Question

I have written a simple perl script to read a line from a .csv file. The code is as per below:

#!/usr/bin/perl

use strict;
use warnings;

use Text::CSV;

my $csv = Text::CSV->new({ binary => 1 });


open my $fh, "<", "testresults.csv" or die "testresults.csv $!";
while ( my $row = $csv->getline( $fh ) ) {
      my @fields = @$row;
}
$csv->eof or $csv->error_diag;
close $fh or die "testresults.csv $!";

And the testresults.csv file looks like this:

Node1,Node2,Node3,Node4,map1,map2,map3,map4,map5,map6,map7,map8,DM,LAT,AVG,product

on the first line followed by the results on each line:

name1,name2,name3,name4,Node1,Node2,Node3,Node4,Node5,Node6,Node7,Node8,0%,
0.002835480002,0.1714008533,4.86003691857886E-04

and so on.

I am getting the following error with my code when I do a ./filename.pl from the command prompt:
CSV_XS ERROR: 2032 – EIF – CR char inside unquoted, not part of EOL @ pos 420

I tried to google for this error but could not fathom much into this error.

Answer 1

It would seem from our conversation in the comments that the error comes of the strings in the input being interlaced with null characters, being made visible by using

use Data::Dumper;
$Data::Dumper::Useqq = 1;
while (<$fh>) {
    print Dumper $_;
}

A quick hack is to strip the null characters in the input file with something like:

perl -i.bak -pwe 'tr/\0//d' testresults.csv

NOTE: But as has been pointed out in comments from people more experienced in encoding matters, this can/could/should be solved by decoding your data instead. Just stripping the bad symbols might break your data in subtle ways, and is not an ideal solution.

I’m sorry, I do not know much about that, but using Text::CSV::Encoded does sound like a good start, like cjm suggested.