According to Christian Hupfer's comment I understood that calculations are possible in a keyval list when the macro is expandable. The macro \numexpr being expandable, calculations are possible.
Hence the interest in knowing which macros are and which are not expandable from TeX, LaTeX, ε-TeX, etc.
I am not asking if there is a test, but if there is a list already made.
Joseph has already answered, interpreting the "expandable` in your question as "fully expandable" (or what is perhaps better named as "safe in an expansion only context" see
Advantages and disadvantages of fully expandable macros
However to answer the question as actually asked, note that \numexpr is neither a macro nor expandable.
All macros are by definition expandable, a macro works by expanding to its replacement text. There is no list of these as it is any command defined via \def or its variants \edef, \gdef, \xdef.
The TeXBook lists all the TeX primitivies, these are marked by a * in the index, although you need to check the description of each individually to see if they are expandable.
In classic TeX, the expandable primitives are (I think:-)
\endinput
\expandafter
\csname
\the
\number
\romannumeral
\if
\ifx
\ifcat
\ifcase
\ifnum
\ifodd
\ifdim
\ifeof
\iftrue
\iffalse
\ifhbox
\ifvbox
\ifvoid
\ifinner
\ifvmode
\ifhmode
\ifmmode
\else
\or
\fi
\input % (\@@input in LaTeX)
\jobname
\meaning
\noexpand
\string
\topmark
\firstmark
\botmark
\fontname
in e-TeX the additional expandable primitives are
\topmarks
\firstmarks
\botmarks
\ifdefined
\ifcsname
\iffontchar
\unless
\eTeXrevision
\unexpanded
\detokenize
\scantokens
in addition to all the above pdfTeX adds the following expandable primitives
\pdfescapestring
\pdfescapename
\pdfescapehex
\pdfstrcmp
\pdfmatch
\ifpdfabsnum
\ifpdfabsdim
\pdfuniformdeviate
\pdfnormaldeviate
\pdffilemoddate
\pdffilesize
\pdfmdfivesum
\pdffiledump
\pdfcolorstackinit
\ifincsname
\ifpdfprimitive
\pdfcreationdate
\pdfinsertht
\pdftexbanner
\pdftexrevision
\expanded
XeTeX has the primitives of etex, and adds the following expandable primitives
\ifincsname
\ifprimitive
\normaldeviate
\uniformdeviate
\creationdate
\filedump
\filemoddate
\filesize
\mdfivesum
\expanded
\XeTeXrevision
\Uchar
\Ucharcat
\strcmp
LuaTeX has the expandable primitives of etex plus (at least)
\luatexbanner
\luatexrevision
\formatname
\Uchar
\directlua
\luaescapestring
\scantextokens
\csstring
\expanded
\ifincsname
\pdfvariable
\pdffeedback
pTeX (Japanese TeX engine) adds the following expandable primitives, which are available in ptex, uptex, eptex and euptex:
\euc
\ifdbox
\ifddir
\ifjfont
\ifmbox
\ifmdir
\iftbox
\iftdir
\iftfont
\ifybox
\ifydir
\jis
\kansuji
\kuten
\ptexfontname
\ptexrevision
\sjis
\tojis
\toucs
\ucs
upTeX (Unicode-aware pTeX) adds the following expandable primitives, which are available in uptex and euptex:
\uptexrevision
e-pTeX (pTeX + e-TeX) adds the following expandable primitives, which are available in eptex and euptex:
\expanded
\ifincsname
\ifpdfprimitive
\pdfcreationdate
\pdffiledump
\pdffilemoddate
\pdffilesize
\pdfmdfivesum
\pdfnormaldeviate
\pdfstrcmp
\pdfuniformdeviate
\Uchar
\Ucharcat
If you find missing ones, feel free to edit this answer here....
\numexpr isn't expandable either)
– David Carlisle
Dec 26 '18 at 10:30
A list of all expandable macros is entirely impossible to provide: there are an open-ended number of cases. What one can do is provide a list of primitives which work by expansion, and general rules for macros.
Essentially, any primitives which carry out assignment or typesetting are not expandable, whilst other primitives are. Thus for example \def, \let, etc. are all non-expandable, as are \hbox, \raise, \setbox, whereas \expandafter, \the and \ifx are all expandable. Note that some primitives will be expandable in the right context: \numexpr is not expandable in itself, but is a valid token coming after \the or \number:
\edef\testa{\numexpr 1+2\relax}\show\testa
\edef\testb{\number\numexpr 1+2\relax}\show\testb
In terms of macros, we have to be clear what we mean by 'expandable'. TeX is a macro expansion language, so if we have for example
\def\baz{}
\def\foo{\def\baz{bar}}
then we can do
\edef\test{\foo}\show\test
although the result is not what is likely wanted/useful. As such, when we talk about expandable macros we normally mean macros which contain only expandable primitives and which thus can be used 'safely' inside an \edef or similar. With e-TeX, it' possible to ensure that macros which don't meet this criterion don't 'blow up':
\def\baz{}
\protected\def\foo{\def\baz{bar}}
\edef\test{\foo}\show\test
Where does that take us? Any macro which contains:
\protected macrois not expandable. That is the majority of 'useful' macros, so as mentioned in comments, if you are not sure, assume non-expandable.
Note that the list of non-expandable primitives is quite long, and we do have to remember about the context. For example, something like \tracingcommands acts like a count register, so is non-expandable unless if follows \the. Thus unless we implement a token-by-token processor, just seeing \tracingcommands doesn't tell us that a macro containing this token is not expandable.
\def\foo{\tracingcommands=1 }% Not expandable
\def\baz{\the\tracingcommands}% Expandable
In LaTeX3/expl3, all macros are either fully expandable or are \protected. Moreover, the expandable ones are all marked as such (with a star) in the documentation. The reason is that it requires some knowledge to see if something is expandable, in particular checking all macro 'dependencies'. As such, one has to work carefully to track expandable macros.
expandable == without side effects?
– dedded
Oct 14 '19 at 13:25
\csname <some-undefined-name>\endcsname adds a \relax to the table of names, plus things like the PDF random number generator step the seed, etc.
– Joseph Wright
Oct 14 '19 at 13:27
Although it is explicitly not asked for a test it may be worth mentioning that a test for the expandability of a token can be based on the circumstance that with expandable tokens and undefined tokens the test
\expandafter\ifx\noexpand⟨token⟩⟨token⟩⟨token not expandable⟩\else⟨token expandable or undefined⟩\fi
delivers the tokens of the \else-branch/delivers the tokens of the ⟨token expandable or undefined⟩-branch. When the ⟨token expandable or undefined⟩-branch is taken, you can evaluate the \meaning of the ⟨token⟩:
\meaning in any case will deliver characters of category code 12(other). The only exception is the space character. \meaning will deliver spaces always with category code 10 (space) (and character code 32).
If and only if the ⟨token⟩ is undefined, then \meaning will deliver the leading catcode-12-token sequence undefined.
If and only if the ⟨token⟩ is a macro, then the tokens delivered by \meaning will contain the sequence ->.
Some expandable primitives may require special treatment/attention in some situations, e.g., \input (renamed to \@@input in LaTeX), \endinput, \noexpand. Probably these, too, can be cranked out by evaluating \meaning.
Besides this in LaTeX there are tokens which sometimes are expandable/unexpandable primitives and sometimes are expandable macros. While writing this sentence I think about the token \protect. Sometimes \protect is let equal to the unexpandable primitive \relax. Sometimes \protect is let equal to the expandable primitive \noexpand. Sometimes \protect is let equal to the macro \@unexpandable@protect which expands to \noexpand\protect\noexpand.
Here is a LaTeX-implementation of a test whether a macro argument does have a first token which is an expandable primitive:
(First you need to know whether the argument is empty. If it is not, you need to know whether its first token can be extracted as an undelimited argument. If and only if it is an explicit space of catcode 10 and charcode 32 or an opening brace/an explicit character token of catcode 1, then it cannot be extracted as an undelimited argument. But then it isn't an expandable primitive either. If it can be extracted, then have TeX extract it and check via \expandafter\ifx\noexpand#1#1 whether it is expandable/undefined. If it is not expandable, it is not an expandable primitive. If it is expandable or undefined check its \meaning for cranking out the case of it being a macro or undefined.)
\errorcontextlines=10000
\documentclass{article}
\makeatletter
%%=============================================================================
%% Paraphernalia:
%% \UD@firstoftwo, \UD@secondoftwo, \UD@Exchange, \UD@CheckWhetherNull,
%% \UD@CheckWhetherBrace, \UD@CheckWhetherLeadingSpace, \UD@ExtractFirstArg
%%=============================================================================
\newcommand\UD@firstoftwo[2]{#1}%
\newcommand\UD@secondoftwo[2]{#2}%
\newcommand\UD@Exchange[2]{#2#1}%
%%-----------------------------------------------------------------------------
%% Check whether argument is empty:
%%.............................................................................
%% \UD@CheckWhetherNull{<Argument which is to be checked>}%
%% {<Tokens to be delivered in case that argument
%% which is to be checked is empty>}%
%% {<Tokens to be delivered in case that argument
%% which is to be checked is not empty>}%
%%
%% The gist of this macro comes from Robert R. Schneck's \ifempty-macro:
%% <https://groups.google.com/forum/#!original/comp.text.tex/kuOEIQIrElc/lUg37FmhA74J>
\newcommand\UD@CheckWhetherNull[1]{%
\romannumeral0\expandafter\UD@secondoftwo\string{\expandafter
\UD@secondoftwo\expandafter{\expandafter{\string#1}\expandafter
\UD@secondoftwo\string}\expandafter\UD@firstoftwo\expandafter{\expandafter
\UD@secondoftwo\string}\UD@firstoftwo\expandafter{} \UD@secondoftwo}%
{\UD@firstoftwo\expandafter{} \UD@firstoftwo}%
}%
%% With eTeX-engines you can also do:
%%\newcommand\UD@CheckWhetherNull[1]{%
%% \romannumeral0\ifcat$\detokenize{#1}$%
%% \expandafter\UD@firstoftwo\else\expandafter\UD@secondoftwo\fi
%% {\UD@firstoftwo\expandafter{} \UD@firstoftwo}%
%% {\UD@firstoftwo\expandafter{} \UD@secondoftwo}%
%%}%
%%-----------------------------------------------------------------------------
%% Check whether argument's first token is a catcode-1-character
%%.............................................................................
%% \UD@CheckWhetherBrace{<Argument which is to be checked>}%
%% {<Tokens to be delivered in case that argument
%% which is to be checked has leading
%% catcode-1-token>}%
%% {<Tokens to be delivered in case that argument
%% which is to be checked has no leading
%% catcode-1-token>}%
\newcommand\UD@CheckWhetherBrace[1]{%
\romannumeral0\expandafter\UD@secondoftwo\expandafter{\expandafter{%
\string#1.}\expandafter\UD@firstoftwo\expandafter{\expandafter
\UD@secondoftwo\string}\UD@firstoftwo\expandafter{} \UD@firstoftwo}%
{\UD@firstoftwo\expandafter{} \UD@secondoftwo}%
}%
%%-----------------------------------------------------------------------------
%% Check whether brace-balanced argument starts with a space-token
%%.............................................................................
%% \UD@CheckWhetherLeadingSpace{<Argument which is to be checked>}%
%% {<Tokens to be delivered in case <argument
%% which is to be checked>'s 1st token is a
%% space-token>}%
%% {<Tokens to be delivered in case <argument
%% which is to be checked>'s 1st token is not
%% a space-token>}%
\newcommand\UD@CheckWhetherLeadingSpace[1]{%
\romannumeral0\UD@CheckWhetherNull{#1}%
{\UD@firstoftwo\expandafter{} \UD@secondoftwo}%
{\expandafter\UD@secondoftwo\string{\UD@CheckWhetherLeadingSpaceB.#1 }{}}%
}%
\newcommand\UD@CheckWhetherLeadingSpaceB{}%
\long\def\UD@CheckWhetherLeadingSpaceB#1 {%
\expandafter\UD@CheckWhetherNull\expandafter{\UD@firstoftwo{}#1}%
{\UD@Exchange{\UD@firstoftwo}}{\UD@Exchange{\UD@secondoftwo}}%
{\UD@Exchange{ }{\expandafter\expandafter\expandafter\expandafter
\expandafter\expandafter\expandafter}\expandafter\expandafter
\expandafter}\expandafter\UD@secondoftwo\expandafter{\string}%
}%
%%-----------------------------------------------------------------------------
%% Extract first inner undelimited argument:
%% \romannumeral0\UD@ExtractFirstArgLoop{<argument>UD@SelDOm}%
%% yields <argument>'s 1st undlimited argument.
%% <argument> must not be blank, i.e., must not be empty or consist of
%% explicit character tokens of catcode 10 and charcode 32 only.
%%.............................................................................
@ifdefinable\UD@RemoveTillUD@SelDOm{%
\long\def\UD@RemoveTillUD@SelDOm#1#2UD@SelDOm{{#1}}%
}%
\newcommand\UD@ExtractFirstArgLoop[1]{%
\expandafter\UD@CheckWhetherNull\expandafter{\UD@firstoftwo{}#1}%
{ #1}%
{\expandafter\UD@ExtractFirstArgLoop\expandafter{\UD@RemoveTillUD@SelDOm#1}}%
}%
%%=============================================================================
%% Check whether token is macro (only with macros the \meaning contains the
%% sequence ->) - the argument of \UD@CheckWhetherMacro must be a single token;
%% the argument of \UD@CheckWhetherMacro must not be empty:
%%=============================================================================
\newcommand\UD@CheckWhetherMacro[1]{%
\expandafter\UD@@CheckWhetherMacro\meaning#1->X%
}%
@ifdefinable\UD@@CheckWhetherMacro{%
\def\UD@@CheckWhetherMacro#1->#2X{%
\UD@CheckWhetherNull{#2}{\UD@secondoftwo}{\UD@firstoftwo}%
}%
}%
%%=============================================================================
%% Check whether token is undefined control sequence (only with undefined
%% control sequences the \meaning has the leading phrase "undefined"; besides
%% this \meaning never delivers no tokens at all) - the argument of
%% \UD@CheckWhetherUndefined must be a single token; the argument of
%% \UD@CheckWhetherUndefined must not be empty:
%%=============================================================================
\begingroup
\def\UD@CheckWhetherUndefined#1{%
\endgroup
\newcommand\UD@CheckWhetherUndefined[1]{%
\expandafter\UD@@CheckWhetherUndefined\meaning##1#1X%
}%
@ifdefinable\UD@@CheckWhetherUndefined{%
\def\UD@@CheckWhetherUndefined##1#1##2X{%
\UD@CheckWhetherNull{##1}{\UD@firstoftwo}{\UD@secondoftwo}%
}%
}%
}%
\escapechar=-1\relax
\expandafter\UD@CheckWhetherUndefined\expandafter{\string\undefined}%
%%=============================================================================
%% Check whether argument has first token which is an expandable primitive
%%=============================================================================
\newcommand\CheckWhetherArgHasFirstTokenWhichIsExpandablePrimitive[1]{%
\romannumeral0%
\UD@CheckWhetherNull{#1}{\UD@firstoftwo\expandafter{} \UD@secondoftwo}{%
\UD@CheckWhetherBrace{#1}{\UD@firstoftwo\expandafter{} \UD@secondoftwo}{%
\UD@CheckWhetherLeadingSpace{#1}{\UD@firstoftwo\expandafter{} \UD@secondoftwo}{%
\expandafter\UD@CheckWhetherArgExamineArgsFirstToken
\romannumeral0\UD@ExtractFirstArgLoop{#1UD@SelDOm}%
}%
}%
}%
}%
\newcommand\UD@CheckWhetherArgExamineArgsFirstToken[1]{%
\expandafter\ifx\noexpand#1#1%
\expandafter\UD@firstoftwo\else\expandafter\UD@secondoftwo\fi{%
\UD@firstoftwo\expandafter{} \UD@secondoftwo
}{%
\UD@CheckWhetherMacro{#1}{\UD@firstoftwo\expandafter{} \UD@secondoftwo}{%
\UD@CheckWhetherUndefined{#1}{\UD@firstoftwo\expandafter{} \UD@secondoftwo}{%
\UD@firstoftwo\expandafter{} \UD@firstoftwo
}%
}%
}%
}%
\makeatother
\begin{document}
These tests are "negative"
\CheckWhetherArgHasFirstTokenWhichIsExpandablePrimitive{}%
{Argument does have a first token which is an expandble primitive}%
{Argument does not have a first token which is an expandble primitive}
\CheckWhetherArgHasFirstTokenWhichIsExpandablePrimitive{{\ifx}\relax}%
{Argument does have a first token which is an expandble primitive}%
{Argument does not have a first token which is an expandble primitive}
\CheckWhetherArgHasFirstTokenWhichIsExpandablePrimitive{\relax\number}%
{Argument does have a first token which is an expandble primitive}%
{Argument does not have a first token which is an expandble primitive}
\CheckWhetherArgHasFirstTokenWhichIsExpandablePrimitive{\TeX}%
{Argument does have a first token which is an expandble primitive}%
{Argument does not have a first token which is an expandble primitive}
\CheckWhetherArgHasFirstTokenWhichIsExpandablePrimitive{ something}%
{Argument does have a first token which is an expandble primitive}%
{Argument does not have a first token which is an expandble primitive}
\CheckWhetherArgHasFirstTokenWhichIsExpandablePrimitive{\UnDeFInEdSelDOMWeiRDStrangeToken}%
{Argument does have a first token which is an expandble primitive}%
{Argument does not have a first token which is an expandble primitive}
These tests are "positive"
\makeatletter
\CheckWhetherArgHasFirstTokenWhichIsExpandablePrimitive{@@input myfilewith weirdcode.tex }%
{Argument does have a first token which is an expandble primitive}%
{Argument does not have a first token which is an expandble primitive}
\makeatother
\CheckWhetherArgHasFirstTokenWhichIsExpandablePrimitive{\number\relax}%
{Argument does have a first token which is an expandble primitive}%
{Argument does not have a first token which is an expandble primitive}
\CheckWhetherArgHasFirstTokenWhichIsExpandablePrimitive{\ifx aa aa \else aa \fi}%
{Argument does have a first token which is an expandble primitive}%
{Argument does not have a first token which is an expandble primitive}
\CheckWhetherArgHasFirstTokenWhichIsExpandablePrimitive{\noexpand\relax}%
{Argument does have a first token which is an expandble primitive}%
{Argument does not have a first token which is an expandble primitive}
\CheckWhetherArgHasFirstTokenWhichIsExpandablePrimitive{\fontname}%
{Argument does have a first token which is an expandble primitive}%
{Argument does not have a first token which is an expandble primitive}
\end{document}
\numexpris not a macro - it's a primitive from the eTeX-extensions.\numexpr...is not expandable.\the\numexpr.../\number\numexpr...is expandable. ;-) – Ulrich Diez Sep 25 '20 at 21:21