' || !metadata.slice(1).includes('base64')) {
  throw new Error('Invalid PDF data URL type')
}

// after — accept the alias some producers emit, keep the contract strict
const mime = metadata[0] === 'pdf' ? 'application/pdf' : metadata[0]
if (mime !== 'application/pdf' || !metadata.slice(1).includes('base64')) {
  throw new Error('Invalid PDF data URL type')
}
Defensive patterns

Strategy: type-guard

Validate before calling

const mime = dataUrl.slice(5, dataUrl.indexOf(','))
const kind = mime.split(';')[0]
switch (kind) {
  case 'application/pdf': openPdfPreview(dataUrl); break
  case 'image/png': case 'image/jpeg': openImagePreview(dataUrl); break
  default: openTextPreview(dataUrl)
}

Type guard

const isPdfDataUrl = (s: string): boolean => /^data:application\/pdf;base64,/i.test(s)

Try / catch

try { blob = dataUrlToBlob(url) } catch (e) { /* check message === 'Invalid PDF data URL type' -> route by MIME to another previewer */ }

Prevention

When it happens

Trigger: Passing a data URL whose MIME is image/*, text/*, or application/octet-stream into the PDF preview; a producer that emitted 'data:pdf;base64' (non-standard type) or omitted ';base64' and inlined raw bytes.

Common situations: Attachment kind detection guessing 'pdf' from a .pdf-ish filename while the actual payload is a PNG screenshot; generic preview code funnelling every data URL through the PDF converter; upstream models/tools emitting non-standard MIME tokens.

Related errors

AI-assisted analysis of NousResearch/hermes-agent@c896c09c42 (2026-08-14). Data as JSON: /api/errors/1c411ca91bf239ad. Report an issue: GitHub.