PDF to formatted text is an imperfect science, as the source PDF has no layout information (margins, word breaks, line spacing, paragraph starts, etc.) so formatting has to be calculated and extrapolated from the physical location of text in the page space. In other words, it's hard to recompute the layout in a way that works for every possible type of PDF document, and more imporantly, it depends highly on the software that was used to create the PDF (for example, Latex strips out all inter-word spacing).

Trapeze works perfectly on many PDFs, but poorly in some (though in the majority of our test cases, it does a fine job). If you find that Trapeze is not working for your files, the best thing you can do is send us some samples. It's the only way we can trace what is happening in the layout engine. Thanks!