JBAS-1437 link-new-auths new options, ID printing

--start-auth-id
--max-auth-count

Also, print-x-ids now prints IDs one per line without any logging
header.

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
index fb8c1fa..c48939e 100755 (executable)
--- a/KCLS/linking/link-new-auth-records.pl
+++ b/KCLS/linking/link-new-auth-records.pl
@@ -19,8 +19,9 @@ my $counter = 0;
 
 # options
 my $help;
-my $new_since;
 my $modified_since;
+my $max_auth_count;
+my $start_auth_id;
 my $print_auth_ids;
 my $print_bib_ids;
 my $link_auths;
@@ -32,8 +33,9 @@ my $db_user = $ENV{PGDATABASE} || 'evergreen';
 my $db_pass = $ENV{PGPASSWORD};
 
 my $opt_result = GetOptions(
-    'new-since=i'       => \$new_since,
     'modified-since=i'  => \$modified_since,
+    'max-auth-count=i'  => \$max_auth_count,
+    'start-auth-id=i'   => \$start_auth_id,
     'print-bib-ids'     => \$print_bib_ids,
     'print-auth-ids'    => \$print_auth_ids,
     'link-bibs'         => \$link_bibs,
@@ -62,16 +64,20 @@ sub help {
             $0 --modified-since 1 --link-auths --link-bibs
         
         Options:
-            
-            --new-since <days>
-                Process authority records created within the last <days> days
-                If --modified-since is also applied, the union of both record
-                sets are processed (via OR query).
 
             --modified-since <days>
-                Process authority records modified within the last <days> days
-                If --new-since is also applied, the union of both record
-                sets are processed (via OR query).
+                Process authority records created or modified within the 
+                last <days> days.
+
+            --max-auth-count <count>
+                Process <count> authority records in total.  Use with
+                --start-auth-id to process batches of records across
+                multiple instances of the script.
+
+            --start-auth-id <id>
+                Process authority records whose ID is equal to or greater
+                than <id>.  Use with --max-auth-count to process batches
+                of records accross multiple runs of the script.
 
             --print-auth-ids
                 Print authority record IDs to process to STDOUT
@@ -109,18 +115,18 @@ my $dbh = DBI->connect($dsn, $db_user, $db_pass)
 $dbh->do('SET statement_timeout = 0');
 
 # ----------------------------------------------------------------------
-# Find the new authority record IDs
-my $where = 'WHERE ';
-
-$where .= "DATE(create_date) >= DATE(NOW() - '$new_since day'::INTERVAL)" 
-    if $new_since;
-
-$where .= ' OR ' if $new_since && $modified_since;
-
-$where .= "DATE(edit_date) >= DATE(NOW() - '$modified_since day'::INTERVAL)"
-    if $modified_since;
+# Load the authority record IDs
+my $where2 = $start_auth_id ? "AND id >= $start_auth_id" : '';
+my $limit = $max_auth_count ? "LIMIT $max_auth_count" : '';
+
+my $sth = $dbh->prepare(<<SQL);
+    SELECT id FROM authority.record_entry 
+    WHERE DATE(edit_date) >= DATE(NOW() - '$modified_since day'::INTERVAL)
+    $where2
+    ORDER BY id 
+    $limit
+SQL
 
-my $sth = $dbh->prepare("SELECT id FROM authority.record_entry $where");
 $sth->execute;
 
 while (my $ref = $sth->fetchrow_hashref()) {
@@ -129,10 +135,14 @@ while (my $ref = $sth->fetchrow_hashref()) {
 $sth->finish;
 
 my $auth_rec_count = scalar(@auth_ids);
-announce("Auth IDs: @auth_ids") if $print_auth_ids;
+print join("\n", @auth_ids) if $print_auth_ids;
+
+# Let the caller know what the last record processed will be, 
+# so the next iteration of the script can start there.
+announce("Final auth ID will be: " . $auth_ids[-1]) if $max_auth_count;
 
 if (!@auth_ids) {
-    announce("No authority records created in the last $new_since days");
+    announce("No authority records edited in the last $modified_since days");
     exit 0;
 }
 
@@ -142,6 +152,7 @@ if (!@auth_ids) {
 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', $db_host,
@@ -238,5 +249,5 @@ if ($link_bibs) {
     }
 }
 
-announce("Bib IDs: @bib_ids") if $print_bib_ids;
+print join("\n", @bib_ids) if $print_bib_ids;