From: Bill Erickson <berickxx@gmail.com>
Date: Wed, 7 Dec 2016 16:04:19 +0000 (-0500)
Subject: JBAS-1437 Backstage importer process bibs first
X-Git-Url: https://old-git.evergreen-ils.org/?a=commitdiff_plain;h=476665999727da03bed67ec4acc6ff280c7944da;p=working%2FEvergreen.git

JBAS-1437 Backstage importer process bibs first

When processing a ZIP file, process bib files before authority files.
On large data sets, processing bibs first allows staff to resume normal
bib editing operations earlier.

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

diff --git a/KCLS/authority-control/backstage/process-backstage-files.pl b/KCLS/authority-control/backstage/process-backstage-files.pl
index c034846b2d..e402b2c81e 100755
--- a/KCLS/authority-control/backstage/process-backstage-files.pl
+++ b/KCLS/authority-control/backstage/process-backstage-files.pl
@@ -26,7 +26,7 @@ my $marc_file;
 my $zip_file;
 my $export_date;
 my $working_dir = '.',
-my $bib_collision_file;
+my $bib_collision_file = 'bib-collisions.mrc'; # in --working-dir
 my $verbose;
 my $db_user = $ENV{PGUSER} || 'evergreen';
 my $db_name = $ENV{PGDATABASE} || 'evergreen';
@@ -158,8 +158,10 @@ sub process_zip_file {
     announce('ERR', "Failed to read $zip_file", 1)
         unless $zip->read($zip_file) == AZ_OK;
 
-    # Avoid processing XLS and HTM files.  
-    # All of the MARC files end in .UTF8.
+    my %marc_files = (bib => [], auth => []);
+
+    # Start by locating the MARC files in the ZIP file
+    # All of the MARC files end in .UTF8 or MRC.
     for my $member ($zip->membersMatching('.*(\.UTF8|\.MRC)')) {
 
         my $basename = basename($member->fileName());
@@ -171,13 +173,18 @@ sub process_zip_file {
         announce('ERR', "Unable to extract to file: $local_file", 1)
             unless $member->extractToFileNamed($local_file) == AZ_OK;
 
-        process_marc_file($local_file);
+        my $key = ($basename =~ /BIB/) ? 'bib' : 'auth';
+        push(@{$marc_files{$key}}, $local_file);
     }
+
+    # Then process bib files first, followed by authority files.
+    process_marc_file($_) for (@{$marc_files{bib}}, @{$marc_files{auth}})
 }
 
 sub process_marc_file {
     my $local_file = shift;
     my $basename = basename($local_file);
+    announce('INFO', "Processing file $basename");
 
     my $marc_batch = MARC::File::USMARC->in($local_file, 'UTF8')
         or announce('ERR', "Unable to read $local_file as MARC", 1);
@@ -406,7 +413,8 @@ sub find_replaced_auths {
     my $matches = $match_auth_sth->fetchall_arrayref;
     my @ids = map {$_->[0]} @$matches;
 
-    announce('INFO', "Auth 010z=$subfield matched records: @ids") if @ids;
+    announce('INFO', "Auth 010z=$subfield matched records: @ids") 
+        if $verbose && @ids;
 
     return @ids;
 }
@@ -426,8 +434,8 @@ sub find_matching_auths {
             $match_auth_001_sth->execute($subfield);
             my $matches = $match_auth_001_sth->fetchall_arrayref;
             my @ids = map {$_->[0]} @$matches;
-            announce('INFO', 
-                "Auth 001=$subfield matched records: @ids") if @ids;
+            announce('INFO', "Auth 001=$subfield matched records: @ids") 
+                if $verbose && @ids;
             return @ids;
         }
     }
@@ -446,7 +454,8 @@ sub find_matching_auths {
     my $matches = $match_auth_sth->fetchall_arrayref;
 
     my @ids = map {$_->[0]} @$matches;
-    announce('INFO', "Auth ${tag}a=$subfield matched records: @ids") if @ids;
+    announce('INFO', "Auth ${tag}a=$subfield matched records: @ids") 
+        if $verbose && @ids;
 
     return @ids;
 }