Expansion in stored csname of restatable theorem

Question

Thinking about this question, I'd like to use thmtools' restatable environment to store theorems with a number in the csname. That way one can loop through them at the end of the document to restate each one. My issue is that the way thmtools stores this csname seems to prevent expansion of commands within.

Here's an example that works. \somecmd steps through a counter and stores its argument in cmd-1, cmd-2, etc.

\documentclass{article}
\newcounter{foo}
\newcommand{\somecmd}[1]{%
    \stepcounter{foo}%
    \expandafter\def\csname cmd-\thefoo\endcsname{#1}%
}
\begin{document}
\somecmd{A}
\somecmd{B}
\somecmd{C}
\UseName{cmd-1}
\UseName{cmd-2}
\UseName{cmd-3}
\ExpandArgs{c}\ShowCommand{cmd-1}
\ExpandArgs{c}\ShowCommand{cmd-2}
\end{document}

The output is A B C and the log shows

> \cmd-1=macro:
->A.
<argument> \cmd-1
l.19 \ExpandArgs{c}\ShowCommand{cmd-1}
> \cmd-2=macro:
->B.
<argument> \cmd-2
l.20 \ExpandArgs{c}\ShowCommand{cmd-2}

as expected.

Now here's a try with restatable where \declaretheoremx works like \declaretheorem but also defines an environment saved<thmname> that stores its contents in a command thm-<num>. Since the environment grabs its contents we need to use the +b argspec.

\documentclass{article}
\usepackage{kantlipsum,amsthm,thmtools}
\newcounter{savedthmcount}
\newcommand{\declaretheoremx}[2][]{
    \declaretheorem[#1]{#2}
    \NewDocumentEnvironment{saved#2}{+b}{%
        \stepcounter{savedthmcount}%
        \begin{restatable}{#2}{thm-\thesavedthmcount}
        ##1
        \end{restatable}
        }{}
    }
\declaretheoremx{theorem}
\begin{document}
\begin{savedtheorem}
\kant[2][1]
\end{savedtheorem}
\begin{savedtheorem}
\kant[3][1]
\end{savedtheorem}
\UseName{thm-1}*
\UseName{thm-2}*
\ExpandArgs{c}\ShowCommand{thm-1}
\ExpandArgs{c}\ShowCommand{thm-2}
\end{document}

Everything works as expected except that each command thm-<num> is just the last theorem called. In the log we get

> \thm-1=macro:
->\@ifstar {\thmt@thisistheonefalse \csname thmt@stored@thm-\thesavedthmcount \
endcsname }{\thmt@thisistheonetrue \csname thmt@stored@thm-\thesavedthmcount \e
ndcsname }.
<argument> \thm-1
l.31 \ExpandArgs{c}\ShowCommand{thm-1}
> \thm-2=macro:
->@ifstar {\thmt@thisistheonefalse \csname thmt@stored@thm-\thesavedthmcount 

endcsname }{\thmt@thisistheonetrue \csname thmt@stored@thm-\thesavedthmcount \e
ndcsname }.
<argument> \thm-2
l.32 \ExpandArgs{c}\ShowCommand{thm-2}

which makes it seem like the stored csname is the literal string thmt@stored@thm-\thesavedthmcount instead of thmt@stored@thm-thm-1, thmt@stored@thm-thm-2, etc.

What is going wrong here? The definition of the restatable environment in thm-restate.sty is somewhat complicated but I don't see anything that would "protect" the argument from expansion.

Edit: This seems to work but I'd still like to know specifically what's going wrong without this trick.

Answer 1

I tried to reproduce your problem, but this code runs fine.

\documentclass{article}
\usepackage{thmtools, thm-restate}
\declaretheorem{theorem}
\newcounter{foo}
%\renewcommand{\thefoo}{\alph{foo}}
\begin{document}
\stepcounter{foo}
\begin{restatable}[Euclid]{theorem}{test\thefoo}
\label{thm:euclid}%
For every prime $p$, there is a prime $p’>p$.
In particular, the list of primes,
\begin{equation}\label{eq:1}
2,3,5,7,\dots
\end{equation}
is infinite.
\end{restatable}
\stepcounter{foo}
\begin{restatable}[Euclid]{theorem}{test\thefoo}
\label{thm:euclid}%
Every integer greater than 1 has a unique prime decomposition.
\end{restatable}
\setcounter{foo}{0}
\loop\ifnum\value{foo}<2
  \stepcounter{foo}%
  \csname test\thefoo\endcsname*
\repeat
\end{document}

Answer 2

Just use the \edef trick mentioned in this answer.

\documentclass{article}
\usepackage{kantlipsum,amsthm,thmtools}
\newcounter{savedthmcount}
\newcommand{\declaretheoremx}[2][]{
    \declaretheorem[#1]{#2}
    \NewDocumentEnvironment{saved#2}{O{}+b}{%
        \stepcounter{savedthmcount}%
        \edef\temp{%
            \noexpand\begin{restatable}[##1]{#2}{thm-\thesavedthmcount}%
          }%
        \temp
        ##2
        \end{restatable}
        }{}
    }
\declaretheoremx{theorem}
\begin{document}
\begin{savedtheorem}
\kant[1][1]
\end{savedtheorem}
\begin{savedtheorem}
\kant[3][1]
\end{savedtheorem}
\UseName{thm-1}*
\UseName{thm-2}*
\end{document}

Answer 3

restatable doesn't perform expansion in its second argument (the label) and you want to expand it. The recent \ExpandArgs facility is your friend.

I made a few changes to the definition of \declaretheoremx:

1. \NewDocumentCommand is more powerful;
2. the counter is allocated as part of it;
3. a trailing optional argument is added (default is the same as the main argument) for the prefix.

\documentclass{article}
\usepackage{kantlipsum,amsthm,thmtools}
\NewDocumentCommand{\declaretheoremx}{O{}mO{#2}}{%
  \newcounter{saved#2}%
  \declaretheorem[#1]{#2}%
  \NewDocumentEnvironment{saved#2}{+b}{%
    \stepcounter{saved#2}%
    \ExpandArgs{nne}\begin{restatable}{#2}{#3-\UseName{thesaved#2}}
    ##1
    \end{restatable}%
  }{}%
}
\declaretheoremx{theorem}[thm]
\begin{document}
\section{Main text}
First a theorem
\begin{savedtheorem}
\kant[2][1]
\end{savedtheorem}
Another theorem
\begin{savedtheorem}
\kant[3][1]
\end{savedtheorem}
\section{The statements}
\UseName{thm-1}*
\UseName{thm-2}*
\end{document}