From 61fc1ecb23681a7ea7c50e5ba7984ee0f67376e3 Mon Sep 17 00:00:00 2001
From: Jason Etheridge <jason@esilibrary.com>
Date: Fri, 19 Dec 2014 16:29:04 -0500
Subject: [PATCH] Option to exclude specific records by id

--exclude-record-id-file=<file containing list of record ids, one per line>

Signed-off-by: Jason Etheridge <jason@esilibrary.com>
---
 Open-ILS/src/support-scripts/marc_export.in | 52 ++++++++++++++++++++++-------
 1 file changed, 40 insertions(+), 12 deletions(-)

diff --git a/Open-ILS/src/support-scripts/marc_export.in b/Open-ILS/src/support-scripts/marc_export.in
index cb09985eff..b8bd45fa13 100755
--- a/Open-ILS/src/support-scripts/marc_export.in
+++ b/Open-ILS/src/support-scripts/marc_export.in
@@ -86,6 +86,7 @@ sub new {
                'exclude-item-id-file=s',
                'exclude-item-barcode-file=s',
                'exclude-item-circmod-file=s',
+               'exclude-record-id-file=s',
                'mfhd',
                'all',
                'replace_001',
@@ -118,18 +119,19 @@ 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 [@sysconfdir@/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          Remains for backward compatibility. No longer used.
- --type or -t       Record type (BIBLIO, AUTHORITY) [BIBLIO]
- --all or -a        Export all records; ignores input list
- --replace_001      Replace the 001 field value with the record ID
- --store            Use the given storage backend to connect to the database.
-                    Choices are (reporter, cstore, storage) [reporter]
- --since            Export records modified since a certain date and time.
+ --help or -h                  This screen.
+ --config or -c                Configuration file [@sysconfdir@/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                     Remains for backward compatibility. No longer used.
+ --type or -t                  Record type (BIBLIO, AUTHORITY) [BIBLIO]
+ --all or -a                   Export all records; ignores input list
+ --replace_001                 Replace the 001 field value with the record ID
+ --store                       Use the given storage backend to connect to the database.
+                               Choices are (reporter, cstore, storage) [reporter]
+ --since                       Export records modified since a certain date and time.
+ --exclude-record-id-file      Specify a file containing list of record ids to exclude
 
  Additional options for type = 'BIBLIO':
  --items or -i                 Include items (holdings) in the output
@@ -203,6 +205,20 @@ HELP
     if ($opts{format} eq 'BRE' && $opts{items}) {
         die "Format BRE is not compatible with exporting holdings."
     }
+    $opts{'exclude-record-ids'} = {};
+    if ($opts{'exclude-record-id-file'}) {
+        if (-e $opts{'exclude-record-id-file'}) {
+            open FILE, $opts{'exclude-record-id-file'};
+            while (my $record_id = <FILE>) {
+                chomp $record_id;
+                $opts{'exclude-record-ids'}->{$record_id} = 1;
+            }
+            close FILE;
+        } else {
+            die "Could not find file for exclude-record-id-file: " . $opts{'exclude-record-id-file'};
+        }
+    }
+
     if ($opts{'exclude-itemless-bibs'} && ! $opts{items}) {
         die "If excluding itemless bibs, must export holdings.";
     }
@@ -468,6 +484,10 @@ sub next {
     # until we run out.  These sres "go with" the previous bib record.
     if ($Marque::config->option_value('mfhd') && $self->{mfhds} && @{$self->{mfhds}}) {
         $r = shift(@{$self->{mfhds}});
+        if (defined $Marque::config->option_value('exclude-record-ids')->{ $r->id() }) {
+            import MARC::File::XML; # Reset SAX Parser.
+            return $self->next(); # skip this record
+        }
         eval {
             $marc = MARC::Record->new_from_xml($r->marc(),
                                                $Marque::config->option_value('encoding'),
@@ -483,6 +503,10 @@ sub next {
         my $data = $self->{sth}->fetchrow_hashref;
         if ($data) {
             $r = $self->{breClass}->from_bare_hash($data);
+            if (defined $Marque::config->option_value('exclude-record-ids')->{ $r->id() }) {
+                import MARC::File::XML; # Reset SAX Parser.
+                return $self->next(); # skip this record
+            }
             if ($Marque::config->option_value('format') eq 'BRE') {
                 $output = OpenSRF::Utils::JSON->perl2JSON($r);
             } else {
@@ -784,6 +808,10 @@ sub next {
     if ($data) {
         my $format = $Marque::config->option_value('format');
         my $are = $self->{fmClass}->from_bare_hash($data);
+        if (defined $Marque::config->option_value('exclude-record-ids')->{ $are->id() }) {
+            import MARC::File::XML; # Reset SAX Parser.
+            return $self->next(); # skip this record
+        }
         if ($format eq 'ARE') {
             $output = OpenSRF::Utils::JSON->perl2JSON($are);
         } else {
-- 
2.11.0