Merge pull request #253 from CSCfi/CSCMETAX-379

CSCMETAX-379:[ADD] special handling of syke datasets for urnresolver

Author: hannu40k

src/metax_api/api/oaipmh/base/metax_oai_server.py

@@ -10,6 +10,8 @@
 from metax_api.models.catalog_record import CatalogRecord
 from metax_api.services import CatalogRecordService as CRS
 
+syke_url_prefix_template = 'http://metatieto.ymparisto.fi:8080/geoportal/catalog/search/resource/details.page?uuid=%s'
+
 
 class MetaxOAIServer(ResumptionOAIPMH):
 
@@ -48,22 +50,41 @@ def _get_filtered_records(self, set, cursor, batch_size, from_=None, until=None)
             query_set = query_set.filter(data_catalog__catalog_json__identifier__in=self._get_default_set_filter())
         return query_set[cursor:batch_size]
 
+    def _handle_syke_urnresolver_metadata(self, record):
+        identifiers = []
+        preferred_identifier = record.research_dataset.get('preferred_identifier')
+        identifiers.append(preferred_identifier)
+        for id_obj in record.research_dataset.get('other_identifier', []):
+            if id_obj.get('notation', '').startswith('{'):
+                uuid = id_obj['notation']
+                identifiers.append(syke_url_prefix_template % uuid)
+        return identifiers
+
     def _get_oai_dc_urnresolver_metadata(self, record):
         """
         Preferred identifier is added only for ida and att catalog records
         other identifiers are added for all.
+
+        Special handling for SYKE catalog.
         """
+
         identifiers = []
-        identifiers.append(settings.OAI['ETSIN_URL_TEMPLATE'] % record.identifier)
 
-        # assuming ida and att catalogs are not harvested
-        if not record.catalog_is_harvested():
-            preferred_identifier = record.research_dataset.get('preferred_identifier')
-            identifiers.append(preferred_identifier)
-        for id_obj in record.research_dataset.get('other_identifier', []):
-            if id_obj.get('notation', '').startswith('urn:nbn:fi:csc-kata'):
-                other_urn = id_obj['notation']
-                identifiers.append(other_urn)
+        data_catalog = record.data_catalog.catalog_json.get('identifier')
+        if data_catalog == 'urn:nbn:fi:att:data-catalog-harvest-syke':
+            identifiers = self._handle_syke_urnresolver_metadata(record)
+
+        else:
+            identifiers.append(settings.OAI['ETSIN_URL_TEMPLATE'] % record.identifier)
+
+            # assuming ida and att catalogs are not harvested
+            if not record.catalog_is_harvested():
+                preferred_identifier = record.research_dataset.get('preferred_identifier')
+                identifiers.append(preferred_identifier)
+            for id_obj in record.research_dataset.get('other_identifier', []):
+                if id_obj.get('notation', '').startswith('urn:nbn:fi:csc-kata'):
+                    other_urn = id_obj['notation']
+                    identifiers.append(other_urn)
 
         meta = {
             'identifier':  identifiers
@@ -189,7 +210,7 @@ def getRecord(self, metadataPrefix, identifier):
         """Implement OAI-PMH verb GetRecord."""
         try:
             record = CatalogRecord.objects.get(
-                data_catalog__catalog_json__identifier__in=self._get_default_set_filter(),
+                # data_catalog__catalog_json__identifier__in=self._get_default_set_filter(),
                 identifier__exact=identifier
             )
         except CatalogRecord.DoesNotExist:

src/metax_api/tests/api/oaipmh/syke.py

@@ -0,0 +1,62 @@
+from django.core.management import call_command
+import lxml.etree
+from rest_framework import status
+from rest_framework.test import APITestCase
+from metax_api.models import CatalogRecord
+from metax_api.api.oaipmh.base.metax_oai_server import syke_url_prefix_template
+
+from metax_api.tests.utils import test_data_file_path, TestClassUtils
+
+
+class SYKEOAIPMHReadTests(APITestCase, TestClassUtils):
+
+    _namespaces = {'o': 'http://www.openarchives.org/OAI/2.0/',
+                   'oai_dc': "http://www.openarchives.org/OAI/2.0/oai_dc/",
+                   'dc': "http://purl.org/dc/elements/1.1/",
+                   'dct': "http://purl.org/dc/terms/",
+                   'datacite': 'http://schema.datacite.org/oai/oai-1.0/'}
+
+    @classmethod
+    def setUpClass(cls):
+        """
+        Loaded only once for test cases inside this class.
+        """
+        call_command('loaddata', test_data_file_path, verbosity=0)
+        super(SYKEOAIPMHReadTests, cls).setUpClass()
+
+    def setUp(self):
+        cr = CatalogRecord.objects.get(pk=1)
+        cr.data_catalog.catalog_json["identifier"] = "urn:nbn:fi:att:data-catalog-harvest-syke"
+        cr.data_catalog.force_save()
+        cr.research_dataset.update({
+            "preferred_identifier": "urn:nbn:fi:csc-kata20170613100856741858",
+            "other_identifier": [
+                {
+                    "notation": "{55AB842F-9CED-4E80-A7E5-07A54F0AE4A4}"
+                }
+            ]
+        })
+        cr.force_save()
+        self.identifier = cr.identifier
+        self.pref_identifier = cr.research_dataset['preferred_identifier']
+        self.dc = cr.data_catalog.catalog_json["identifier"]
+        self._use_http_authorization()
+
+    def _get_results(self, data, xpath):
+        root = data
+        if isinstance(data, bytes):
+            root = lxml.etree.fromstring(data)
+        return root.xpath(xpath, namespaces=self._namespaces)
+
+    def test_get_urn_resolver_record(self):
+        response = self.client.get(
+            '/oai/?verb=GetRecord&identifier=%s&metadataPrefix=oai_dc_urnresolver' % self.identifier)
+        self.assertEqual(response.status_code, status.HTTP_200_OK)
+        identifiers = self._get_results(response.content,
+            '//o:record/o:metadata/oai_dc:dc/dc:identifier[text()="%s"]' % self.pref_identifier)
+        self.assertTrue(len(identifiers) == 1, response.content)
+
+        syke_url = syke_url_prefix_template % '{55AB842F-9CED-4E80-A7E5-07A54F0AE4A4}'
+        identifiers = self._get_results(response.content,
+            '//o:record/o:metadata/oai_dc:dc/dc:identifier[text()="%s"]' % syke_url)
+        self.assertTrue(len(identifiers) == 1, response.content)