From 2d639dd1b58ba4ad96593be6bb8914ff0fd8c0b8 Mon Sep 17 00:00:00 2001
From: Gerhard Gonter <ggonter@gmail.com>
Date: Fri, 13 Oct 2023 15:54:17 +0200
Subject: [PATCH] added code to inject metadata fields "publisher" and
"publicationYear", if missing
---
dcd.pl | 32 ++++++++++++++++++++++++++++++--
1 file changed, 30 insertions(+), 2 deletions(-)
diff --git a/dcd.pl b/dcd.pl
index 8458566..30b4de2 100755
--- a/dcd.pl
+++ b/dcd.pl
@@ -291,6 +291,8 @@ print __LINE__, " prod_doi_string=[$prod_doi_string] pfx=[$pfx] ns=[$ns] nr=[$nr
unshift (@$md, $doi_element);
print __LINE__, " md: ", main::Dumper ($md);
+ fixup_phaidra_metadata($md);
+
my $xml_new= $repo_obj->json_2_xml ($md);
print __LINE__, " xml_new=[$xml_new]\n";
@@ -311,8 +313,10 @@ print __LINE__, " prod_doi_string=[$prod_doi_string] pfx=[$pfx] ns=[$ns] nr=[$nr
syswrite (DC_XML, $prod_xml_new);
close (DC_XML);
- print "metatada fetched: ", get_ts(), "\n";
- if ($datacite_res->{status} ne 'OK')
+ print __LINE__, " metatada fetched: ", get_ts(), "\n";
+ if ($datacite_res->{status} ne 'OK'
+ && $datacite_res->{status} ne 'INCOMPLETE' # maybe we should check, what was missing
+ )
{
print "Metadata not ok; status=[$datacite_res->{status}] errors: ", Dumper ($datacite_res->{errors});
return undef;
@@ -352,6 +356,30 @@ print __LINE__, " prod_doi_string=[$prod_doi_string] pfx=[$pfx] ns=[$ns] nr=[$nr
}
+# TODO: shema checking could work differently
+sub fixup_phaidra_metadata
+{
+ my $md= shift; # should be an array
+
+ # print __LINE__, " fixup_phaidra_metadata: md= ", Dumper($md);
+ die 'not an array reference' unless (ref($md) eq 'ARRAY');
+
+ my $cnt= @$md;
+ print __LINE__, " md element count=[$cnt]\n";
+ my %pos;
+ for (my $i= 0; $i < $cnt; $i++)
+ {
+ my $e= $md->[$i];
+ print __LINE__, " fpm: i=[$i] e= ", Dumper($e);
+ $pos{$e->{xmlname}}= $i;
+ }
+
+ print __LINE__, " pos: ", Dumper(\%pos);
+
+ splice(@$md, 3, 0, { xmlname => 'publisher', 'value' => ':none' }) unless (exists ($pos{publisher}));
+ splice(@$md, 4, 0, { xmlname => 'publicationYear', 'value' => '2023' }) unless (exists ($pos{publicationYear}));
+}
+
sub get_ts
{
my $time= shift || time ();
--
GitLab