LaTeX to plain text for e.g. generation of statistics

I would like to convert a large LaTeX project (i.e. spanning multiple files) into plain text. The purpose is generation of statistics, so representing mathematics is not an issue. In fact, all mathematics is ideally ignored.

I have found http://code.google.com/p/textricks/ but could not get it to run. It seems unfinished, but is exactly what I am looking for otherwise.

-
(if anyone knows a good tag, feel free to edit the question) –  Ben Oct 17 '11 at 20:00
Done. plain-tex was definitely wrong, which is about plain-TeX not plain-text files. I added the conversion tag and created a new text-only tag (which might be renamed) –  Martin Scharrer Oct 17 '11 at 20:10
I would compile the document into a PDF and then use `pdftotext` to convert it to a text file. This ensures that you are using the LaTeX output not the input which might differ. Of course, if you want to do statistics about the LaTeX files and not the document generated by it, then you need to convert the source instead. –  Martin Scharrer Oct 17 '11 at 20:18
The UK Tex FAQ lists some options for this: tex.ac.uk/cgi-bin/texfaq2html?label=fmtconv+toascii –  Daniel Oct 17 '11 at 20:23
pdftotext is good enough for my purposes. I can post-process the file if needed. The question might still be interesting, though. Can it be done without going via PDF? –  Ben Oct 17 '11 at 20:27

I would compile the document into a PDF and then use `pdftotext` to convert it to a text file. You should disable all hyphenation and remove the page header (`\pagestyle{empty}`) to get only the raw text. This ensures that you are using the LaTeX output not the input which might differ.