[OPEN-ILS-GENERAL] How to Export Authority Records

Dan Scott dan at coffeecode.net
Thu May 24 11:52:27 EDT 2012


On Thu, May 24, 2012 at 08:43:02AM -0400, Sharp, Chris wrote:
> Hi Arayik,
> 
> I honestly don't know if this would work with a 1.4.0.4 database, but you might try using the latest version of the built-in MARC export script that comes with more recent versions of Evergreen:
> 
> http://git.evergreen-ils.org/?p=Evergreen.git;a=blob_plain;f=Open-ILS/src/support-scripts/marc_export.in;h=588f8ddd7f022b7d42fbf0b4c377ba4089cf58bc;hb=HEAD
> 
> I've had success using this on our 2.1.1 database.
> 
> Hope that helps!

I think it will work, having taken just a quick peek at the source code.

To make Araik's life easier, I'm attaching a built version of the script
(otherwise you would need to configure && make the code to get the
defaults for /openils/conf etc in place).

Hope this helps!

Dan
-------------- next part --------------
#!/usr/bin/perl
# vim:et:sw=4:ts=4:
use strict;
use warnings;
use bytes;

use OpenSRF::System;
use OpenSRF::EX qw/:try/;
use OpenSRF::AppSession;
use OpenSRF::Utils::JSON;
use OpenSRF::Utils::SettingsClient;
use OpenILS::Application::AppUtils;
use OpenILS::Utils::Fieldmapper;
use OpenILS::Utils::CStoreEditor;

use MARC::Record;
use MARC::File::XML ( BinaryEncoding => 'UTF-8' );
use UNIVERSAL::require;

use Time::HiRes qw/time/;
use Getopt::Long;


my @formats = qw/USMARC UNIMARC XML BRE ARE/;

my $config = '/openils/conf/opensrf_core.xml';
my $format = 'USMARC';
my $encoding = 'MARC8';
my $location = '';
my $dollarsign = '$';
my $idl = 0;
my $help = undef;
my $holdings = undef;
my $timeout = 0;
my $export_mfhd = undef;
my $type = 'biblio';
my $all_records = undef;
my $replace_001 = undef;
my @library = ();

GetOptions(
        'help'       => \$help,
        'items'      => \$holdings,
        'mfhd'       => \$export_mfhd,
        'all'        => \$all_records,
        'replace_001'=> \$replace_001,
        'location=s' => \$location,
        'money=s'    => \$dollarsign,
        'config=s'   => \$config,
        'format=s'   => \$format,
        'type=s'     => \$type,
        'xml-idl=s'  => \$idl,
        'encoding=s' => \$encoding,
        'timeout=i'  => \$timeout,
        'library=s'  => \@library,
);

if ($help) {
print <<"HELP";
This script exports MARC authority, bibliographic, and serial holdings
records from an Evergreen database. 

Input to this script can consist of a list of record IDs, with one record ID
per line, corresponding to the record ID in the Evergreen database table of
your requested record type.

Alternately, passing the --all option will attempt to export all records of
the specified type from the Evergreen database. The --all option starts at
record ID 1 and increments the ID by 1 until the largest ID in the database
is retrieved. This may not be very efficient for databases with large gaps
in their ID sequences.

Usage: $0 [options]
 --help or -h       This screen.
 --config or -c     Configuration file [/openils/conf/opensrf_core.xml]
 --format or -f     Output format (USMARC, UNIMARC, XML, BRE, ARE) [USMARC]
 --encoding or -e   Output encoding (UTF-8, ISO-8859-?, MARC8) [MARC8]
 --xml-idl or -x    Location of the IDL XML
 --timeout          Timeout for exporting a single record; increase if you
                    are using --holdings and are exporting records that
                    have a lot of items attached to them.
 --type or -t       Record type (BIBLIO, AUTHORITY) [BIBLIO]
 --all or -a        Export all records; ignores input list
 --library          Export the bibliographic records that have attached
                    holdings for the listed library or libraries as
                    identified by shortname
 --replace_001      Replace the 001 field value with the record ID

 Additional options for type = 'BIBLIO':
 --items or -i      Include items (holdings) in the output
 --money            Currency symbol to use in item price field [\$]
 --mfhd             Export serial MFHD records for associated bib records
                    Not compatible with --format=BRE
 --location or -l   MARC Location Code for holdings from
                    http://www.loc.gov/marc/organizations/orgshome.html

Examples:

To export a set of USMARC records in a file named "output_file" based on the
IDs contained in a file named "list_of_ids":
  cat list_of_ids | $0 > output_file

To export a set of MARC21XML authority records in a file named "output.xml"
for all authority records in the database:
  $0 --format XML --type AUTHORITY --all > output.xml

To export a set of USMARC bibliographic records encoded in UTF-8 in a file
named "sys1_bibs.mrc" based on records which have attached callnumbers for the
libraries with the short names "BR1" and "BR2":

  $0 --library BR1 --library BR2 --encoding UTF-8 > sys1_bibs.mrc

HELP
    exit;
}

if ($all_records && @library) {
    die('Incompatible arguments: you cannot combine a request for all ' .
        'records with a request for records by library');
}

$type = lc($type);
$format = uc($format);
$encoding = uc($encoding);

binmode(STDOUT, ':raw') if ($encoding ne 'UTF-8');
binmode(STDOUT, ':utf8') if ($encoding eq 'UTF-8');

if (!grep { $format eq $_ } @formats) {
    die "Please select a supported format.  ".
        "Right now that means one of [".
        join('|', at formats). "]\n";
}

if ($format ne 'XML') {
    my $type = 'MARC::File::' . $format;
    $type->require;
}

if ($timeout <= 0) {
    # set default timeout and/or correct silly user who 
    # supplied a negative timeout; default timeout of
    # 300 seconds if exporting items determined empirically.
    $timeout = $holdings ? 300 : 1;
}

OpenSRF::System->bootstrap_client( config_file => $config );

if (!$idl) {
    $idl = OpenSRF::Utils::SettingsClient->new->config_value("IDL");
}

Fieldmapper->import(IDL => $idl);

my $ses = OpenSRF::AppSession->create('open-ils.cstore');
OpenILS::Utils::CStoreEditor::init();
my $editor = OpenILS::Utils::CStoreEditor->new();

print <<HEADER if ($format eq 'XML');
<?xml version="1.0" encoding="$encoding"?>
<collection xmlns='http://www.loc.gov/MARC21/slim'>
HEADER

my %orgs;
my %shelves;

my $flesh = {};

if ($holdings) {
    get_bib_locations();
}

my $start = time;
my $last_time = time;
my %count = ('bib' => 0, 'did' => 0);
my $speed = 0;

if ($all_records) {
    my $top_record = 0;
    if ($type eq 'biblio') {
        $top_record = $editor->search_biblio_record_entry([
            {deleted => 'f'},
            {order_by => { 'bre' => 'id DESC' }, limit => 1}
        ])->[0]->id;
    } elsif ($type eq 'authority') {
        $top_record = $editor->search_authority_record_entry([
            {deleted => 'f'},
            {order_by => { 'are' => 'id DESC' }, limit => 1}
        ])->[0]->id;
    }
    for (my $i = 0; $i++ < $top_record;) {
        export_record($i);
    }
} elsif (@library) {
    my $recids = $editor->json_query({
        select => { bre => ['id'] },
        from => { bre => 'acn' },
        where => {
            '+bre' => { deleted => 'f' },
            '+acn' => { 
                deleted => 'f', 
                owning_lib => {
                    in => {
                        select => {'aou' => ['id'] },
                        from => 'aou',
                        where => { shortname => { in => \@library } }
                    } 
                }
            }
        },
        distinct => 1,
        order_by => [{
            class => 'bre',
            field => 'id',
            direction => 'ASC' 
        }]
    });

    foreach my $record (@$recids) {
        export_record($record->{id});
    }; 
} else {
    while ( my $i = <> ) {
        export_record($i);
    }
}
print "</collection>\n" if ($format eq 'XML');

$speed = $count{did} / (time - $start);
my $time = time - $start;
print STDERR <<DONE;

Exports Attempted : $count{bib}
Exports Completed : $count{did}
Overall Speed     : $speed
Total Time Elapsed: $time seconds

DONE

sub export_record {
    my $id = int(shift);

    my $bib; 

    my $r = $ses->request( "open-ils.cstore.direct.$type.record_entry.retrieve", $id, $flesh );
    my $s = $r->recv(timeout => $timeout);
    if (!$s) {
        warn "\n!!!!! Failed trying to read record $id\n";
        return;
    }
    if ($r->failed) {
        warn "\n!!!!!! Failed trying to read record $id: " . $r->failed->stringify . "\n";
        return;
    }
    if ($r->timed_out) {
        warn "\n!!!!!! Timed out trying to read record $id\n";
        return;
    }
    $bib = $s->content;
    $r->finish;

    $count{bib}++;
    return unless $bib;

    if ($format eq 'ARE' or $format eq 'BRE') {
        print OpenSRF::Utils::JSON->perl2JSON($bib);
        stats();
        $count{did}++;
        return;
    }

    try {

        my $r = MARC::Record->new_from_xml( $bib->marc, $encoding, $format );
        if ($type eq 'biblio') {
            add_bib_holdings($bib, $r);
        }

        if ($replace_001) {
            my $tcn = $r->field('001');
            if ($tcn) {
                $tcn->update($id);
            } else {
                my $new_001 = MARC::Field->new('001', $id);
                $r->insert_fields_ordered($new_001);
            }
        }

        if ($format eq 'XML') {
            my $xml = $r->as_xml_record;
            $xml =~ s/^<\?.+?\?>$//mo;
            print $xml;
        } elsif ($format eq 'UNIMARC') {
            print $r->as_usmarc;
        } elsif ($format eq 'USMARC') {
            print $r->as_usmarc;
        }

        $count{did}++;

    } otherwise {
        my $e = shift;
        warn "\n$e\n";
        import MARC::File::XML; # reset SAX parser so that one bad record doesn't kill the entire export
    };

    if ($export_mfhd and $type eq 'biblio') {
        my $mfhds = $editor->search_serial_record_entry({record => $id, deleted => 'f'});
        foreach my $mfhd (@$mfhds) {
            try {
                my $r = MARC::Record->new_from_xml( $mfhd->marc, $encoding, $format );

                if ($format eq 'XML') {
                    my $xml = $r->as_xml_record;
                    $xml =~ s/^<\?.+?\?>$//mo;
                    print $xml;
                } elsif ($format eq 'UNIMARC') {
                    print $r->as_usmarc;
                } elsif ($format eq 'USMARC') {
                    print $r->as_usmarc;
                }
            } otherwise {
                my $e = shift;
                warn "\n$e\n";
                import MARC::File::XML; # reset SAX parser so that one bad record doesn't kill the entire export
            };
        }
    }

    stats() if (! ($count{bib} % 50 ));
}

sub stats {
    try {
        no warnings;

        $speed = $count{did} / (time - $start);

        my $speed_now = ($count{did} - $count{did_last}) / (time - $count{time_last});
        my $cn_speed = $count{cn} / (time - $start);
        my $cp_speed = $count{cp} / (time - $start);

        printf STDERR "\r  $count{did} of $count{bib} @  \%0.4f/s ttl / \%0.4f/s rt ".
                "($count{cn} CNs @ \%0.4f/s :: $count{cp} CPs @ \%0.4f/s)\r",
                $speed,
                $speed_now,
                $cn_speed,
                $cp_speed;
    } otherwise {};
    $count{did_last} = $count{did};
    $count{time_last} = time;
}

sub get_bib_locations {
    print STDERR "Retrieving Org Units ... ";
    my $r = $ses->request( 'open-ils.cstore.direct.actor.org_unit.search', { id => { '!=' => undef } } );

    while (my $o = $r->recv) {
        die $r->failed->stringify if ($r->failed);
        $o = $o->content;
        last unless ($o);
        $orgs{$o->id} = $o;
    }
    $r->finish;
    print STDERR "OK\n";

    print STDERR "Retrieving Shelving locations ... ";
    $r = $ses->request( 'open-ils.cstore.direct.asset.copy_location.search', { id => { '!=' => undef } } );

    while (my $s = $r->recv) {
        die $r->failed->stringify if ($r->failed);
        $s = $s->content;
        last unless ($s);
        $shelves{$s->id} = $s;
    }
    $r->finish;
    print STDERR "OK\n";

    $flesh = { flesh => 2, flesh_fields => { bre => [ 'call_numbers' ], acn => [ 'copies' ] } };
}

sub add_bib_holdings {
    my $bib = shift;
    my $r = shift;

    my $cn_list = $bib->call_numbers;
    if ($cn_list && @$cn_list) {

        $count{cn} += @$cn_list;
    
        my $cp_list = [ map { @{ $_->copies } } @$cn_list ];
        if ($cp_list && @$cp_list) {

            my %cn_map;
            push @{$cn_map{$_->call_number}}, $_ for (@$cp_list);
                            
            for my $cn ( @$cn_list ) {
                my $cn_map_list = $cn_map{$cn->id};

                for my $cp ( @$cn_map_list ) {
                    $count{cp}++;
                            
                    $r->insert_grouped_field( MARC::Field->new( '852', '4', ' ',
                            ($location ? ( 'a' => $location ) : ()),
                            b => $orgs{$cn->owning_lib}->shortname,
                            b => $orgs{$cp->circ_lib}->shortname,
                            c => $shelves{$cp->location}->name,
                            j => $cn->label,
                            ($cp->circ_modifier ? ( g => $cp->circ_modifier ) : ()),
                            p => $cp->barcode,
                            ($cp->price ? ( y => $dollarsign.$cp->price ) : ()),
                            ($cp->copy_number ? ( t => $cp->copy_number ) : ()),
                            ($cp->ref eq 't' ? ( x => 'reference' ) : ()),
                            ($cp->holdable eq 'f' ? ( x => 'unholdable' ) : ()),
                            ($cp->circulate eq 'f' ? ( x => 'noncirculating' ) : ()),
                            ($cp->opac_visible eq 'f' ? ( x => 'hidden' ) : ()),
                        )
                    );

                    stats() if (! ($count{cp} % 100 ));
                }
            }
        }
    }
}


More information about the Open-ils-general mailing list