You should look at xpdf. It contains pdf2txt that converts pdf to text. This is used by the python tool pdfSearch that seems to come close to what you want.
I'm not sure this is useful - I already use pdf2txt in another context.
The problem is that I need to know the page number and offset (ie nth char) of the words to highlight. pdf2txt doesn't retain this information - it simply returns all the text in the PDF.
I know that pdf2txt only outputs the text. Absent another solution, though, that code may for the basis for a perl module you could write that would preserve the necessary information.