Function parseMultiformats

apps/sim/lib/pptx-renderer/utils/emf-parser.ts:133–166

* Parse MULTIFORMATS GDI comment — contains format descriptors pointing to embedded data.

(
  data: Uint8Array,
  view: DataView,
  offset: number,
  _recordSize: number
)

 * Parse MULTIFORMATS GDI comment — contains format descriptors pointing to embedded data.
 */
function parseMultiformats(
  data: Uint8Array,
  view: DataView,
  offset: number,
  _recordSize: number
): EmfContent | null {
  // Layout from record start:
  // +12: commentIdentifier(4), +16: publicCommentIdentifier(4)
  // +20: outputRect(16 = RECTL)
  // +36: countFormats(4)
  // +40: format descriptors array, each: { signature(4), version(4), cbData(4), offData(4) }
  if (offset + 40 > data.length) return null

  const countFormats = view.getUint32(offset + 36, true)
  const descriptorStart = offset + 40

  for (let i = 0; i < countFormats && i < 10; i++) {
    const descOff = descriptorStart + i * 16
    if (descOff + 16 > data.length) break

    const cbData = view.getUint32(descOff + 8, true)
    const offData = view.getUint32(descOff + 12, true)

    // offData is relative to the start of the record
    const dataStart = offset + offData
    if (dataStart + cbData > data.length || cbData === 0) continue

    const formatData = data.subarray(dataStart, dataStart + cbData)
    const pdf = extractPdfFromBuffer(formatData)
    if (pdf) return { type: 'pdf', data: pdf }
  }

  return null
}

/**
 * Search for %PDF...%%EOF in a buffer and extract the PDF bytes.