CarLaTeX
This simple example works:
```
\begin{filecontents*}{example.csv}
A
{51,502}
{121,151}
\end{filecontents*}
\documentclass{article}
\usepackage{tabularray}
\usepackage{csvsimple-l3}
\begin{document}
\csvreader[
head=true,
tabularray = {
width=\linewidth,
colspec={r},
},
table head = {First\\},
]
{example.csv}
{}
{\csvlinetotablerow}
\end{document}
```
In case my `csv` has `"..."` around the numbers with comma as decimal point, in the [`csvsimple-l3` manual](https://ctan.mirror.garr.it/mirrors/ctan/macros/latex/contrib/csvsimple/csvsimple-l3.pdf) there is an hook to trasform `"..."` into `{...}`.
But it does not work:
```
\begin{filecontents*}{example.csv}
A
"51,502"
"121,151"
\end{filecontents*}
\documentclass{article}
\usepackage{tabularray}
\usepackage{csvsimple-l3}
\AddToHook{csvsimple/csvline}
{
\tl_set_eq:NN \l_tmpa_tl \csvline
\regex_replace_all:nnN { "([^"]+)" } { {\1} } \l_tmpa_tl
\tl_gset_eq:NN \csvline \l_tmpa_tl
}
\begin{document}
\csvreader[
head=true,
tabularray = {
width=\linewidth,
colspec={r},
},
table head = {First\\},
]
{example.csv}
{}
{\csvlinetotablerow}
\end{document}
```
The error is:
```
! Undefined control sequence.
\__hook_toplevel csvsimple/csvline -> \tl
_set_eq:NN \l _tmpa_tl \csvline \r...
l.26 {\csvlinetotablerow}
```
Is the hook correct? Or am I doing something wrong?
Top Answer
samcarter
Adding `\ExplSyntaxOn...\ExplSyntaxOff` seems to do the trick:
```
\begin{filecontents*}{example.csv}
A
"51,502"
"121,151"
\end{filecontents*}
\documentclass{article}
\usepackage{tabularray}
\usepackage{csvsimple-l3}
\ExplSyntaxOn
\AddToHook{csvsimple/csvline}
{
\tl_set_eq:NN \l_tmpa_tl \csvline
\regex_replace_all:nnN { "([^"]+)" } { {\1} } \l_tmpa_tl
\tl_gset_eq:NN \csvline \l_tmpa_tl
}
\ExplSyntaxOff
\begin{document}
\csvreader[
head=true,
tabularray = {
width=\linewidth,
colspec={r},
},
table head = {First\\},
]
{example.csv}
{}
{\csvlinetotablerow}
\end{document}
```
Answer #2
Skillmon
samcarter provided a valid answer for the problem at hand. However, running a regular expression over *all* data of a csv is *very* wasteful. Therefore this answer doesn't face the problem at hand but instead defines alternative code paths that should have better performance
# Assuming fixed category (best performance)
When we assume fixed category codes we can write a very tight and fast loop to simply replace all quoted blocks to be inside of `{}`. Whichever category code applies during reading the CSV file must be the one being active during the definition of `\cleanQuotes`:
```
\begin{filecontents*}{example.csv}
A
"51,502"
"121,151"
\end{filecontents*}
\documentclass{article}
\usepackage{tabularray}
\usepackage{csvsimple-l3}
\makeatletter
\newcommand\cleanQuotes[1]
{%
\edef\csvline{\cleanQuotes@\@empty#1"\cleanQuotes@"}%
}
\long\def\cleanQuotes@ifend#1\cleanQuotes@{}
\long\def\cleanQuotes@#1"%
{%
\unexpanded\expandafter{#1}%
\cleanQuotes@b\@empty
}
\long\def\cleanQuotes@b#1"%
{%
\cleanQuotes@ifend#1\cleanQuotes@end\cleanQuotes@
{\unexpanded\expandafter{#1}}%
\cleanQuotes@\@empty
}
\long\def\cleanQuotes@end\cleanQuotes@#1\cleanQuotes@\@empty{}%
\makeatother
\AddToHook{csvsimple/csvline}
{%
\expandafter\cleanQuotes\expandafter{\csvline}%
}
\begin{document}
\csvreader[
head=true,
tabularray = {
width=\linewidth,
colspec={r},
},
table head = {First\\},
]
{example.csv}
{}
{\csvlinetotablerow}
\end{document}
```
# Supporting different category codes (still better than `l3regex`)
Since we only care for a single character code we can still be faster than `l3regex` by using a token-by-token (well, technically an item-by-item) parser, like the one defined in `etl` (*disclaimer:* I'm the author -- caveat: the used mechanism doesn't handle the edge case of an implicit `"`-token (`\let` to a token with the charcode of `"`)):
```
\begin{filecontents*}{example.csv}
A
"51,502"
"121,151"
\end{filecontents*}
\documentclass{article}
\usepackage{tabularray}
\usepackage{csvsimple-l3}
\usepackage{etl}
\ExplSyntaxOn
\cs_new_protected:Npn \cleanQuotes #1
{
\tl_set:Ne \csvline
{
\etl_act:nnnnn
\__my_cleanQuotes:nN
\__my_cleanQuotes:n
\__my_cleanQuotes:nn
{ \__my_cleanQuotes_normal:NN \exp_not:n }
{#1}
}
}
\cs_new:Npn \__my_cleanQuotes:nN #1#2
{ #1 {#2} }
\cs_new:Npn \__my_cleanQuotes:n #1
{ \use_ii:nn #1 { ~ } }
\cs_new:Npn \__my_cleanQuotes:nn #1#2
{ \use_ii:nn #1 { {#2} } }
\cs_new:Npn \__my_cleanQuotes_normal:NN #1#2
{
\token_if_eq_charcode:NNTF " #2
{
\etl_act_status:n
{ \__my_cleanQuotes_quote:NnN \__my_cleanQuotes_add:nn {} }
}
{ \exp_not:N #2 }
}
\cs_new:Npn \__my_cleanQuotes_add:nn #1#2
{
\etl_act_status:n
{ \__my_cleanQuotes_quote:NnN \__my_cleanQuotes_add:nn {#1#2} }
}
\cs_new:Npn \__my_cleanQuotes_quote:NnN #1#2#3
{
\token_if_eq_charcode:NNTF " #3
{ { \exp_not:n {#2} } }
{
\etl_act_status:n
{ \__my_cleanQuotes_quote:NnN \__my_cleanQuotes_add:nn {#2#3} }
}
}
\ExplSyntaxOff
\AddToHook{csvsimple/csvline}
{%
\expandafter\cleanQuotes\expandafter{\csvline}%
}
\begin{document}
\csvreader[
head=true,
tabularray = {
width=\linewidth,
colspec={r},
},
table head = {First\\},
]
{example.csv}
{}
{\csvlinetotablerow}
\end{document}
```