update discovery metadata DOI definitions (#129)

tomkralidis · web-flow · commit 9106509ca3de · 2025-11-03T22:40:33.000-05:00
diff --git a/data/init/discovery-metadata/Broad-band.yml b/data/init/discovery-metadata/Broad-band.yml
@@ -3,6 +3,9 @@ mcf:
 
 metadata:
     identifier: Broad-band
+    additional_identifiers:
+        - identifier: 10.14287/10000012
+          scheme: https://doi.org
     language: en
     language_alternate: fr
     charset: utf8
@@ -27,7 +30,6 @@ identification:
         creation: 2025-04-30
         publication: 2025-04-30
     wmo_data_policy: core
-    doi: 10.14287/10000012
     keywords:
         default:
             keywords:
diff --git a/data/init/discovery-metadata/Lidar.yml b/data/init/discovery-metadata/Lidar.yml
@@ -3,6 +3,9 @@ mcf:
 
 metadata:
     identifier: Lidar
+    additional_identifiers:
+        - identifier: 10.14287/10000007
+          scheme: https://doi.org
     language: en
     language_alternate: fr
     charset: utf8
@@ -27,7 +30,6 @@ identification:
         creation: 2025-04-30
         publication: 2025-04-30
     wmo_data_policy: core
-    doi: 10.14287/10000007
     keywords:
         default:
             keywords:
diff --git a/data/init/discovery-metadata/Multi-band.yml b/data/init/discovery-metadata/Multi-band.yml
@@ -3,6 +3,9 @@ mcf:
 
 metadata:
     identifier: Multi-band
+    additional_identifiers:
+        - identifier: 10.14287/10000010
+          scheme: https://doi.org
     language: en
     language_alternate: fr
     charset: utf8
@@ -27,7 +30,6 @@ identification:
         creation: 2025-04-30
         publication: 2025-04-30
     wmo_data_policy: core
-    doi: 10.14287/10000010
     keywords:
         default:
             keywords:
diff --git a/data/init/discovery-metadata/OzoneSonde.yml b/data/init/discovery-metadata/OzoneSonde.yml
@@ -3,6 +3,9 @@ mcf:
 
 metadata:
     identifier: OzoneSonde
+    additional_identifiers:
+        - identifier: 10.14287/10000008
+          scheme: https://doi.org
     language: en
     language_alternate: fr
     charset: utf8
@@ -27,7 +30,6 @@ identification:
         creation: 2025-04-30
         publication: 2025-04-30
     wmo_data_policy: core
-    doi: 10.14287/10000008
     keywords:
         default:
             keywords:
diff --git a/data/init/discovery-metadata/RocketSonde.yml b/data/init/discovery-metadata/RocketSonde.yml
@@ -3,6 +3,9 @@ mcf:
 
 metadata:
     identifier: RocketSonde
+    additional_identifiers:
+        - identifier: 10.14287/10000009
+          scheme: https://doi.org
     language: en
     language_alternate: fr
     charset: utf8
@@ -27,7 +30,6 @@ identification:
         creation: 2025-04-30
         publication: 2025-04-30
     wmo_data_policy: core
-    doi: 10.14287/10000009
     keywords:
         default:
             keywords:
diff --git a/data/init/discovery-metadata/Spectral.yml b/data/init/discovery-metadata/Spectral.yml
@@ -3,6 +3,9 @@ mcf:
 
 metadata:
     identifier: Spectral
+    additional_identifiers:
+        - identifier: 10.14287/10000011
+          scheme: https://doi.org
     language: en
     language_alternate: fr
     charset: utf8
@@ -28,7 +31,6 @@ identification:
         creation: 2025-04-30
         publication: 2025-04-30
     wmo_data_policy: core
-    doi: 10.14287/10000011
     keywords:
         default:
             keywords:
diff --git a/data/init/discovery-metadata/TotalOzone.yml b/data/init/discovery-metadata/TotalOzone.yml
@@ -3,6 +3,9 @@ mcf:
 
 metadata:
     identifier: TotalOzone
+    additional_identifiers:
+        - identifier: 10.14287/10000004
+          scheme: https://doi.org
     language: en
     language_alternate: fr
     charset: utf8
@@ -27,7 +30,6 @@ identification:
         creation: 2025-04-30
         publication: 2025-04-30
     wmo_data_policy: core
-    doi: 10.14287/10000004
     keywords:
         default:
             keywords:
diff --git a/data/init/discovery-metadata/Totalozoneobs.yml b/data/init/discovery-metadata/Totalozoneobs.yml
@@ -3,6 +3,9 @@ mcf:
 
 metadata:
     identifier: TotalOzoneObs
+    additional_identifiers:
+        - identifier: 10.14287/10000003
+          scheme: https://doi.org
     language: en
     language_alternate: fr
     charset: utf8
@@ -27,7 +30,6 @@ identification:
         creation: 2025-04-30
         publication: 2025-04-30
     wmo_data_policy: core 
-    doi: 10.14287/10000003
     keywords:
         default:
             keywords:
diff --git a/data/init/discovery-metadata/UmkehrN14_1.yml b/data/init/discovery-metadata/UmkehrN14_1.yml
@@ -3,6 +3,9 @@ mcf:
 
 metadata:
     identifier: UmkehrN14_1.0
+    additional_identifiers:
+        - identifier: 10.14287/10000005
+          scheme: https://doi.org
     language: en
     language_alternate: fr
     charset: utf8
@@ -27,7 +30,6 @@ identification:
         creation: 2025-04-30
         publication: 2025-04-30
     wmo_data_policy: core
-    doi: 10.14287/10000005
     keywords:
         default:
             keywords:
@@ -114,4 +116,4 @@ distribution:
         description:
             en: *abstract_en
             fr: *abstract_fr
-        function: download
+        function: download
diff --git a/data/init/discovery-metadata/UmkehrN14_2.yml b/data/init/discovery-metadata/UmkehrN14_2.yml
@@ -3,6 +3,9 @@ mcf:
 
 metadata:
     identifier: UmkehrN14_2.0
+    additional_identifiers:
+        - identifier: 10.14287/10000006
+          scheme: https://doi.org
     language: en
     language_alternate: fr
     charset: utf8
@@ -27,7 +30,6 @@ identification:
         creation: 2025-04-30
         publication: 2025-04-30
     wmo_data_policy: core
-    doi: 10.14287/10000006
     keywords:
         default:
             keywords:
@@ -114,4 +116,4 @@ distribution:
         description:
             en: *abstract_en
             fr: *abstract_fr
-        function: download
+        function: download
diff --git a/data/init/discovery-metadata/template.yml.sample b/data/init/discovery-metadata/template.yml.sample
@@ -5,6 +5,9 @@ mcf:
 # fixed except for identifier
 metadata:
     identifier: totalozone
+    additional_identifiers:
+        - identifier: 10.14287/10000004
+          scheme: https://doi.org
     language: en
     language_alternate: fr
     charset: utf8
@@ -30,7 +33,6 @@ identification:
         creation: 2025-04-30
         publication: 2025-04-30
     wmo_data_policy: core  # fixed
-    doi: 10.14287/10000004  # doi
     keywords:
         default:
             keywords:
diff --git a/data/init/discovery-metadata/uv_index_hourly.yml b/data/init/discovery-metadata/uv_index_hourly.yml
@@ -3,6 +3,9 @@ mcf:
 
 metadata:
     identifier: uv_index_hourly
+    additional_identifiers:
+        - identifier: 10.14287/10000013
+          scheme: https://doi.org
     language: en
     language_alternate: fr
     charset: utf8
@@ -27,7 +30,6 @@ identification:
         creation: 2025-04-30
         publication: 2025-04-30
     wmo_data_policy: core
-    doi: 10.14287/10000013
     keywords:
         default:
             keywords:
@@ -102,4 +104,4 @@ distribution:
         description:
             en: *abstract_en
             fr: *abstract_fr
-        function: download
+        function: download
diff --git a/woudc_data_registry/generate_metadata.py b/woudc_data_registry/generate_metadata.py
@@ -123,7 +123,10 @@ def generate_metadata(woudc_yaml):
                                 value2['description_en']
                             dataset_md['properties']['abstract_fr'] = \
                                 value2['description_fr']
-                            dataset_md["properties"]['doi'] = value2['doi']
+                            dataset_md["properties"]['externalIds'] = [{
+                                'value': value2['doi'],
+                                'scheme': 'https://doi.org'
+                            }]
                             dataset_md['properties']['keywords_en'] = \
                                 value2['keywords_en']
                             dataset_md['properties']['keywords_fr'] = \
diff --git a/woudc_data_registry/search.py b/woudc_data_registry/search.py
@@ -226,10 +226,6 @@
                 'fields': {'raw': typedefs['keyword']}
             },
             'dataset_snapshots': dataset_links,
-            'doi': {
-                'type': 'text',
-                'fields': {'raw': typedefs['keyword']}
-            },
             'keywords': {
                 'type': 'text',
                 'fields': {'raw': typedefs['keyword']}
