#!perl

=head2

Compares two lists, which are compiled from a file of lines.
Each line is put in an array slot. 
Each item in one array is compared with each item in the second.
Prints out items that are not in one list or the other (separately).

=cut

###########################
### Initialize includes ###
### and basic needs     ###
###########################
use strict;
use MARC::BBMARC;
use List::Compare;

##########################
## Time coding routines ##
## Print start time and ##
## set start variable   ##
##########################

use Time::HiRes qw(  tv_interval );
# measure elapsed time 
my $t0 = [Time::HiRes::time()];
my $startingtime = MARC::BBMARC::startstop_time();
#########################
### Start main program ##
#########################

print ("Welcome to list comparison\n");

##### File handling initialization ######
#prompt for input file
print ("What is the first file (list1)? ");
my $file1=<>;
chomp $file1;
$file1 =~ s/^\"(.*)\"$/$1/;
open(IN1, "<$file1");

print ("What is the second file (list2)? ");
my $file2 = <>;
chomp $file2;
$file2 =~ s/^\"(.*)\"$/$1/;
open(IN2, "<$file2");

print ("What is the export file? ");
my $exportfile = <>;
chomp $exportfile;
$exportfile =~ s/^\"(.*)\"$/$1/;
open(OUT, ">$exportfile");


#if using MacPerl, set creator and type to BBEdit and Text
if ($^O eq 'MacOS') {
MacPerl::SetFileInfo('R*ch', 'TEXT', $exportfile);
}

############################################
# Set start time for main calculation loop #
############################################
my $t1 = [Time::HiRes::time()];
my $runningrecordcount=0;
###################################################
#### put lines from first file in array #####
my @list1 = ();
while (my $line = <IN1>) {
chomp $line;
push @list1, ($line);
###################################################
### add to count for user notification ###
$runningrecordcount++;
MARC::BBMARC::counting_print ($runningrecordcount);
###################################################
} # while

my @list2 = ();
while (my $line = <IN2>) {
chomp $line;
push @list2, ($line);
###################################################
### add to count for user notification ###
$runningrecordcount++;
MARC::BBMARC::counting_print ($runningrecordcount);
###################################################
} # while


########## Start comparison #########

my $lc = List::Compare->new(\@list1, \@list2);

#present in first file but not in second
my @list1only = $lc->get_unique;

#present in second file but not in first
my @list2only = $lc->get_complement;

#in one file or the other but not both
my @LorRonly = $lc->get_symmetric_difference;

#print the results
print OUT "The following were in $file1 only:\n";
print OUT join "\n", @list1only, "\n";

print OUT "\n", "-"x20;
print OUT "The following were in $file2 only:\n";
print OUT join "\n", @list2only, "\n";

print OUT "\n", "-"x20;
print OUT "The following were in $file1 or $file2 but not both:\n";
print OUT join "\n", @LorRonly, "\n";

print "Counts, list1 only: ", scalar @list1only, "\nlist2only: ", scalar @list2only, "\none or the other: ", scalar @LorRonly, "\n";

print OUT "\n", "-"x20;

#######################
#look for duplicate records in file1
print OUT "Duplicates in $file1:\n";
map {$_ =~ s/\s//g} @list1;
my %counts;
foreach my $list1item (@list1) {
$counts{$list1item}++;
}
foreach my $key (sort keys %counts) {
if ($counts{$key} > 1) {print OUT "$key\n";}
}

print OUT "\n", "-"x20;
#######################
#look for duplicate records in file2
print OUT "Duplicates in $file2:\n";
map {$_ =~ s/\s//g} @list2;
my %counts;
foreach my $list2item (@list2) {
$counts{$list2item}++;
}
foreach my $key (sort keys %counts) {
if ($counts{$key} > 1) {print OUT "$key\n";}
}

print OUT "\n", "-"x20;

close $file1;
close $file2;

close OUT;



##########################
### Main program done.  ##
### Report elapsed time.##
##########################

my $elapsed = tv_interval ($t0);
my $calcelapsed = tv_interval ($t1);
print sprintf ("%.4f %s\n", "$elapsed", "seconds from execution\n");
print sprintf ("%.4f %s\n", "$calcelapsed", "seconds to calculate\n");
my $endingtime = MARC::BBMARC::startstop_time();
print "Started at $startingtime\nEnded at $endingtime";

if ($^O eq 'MacOS') {
#set creator and type to BBEdit and Text
MacPerl::SetFileInfo('R*ch', 'TEXT', $exportfile);
}

print "\n\nPress Enter to quit";
<>;


#####################
### END OF PROGRAM ##
#####################

=head1 LICENSE

This code may be distributed under the same terms as Perl itself. 

Please note that this code is not a product of or supported by the 
employers of the various contributors to the code.

=head1 AUTHOR

Bryan Baldus
eija [at] inwave [dot] com

Copyright (c) 2003-2004

=cut