per page processing?

[Allison, Timothy B.]

All,
  Raymond Wu recently opened TIKA-1679 and recommended that we switch to 
per-page processing so that if there's an exception on one page, we'll still be 
able to extract contents from other pages.

  The proposed fix is along these lines:

             int nop = document.getNumberOfPages();
            for(int i=1;i<=nop;i++) {
                PDF2XHTML pdf2XHTML = new PDF2XHTML(handler, metadata,
                extractAnnotationText, enableAutoSpace,
                suppressDuplicateOverlappingText, sortByPosition);
                try {
                    pdf2XHTML.setStartPage(i);
                    pdf2XHTML.setEndPage(i);
                    pdf2XHTML.writeText(document, dummyWriter);
                } catch(Exception e) {
                    // TODO ...
                }

  Does this seem reasonable?  Any gut reaction/estimates on the performance 
hit?  Perhaps we should make this mode configurable?

Thank you.

             Best,

                        Tim