For context, I'm a graduate student and I rely on the indexes of digitized books pretty heavily. It is annoying that the page numbers in djvu/pdf files almost always fail to align with those of the print text. One work around I've found is a bit brutal: chop the document into two pieces (one indexed by Roman numerals and one indexed by Arabic numerals) then permute the order of the two pieces. This works, but is unsatisfying and tedious to implement.
Is there a way to somehow designate what the first (Arabic numerally indexed) page of a pdf/djvu document is, allowing you to search the contents/introduction/acknowledgement sections using the Roman number indexing common in print text and leaving arabic numerals for searching the body of the text?