They OCR it, so with the PDF you get whatever text data they are able to extract from it (don't know about the encoding, but that is easy to convert), plus additional information.