JBAS-1417 New authority record linking script

Runs auth-to-auth and bib-to-auth linking for authority records created
since a specified date.

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

diff --git a/KCLS/linking/link-new-auth-records.pl b/KCLS/linking/link-new-auth-records.pl
new file mode 100755 (executable)
index 0000000..9a0f4a0
--- /dev/null
+++ b/KCLS/linking/link-new-auth-records.pl
@@ -0,0 +1,185 @@
+#!/usr/bin/perl
+# ----------------------------------------------------------------------
+# Find authority records newer than a specified age.  Once found,
+# run each through the auth-to-auth linking process.  Then locate
+# bib records that we might want to link to the new records and
+# pass them off to the bib-to-auth linker.
+# ----------------------------------------------------------------------
+use strict;
+use warnings;
+use DBI;
+use Getopt::Long;
+use DateTime;
+use Pod::Usage qw/pod2usage/;
+use Time::HiRes qw/usleep/;
+
+my @auth_ids;
+my @bib_ids;
+my $counter = 0;
+
+# options
+my $help;
+my $new_since;
+my $print_auth_ids;
+my $print_bib_ids;
+my $link_auths;
+my $link_bibs;
+my $progress;
+my $db_host = $ENV{PGHOST} || 'localhost';
+my $db_port = $ENV{PGPORT} || '5432';
+my $db_user = $ENV{PGDATABASE} || 'evergreen';
+my $db_pass = $ENV{PGPASSWORD};
+
+my $opt_result = GetOptions(
+    'new-since=i'       => \$new_since,
+    'print-bib-ids'     => \$print_bib_ids,
+    'print-auth-ids'    => \$print_auth_ids,
+    'link-bibs'         => \$link_bibs,
+    'link-auths'        => \$link_auths,
+    'progress'          => \$progress,
+    "db-host=s" => \$db_host,
+    "db-user=s" => \$db_user,
+    "db-pass=s" => \$db_pass,
+    "db-port=s" => \$db_port,
+    'help'      => \$help
+);
+
+sub announce {
+    my $msg = shift;
+    print DateTime->now(time_zone => 'local')->strftime('%F %T')." $msg\n";
+}
+
+pod2usage(0) if !$opt_result || $help;
+
+my $dsn = "dbi:Pg:database=evergreen;host=$db_host;port=$db_port";
+my $dbh = DBI->connect($dsn, $db_user, $db_pass) 
+    or die "Cannot connect to database: $dsn\n";
+
+$dbh->do('SET statement_timeout = 0');
+
+# ----------------------------------------------------------------------
+# Find the new authority record IDs
+
+my $sth = $dbh->prepare(<<SQL);
+    SELECT id FROM authority.record_entry 
+    WHERE DATE(create_date) >= DATE(NOW() - '$new_since day'::INTERVAL)
+SQL
+
+$sth->execute;
+while (my $ref = $sth->fetchrow_hashref()) {
+    push(@auth_ids, $ref->{id});
+}
+$sth->finish;
+
+my $auth_rec_count = scalar(@auth_ids);
+announce("Auth IDs: @auth_ids") if $print_auth_ids;
+
+if (!@auth_ids) {
+    announce("No authority records created in the last $new_since days");
+    exit 0;
+}
+
+# ----------------------------------------------------------------------
+# Auth-to-Auth linking
+
+if ($link_auths) {
+    # Pass all new authority records to the auth-to-auth linker
+    for my $rec_id (@auth_ids) {
+        system(
+            './authority_authority_linker.pl',
+            '--db-host', $db_host,
+            '--db-user', $db_user,
+            '--db-pass', ($db_pass || ''),
+            '--record', $rec_id
+        );
+
+        usleep(250000); # 1/4 second; allow ctrl-c to penetrate
+        announce("Auth records processed: $counter/$auth_rec_count") 
+            if $progress && ++$counter % 10 == 0;
+    }
+}
+$counter = 0;
+
+# Exit if there is nothing left to do.
+exit unless $print_bib_ids || $link_bibs;
+
+# ----------------------------------------------------------------------
+# Find bib records that we might want to link to the new authority 
+# record.
+#
+# Query: give me bib records that link to browse entries that also
+# link to exactly one authority record, specifically the new authority
+# records we are processing via this script.  Only include bib records
+# that are not already linked via bib_linking to said authority record.
+# This represents the set of bib records that might need to be linked 
+# to our new authority records.
+# ----------------------------------------------------------------------
+my %bib_ids; # de-dupe by record ID.
+my $auth_ids_param = join(',', @auth_ids);
+
+for my $axis (qw/author subject series title/) {
+    my $query = <<SQL;
+SELECT 
+    entry.id, 
+    are.id AS auth_record,
+    def.source AS bib_record
+FROM metabib.browse_${axis}_entry entry
+    JOIN metabib.browse_${axis}_entry_simple_heading_map map
+        ON (map.entry = entry.id)
+    JOIN authority.simple_heading ash ON (ash.id = map.simple_heading)
+    JOIN authority.record_entry are ON (are.id = ash.record)
+    JOIN metabib.browse_${axis}_entry_def_map def ON (def.entry = entry.id)
+    JOIN biblio.record_entry bre ON (bre.id = def.source)
+    JOIN (
+        -- we only care about browse entries that link to 
+        -- exactly one auth record, the auth record in question.
+        SELECT entry.id, COUNT(are.id)
+        FROM metabib.browse_${axis}_entry entry
+            JOIN metabib.browse_${axis}_entry_simple_heading_map map
+                ON (map.entry = entry.id)
+            JOIN authority.simple_heading ash 
+                ON (ash.id = map.simple_heading)
+            JOIN authority.record_entry are 
+                ON (are.id = ash.record)
+        WHERE NOT are.deleted
+        GROUP BY 1
+        HAVING COUNT(are.id) = 1
+    ) singles ON (singles.id = entry.id)
+    LEFT JOIN authority.bib_linking link 
+        ON (link.bib = def.source AND link.authority = are.id)
+WHERE 
+    bre.deleted IS FALSE
+    AND link.authority IS NULL -- unlinked records
+    AND are.id IN ($auth_ids_param)
+SQL
+
+    $sth = $dbh->prepare($query);
+    $sth->execute;
+    while (my $ref = $sth->fetchrow_hashref()) {
+        $bib_ids{$ref->{bib_record}} = 1; # de-dupe
+    }
+    $sth->finish;
+}
+
+@bib_ids = sort(keys(%bib_ids));
+my $bib_rec_count = scalar(@bib_ids);
+
+if ($link_bibs) {
+    for my $rec_id (@bib_ids) {
+        # fire off the linker for each of the records identied
+        system('./authority_control_fields.pl', 
+            '--db-host', $db_host,
+            '--db-user', $db_user,
+            '--db-pass', ($db_pass || ''),
+            '--record', $rec_id, 
+            '--refresh'
+        );
+
+        usleep(250000); # 1/4 second; allow ctrl-c to penetrate
+        announce("Bib records processed: $counter/$bib_rec_count") 
+            if $progress && ++$counter % 10 == 0;
+    }
+}
+
+announce("Bib IDs: @bib_ids") if $print_bib_ids;
+