JBAS-980 bib linking force thesaurus; local thesaurus

Preliminary support for mapping local thesaurus values (ind2=7) to one
of the supported authority leader/11 options.

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

diff --git a/KCLS/linking/authority_control_fields.pl b/KCLS/linking/authority_control_fields.pl
index 61d737e..230e9a3 100755 (executable)
--- a/KCLS/linking/authority_control_fields.pl
+++ b/KCLS/linking/authority_control_fields.pl
@@ -567,6 +567,15 @@ my %AUTH_TO_BIB_IND2 = (
     'z' => '8'  # Other
 );
 
+# Maps 6XX subfield $2 values to their equivalent authority leader/11 value.
+# These occur when the indicator 2 == 7 ("Source specified in subfield $2")
+my %BIB_SF2_TO_AUTH = (
+    lcsh => 'a',
+    mesh => 'c',
+    nal  => 'd',
+    rvm  => 'v'
+);
+
 my $start_time = localtime();
 
 if($input_file) {
@@ -581,7 +590,25 @@ if($input_file) {
 # indicator 2 (thesaurus) value, returns the ID of the first 
 # authority record in the set that matches the thesaurus.
 sub find_matching_auth_for_thesaurus {
-    my ($e, $cfield_ind2, $auth_ids) = @_;
+    my ($e, $bib_field, $auth_ids) = @_;
+
+    my $cfield_ind2 = $bib_field->indicator(2) || '';
+
+    if ($cfield_ind2 eq '7') {
+        # subject thesaurus code is embedded in the bib field subfield 2
+
+        my $thesaurus = $bib_field->subfield('2') || '';
+        announce("Found local thesaurus value $thesaurus");
+
+        $cfield_ind2 = $BIB_SF2_TO_AUTH{$thesaurus} || '';
+
+        if ($cfield_ind2) {
+            announce("Mapped local thesaurus '$thesaurus' to '$cfield_ind2'");
+        } else {
+            announce("No mapping found for local thesaurus '$thesaurus'");
+            return undef;
+        }
+    }
 
     my $auth_leaders = $e->json_query({
         select => {afr => ['record', 'value']},
@@ -646,7 +673,7 @@ foreach my $rec_id (@records) {
                 }
                 next if !$match_tag;
 
-                announce("Searching for matches (auth tag=$match_tag): ".  
+                announce("Searching for matches (auth tag=$match_tag): \n\t".  
                     Dumper(\@searches));
 
                 my @tags = ($match_tag);
@@ -681,40 +708,31 @@ foreach my $rec_id (@records) {
                 # Find the best authority record to use for linking.
 
                 my $auth_id;
-                my $stop_looking = 0;
 
-                if ($bib_field->tag() ge 650 and $bib_field->tag() le 659) {
-                    # If the controlled bib field has an indicator 2 value 
-                    # indicating the thesarus, use the first authority
-                    # record in the set with the same thesaurus.
+                if ($bib_field->tag() =~ /^65[015]/) {
+                    # Using the indicator-2 value from the  controlled bib 
+                    # field, find the first authority in the list of matches
+                    # that uses the same thesaurus.  If no such authority 
+                    # is found, no matching occurs.
                     # TODO: perhaps this step should be part of the
-                    # validateion API search call above.
+                    # validation API search call above.
                    
-                    my $cfield_ind2 = $bib_field->indicator(2);
-
-                    if (defined $cfield_ind2) {
-
-                        $auth_id = find_matching_auth_for_thesaurus(
-                            $e, $cfield_ind2, $validates) || '';
-
-                        announce("Thesaurus match for ind2=$cfield_ind2 returned $auth_id");
+                    $auth_id = find_matching_auth_for_thesaurus(
+                        $e, $bib_field, $validates) || '';
 
-                        # If we don't find a matching authority record using 
-                        # the same thesuarus, we have nothing to link to.
-                        $stop_looking = 1;
-                    }
-                } 
-
-                # No best record found above.  Use the first in the list.
-                $auth_id = $validates->[0] unless $auth_id or $stop_looking;
+                } else {
+                    # For all other controlled fields, use the first 
+                    # authority record in the result set.
+                    $auth_id = $validates->[0];
+                }
 
                 # Don't exit here just because we have no $auth_id.  The
                 # bib field could have been changed above in the cleanup / 
                 # delete phase.
                 
                 if ($auth_id) {
-                    # Add the control number agency info from the matching
-                    # authority record to the controlled bib field.
+                    # Add the auth ID and control number agency info from the 
+                    # matching authority record to the controlled bib field.
 
                     my $auth_003 = $e->json_query({
                         select => {afr => ['value']},