I'm talking about a pdf document that was created via LaTeX (probably pdflatex) but the tex source of which is not available. I'd appreciate a solution that is less manual than trying to convert the pdf to tex and then changing the style (if that even works, the only answer there suggests it's far from functional).
edit with more readable I meaning changing the style in the original but unknown tex source such that I obtain a pdf that looks like e.g. a publication instead. This would of course in the ideal case include the pages renumbered etc. - so maybe it's impossible to achieve without texifying the pdf somehow first...
edit2 For those interested in this, a SU question about batch cropping pdfs mentions two tools that might be useful: briss, which crops borders (so that a two pages per page printing remains readable) and k2pdfopt which even rearranges a document's structure (untested).
pdftkandpdfcropare included in TeXLive and can do what you describe and more. But that's not what you originally asked for. All of that type of tool treats a page as a single entity and doesn't try to deconstruct it. – Andrew Stacey Oct 26 '11 at 09:10