Fix tests

udata/core/dataservices/rdf.py

@@ -3,13 +3,12 @@
 from typing import List, Optional
 from rdflib import RDF, Graph
 
-from udata.core.dataservices.models import Dataservice, HarvestMetadata
+from udata.core.dataservices.models import Dataservice, HarvestMetadata as HarvestDataserviceMetadata
 from udata.core.dataset.models import Dataset, License
 from udata.core.dataset.rdf import sanitize_html
 from udata.harvest.models import HarvestSource
 from udata.rdf import DCAT, DCT, contact_point_from_rdf, rdf_value, theme_labels_from_rdf, themes_from_rdf, url_from_rdf
 
-
 def dataservice_from_rdf(graph: Graph, dataservice: Dataservice, node, all_datasets: List[Dataset]) -> Dataservice :
     '''
     Create or update a dataset from a RDF/DCAT graph
@@ -46,6 +45,9 @@ def dataservice_from_rdf(graph: Graph, dataservice: Dataservice, node, all_datas
     if license is not None:
         dataservice.license = License.guess(license)
 
+    if not harvest:
+        harvest = HarvestDataserviceMetadata()
+
     dataservice.harvest.created_at = rdf_value(d, DCT.issued)
     dataservice.metadata_modified_at = rdf_value(d, DCT.modified)
 

udata/harvest/backends/base.py

@@ -194,9 +194,6 @@ def process_dataset(self, remote_id: str, **kwargs):
             if not remote_id:
                 raise HarvestSkipException("missing identifier")
 
-            if not dataset.harvest:
-                dataset.harvest = HarvestDatasetMetadata()
-
             current_app.logger.addHandler(log_catcher)
             dataset = self.inner_process_dataset(item, **kwargs)
 
@@ -254,9 +251,6 @@ def process_dataservice(self, remote_id: str, **kwargs) -> bool :
             if not remote_id:
                 raise HarvestSkipException("missing identifier")
 
-            if not dataservice.harvest:
-                dataservice.harvest = HarvestDataserviceMetadata()
-
             dataservice = self.inner_process_dataservice(item, **kwargs)
 
             dataservice.harvest = self.update_dataservice_harvest_info(dataservice.harvest, remote_id)
@@ -290,7 +284,10 @@ def process_dataservice(self, remote_id: str, **kwargs) -> bool :
             item.ended = datetime.utcnow()
             self.save_job()
 
-    def update_dataset_harvest_info(self, harvest: HarvestDatasetMetadata, remote_id: int):
+    def update_dataset_harvest_info(self, harvest: Optional[HarvestDatasetMetadata], remote_id: int):
+        if not harvest:
+            harvest = HarvestDatasetMetadata()
+
         harvest.backend = self.display_name
         harvest.source_id = str(self.source.id)
         harvest.remote_id = remote_id
@@ -303,7 +300,10 @@ def update_dataset_harvest_info(self, harvest: HarvestDatasetMetadata, remote_id
 
         return harvest
 
-    def update_dataservice_harvest_info(self, harvest: HarvestDataserviceMetadata, remote_id: int):
+    def update_dataservice_harvest_info(self, harvest: Optional[HarvestDataserviceMetadata], remote_id: int):
+        if not harvest:
+            harvest = HarvestDataserviceMetadata()
+
         harvest.backend = self.display_name
         harvest.domain = self.source.domain