I would like to read in data from a CSV file into a plot. Unfortunately, the data contains commas itself:

Name, Description, Data
hello, "hello, world", 1


Note that the data is quoted using quotation marks, which is quite common when it comes to CSV files. I read in the file using the following:

\pgfplotstableread[col sep=comma]{data.csv}{\datatable}


Problem is: the pgfplotstable package does not seem to be able to handle quotes, specifically, I get an error message because there are apparently too many entries in the second row. I edited the file in the following way:

Name, Description, Data
hello, hello\, world, 1


This seems to work OK. I then tried to use the Description column as symbolic coordinates in a plot:

  \begin{axis}[yticklabels from table={\datatable}{Description}]
# ....
\end{axis}


Problem in this case: the commas disappear, the label only reads "hello world".

Is there a way to read the quoted entries (I am stuck using pgfplotstable), or to quote the file in such a way that I can read the column into the labels?

• Providing an MWE would help. LaTeX understands groups, rather than quotes. So maybe hello, {hello, world}, 1 would work. Note that \, is a LaTeX macro for a small horizontal space, not a quoted comma, which would explain why your test did not produce a comma. – Steven B. Segletes Nov 2 '17 at 11:51
• I tested my theory on another question's pgfplotstable. at tex.stackexchange.com/questions/375737/…, changing "Grapes", 22 to "{Grapes, red}", 22 and it works fine with the grouped comma. Without the grouping braces, it errs just as in your case. – Steven B. Segletes Nov 2 '17 at 11:59
• If you can manually place braces then that's the easiest solution. – percusse Nov 2 '17 at 12:01
• Another way is to define, let's say, \let\comma, and then in your data set use hello, hello\comma{} world, 1, but this seems more difficult than the group approach. Note though, that with the group approach, even hello, hello{,} world, 1 would be sufficient. – Steven B. Segletes Nov 2 '17 at 12:10

Per the comments: since I am able to modify the CSV directly, an acceptable way to escape commas is be the fifth option, i.e. surrounding the cell by curly braces:

\documentclass{minimal}
\usepackage{pgfplots}
\usepackage{filecontents}

\begin{filecontents*}{data.csv}
Name,Description,Data
hello,hello,1
second,second \, option,2
third,third \, option,3
fourth,"{fourth, option}",4
fifth,{fifth, option},5
\end{filecontents*}

\begin{document}

\begin{tikzpicture}
\begin{axis}[ymin=-5,ymax=5,yticklabels from table={\datatable}{Description}]
\end{axis}
\end{tikzpicture}

\end{document}


Just as a note: I used the \, not to produce extra spacing but to escape the comma. After all, a line of the form first, "first, option", 1 will cause latex to fail with the error message

! Package pgfplots Error: Table 'data.csv' appears to have too many columns


So the parsing options for CSV files using pgfplotstable seem to be rather limited...