From d96ac51b83bb34b4315e9c2be6368011478f4fef Mon Sep 17 00:00:00 2001
From: miker <miker@dcc99617-32d9-48b4-a31d-7c20da2025e4>
Date: Mon, 7 Nov 2005 18:00:43 +0000
Subject: [PATCH] script to create SQL import file from MARC authority data

git-svn-id: svn://svn.open-ils.org/ILS/trunk@1973 dcc99617-32d9-48b4-a31d-7c20da2025e4
---
 .../src/extras/import/import_authority.pl     | 89 +++++++++++++++++++
 1 file changed, 89 insertions(+)
 create mode 100755 Evergreen/src/extras/import/import_authority.pl

diff --git a/Evergreen/src/extras/import/import_authority.pl b/Evergreen/src/extras/import/import_authority.pl
new file mode 100755
index 0000000000..c8559c3700
--- /dev/null
+++ b/Evergreen/src/extras/import/import_authority.pl
@@ -0,0 +1,89 @@
+#!/usr/bin/perl -w
+use strict;
+use XML::LibXML;
+use Time::HiRes qw/time/;
+use Getopt::Long;
+use Data::Dumper;
+use Error qw/:try/;
+use open qw/:utf8/;
+
+$|=1;
+
+my ($userid,$sourceid,$rec_id,$entry_file,$id_tag) = (1,2,1,'authority_entry.sql','//*[@tag="035"][1]');
+
+GetOptions (	
+	"sourceid"		=> \$sourceid,
+	"sql_output=s"		=> \$entry_file,
+	"userid=i"		=> \$userid,
+	"first=i"		=> \$rec_id,
+	"id_tag_xpath=s"	=> \$id_tag,
+);
+
+my $tcn_map;
+
+open RE, ">$entry_file" or die "Can't open $entry_file!  $!\n";
+
+print RE <<SQL;
+SET CLIENT_ENCODING TO 'UNICODE';
+COPY authority.record_entry (id,editor,creator,arn_value,marc,last_xact_id) FROM STDIN;
+SQL
+
+my $xact_id = time;
+
+my $parser = XML::LibXML->new;
+
+my $xml = '';
+while ( $xml .= <STDIN> ) {
+	chomp $xml;
+	next unless $xml;
+
+	my $tcn = '';
+	my $success = 0;
+	try {
+		my $doc = $parser->parse_string($xml);
+		my @nodes = $doc->documentElement->findnodes( $id_tag );
+		for my $n (@nodes) {
+			$tcn .= $n->textContent;
+		}
+		$tcn =~ s/^\s*(\.+)\s*/$1/o;
+		$tcn =~ s/\s+/_/go;
+		$success = 1;
+	} catch Error with {
+		my $e = shift;
+		warn $e;
+		warn $xml;
+	};	
+	next unless $success;
+
+	$xml =~ s/\\/\\\\/go;
+	$xml =~ s/\t/\\t/go;
+
+	$tcn =~ s/^.*?(\w+)\s*$/$1/go;
+	
+	unless ($tcn) {
+		warn "\nNo TCN found for rec # $rec_id\n";
+		$xml = '';
+		$rec_id++;
+		next;
+	}
+
+	if (exists($$tcn_map{$tcn})) {
+		warn "\n !! TCN $tcn already exists!\n";
+		$xml = '';
+		next;
+	}
+
+	print ".";
+	$$tcn_map{$tcn} = $rec_id;
+
+	print RE join("\t", ($rec_id,$userid,$userid,$tcn,$xml,$xact_id))."\n";
+
+	$rec_id++;
+	$xml = '';
+}
+
+print RE "\\.\n";
+print RE "SELECT setval('authority.record_entry_id_seq'::TEXT, $rec_id);\n";
+
+
+
-- 
2.43.2