Cfrgtkky

I have the following situation: I want to use the argument of newcommand to be the name of a new variable, which I define within newcommand using def.

When I try to compile it always says that my newly defined variable doesn't exist. I think the problem is the "#"-symbol of the argument within the def command.

documentclass{article}

newcounter{src}

newcommand{src}[1]{stepcounter{src}def #1 {arabic{src}} [arabic{src}]}

begin{document}

src{testone}

src{testtwo}

first test: testone

second test: testtwo

end{document}

My expected output would be:

1
2
first test: 1
second test: 2

But instead I get:

! Undefined control sequence. testone

! Undefined control sequence. testtwo

Does anyone have a solution for how to use the # for a variable name?

Thank you in advance

The construction of command sequence names can be done with

csname #1endcsname

but this is not sufficient for def, it must be prepended with expandafter, to expand the sequence name first, then use def (the same holds for edef etc.)

i.e.:

expandafterdefcsname #1endcsname{some expansion stuff}

will expand to deffoo{some expansion stuff} if #1 contains foo.

I used edef here to use the counter number at the time of definition of the macro name, otherwise arabic{src} would print the current number.

documentclass{article}

newcounter{src}



newcommand{src}[1]{stepcounter{src}expandafteredefcsname#1endcsname {arabic{src}} [arabic{src}]}

begin{document}

src{testone}



src{testtwo}



first test: testone



second test: testtwo

end{document}

Christian Hupfer has well explained what happens. You might enjoy a different approach based on expl3.

The main function used is cs_new:cpx, which is a variant of cs_new:Npn.

1. 
   cs stands for “control sequence”


2. 
   new is self explaining


3. 
   : separates the function name from the signature, that is, the list of its arguments


4. 
   N stands for a single token


5. 
   p stands for “parameter text” (in the example case below nothing)


6. 
   n stands for “braced argument” (the replacement text)

The variant is cs_new:cpx where c means that we expect a braced argument which will be transformed into a single symbolic token with its backslash at the beginning; x instead means that the corresponding argument will be fully expanded. We need full expansion because we want to store the current value of the counter, not the instructions to produce the value.

The other functions are int_new:N that allocates a new variable of type int (integer) and int_gincr:N that globally increments the variable.

documentclass{article}

usepackage{xparse}



ExplSyntaxOn



% allocate a new integer variable

int_new:N g_philipp_src_int



% the main command

NewDocumentCommand{src}{m}

 {

  % step the integer variable

  int_gincr:N g_philipp_src_int

  % define #1 with expansion of the replacement text

  cs_new:cpx { #1 } { int_to_arabic:n { g_philipp_src_int } }

  % print the current value

  [int_to_arabic:n { g_philipp_src_int }]

 }

ExplSyntaxOff



begin{document}



src{testone}

src{testtwo}



first test: testone



second test: testtwo



end{document}

As an other option you can use the package etoolbox. The package provides a lot of useful commands and hooks.

In your case the command csdef is useful which is defined by etoolbox. This is similar to the solution of Christian.

documentclass{article}

newcounter{src}

usepackage{etoolbox}

newcommand{src}[1]{stepcounter{src}csedef{#1}{arabic{src}}[arabic{src}]}

begin{document}

src{testone}



src{testtwo}



first test: testone



second test: testtwo

end{document}