From 4ec0f4a9a6bd90429aa8f31e5127e7352fa362d5 Mon Sep 17 00:00:00 2001
From: Jordan Padams <jordan.h.padams@jpl.nasa.gov>
Date: Mon, 16 May 2022 14:16:28 -0700
Subject: [PATCH] Add sync script and update identifier handling

* Add sync_dois.sh script per #309
* Update identifier handling for NoneType identifiers #331
---
 scripts/sync_dois.sh                          | 47 +++++++++++++++++++
 .../outputs/datacite/datacite_web_parser.py   |  6 ++-
 2 files changed, 52 insertions(+), 1 deletion(-)
 create mode 100755 scripts/sync_dois.sh

diff --git a/scripts/sync_dois.sh b/scripts/sync_dois.sh
new file mode 100755
index 00000000..582815a0
--- /dev/null
+++ b/scripts/sync_dois.sh
@@ -0,0 +1,47 @@
+#!/bin/bash
+
+while [[ $# -gt 0 ]]; do
+  case $1 in
+    -h|--help)
+      echo "sync_dois.sh [-h|--help] [-p|--prefix <doi_prefix>] [-s|--submitter <submitter email>]"
+      exit 0
+      ;;
+    -p|--prefix)
+      PREFIX="$2"
+      shift
+      shift
+      ;;
+    -s|--submitter)
+      SUBMITTER="$2"
+      shift
+      shift
+      ;;
+    *)
+      echo "Unknown option $1"
+      exit 1
+      ;;
+  esac
+done
+
+if [[ -z ${PREFIX} ]]; then
+  PREFIX="10.17189"
+fi
+
+if [[ -z ${SUBMITTER} ]]; then
+  SUBMITTER="pds-operator@jpl.nasa.gov"
+fi
+
+echo "=================================================="
+echo "Starting DOI sync for $(date)"
+echo
+echo "PREFIX=${PREFIX}"
+echo "SUBMITTER=${SUBMITTER}"
+
+source /home/pds4/pds-doi-service/bin/activate
+
+pds-doi-init --service datacite --prefix ${PREFIX} --submitter ${SUBMITTER}
+
+echo "Sync complete"
+echo
+
+exit 0
diff --git a/src/pds_doi_service/core/outputs/datacite/datacite_web_parser.py b/src/pds_doi_service/core/outputs/datacite/datacite_web_parser.py
index b26cf1b8..e75492ce 100644
--- a/src/pds_doi_service/core/outputs/datacite/datacite_web_parser.py
+++ b/src/pds_doi_service/core/outputs/datacite/datacite_web_parser.py
@@ -108,7 +108,11 @@ def _parse_identifiers(record):
             identifiers = record["identifiers"]
 
             for identifier in identifiers:
-                identifier["identifier"] = identifier["identifier"].strip()
+                if identifier["identifier"] is None:
+                    logger.warn(f"Odd metadata. NoneType identifier in record: {json.dumps(record, indent=4, sort_keys=True)}")
+                    identifiers.remove(identifier)
+                else:
+                    identifier["identifier"] = identifier["identifier"].strip()
 
             return identifiers
         except KeyError: