[NTG-context] counting the words in a TeX document

[gnwiii at gmail.com]

On 8/5/06, Mojca Miklavec <mojca.miklavec.lists at gmail.com> wrote:
> Hello,
>
> I would like to ask how difficult it would be to count the number of
> words in a TeX/ConTeXt document. If it's too complex, please ignore
> the rest of the message.

It wasn't too complex for Michael Downes using LaTeX:

\ProvidesFile{wordcount.tex}[2000/09/27 v1.5 Michael Downes]
% Copyright 2000 Michael John Downes
% This file has no restrictions on its use, distribution, or sale.
%
% If you run LaTeX on wordcount.tex it will prompt you for the name of a
% document to be counted. For most people, however, it will be more
% convenient to run the shell script wordcount.sh, giving the document
% name as the first argument. The comments in wordcount.sh
% give further information about the usage and limitations of this tool.

% The fundamental idea is to mark each character and interword space
% with a unique tag that will show up in TeX "showbox" output. Then
% arrange to make the output routine trigger a TeX overfull vbox message
% for the page box so that everything gets reported in the TeX log.
% Then run grep -c (or an equivalent text search utility, e.g., perl) on
% the log file to count the occurrences.
% [....]

> Most recipes for LaTeX say that it's best to do something like
> "pdftotext" and then issue "wc" to count the words in the resulting
> text file, but windows users don't have "wc" and sometimes you only
> need to know the length of the abstract or so ...

Many GNU utilities have been ported (GNUWin32), or can be implemented
in perl/ruby which context uses anyway.

> Some time ago Hans mentioned that he counts the number of appearance
> of single charactres, but I don't know how difficult it would be to
> extend it to count the number of words.
>
> The problem is not that well defined (how to handle equations, some
> would probably want to exclude headers, footers, buttons, ...), but it
> only needs to be an approximation and "backward compatibility" (in the
> sense that counter would have to result in the same number after some
> years) is not needed at all since algorithms might improve with time
> and the resulting document doesn't really depend on that number, it
> would only be written to the log file.
>
> My idea for the interface would be something like
>
> \startwordcount[abstract]
> \startframedtext
> Bla bla.
> \stopframedtext
> \stopwordcount
>
> which would write something like "abstract: 2 words" to the log file
>
> or
>
> \startstatistics[abstract][words]
> \startframedtext
> Bla bla.
> \stopframedtext
> \stopstatistics
>
> But this is really a low priority. I'm currently using Acrobat to copy
> the text, then I paste it into Office and take a look at statistics
> there when I need to obey some limitations.
>
> So, if there's a simple solution, I would be glad to use it, but if it
> takes too much time to implement it, it's probably not worth the
> effort.

ConTeXt already analyzes the "scratch" files with perl or ruby, so if
you can adapt MD's idea it shouldn't be a big deal to have texexec
print the result.

-- 
George N. White III <aa056 at chebucto.ns.ca>
Head of St. Margarets Bay, Nova Scotia