I made good experiences by using an external pdf2txt-converter and the parsing the output - but this of course depends on your input-document.

As a side note, if you go down this route, make absolutely certain that your external program will extract the text with some sort of X/Y position.

Unless you have full and complete control over the PDF and its generation, parsing PDF text by fixed position row/column is pretty much guaranteed to end in failure, frustration, and an absolutely massive nest of exceptions and special parsing cases