JBAS-980 Make auth-2-auth more aggressive

Remove all existing links for each processed auth-2-auth field before
rebuilding the links.  This lets us remove bogus links.

Signed-off-by: Bill Erickson <berickxx@gmail.com>

diff --git a/KCLS/linking/authority_authority_linker.pl b/KCLS/linking/authority_authority_linker.pl
index 97e6f01..0aecb2f 100755 (executable)
--- a/KCLS/linking/authority_authority_linker.pl
+++ b/KCLS/linking/authority_authority_linker.pl
@@ -51,6 +51,8 @@ my $db_host = $ENV{PGHOST} || 'localhost';
 my $db_port = $ENV{PGPORT} || '5432';
 my $db_user = $ENV{PGDATABASE} || 'evergreen';
 my $db_pass = $ENV{PGPASSWORD};
+my $links_removed = 0;
+my $links_added = 0;
 
 my %options;
 my $result = GetOptions(
@@ -101,7 +103,9 @@ my $query = q{
                     (sh2.atag = af2.id AND af2.main_entry IS NOT NULL 
                         AND af2.linking_subfield IS NOT NULL)
                 %s -- where clause here
-            EXCEPT SELECT target, source, field FROM authority.authority_linking
+            -- Ignore authority.authority_linking rows since we want to
+            -- rebuild all links, which may mean deleting bogus links.
+            -- EXCEPT SELECT target, source, field FROM authority.authority_linking
     ) x GROUP BY 1
 };
 
@@ -161,6 +165,8 @@ while (my ($src, $links) = $sth->fetchrow_array) {
         for my $link (split ';', $links) {
             my ($target, $field_id) = split ',', $link;
 
+            next if $target eq $src_rec->id;
+
             announce("Target: $target, field_id: $field_id");
 
             my $target_rec = $e->retrieve_authority_record_entry($target);
@@ -173,11 +179,13 @@ while (my ($src, $links) = $sth->fetchrow_array) {
 
             my $auth_target_thesaurus = 
                 substr($target_marc->field('008')->data(), 11, 1);
+
             announce("Target record thesaurus value=$auth_target_thesaurus");
 
-            # warn here, cleanup invalid links below
-            announce("Thesauri for source/target records do not match")
-                if $auth_src_thesaurus ne $auth_target_thesaurus;
+            if ($auth_src_thesaurus ne $auth_target_thesaurus) {
+                announce("Thesauri for source/target records do not match.  Skipping..");
+                next;
+            }
 
             my $cni = $target_marc->field('003')->data;
             my $acsaf = get_acsaf($e, $field_id);
@@ -186,19 +194,18 @@ while (my ($src, $links) = $sth->fetchrow_array) {
                 next;
             }
 
+            # start by removing all existing links for the current tag
             for my $field ($src_marc->field($acsaf->tag)) {
-
-                if ($auth_src_thesaurus ne $auth_target_thesaurus) {
-                    my @zeros = $field->subfield('0');
-                    announce("Existing links: @zeros");
-                    if (grep { $_ =~ qr/\)$target$/ } @zeros) {
-                        announce("Removing link(s) on ".$field->tag.
-                            " for src/target thesaurus mismatch");
-                        $field->delete_subfield(code => '0', match => qr/\)$target$/);
-                        $changed = 1;
-                    }
-                    next;
+                if (my $val = $field->subfield('0')) {
+                    announce("Removing existing subfield 0 : $val");
+                    $field->delete_subfield(code => '0');
+                    $changed = 1;
+                    $links_removed++;
                 }
+            }
+
+            # rebuild the links for the current tag
+            for my $field ($src_marc->field($acsaf->tag)) {
 
                 my $src_string = matchable_string(
                     $field, $acsaf->main_entry->display_sf_list, 
@@ -212,6 +219,7 @@ while (my ($src, $links) = $sth->fetchrow_array) {
                 my ($tfield) = $target_marc->field($acsaf->main_entry->tag);
 
                 if(defined $tfield) {
+
                     my $target_string = matchable_string(
                         $tfield, $acsaf->main_entry->display_sf_list, 
                         $acsaf->main_entry->joiner
@@ -221,6 +229,7 @@ while (my ($src, $links) = $sth->fetchrow_array) {
                         announce("Got a match");
                         $field->update('0' => "($cni)$target");
                         $changed = 1;
+                        $links_added++;
                     }
                 }
             }
@@ -257,6 +266,10 @@ if (@records) {
     announce("Processed all records; processed=$total_records; problems=$problems");
 }
 
+announce("links removed: $links_removed");
+announce("links added: $links_added");
+announce("delta added: ".($links_added - $links_removed));
+
 exit ($problems > 0);
 
 __END__