Is it possible to redefine the white space command?

Question

I would like to replace the white space (full space entered by pressing "space" any number of times on the keyboard or by hitting enter) by something else. Let's say, a rule of the same width.

Is there some command I could redefine? The easiest way in order to keep the stretching of the white spaces would be to highlight them.

I already know that I can get something like this with the package lua-visual-debug and debugging is the purpose for that redefinition, but I would like to do that in pdfLaTeX.

If we can find a simple answer to this, we might get some reference for debugging "spurious white-space errors".

Of course, I am giving an MWE for this. The first line should result in the second line after inserting the redefinition (hopefully not with hard-coded width of the rules.)

% arara: pdflatex

\documentclass{article}
\usepackage{xcolor}
\definecolor{spacecolor}{named}{red}
\newcommand*{\test}{%
    Test}
\newcommand{\yatest}{
    Test
    }
\newcommand*{\here}{\textcolor{spacecolor}{\rule{.335em}{1.5ex}}}
\usepackage{lua-visual-debug}

\begin{document}
Test Test    Test \test\yatest\test 

Test\here{}Test\here{}Test\here{}Test\here{}Test\here{}Test
\end{document}

Answer 1

Borrowing code by Marcin Woliński at "How to make a box disappear at a line break", already used for my answer at Check if at begin of a line

\documentclass{article}
\newcommand{\AND}{%
  \leaders\hrule height 1.5ex
  \hskip\fontdimen2\font plus \fontdimen3\font minus \fontdimen4\font
  }

\newenvironment{showspaces}
 {\par\obeyspaces\obeylines
  \begingroup\lccode`\~=`\ \lowercase{\endgroup\let~}\AND
  \begingroup\lccode`\~=`\^^M\lowercase{\endgroup\let~}\AND}
 {\par}

\begin{document}

\begin{showspaces}%
some text some text someveryverylongword text text
some text some text someveryverylongword text text
some text some text someveryverylongword text text
some text some text someveryverylongword text text
some text some text someveryverylongword text text
\end{showspaces}

\end{document}

Answer 2

The new feature \pdfinterwordspaceon of pdfTeX 1.40.15 can be exploited to get a tiny visible space, where TeX would put an invisible space. The feature has the purpose to have a real space character in the output instead of pure white space by moving to the next character position. By changing the font to cmtt10, which has a visible space at the space position:

\pdfmapline{=dummy-space <cmtt10.pfb}
\pdfglyphtounicode{space}{0042}

\documentclass{article}

\newcommand{\foo}{Some text with }

\begin{document}
\pdfinterwordspaceon
\foo faked interword spaces.
\end{document}

Disadvantages:

• The character is scaled so tiny, that it is nearly invisible, even with an visible character.

• Several spaces in a row are merged to one space character.

• Even worse, the space character additions can also happen for \hskip or \kern commands.

Advantage:

• At least this method gets some useful visual feedback on unwanted spaces hidden in macros, where category code changes have too much side effects.

At least, the appearance of the visible space could be improved by post-processing the PDF file.

Answer 3

Knuth's CWEB program inserts visible space characters automatically for string constants in a C program, as this example shows (save as space.w, compile as cweave space && pdftex space).

@ \.{CWEB}, show me how you format spaces in string constants.

@p
#include <stdio.h>
static const char greeting[] = "H e l l o  ,   w o r l d !\n";
int main(void)
{
    printf("%s", greeting);
    return(0);
}

This behavior is defined in cwebmac.tex, and I have modified only the part that shows spaces for LaTeX in this example:

\documentclass{article}
\newcommand{\SP}{\texttt{\char`\ }} % (visible) space in a string
\newcommand{\showspaces}[1]{\begingroup\let\ =\SP #1\endgroup}
\begin{document}

Test\SP Test\SP\SP % method one

\showspaces{Test\ Test\ \ } % method two

\end{document}

CWEB does this automatically by pre-processing the input .w file, which only contains "H e l l o" with spaces, and producing a .tex file which contains H\ e\ l\ l\ o. Especially if this is for testing purposes, perhaps a preprocessor approach might work for you as well. For example, a fairly simple C program could read a .tex file in one character at a time and every time it reads a space it could substitute \SP.

Something like this could be the core algorithm:

if ((c = fgetc(infile)) == ' ') 
    fprintf(outfile, "\\SP");
else fputc(c, outfile);