#!/usr/bin/perl
use warnings;
use strict;

use Getopt::Long;
#use MARC::Record;
#use MARC::File::XML ( BinaryEncoding => 'utf-8' );

# configuration hashref
my $conf  = {};
initialize($conf);

my %id;

open F, "<", $conf->{idfile};
while (<F>) {
	chomp;
	$id{$_} = 1;
}
close F;

my $M; my $I; my $S;
open $M, '<:utf8', $conf->{marcfile};
open $I, '>:utf8', $conf->{'output-import'};
open $S, '>:utf8', $conf->{'output-shelve'};

while (<$M>) {
    my $tag = $conf->{tag};
    my $sub = $conf->{subfield};

    /tag="$tag" ind1=" " ind2=" ">.*?<subfield code="$sub">(\d+)</;
    if ($conf->{incoming}) {
        print $S $_ if ($id{$1});
        print $I $_ unless ($id{$1});;
    } else {
        print $S $_ unless ($id{$1});
        print $I $_ if ($id{$1});;
    }
    $conf->{count}++;

    unless ($conf->{count} % 100) {
        print STDERR "\rProcessed: ",$conf->{count};
    }
}

=head2 initialize

Performs boring script initialization. Handles argument parsing,
mostly.

=cut

sub initialize {
    my ($c) = @_;
    my @missing = ();

    # set mode on existing filehandles
    binmode(STDIN, ':utf8');

    my $rc = GetOptions( $c,
                         'incoming',
                         'incumbent',
                         'tag|t=i',
                         'subfield|s=s',
                         'idfile|i=s',
                         'marcfile|m=s',
                         'outputimport|oi=s',
                         'outputshelved|os=s',
                         'help|h',
                       );
    show_help() unless $rc;
    show_help() if ($c->{help});

    $c->{'incoming-tag'}         = 903;
    $c->{'incoming-subfield'}    = 'a';
    $c->{'incumbent-tag'}         = 901;
    $c->{'incumbent-subfield'}    = 'c';
    my @keys = keys %{$c};
    unless ($c->{incoming} or $c->{incumbent}) {
        print "One of --incoming or --incumbent is required.\n";
        show_help();
    }
    if ($c->{incoming} and $c->{incumbent}) {
        print "Only one of --incoming or --incumbent can be specified.\n";
        show_help();
    }
    for my $key ('idfile', 'marcfile', 'output-import', 'output-shelved')
      { push @missing, $key unless $c->{$key} }
    if (@missing) {
        print "Required option: ", join(', ', @missing), " missing!\n";
        show_help();
    }

}


=head2 show_help

Display usage message when things go wrong

=cut

sub show_help {
print <<HELP;

The purpose of this utility is to split a MARCXML file in twain,
producing a set of records which will imported into Evergreen, and a
set of records which will not.

Usage is: $0 [REQUIRED ARGS]
Req'd Arguments
  --incoming  \\___ One (and only one) of these two must
  --incumbent /    be specified

  If --incoming is specified, the record ids in the file specified by
  --idfile will be used as EXCLUSION data. That is, the given record
  ids will be treated as records which match incumbent records and are
  being compressed into existing data, and so WILL NOT be
  imported. The --output-import file will contain records whose ids DO
  NOT occur in --idfile; --output-shelve will contain the records
  which DO occur.

  If --incumbent is specified, the reverse occurs.

  --idfile         -i   File of record ids to use as source for matchpoints
  --marcfile       -m   MARCXML source file
  --output-import  -oi  Output MARCXML file for records to be imported
  --output-shelve  -os  Output MARCXML file for records to be ignored

Optional Arguments
  --tag            -t   MARC tag to use as matchpoint (default 903 for incoming,
                        901 for incumbent)
  --subfield       -s   Subfield of tag to use (default 'c' for incoming, 'a'
                        for incumbent)
HELP
exit 1;
}