Re: Text::CSV - parsing

[Eily]

Actually Text::CSV handles quoted fields pretty well, but the default quote character is ". See the quote_char section of the Text::CSV. Note that the sep_char is a single character though, so spaces are not removed, which would make 101, '1997' invalid CSV, because the quote_char should be the next thing after the sep_char. To allow spaces, set allow_whitespace to 1:

use strict; use Text::CSV;
use Data::Dumper; 
my @lines=<DATA>;
my $csv = Text::CSV->new ({sep_char  => ',' , quote_char => "'", allow
+_whitespace => 1});

my @AoA;

for (@lines) { if    (/^\((.+?)\).?/){my $con=$1; 
  if ($csv->parse($con)) {
 
      my @fields = $csv->fields();   push @AoA,[@fields];
 
  } else {     warn "Line could not be parsed: $_\n";  }
}
}

print Dumper \@AoA;


__DATA__
(101, '1997-02-25', 'S1', 31.00, NULL, 0.00, 'this becomes two fields,
+ so no go', 5.11),
(102, '1998-03-26', 'S1', 31.00, NULL, 0.00, 'this will remain one fie
+ld', 6.11),

$VAR1 = [
          [
            '101',
            '1997-02-25',
            'S1',
            '31.00',
            'NULL',
            '0.00',
            'this becomes two fields, so no go',
            '5.11'
          ],
          [
            '102',
            '1998-03-26',
            'S1',
            '31.00',
            'NULL',
            '0.00',
            'this will remain one field',
            '6.11'
          ]
        ];

[haukex]

Good post, just a note: this will not work right if the quoted strings happen to be broken onto two lines. In that case, Text::CSV's getline is better (see the section "Embedded newlines" in the doc):

use warnings;
use strict;
use Text::CSV;
use Data::Dump;

my $csv = Text::CSV->new ({ binary=>1, auto_diag=>2,
    quote_char=>"'", allow_whitespace=>1 });
while ( my $row = $csv->getline( \*DATA ) ) {
    dd $row;
}
$csv->eof;

__DATA__
(101, '1997-02-25', 'S1', 31.00, NULL, 0.00, 'this becomes two fields,
so no go', 5.11),
(102, '1998-03-26', 'S1', 31.00, NULL, 0.00, 'this
will remain one field', 6.11),

Output:

[
  "(101",
  "1997-02-25",
  "S1",
  "31.00",
  "NULL",
  "0.00",
  "this becomes two fields,\nso no go",
  "5.11)",
  "",
]
[
  "(102",
  "1998-03-26",
  "S1",
  "31.00",
  "NULL",
  "0.00",
  "this\nwill remain one field",
  "6.11)",
  "",
]

[Eily]

Well, this doesn't deal with the parentheses properly, so this would make your ++solution a better one then. You don't have to worry about embedded newlines, enclosing parentheses, the non default quote character or extra white space. And parsing DB data as SQL makes sense :)