oai: fix error on document type processing

* Closes #789. Co-Authored-by: Bertrand Zuchuat <[email protected]> Co-Authored-by: Johnny Mariéthoz <[email protected]>
rero · Mar 9, 2022 · 8f3e748 · 8f3e748
1 parent 47b402e
commit 8f3e748
Show file tree

Hide file tree

Showing 5 changed files with 82 additions and 70 deletions.
diff --git a/sonar/config.py b/sonar/config.py
@@ -851,7 +851,7 @@ def _(x):
     'oai_dc': {
         'namespace': 'http://www.openarchives.org/OAI/2.0/oai_dc/',
         'schema': 'http://www.openarchives.org/OAI/2.0/oai_dc.xsd',
-        'serializer': 'sonar.modules.documents.serializers.oaipmh_oai_dc',
+        'serializer': 'sonar.modules.documents.serializers.dc.sonar_dublin_core',
     }
 }
 #: Number of records to return per page in OAI-PMH results.

diff --git a/sonar/modules/documents/serializers/__init__.py b/sonar/modules/documents/serializers/__init__.py
@@ -21,24 +21,20 @@
 
 from datetime import datetime
 
-from flask import current_app, request
+from flask import request
 from invenio_records_rest.serializers.response import record_responsify, \
     search_responsify
 
 from sonar.modules.collections.api import Record as CollectionRecord
-from sonar.modules.documents.serializers.dc import SonarDublinCoreSerializer
 from sonar.modules.documents.serializers.google_scholar import \
     SonarGoogleScholarSerializer
 from sonar.modules.documents.serializers.schemaorg import \
     SonarSchemaOrgSerializer
-from sonar.modules.documents.serializers.schemas.dc import DublinCoreV1
 from sonar.modules.documents.serializers.schemas.google_scholar import \
     GoogleScholarV1
 from sonar.modules.documents.serializers.schemas.schemaorg import SchemaOrgV1
 from sonar.modules.organisations.api import OrganisationRecord
 from sonar.modules.serializers import JSONSerializer as _JSONSerializer
-from sonar.modules.subdivisions.api import Record as SubdivisionRecord
-from sonar.modules.users.api import current_user_record
 from sonar.modules.utils import get_language_value
 
 from ..marshmallow import DocumentSchemaV1
@@ -84,8 +80,6 @@ def post_process_serialize_search(self, results, pid_fetcher):
 # ===========
 #: JSON serializer definition.
 json_v1 = JSONSerializer(DocumentSchemaV1)
-#: Dublin Core serializer
-dc_v1 = SonarDublinCoreSerializer(DublinCoreV1, replace_refs=True)
 #: schema.org serializer
 schemaorg_v1 = SonarSchemaOrgSerializer(SchemaOrgV1, replace_refs=True)
 #: google scholar serializer
@@ -104,8 +98,3 @@ def post_process_serialize_search(self, results, pid_fetcher):
     'json_v1_response',
     'json_v1_search',
 )
-
-# OAI-PMH record serializers.
-# ===========================
-#: OAI-PMH OAI Dublin Core record serializer.
-oaipmh_oai_dc = dc_v1.serialize_oaipmh
diff --git a/sonar/modules/documents/serializers/dc.py b/sonar/modules/documents/serializers/dc.py
@@ -17,16 +17,34 @@
 
 """Dublin Core serializer."""
 
-from invenio_records_rest.serializers.dc import DublinCoreSerializer
+from dcxml import simpledc
+from flask_resources.serializers import SerializerMixin
 
+from sonar.modules.documents.serializers.schemas.dc import DublinCoreSchema
 
-class SonarDublinCoreSerializer(DublinCoreSerializer):
-    """Marshmallow based DublinCore serializer for records."""
 
-    def dump(self, obj, context=None):
-        """Serialize object with schema.
+class SonarDublinCoreXMLSerializer(SerializerMixin):
+    """DublinCore serializer for records."""
 
-        Mandatory to override this method, as invenio-records-rest does not
-        use the right way to dump objects (compatible with marshmallow 3.9).
+    def __init__(self, **options):
+        """Constructor."""
+        self.schema_class = DublinCoreSchema
+
+    def transform_record(self, obj):
+        """Tranform record."""
+        # TODO: Remove this hack after migrate to invenio ressources
+        return self.schema_class().dump(dict(metadata=obj))
+
+    def serialize_object_xml(self, obj):
+        """Serialize a single record and persistent identifier to etree.
+
+        :param obj: Record instance
         """
-        return self.schema_class(context=context).dump(obj)
+        json = self.transform_record(obj["_source"])
+        return simpledc.dump_etree(json)
+
+
+def sonar_dublin_core(pid, record):
+    """Get DublinCore XML for OAI-PMH."""
+    return SonarDublinCoreXMLSerializer()\
+        .serialize_object_xml(record)
diff --git a/sonar/modules/documents/serializers/schemas/dc.py b/sonar/modules/documents/serializers/schemas/dc.py
@@ -28,7 +28,7 @@
 from .base_schema import BaseSchema
 
 
-class DublinCoreV1(BaseSchema):
+class DublinCoreSchema(BaseSchema):
     """Schema for records v1 in JSON."""
 
     contributors = fields.Method('get_contributors')
@@ -249,11 +249,12 @@ def get_titles(self, obj):
 
     def get_types(self, obj):
         """Get types."""
-        if obj['metadata'].get('documentType'):
-            return [
-                'http://purl.org/coar/resource_type/{type}'.format(
-                    type=obj['metadata']['documentType'].split(':')[1])
-            ]
+        if obj['metadata'].get('documentType', ''):
+            types = obj['metadata'].get('documentType', '').split(':')
+            if len(types) == 1:
+                return [f'{types[0]}']
+            if len(types) == 2:
+                return [f'http://purl.org/coar/resource_type/{types[1]}']
 
         return []