sjvr767: I was going to work on this very idea, but you beat me to it
First, there are two very good projects which already implement this: pdfcrop
(the latter has a very good fork at pdfcrop2
). All of them have the same disadvantage: they detect the bounding box using ghostscript (which is very good and accurate) but then they don't update the PDF in-place: they re-create the PDF using pdftex or other software.
I'd already done a proof-of-concept that it worked using pyPdf [I've contributed to it in the past] but other projects (notably ebookutils) took my time
Would you be interested in taking it further using gs? The command line to generate a bbox is
gs -dBATCH -dSAFER -dNOPAUSE -dUseCropBox -sDEVICE=bbox <input.pdf>
You can capture the output using the subprocess module and then use it for setting the cropbox.
EDIT: just saw that you posted this much earlier. My apologies