expl3: How to remove the final comma in a clist mapping

Question

I want to process a clist, and I can't use \clist_use:Nn as I have to some rather complicated things with the items. But how do you remove the final separator without to much fuss when using \clist_map_inline:Nn?

\documentclass[a4paper]{article}
\usepackage{expl3}
\begin{document}
\ExplSyntaxOn
\clist_const:Nn\c_UF_clist {a,b,c,d}

\clist_map_inline:Nn\c_UF_clist
{\textbf{#1},~} %no comma after d!!!!
\ExplSyntaxOff
\end{document}

enter image description here

I've seen egreg first poping the left item, and then use {,~\textbf{#1} mapping the rest of the clist. Still, I don't know what's the “official way”. May be something similar to \seq_set_map:NNn \foo \bar {\textbf{#1}} and then \seq_use:Nn \foo {,~}? — Manuel, Mar 09 '15 at 19:50

egreg · Answer 1 · 2015-03-12T17:30:26.620

Without using sequences, you can pop the leftmost item in the clist and map the remaining clist:

\documentclass{article}
\usepackage{expl3}

\ExplSyntaxOn

\tl_new:N \l__UF_clist_use_head_tl
\clist_new:N \l__UF_clist_use_tail_clist
\cs_new_protected:Npn \UF_clist_use:NNn #1 #2 #3
 {
  \clist_set_eq:NN \l__UF_clist_use_tail_clist #1
  \clist_pop:NN \l__UF_clist_use_tail_tl \l__UF_clist_use_head_tl
  \exp_args:NV #2 \l__UF_clist_use_head_tl
  \clist_map_inline:Nn \l__UF_clist_use_tail_clist
   {
    #3 #2 { ##1 }
   }
 }

\clist_const:Nn \c_UF_clist {a,b,c,d}

\begin{document}

\UF_clist_use:NNn \c_UF_clist \textbf { ,~ }

\end{document}

The first argument of \UF_clist_xuse:NNn is a clist variable, the second argument is a one parameter function (here \textbf) and the third argument is the separator between arguments. This function can't be nested, but it doesn't seem to be a big limitation.

enter image description here

If you plan to emulate \clist_use:Nnnn, then transforming the clist into a sequence and applying \seq_set_map:NNn seems the best approach.

Kernel functions called \seq_use:NNnnn, \seq_use:NNn, \clist_use:NNnnn and \clist_use:NNn, where the second argument is a one parameter function to be applied to each item, would be welcome, in my opinion (even if they aren't fully expandable).

I'll take a look at the suggested use:NN syntax at some point — Joseph Wright, Mar 10 '15 at 14:25
Thanks. This works fine too. I accepted the one from Manual as it is more in the spirit of "natural and not much fuss", at least more (in my eyes) than "popping something to the left, to avoid something on the right". — Ulrike Fischer, Mar 11 '15 at 19:51
Wouldn't if be cleaner to use a clist for the tail-part instead of a token list? — Ulrike Fischer, Mar 12 '15 at 17:12

Manuel · Accepted Answer · 2015-03-10T12:13:15.020

5

This, seems to me, is natural, although a bit tedious. (I leave the \seq_pop_left:NN option for another answer).

\documentclass[a4paper]{article}
\usepackage{expl3}
\begin{document}
\ExplSyntaxOn
\seq_set_from_clist:Nn \l_tmpa_seq { a , b , c , d }
\seq_set_map:NNn \l_tmpb_seq \l_tmpa_seq { \exp_not:n { \textbf { #1 } } }
\seq_use:Nn \l_tmpb_seq { ,~ }
\ExplSyntaxOff
\end{document}

In case you can't use \seq_set_map:NNn because of its experimentalness, you can define your own function

\seq_new:N \l__map_and_use_seq
\cs_new_protected:Npn \uf_seq_map_use:Nnn #1 #2 #3
 {
  \seq_clear:N \l__map_and_use_seq
  \seq_map_inline:Nn #1 { \seq_put_right:Nn \l__map_and_use_seq { #2 } }
  \seq_use:Nn \l__map_and_use_seq { #3 }
 }

\seq_set_from_clist:Nn \l_tmpa_seq { a , b , c , d }    
\uf_seq_map_use:Nnn \l_tmpa_seq { \textbf { [#1] } } { ,~ }

Or even clist directly if you prefer

\seq_new:N \l__map_and_use_seq
\seq_new:N \l__set_and_map_seq
\cs_new_protected:Npn \uf_clist_map_use:nnn #1 #2 #3
 {
  \seq_clear:N \l__map_and_use_seq
  \seq_set_from_clist:Nn \l__set_and_map_seq { #1 }
  \seq_map_inline:Nn \l__set_and_map_seq { \seq_put_right:Nn \l__map_and_use_seq { #2 } }
  \seq_use:Nn \l__map_and_use_seq { #3 }
 }

\uf_clist_map_use:nnn { a , b , c , d } { \textbf { [#1] } } { ,~ }

edited Mar 10 '15 at 12:13

answered Mar 09 '15 at 20:02

Manuel

27,118

This looks quite natural (apart from the fact that one has to convert to a sequence first, perhaps I should create sequences from start), but I can't use \seq_set_map:NNn as it is marked as "experimental" in source3.pdf. – Ulrike Fischer Mar 10 '15 at 08:50
You could declare a new function. – Manuel Mar 10 '15 at 12:05
@UlrikeFischer Part of the reason for having experimental functions is to see if they are useful! We don't remove stuff unless we have first checked for use in TeX Live. – Joseph Wright Mar 10 '15 at 14:24
@JosephWright: The code will be in a class of a customer. It will be never seen by texlive or CTAN. And I agree with egreg that more "use"-functions would be useful. It is imho very natural (in typesetting) that you want to do something different with the first and the last item of a list. – Ulrike Fischer Mar 10 '15 at 14:44
@UlrikeFischer Non-released code is of course a bit more tricky, but the point 'if it doesn't get tested we don't know it's useful' remains true! The present use functions are not exactly for typesetting or at least not focussed exclusively on that area. However, I can see the point: I need to work out the naming, etc. – Joseph Wright Mar 10 '15 at 15:47
Well now I really have the choice ;-). – Ulrike Fischer Mar 11 '15 at 19:48

expl3: How to remove the final comma in a clist mapping

2 Answers2