I also am very curious what you did to make Text::CSV_XS be 4 time slower than your code.
For what it is worth, I tried to come up with the most efficient Text::CSV_XS code to attack your situation, only to find there was a bug in the module, which is now fixed in version 1.19. The code would be something like this:
use 5.20.0;
use warnings;
use Text::CSV_XS;
my $csv = Text::CSV_XS->new ({
binary => 1,
auto_diag => 1,
keep_meta_info => 11,
});
my $row = $csv->getline (*DATA); # get the first line to count number
+of fields
my @row = @$row;
$csv->bind_columns (\(@row)); # speed up all remaining fetches
do {
$row[$_] =~ tr{,}{-} for grep { $csv->is_quoted ($_) } 0..$#row;
$csv->say (*STDOUT, \@row);
} while ($csv->getline (*DATA));
__END__
65722417,"1193,1",7980,1133566,4169735,035,FEDERAL UNIVERSAL SERVICE F
+UND,0.12998
65722417,"1193,1",1012,1132900,4150053,C2,Carrier Cost Recovery Fee,0.
+0273
$ perl test.pl
65722417,"1193-1",7980,1133566,4169735,035,"FEDERAL UNIVERSAL SERVICE
+FUND",0.12998
65722417,"1193-1",1012,1132900,4150053,C2,"Carrier Cost Recovery Fee",
+0.0273
Not being dynamic, not forcing fields with a space to be quoted, and using a file instead of data, that would be:
use 5.20.0;
use warnings;
use Text::CSV_XS;
my $csv = Text::CSV_XS->new ({
binary => 1,
auto_diag => 1,
quote_space => 0,
keep_meta_info => 11,
});
open my $io, "<", "test.csv";
my @row = ("") x 8; # The CSV has 8 columns
$csv->bind_columns (\(@row));
while ($csv->getline ($io)) {
$row[$_] =~ tr{,}{-} for grep { $csv->is_quoted ($_) } 0..$#row;
$csv->say (*STDOUT, \@row);
}
I'd be very interested in how much worse that performs on big datasets over your code.
Enjoy, Have FUN! H.Merijn