[RT17143] Retrieving relevant record ids

This code now retrieves record ids of records with bad date1 values and
proper 260 $c or 264 $c that can replace them.  Next up, retrieve MARC
via OpenSRF, edit it via MARC::Record and save the updated MARC record
via OpenSRF.

diff --git a/data_cleanup/date1/date1_cleanup.pl b/data_cleanup/date1/date1_cleanup.pl
index df549fb..236965a 100644 (file)
--- a/data_cleanup/date1/date1_cleanup.pl
+++ b/data_cleanup/date1/date1_cleanup.pl
@@ -1,5 +1,5 @@
 #!/usr/bin/perl
-# vim:et:ts=4:
+# vim:et:ts=4:sw=4:
 use strict;
 use warnings;
 
@@ -32,10 +32,36 @@ my $confxml = $xmlparser->parse_file($core_config);
 my $confxpc = XML::LibXML::XPathContext->new($confxml);
 my $osrfxml = $xmlparser->parse_file($settings_config);
 
-check_all_database_connections();
+my $dbh = init_database_connections();
 
-sub check_all_database_connections {
-    print "\nChecking database connections\n";
+get_date1_records($dbh);
+
+$dbh->disconnect;
+
+sub get_date1_records {
+    my ($dbh) = @_;
+
+    # Get list of server languages
+    my $sth = $dbh->prepare("SELECT DISTINCT mrfr.record
+        FROM metabib.real_full_rec mrfr INNER JOIN biblio.record_entry bre ON mrfr.record = bre.id
+        INNER JOIN asset.call_number acn ON bre.id = acn.record
+        INNER JOIN asset.copy ac ON acn.id = ac.call_number
+        INNER JOIN actor.org_unit aou ON ac.circ_lib = aou.id
+        WHERE mrfr.record IN (SELECT record
+            FROM metabib.real_full_rec
+            WHERE tag = '008' AND substring(value, 8, 4) ~ '(^\\s*\$|^\\d{1,3}\$|[^0-9\\su]|203[1-9])')
+        AND (tag = '260' OR tag = '264') AND subfield = 'c' AND value ~ '\\d{1,4}' AND bre.deleted = false AND aou.shortname NOT IN ('MWP', 'AB', 'LB', 'ITC', 'HBCA')");
+    $sth->execute;
+    my $records = $sth->fetchall_arrayref([0]);
+    $sth->finish;
+
+    for (@$records) {
+        print $_->[0] . "\n";
+    }
+}
+
+sub init_database_connections {
+    print "\nInitializing database connection\n";
     # Check database connections
     my @databases = $osrfxml->findnodes('//database');
 
@@ -71,25 +97,26 @@ sub check_all_database_connections {
             next unless $node->nodeType == XML::LibXML::XML_ELEMENT_NODE;
             $osrf_xpath .= "/" . $node->nodeName;
         }
-        $output .= test_db_connect($db_name, $db_host, $db_port, $db_user, $db_pw, $osrf_xpath);
+
+        my $dbh = db_connect($db_name, $db_host, $db_port, $db_user, $db_pw, $osrf_xpath);
+
+        return $dbh;
     }
 }
 
-sub test_db_connect {
+sub db_connect {
     my ($db_name, $db_host, $db_port, $db_user, $db_pw, $osrf_xpath) = @_;
 
     my $dsn = "dbi:Pg:dbname=$db_name;host=$db_host;port=$db_port";
-    my $de = undef;
-    my ($dbh, $encoding, $langs);
+    my $dbh;
+
     $dbh = DBI->connect($dsn, $db_user, $db_pw);
 
     # Short-circuit if we didn't connect successfully
     unless($dbh) {
-        $de = "* $osrf_xpath :: Unable to connect to database $dsn, user=$db_user, password=$db_pw\n";
         warn "* $osrf_xpath :: Unable to connect to database $dsn, user=$db_user, password=$db_pw\n";
-        return $de;
+        return -1;
     }
 
-    $dbh->disconnect;
-    print "* $osrf_xpath :: Successfully connected to database $dsn\n" unless ($de);
+    return $dbh;
 }