Storing a CSV row columns in a list/array and indexing into it

Question

Following up this answer, how to extend the code of its second method to store each row value columns in a list or array and be able to index into it?

In other words, I need another macro (along with \getValue) that stores all the values of the columns of the same row (except the first column that has the row name) in the following way:

\getRow{<row name located in the first column>}{<list/array name of the whole row value columns>}

For example, consider the following file

\begin{filecontents*}{test.csv}
First Parameter , 7 , 9  ,  ,
Third Parameter , 5 , 10 ,  ,
Fifth Parameter , 3 , 6  ,  , 44
\end{filecontents*}

• I can have the value of 44 by first storing the row of Fifth Parameter in \myList by \getRow{Fifth Parameter}{\myList}. Then \myList[4] will get me the value 44.

• The first value of the row (e.g. 3) can be accessed by \myList or \myList[1].

Answer 1

I've added a \getRow command to the “Second method” of the existing code, as you asked. Its syntax is:

\getRow[*] {\macro} {key} {label}

• The star form performs a global assignment to \macro (otherwise, the assignment is local).

• The label must identify a file you've previously read using \ReadCSV (it corresponds to the first mandatory argument of \ReadCSV).

• The key is used to select a particular row from that file (it has same meaning as with \getValue and \CSVItem).

After such a \getRow call, \macro is safe to use in expansion-only contexts (e.g., inside \edef, \write, \num from siunitx, etc.): it wraps the desired value using the \unexpanded e-TeX primitive, just like \tl_item:nn does.

Negative indices are supported and count from the end:

• \macro[-1] expands to the last item of the row stored in \macro;

• \macro[-2] expands to its predecessor;

• etc.

As a special case, \macro[non-empty] expands to the number of non-empty items in the row stored inside \macro.

\begin{filecontents*}{test.csv}
Third Parameter  , 7 , 9          ,
First Parameter  , 5 , {foo, bar} ,
Second Parameter , 3 , 6          , 44
\end{filecontents*}

\documentclass{article}
\usepackage{xparse}

\ExplSyntaxOn
% Step 1: reading the file
\ior_new:N \l__diaa_csv_ior
\bool_new:N \l__diaa_csv_str_bool
\seq_new:N \l__diaa_csv_tmp_seq

% str mode (bool/star), key column, label, value columns, file
\NewDocumentCommand \ReadCSV { s O{1} m O{} m }
  {
    \IfBooleanTF {#1}
      { \bool_set_true:N \l__diaa_csv_str_bool }
      { \bool_set_false:N \l__diaa_csv_str_bool }
    \diaa_csv_read:nnnn {#3} {#2} {#4} {#5}
  }

% label, key column, value columns, file
\cs_new_protected:Npn \diaa_csv_read:nnnn #1 #2 #3 #4
  {
    \tl_if_blank:nTF {#3}       % Detect number of columns and use 2 to last
      {
        \ior_open:NnTF \l__diaa_csv_ior {#4}
          {
            \bool_if:NTF \l__diaa_csv_str_bool
              { \ior_str_get:NN }
              { \ior_get:NN }
              \l__diaa_csv_ior \l_tmpa_tl

            \ior_close:N \l__diaa_csv_ior
            \seq_set_split:NnV \l_tmpa_seq { , } \l_tmpa_tl
            \seq_clear:N \l__diaa_csv_tmp_seq
            \int_step_inline:nnn { 2 } { \seq_count:N \l_tmpa_seq }
              { \seq_put_right:Nn \l__diaa_csv_tmp_seq {##1} }
          }
          { \msg_error:nnn { diaa } { file-not-found } {#4} }
      }
      { \seq_set_split:Nnn \l__diaa_csv_tmp_seq { , } {#3} } % explicit columns

    \ior_open:NnTF \l__diaa_csv_ior {#4}
      {
        \prop_new:c { g__diaa_csv_#1_prop }
        \__diaa_csv_read:nn {#1} {#2}
        \ior_close:N \l__diaa_csv_ior
      }
      { \msg_error:nnn { diaa } { file-not-found } {#4} }
  }

\msg_new:nnn { diaa } { file-not-found }
  { File~`#1'~not~found. }

\cs_generate_variant:Nn \prop_put:Nnn { cxV }

% label, key column
\cs_new_protected:Npn \__diaa_csv_read:nn #1 #2
  {
    \bool_if:NTF \l__diaa_csv_str_bool
      { \ior_str_map_inline:Nn }
      { \ior_map_inline:Nn }
        \l__diaa_csv_ior
        {
          \seq_set_split:Nnn \l_tmpa_seq { , } {##1} % split one CSV row
          \tl_clear:N \l_tmpa_tl
          \seq_map_inline:Nn \l__diaa_csv_tmp_seq
            {
              \tl_put_right:Nx \l_tmpa_tl { { \seq_item:Nn \l_tmpa_seq {####1} } }
            }

          \prop_put:cxV { g__diaa_csv_#1_prop }
            { \seq_item:Nn \l_tmpa_seq {#2} }
            \l_tmpa_tl
        }
  }

% Step 2: getting the values
% star → global assignment, macro or tl var, value column, key, label
\NewDocumentCommand \getValue { s m O{1} m m }
  {
    \IfBooleanTF {#1} { \tl_gset:Nx } { \tl_set:Nx }
      #2 { \diaa_csv_item:nnn {#4} {#3} {#5} }
  }

% key, value column, label
\NewExpandableDocumentCommand \CSVItem { m O{1} m }
  { \diaa_csv_item:nnn {#1} {#2} {#3} }

\cs_generate_variant:Nn \tl_item:nn { f }

% key, value column, label
\cs_new:Npn \diaa_csv_item:nnn #1 #2 #3
  {
    \tl_item:fn { \prop_item:cn { g__diaa_csv_#3_prop } {#1} } {#2}
  }

% star → global assignment, macro, key, label
\NewDocumentCommand \getRow { s m m m }
  {
    \prop_get:cnN { g__diaa_csv_#4_prop } {#3} \l_tmpa_tl
    \IfBooleanTF {#1} { \cs_gset_nopar:Npx } { \cs_set_nopar:Npx } #2 [ ##1 ]
      {
        \exp_not:N \str_if_eq:nnTF {##1} { non-empty }
          {
            \exp_not:N \__diaa_nb_nonempty_items_in_row:nw { 0 }
            \exp_not:V \l_tmpa_tl
            \exp_not:n { \q_recursion_tail \q_recursion_stop }
          }
          { \exp_not:N \tl_item:nn { \exp_not:V \l_tmpa_tl } {##1} }
      }
  }

\cs_new:Npn \__diaa_nb_nonempty_items_in_row:nw #1#2
  {
    \quark_if_recursion_tail_stop_do:nn {#2} { \int_eval:n {#1} }
    \tl_if_empty:nTF {#2}
      { \__diaa_nb_nonempty_items_in_row:nw {#1} }
      { \__diaa_nb_nonempty_items_in_row:nw { #1 + 1 } }
  }
\ExplSyntaxOff

\begin{document}

% Use default column for the key (1). The second empty optional argument (list
% of value columns) means we want to autodetect the value columns; then, the
% first column is for keys and all other columns are used as value columns.
\ReadCSV{mydata}{test.csv}

\getValue\rdPar{Second Parameter}{mydata}
\rdPar                         % 3

\getValue\rdPar[2]{Second Parameter}{mydata}
\rdPar                         % 6

\getValue\rdPar[3]{Second Parameter}{mydata}
\rdPar                         % 44

\getValue\rdPar{Third Parameter}{mydata}
\rdPar                         % 7

\edef\rdPar{\CSVItem{First Parameter}{mydata}}%
\rdPar                         % 5

\edef\rdPar{\CSVItem{First Parameter}[2]{mydata}}%
\rdPar                         % foo, bar

\edef\rdPar{\CSVItem{First Parameter}[3]{mydata}}%
\ifx\rdPar\empty
  \textlangle empty\textrangle
\else
  \rdPar
\fi

\getRow{\RowA}{First Parameter}{mydata}
\getRow{\RowB}{Second Parameter}{mydata}
\getRow{\RowC}{Third Parameter}{mydata}

``\RowA[1]'', ``\RowA[2]'', ``\RowA[3]'' ($\RowA[non-empty]$ non-empty)\par
``\RowB[1]'', ``\RowB[2]'', ``\RowB[3]'' ($\RowB[non-empty]$ non-empty)\par
``\RowC[1]'', ``\RowC[2]'', ``\RowC[3]'' ($\RowC[non-empty]$ non-empty)

{% Enter a group and perform a global assignment
  \getRow*{\globallyDefined}{First Parameter}{mydata}%
}% Leave the group
``\globallyDefined[1]'', ``\globallyDefined[2]'', ``\globallyDefined[3]''

\edef\zzz{abc'\RowA[2]'def}
%\show\zzz % prints \zzz=macro:->abc'foo, bar'def.
\edef\zzz{\RowA[non-empty]}
%\show\zzz % prints \zzz=macro:->2.

\end{document}