trunk/bin/isi-merge.pl

#!/usr/bin/perl

use warnings;
use strict;
use autodie;
use Digest::MD5 qw(md5_hex);
use Digest::SHA1 qw(sha1_hex);
use Data::Dump qw(dump);

my $use_sha1 = $ENV{SHA1} || 1;

my @files = @ARGV;
@files = glob '/tmp/isi.*-*.txt' unless @files;

my $path = '/tmp/isi.full.txt';

warn "# ", $#files + 1, " files to $path sha:$use_sha1\n";

open(my $out_fh, '>', $path);
print $out_fh "FN ISI Export Format\nVR 1.0\n";

my $rec;
my $nr = 0;

my $md5;

my $report;

foreach my $file ( sort {
        my $a_r = $1 if $a =~ m{(\d+)-\d+};
        my $b_r = $1 if $b =~ m{(\d+)-\d+};
        $a_r <=> $b_r;
} @files ) {
        print STDERR $file;

        push @{ $report->{files} }, $file;

        open(my $fh, '<', $file);
        while(<$fh>) {
                next if m/^(FN|VR|EF)/;

                if ( ! m/^[\r\n]+$/s ) {
                        $rec .= $_;
                } else {
                        $nr++;
                        my $digest = $use_sha1 ? sha1_hex $rec : md5_hex $rec;
                        if ( my $times = $md5->{$digest} ) {
                                print STDERR 'd';
                                $report->{file}->{$file}->{duplicates}++;
                        } else {
                                print $out_fh $rec . $_;
                                $report->{file}->{$file}->{records}++;
                                $report->{total_records}++;
                                print STDERR '.';
                        }

                        $md5->{$digest}++;
                        $rec = '';
                }
        }
        warn "\n";
}

print $out_fh "EF\n";
close $out_fh;


warn "# $path ", -s $path, dump $report;

1	#!/usr/bin/perl
2
3	use warnings;
4	use strict;
5	use autodie;
6	use Digest::MD5 qw(md5_hex);
7	use Digest::SHA1 qw(sha1_hex);
8	use Data::Dump qw(dump);
9
10	my $use_sha1 = $ENV{SHA1} \|\| 1;
11
12	my @files = @ARGV;
13	@files = glob '/tmp/isi.-.txt' unless @files;
14
15	my $path = '/tmp/isi.full.txt';
16
17	warn "# ", $#files + 1, " files to $path sha:$use_sha1\n";
18
19	open(my $out_fh, '>', $path);
20	print $out_fh "FN ISI Export Format\nVR 1.0\n";
21
22	my $rec;
23	my $nr = 0;
24
25	my $md5;
26
27	my $report;
28
29	foreach my $file ( sort {
30	my $a_r = $1 if $a =~ m{(\d+)-\d+};
31	my $b_r = $1 if $b =~ m{(\d+)-\d+};
32	$a_r <=> $b_r;
33	} @files ) {
34	print STDERR $file;
35
36	push @{ $report->{files} }, $file;
37
38	open(my $fh, '<', $file);
39	while(<$fh>) {
40	next if m/^(FN\|VR\|EF)/;
41
42	if ( ! m/^[\r\n]+$/s ) {
43	$rec .= $_;
44	} else {
45	$nr++;
46	my $digest = $use_sha1 ? sha1_hex $rec : md5_hex $rec;
47	if ( my $times = $md5->{$digest} ) {
48	print STDERR 'd';
49	$report->{file}->{$file}->{duplicates}++;
50	} else {
51	print $out_fh $rec . $_;
52	$report->{file}->{$file}->{records}++;
53	$report->{total_records}++;
54	print STDERR '.';
55	}
56
57	$md5->{$digest}++;
58	$rec = '';
59	}
60	}
61	warn "\n";
62	}
63
64	print $out_fh "EF\n";
65	close $out_fh;
66
67
68	warn "# $path ", -s $path, dump $report;
69