how to unpack little or big endian in Convert::IBM390

Question

I'm using Convert::IBM390 to convert EBCDIC files to ASCII files.

#!/usr/bin/perl -w

use warnings;
use Convert::IBM390 qw(:all);

open EBCDIC, "<D:/EBCDIC.txt" or die "error -> opening $!";
open ASCII, ">D:/ASCII.txt" or die "error -> opening $!";

my $text;
my $template = 'e15.0 e15 z4 I2 I2 i2 N16.0 p11.0';
binmode EBCDIC;
while (read (EBCDIC, $buffer, 67))
{
   @fields = unpackeb($template, $buffer);
   $text= join(",",@fields);
   print ASCII $text."\n";
}

close EBCDIC;
close ASCII;

I got this script in this link

I have problems when the EBCDIC data contain Little or Big Endian integers.

I had searched for unpacking those character and used N/n V/v but those things are not accepted in this module. got error as

Invalid type in unpackeb: 'N'

The EBCDIC FILE from the mainframe consists of following columns:

EBCDIC Decimal(15,0)
EBCDIC String(15)
Zoned Decimal(4)
unsigned little endian integer(2)
unsigned big endian integer(2)
signed big endian integer(2)
big endian decimal(16,0)
packed decimal(11,0)

Any suggestions ?

Answer 1

Do you realise that all multi-byte binary numbers must be either little- or big-endian?

I assume it's an i/I or s/S field that you want to reverse the endianness of? It doesn't appear that Convert::IBM390 will let you do that, but I'm not surprised as IBM System/360, System/370 and ESA/390 are all big-endian and you should never get any data that is little-endian. Where did your data come from?

My best suggestion is that you write a simple conversion routine that takes a 32-bit value and returns the same value with its endianness reversed. Here's a program that implements swap_endianness_32 and demonstrates its functionality.

use strict;
use warnings;

printf "%08X\n", swap_endianness_32(0x12345678);

sub swap_endianness_32 {
    unpack 'N', pack 'V', shift;
}

output

78563412