#!/usr/bin/perl

# Author: Kevin P. Inscoe <kevin@inscoe.org>.
# File: split.pl
# Date of creation: May 18, 2006.
# Warranty: None expressed or implied.
# License: The Open Software License. V1.1
# http://www.opensource.org/licenses/osl.php
#
# OSI Certified Open Source Software. http://www.opensource.org/licenses/
#
# Prerequisites: Perl 5.004 (minimum).

# The purpose of this program is to take large input files and split them
# into smaller files by size in sequential order ending in .nnn zero padded
#
# Usage: split.pl file <-b sizem/g> <-v>
#
#       -b split at number of bytes (or megabtes/m or gigabytes/g)
#               Default is 100mb files.
#       -v Turns on logging
# 

# Requirements
require 5.004;
use strict;
use Getopt::Std;

# Globals
use vars qw($opt_b $opt_v);
# default to 100 mb files
my $sizebreak=100000000;
my $size=0;
my $tmp=0;
my $filename;
my $currentsize=0;
my $bytecount=0;
my $recs=0;
my $files=1;
my $ctr=1;
my $newfile;
my $pad;

# Get our parameters
print "Raw parameters: @ARGV\n" if ($opt_v);

my $status = getopts('vb:');
($status == 0 or @ARGV < 1) and die ("Usage: $0 [-v] [-b size (m/g)] file\n");

getopts('vb:');                 # -v is a boolean flags, -b takes an argument
                                # Sets opt_* as a side effect.

print "Split size is $opt_b\n" if ($opt_v);

# does the file exists?
$filename = $ARGV[$#ARGV];

print "Input file is $filename\n" if ($opt_v);

# if opt_b was set else use the default set above
if ($opt_b ne '') {
        $tmp=$opt_b;

# Convert size nomenclature to numbers
        if ($opt_b =~ m/m$/) {
# it's megabytes
                $tmp =~ s/m//g; 
                $sizebreak = $tmp * 1000000;
        }

        if ($opt_b =~ m/g$/) {
# it's gigabytes
                $tmp =~ s/g//g;
                $sizebreak = $tmp * 1000000000;
        }
}

print "Size break is $sizebreak;\n" if ($opt_v);

if (-e $filename) {

        $newfile=$filename . ".001";
        open(IN, "<$filename");
        open(OUT, ">$newfile");
        while(<IN>) {
                $recs++;
                # Get the number of bytes in current buffer
                $size = length($_);
                if ( $currentsize > $sizebreak ) {
                        $currentsize=$size;
                        $ctr++;
                        close (OUT);
                        $pad = &pad_zeros($ctr);
                        $newfile=$filename . "." . $pad;
                        open(OUT, ">$newfile");
                        $files++;
                        print OUT;
                } else {
                        $currentsize = $currentsize + $size;
                        $bytecount = $bytecount + $size;
                        print OUT;
                }
        }

# Cleanup and summary report
close(IN);
close(OUT);
print "\n$recs lines were processed into $files file(s), $bytecount total
bytes.\n";
exit;
}
        else {
                print "$0 file not found - $filename, aborting...\n";
                exit -1;
        }

exit 0;

sub pad_zeros ()

{
        my ($unpadded, $size);
        ($unpadded) = @_;
        $size = length($unpadded);
        if ( $size > 3 ) { 
                print "ERROR: too many files!\n";
                return $unpadded;
        }

        if ( $size == 3 ) { return $unpadded; }
        if ( $size == 2 ) { return "0" . $unpadded; }
        if ( $size == 1 ) { return "00" . $unpadded; }
        if ( $size == 0 ) { 
                print "ERROR: empty counter at pad_zeros()!\n";
                exit -1;
        }
}