20
\documentclass{article}
\usepackage{tikz}

\def\firstlist{0,1,2}
\def\secondlist{0,1,2}

\newcommand{\testa}{
 \foreach \x/\y in {\firstlist/\secondlist} {
    \draw(0,\x)--(1,\y);
    }
}

\newcommand{\testb}{
 \foreach \x/\y in {0/0,1/1,2/2} {
    \draw(0,\x)--(1,\y);
    }
}
\begin{document}
\begin{tikzpicture}
\testa
\end{tikzpicture}
\end{document}

How can I make the output of the \testa command equal to the output of the \testb command?

Carlitos
  • 225
  • Welcome to TeX.sx! This is a nice minimal example. I think you want to have another picture with \testb in it though, right? – Alan Munn Jul 29 '14 at 22:12
  • @AlanMunn yes, to see the difference between the outputs. \testb is the desired output. – Carlitos Jul 29 '14 at 22:14

4 Answers4

20

Here is a very simple solution:

\documentclass{article}
\usepackage{tikz}

\def\firstlist{0,1,2}
\def\secondlist{0,1,2}

\newcommand{\testa}{
  \foreach \x [count=\c,evaluate=\c as \y using {{\secondlist}[\c-1]}]  in \firstlist {
    \draw(0,\x)--(1,\y);
  }
}

\begin{document}
\begin{tikzpicture}
\testa
\end{tikzpicture}
\end{document}

An extended solution defining the new style parallel foreach. You can use multiple parallel lists:

\documentclass{standalone}
\usepackage{tikz}

\pgfset{
  foreach/parallel foreach/.style args={#1in#2via#3}{evaluate=#3 as #1 using {{#2}[#3-1]}},
}

\def\firstlist{0,1,2,10,11}
\def\secondlist{0,1,2,10,11}
\def\thirdlist{1,2,10,11,0}

\newcommand{\testa}{
  \foreach \x [count=\c,
  parallel foreach=\y in \secondlist via \c,
  parallel foreach=\z in \thirdlist via \c]
  in \firstlist
  {
    \node[left] at (0,\x) {\x};
    \draw(0,\x)--(1,\y);
    \draw[red](0,\x)--(1,\z);
  }
}

\begin{document}
\begin{tikzpicture}
\testa
\end{tikzpicture}
\end{document}

Both methods use the array features of pgfmath: they evaluate each element as a math formula. If you want string elements, use quotes ("...").

\documentclass[margin=1mm]{standalone}
\usepackage{tikz}

\pgfset{
  foreach/parallel foreach/.style args={#1in#2via#3}{
    evaluate=#3 as #1 using {{#2}[#3-1]}
  },
}

\def\firstlist{0,1,2,10,11}
\def\secondlist{0,1,2,10,11}
\def\thirdlist{1,2,10,11,0}
\def\fourthlist{"label a","label b","label c","label $\delta$","label e"}
\newcommand{\testa}{
  \foreach \x [count=\c,
  parallel foreach=\y in \secondlist via \c,
  parallel foreach=\z in \thirdlist via \c,
  parallel foreach=\lab in \fourthlist via \c]
  in \firstlist
  {
    \node[left] at (0,\x) {\lab};
    \draw(0,\x)--(1,\y);
    \draw[red](0,\x)--(1,\z);
  }
}

\begin{document}
\begin{tikzpicture}
\testa
\end{tikzpicture}
\end{document}

enter image description here

Paul Gaborit
  • 70,770
  • 10
  • 176
  • 283
  • This doesn't work if element of secondlist is of more than one character! – taninamdar Dec 11 '15 at 21:04
  • 1
    @taninamdar Both solution may use {0,1,2,10,11} (as in my second edited example). If you want string elements, use quotes "..." (as in my third example). – Paul Gaborit Dec 12 '15 at 10:41
10

You can count how many elements there are in one of the arrays and use that number as the foreach limit. Then every spin accesses one element of the arrays;

\documentclass[tikz]{standalone}
\def\firstlist{{0,1,2}} % <== Notice the double brace for array notation
\def\secondlist{{0,1,2}}

\newcommand{\testa}{
\pgfmathdim{\firstlist}%Get the number of elements in array
  \foreach \x in {0,...,\numexpr\pgfmathresult-1\relax}{%array index starts from zero
    \draw(0,{array(\firstlist,\x)})--(1,{array(\secondlist,\x)});
    }
}

\newcommand{\testb}{
 \foreach \x/\y in {0/0,1/1,2/2} {
    \draw(0,\x)--(1,\y);
    }
}
\begin{document}
\begin{tikzpicture}
\testa
\begin{scope}[shift={(2,0)}] %Test the result with \testb next to it
\testb
\end{scope}
\end{tikzpicture}
\end{document}

enter image description here

percusse
  • 157,807
9

I'm not so sure you want to do it. ;-)

\documentclass{article}
\usepackage{xparse}

\def\xforeach#1#{\xforeachaux{#1}}

\ExplSyntaxOn
\NewDocumentCommand{\xforeachaux}{mm}
 {
  \carlitos_xforeach:nn { #1 } { #2 }
 }

\seq_new:N \l_carlitos_xf_first_seq
\seq_new:N \l_carlitos_xf_second_seq
\seq_new:N \l_carlitos_xf_list_seq

\cs_new_protected:Npn \carlitos_xforeach:nn #1 #2
 {
  \seq_clear:N \l_carlitos_xf_list_seq
  \__carlitos_split:n { #2 }
  \seq_mapthread_function:NNN
   \l_carlitos_xf_first_seq
   \l_carlitos_xf_second_seq
   \__carlitos_xf_additem:nn
  \__carlitos_xf_do:nx { #1 } { \seq_use:Nn \l_carlitos_xf_list_seq { , } }
 }
\cs_new_protected:Npn \__carlitos_split:n #1
 {
  \__carlitos_split_aux:w #1 \q_stop
 }
\cs_new_protected:Npn \__carlitos_split_aux:w #1/#2 \q_stop
 {
  \seq_set_split:Nno \l_carlitos_xf_first_seq { , } { #1 }
  \seq_set_split:Nno \l_carlitos_xf_second_seq { , } { #2 }
 }
\cs_new_protected:Npn \__carlitos_xf_additem:nn #1 #2
 {
  \seq_put_right:Nn \l_carlitos_xf_list_seq { #1/#2 }
 }
\cs_new_protected:Npn \__carlitos_xf_do:nn #1 #2
 {
  \foreach #1 { #2 }
 }
\cs_generate_variant:Nn \seq_set_split:Nnn { Nno }
\cs_generate_variant:Nn \__carlitos_xf_do:nn { nx }
\ExplSyntaxOff

\usepackage{tikz}

\def\firstlist{0,1,2}
\def\secondlist{0,1,2}

\newcommand{\testa}{
 \xforeach \x/\y in {\firstlist/\secondlist} {
    \draw(0,\x)--(1,\y);
    }
}

\newcommand{\testb}{
 \foreach \x/\y in {0/0,1/1,2/2} {
    \draw(0,\x)--(1,\y);
    }
}
\begin{document}
\begin{tikzpicture}
\testa
\end{tikzpicture}
\qquad
\begin{tikzpicture}
\testb
\end{tikzpicture}
\end{document}

enter image description here

A completely different implementation, where you don't even need \x/\y; it supports up to three slash separated arguments, which can be either comma separated lists (spaces are ignored before and after the commas) or macros expanding to comma separated lists.

Each cycle in the foreach loop is performed using the second argument, which uses #1, #2 and #3 to denote the current list element. The number of cycles is determined by the first list.

Probably a better error recovery should be provided if the given items are more than three, but this is left as an exercise.

\documentclass{article}
\usepackage{xparse,tikz}

\ExplSyntaxOn

\NewDocumentCommand{\Xforeach}{mm}
 {
  \carlitos_Xforeach:nn { #1 } { #2 }
 }

\seq_new:N \l__carlitos_Xf_input_seq
\seq_new:N \l__carlitos_Xf_output_seq
\seq_new:N \l__carlitos_Xf_lista_seq
\seq_new:N \l__carlitos_Xf_listb_seq
\seq_new:N \l__carlitos_Xf_listc_seq
\tl_new:N \l__carlitos_Xf_tmp_tl
\int_new:N \l__carlitos_Xf_cycle_int

\cs_new_protected:Npn \carlitos_Xforeach:nn #1 #2
 {
  \seq_set_split:Nnn \l__carlitos_Xf_input_seq { / } { #1 }
  \seq_clear:N \l__carlitos_Xf_output_seq
  \seq_map_inline:Nn \l__carlitos_Xf_input_seq
   {
    \seq_put_right:No \l__carlitos_Xf_output_seq { ##1 }
   }
  \int_case:nnF { \seq_count:N \l__carlitos_Xf_input_seq }
   {
    { 1 } { \__carlitos_Xf_one:n   { #2 } }
    { 2 } { \__carlitos_Xf_two:n   { #2 } }
    { 3 } { \__carlitos_Xf_three:n { #2 } }
   }
   {
    Only~one,~two~or~three!
   }
 }

\cs_new_protected:Npn \__carlitos_Xf_one:n #1
 {
  \seq_pop_left:NN \l__carlitos_Xf_output_seq \l__carlitos_Xf_tmp_tl
  \seq_set_split:NnV \l__carlitos_Xf_first_seq { , } \l__carlitos_Xf_tmp_tl
  \cs_set_protected:Npn \__carlitos_Xf_loop:n ##1 { #1 }
  \cs_generate_variant:Nn \__carlitos_Xf_loop:n { x }
  \int_zero:N \l__carlitos_Xf_cycle_int
  \seq_map_inline:Nn \l__carlitos_Xf_first_seq
   {
    \int_incr:N \l__carlitos_Xf_cycle_int
    \__carlitos_Xf_loop:x { \seq_item:Nn \l__carlitos_Xf_first_seq { \l__carlitos_Xf_cycle_int } }
   }
 }
\cs_new_protected:Npn \__carlitos_Xf_two:n #1
 {
  \seq_pop_left:NN \l__carlitos_Xf_output_seq \l__carlitos_Xf_tmp_tl
  \seq_set_split:NnV \l__carlitos_Xf_first_seq { , } \l__carlitos_Xf_tmp_tl
  \seq_pop_left:NN \l__carlitos_Xf_output_seq \l__carlitos_Xf_tmp_tl
  \seq_set_split:NnV \l__carlitos_Xf_second_seq { , } \l__carlitos_Xf_tmp_tl
  \cs_set_protected:Npn \__carlitos_Xf_loop:nn ##1 ##2 { #1 }
  \cs_generate_variant:Nn \__carlitos_Xf_loop:nn { xx }
  \int_zero:N \l__carlitos_Xf_cycle_int
  \seq_map_inline:Nn \l__carlitos_Xf_first_seq
   {
    \int_incr:N \l__carlitos_Xf_cycle_int
    \__carlitos_Xf_loop:xx
     { \seq_item:Nn \l__carlitos_Xf_first_seq { \l__carlitos_Xf_cycle_int } }
     { \seq_item:Nn \l__carlitos_Xf_second_seq { \l__carlitos_Xf_cycle_int } }
   }
 }
\cs_new_protected:Npn \__carlitos_Xf_three:n #1
 {
  \seq_pop_left:NN \l__carlitos_Xf_output_seq \l__carlitos_Xf_tmp_tl
  \seq_set_split:NnV \l__carlitos_Xf_first_seq { , } \l__carlitos_Xf_tmp_tl
  \seq_pop_left:NN \l__carlitos_Xf_output_seq \l__carlitos_Xf_tmp_tl
  \seq_set_split:NnV \l__carlitos_Xf_second_seq { , } \l__carlitos_Xf_tmp_tl
  \seq_pop_left:NN \l__carlitos_Xf_output_seq \l__carlitos_Xf_tmp_tl
  \seq_set_split:NnV \l__carlitos_Xf_third_seq { , } \l__carlitos_Xf_tmp_tl
  \cs_set_protected:Npn \__carlitos_Xf_loop:nnn ##1 ##2 ##3 { #1 }
  \cs_generate_variant:Nn \__carlitos_Xf_loop:nnn { xxx }
  \int_zero:N \l__carlitos_Xf_cycle_int
  \seq_map_inline:Nn \l__carlitos_Xf_first_seq
   {
    \int_incr:N \l__carlitos_Xf_cycle_int
    \__carlitos_Xf_loop:xxx
     { \seq_item:Nn \l__carlitos_Xf_first_seq { \l__carlitos_Xf_cycle_int } }
     { \seq_item:Nn \l__carlitos_Xf_second_seq { \l__carlitos_Xf_cycle_int } }
     { \seq_item:Nn \l__carlitos_Xf_third_seq { \l__carlitos_Xf_cycle_int } }
   }
 }

\ExplSyntaxOff
\begin{document}
\def\firstlist{0,1,2}
\def\secondlist{0,1,2}
\def\thirdlist{a,b,c}

\Xforeach{\firstlist}{--#1-- }

\bigskip

\begin{tikzpicture}
\Xforeach{\firstlist/\secondlist}{\draw(0,#1)--(1,#2);}
\end{tikzpicture}\qquad
\begin{tikzpicture}
\Xforeach{0,1,2/\secondlist}{\draw(0,#1)--(1,#2);}
\end{tikzpicture}

\bigskip

\Xforeach{\firstlist/\secondlist/\thirdlist}{#1/#2/#3 }

\Xforeach{\firstlist/\secondlist/\thirdlist}{#1/#2/#3 }

\Xforeach{\firstlist/0,1,2/a,b,c}{#1/#2/#3 }
\end{document}

enter image description here

egreg
  • 1,121,712
  • you managed to scare me :) – Carlitos Jul 30 '14 at 09:01
  • @Carlitos It just shows that keeping your proposed syntax is not really easy. I tried thinking to a better syntax, but it was too late in the night. Stay focused. – egreg Jul 30 '14 at 09:04
  • Well, in the end I discarded the list thing and opted for a completely different approach, since my real world problem was actually more complicated than that in the MWE. – Carlitos Jul 30 '14 at 09:09
7

You could always "roll your own" solution, but it would require considerable work to make it look as nice as \foreach:

\documentclass[tikz,border=5]{standalone}
\makeatletter
\def\attxt{@}
\long\def\frch#1#2#3{\long\def\frchaction##1##2{#3}%
\def\lsta{}\def\lstb{}%
\edef\lst{#1,@,/#2,@,}\expandafter\@frch\lst\@}
\def\@frch#1,#2/#3,#4\@{%
\def\tmpa{#1}\def\tmpb{#3}%
\ifx\tmpa\attxt%
  \ifx\tmpb\attxt%
  \else%
    \def\lstb{#3}\frchaction{\lsta}{#3}\@frch @,/#4,\@%
  \fi%
\else%
  \ifx\tmpb\attxt%
    \def\lsta{#1}\frchaction{#1}{\lstb}\@frch#2/@,\@%
  \else%
    \def\lsta{#1}\def\lstb{#3}\frchaction{#1}{#3}\@frch#2/#4\@%
  \fi%
\fi}

\begin{document}
\begin{tikzpicture}
\frch{0,1,2}{0,1,2}{ \draw (0,#1) -- (1,#2); }
\frch{0,1,2}{0,1,2,3,4,5}{ \draw (2,#1) -- (3,#2); }
\frch{0,1,2,3,4,5}{0,1,2}{ \draw (4,#1) -- (5,#2); }
\end{tikzpicture}
\end{document}

enter image description here

Mark Wibrow
  • 70,437