As Bernard suggested, why don't use csvsimple?
If you have commas in your cells, just put the contents within curly brackets.
If you have & in your cells, just write \&.
\documentclass{article}
\usepackage{filecontents}
\usepackage{csvsimple}
\begin{document}
\begin{filecontents*}{mydata.csv}
a,b,c
\end{filecontents*}
\csvreader[
tabular=|c|c|c|,
nohead
]{mydata.csv}% filename
{}{\csvcoli & \csvcolii & \csvcoliii}
\vspace{3ex}
and if you have commas or \& in your cells:
\vspace{3ex}
\begin{filecontents*}{mydata.csv}
a,b,c
{c, d},{e, f, g},{h, i, j, k}
l \& m,{n \& o, p},q \& r \& s \& t
\end{filecontents*}
\csvreader[
tabular=|c|c|c|,
nohead
]{mydata.csv}% filename
{}{\csvcoli & \csvcolii & \csvcoliii}
\end{document}

tabularinlatex.ltx. That is a real problem. – Mar 04 '19 at 14:53csvsimplepackage, or the more powerfuldatatool. – Bernard Mar 04 '19 at 15:00&s? Ultimately it depends on the nature of the data whether commas or ampersands are more common in the cells. (I'm not recommending trying to achieve the OP's goal in LaTeX -- my preference would be to run a preprocessor on the comma-separated data to produce valid LaTeX syntax -- just pointing out that the LaTeX syntax just happens to be what Knuth/Lamport decided to adopt for their use-cases, and isn't objectively better for all applications.) – ShreevatsaR Mar 04 '19 at 18:16\&)whereas for the comma you can't . Also, using a comma is certainly mor commen (for instance if a cell contains a list, or the decimal comma for numeric values – not all countries use a dot for that. – Bernard Mar 04 '19 at 18:47\&and ,except that the former happens to have an appropriate definition in LaTeX by default, and the latter doesn't (and would have to be redefined after finding a new name for what is currently,). And yes, commas may indeed be more common in many/most "typical" applications, but we don't know what's more common in the OP's application. Question is whether to bend the tool for the sake of our task or vice-versa… IMO the only reason for preferring&` here is because LaTeX is more suited for it, not anything intrinsic to the syntax. – ShreevatsaR Mar 04 '19 at 22:07results.csvinto the LaTeX report. However, I realised it was easier to export to CSV using&as a delimiter. For example, if usingpandas, you can dodataframe.to_csv('results.csv', sep='&', line_terminator='\\\\\n')– Ciprian Tomoiagă Mar 05 '19 at 14:02