I needed to extract text out of PDF files about 3 years ago. I used an open source project named XPDF. I first shelled out a call to XPDF for the given file. It created a text file corresponding to the PDF file, closely following the format of the original.

This only works on *searchable* PDFs (text behind the image).