pdf2djvu creates DjVu files from PDF files.

It's able to extract:

 * graphics
 * text layer
 * hyperlinks
 * document outline (bookmarks)
 * metadata (including XMP metadata)