Format, empty vars, and fixed-width records

[kwaping]

I need to output data to a fixed-width record file. The lines need to be exactly 129 characters long. I thought format would be great for this, and it has been - except for one issue... The line only prints up to the last variable that contains a non-pad character. My current code runs out of data at position 92, after which everything is undef. That is also where the output line ends. Unfortunately, I need it to pad out to the 129 character requirement.

Can anyone help? Here's what I've tried already:

• Making the final variable an empty string
• Making the final variable a single space
• Removing everything in the format picture after the last data-containing variable and hard-coding a run of spaces

None of those ideas worked - the line was still truncated at 92 characters in each instance.

I also tried using sprintf, but if the data is longer than intended, the line becomes too long. I'd rather not substr and sprintf every piece of data, but I'll do it if that's what's required.

Example code (disregard unintialized and "not a number" warnings):

#!/usr/bin/perl

use strict;
use warnings;

my (
    $emp_number,
    $name,
    $override_dept,
    $job,
    $shift,
    $d_e,
    $d_e_code,
    $override_rate,
    $hours,
    $year,
    $month,
    $day,
    $filler1,
    $filler2,
    $amount,
    $seq_num,
    $override_div,
    $override_branch,
    $override_state,
    $override_local,
    $state_misc,
    $rate,
    $ssn,
) = 'A' .. 'P';

format XXX =
@>>>>>@>>>>>>>>>>>>>>>>>>>>>>>>@>>>>>@>>>>>>>>>>>@@@>@#####.##@####.##
+@>>>@>@>@>@>@#####.##@@>>>>>@>>>>>@>@>>>>>>>>>@@@>>>>>>>>>>
$emp_number,$name,$override_dept,$job,$shift,$d_e,$d_e_code,$override_
+rate,$hours,$year,$month,$day,$filler1,$filler2,$amount,$seq_num,$ove
+rride_div,$override_branch,$override_state,$override_local,$state_mis
+c,$rate,$ssn
.

my $record = '';
open my ($out), '>', \$record or die $!;

my $oldfh = select($out);
$~ = 'XXX';

write;

select($oldfh);
close($out) or die $!;

print $record;

Update: I've decided to go with the substr(sprintf()) method after all. Not the prettiest or most elegant, but it works.

[BrowserUk]

I also tried using sprintf, but if the data is longer than intended, the line becomes too long.

You do know you can specify the maximum width?

printf "%2.2s>%32.32s<\n", 
    $_, 'x' x $_ for 30 .. 34, 99..101;;
30>  xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx<
31> xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx<
32>xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx<
33>xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx<
34>xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx<
99>xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx<
10>xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx<
10>xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx<

[kwaping]

Thank you for mentioning that, as I did not know about that detail. However, after reading the perldoc on it, I don't believe that will limit the lengths of my numeric fields, which may be a problem.

[BrowserUk]

As I showed with the decimal field at the beginning of the records above, you can truncate integer fields by using %N.Ns. Unfortunately, the truncation always occurs on the right regardless of whether you use '-' or not:

[0] Perl> printf "%2.2s\n", $_ for 9, 99 .. 101;;
 9
99
10
10
[0] Perl> printf "%-2.2s\n", $_ for 9, 99 .. 101;;
9
99
10
10

Which may not be what you want. The real problem (sic) is floating point numbers, there's just no sensible way to squeeze them into a fixed width field if they stray beyond a limited range:

printf "%-8.8s\n", sprintf '%8f', $_ 
    for map{ 1.23456789 * $_} map{ "1e$_" } -10 .. 10;;
0.000000
0.000000
0.000000
0.000000
0.000001
0.000012
0.000123
0.001235
0.012346
0.123457
1.234568
12.34567
123.4567
1234.567
12345.67
123456.7
1234567.
12345678
12345678
12345678
12345678

And its worse if you want or need to maintain point alignment.

[runrig]

I would probably use Parse::FixedLength for this...at least for the debugging phase...but I'm biased (since I'm the author). You could just use pack (with "A" for the format types) though.

[kwaping]

Thanks for the suggestions! I'm going to research your proposed solutions. I did see your module when doing a CPAN search, but the naming didn't make it obvious that it could create fixed-length strings as well as read them.

I have a related suggestion for your module. Please consider creating a new parameter to new called "justify" to ease the creation of records where every field must be right-justified.

[state-o-dis-array]

When I change the line
) = 'A' .. 'P';
to
) = 'A' .. 'W';
it seems to work for me as you are seeking.

[kwaping]

Unfortunately, I can't change the data to make the line print out to character 129. I have to have nothing but pad characters (spaces) from the last real data out to the end of the record.

[state-o-dis-array]

Sorry, that is evident after a more careful reading of the original post.