Merge pull request #11735 from Vlad0n20/feature/ENG-9827

[ENG-9827] - save collection-submission custom metadata to CedarMetadataRecord on   create/update

Author: adlius

api_tests/collections/test_views.py

@@ -10,6 +10,7 @@
 from api_tests.subjects.mixins import UpdateSubjectsMixin, SubjectsFilterMixin, SubjectsListMixin, \
     SubjectsRelationshipMixin
 from api_tests.utils import disconnected_from_listeners
+from tests.utils import capture_notifications
 from framework.auth.core import Auth
 from osf import features
 from osf.models import Collection, VersionedGuidMixin
@@ -4450,16 +4451,70 @@ def test_switch_active_no_provider_submission_succeeds(self, app, user_one, proj
                 )
         assert res.status_code == 201
 
-    def test_switch_active_missing_cedar_record_submission_fails(self, app, user_one, project, url, payload):
-        with override_switch(features.COLLECTION_SUBMISSION_WITH_CEDAR, active=True):
-            res = app.post_json_api(
-                url,
-                payload(guid=project._id),
-                auth=user_one.auth,
-                expect_errors=True,
-            )
+    def test_switch_active_submission_without_cedar_record_fails(
+            self, app, user_one, project, url, payload, cedar_template):
+        with capture_notifications(expect_none=True):
+            with mock_update_share():
+                with override_switch(features.COLLECTION_SUBMISSION_WITH_CEDAR, active=True):
+                    res = app.post_json_api(
+                        url,
+                        payload(guid=project._id),
+                        auth=user_one.auth,
+                        expect_errors=True,
+                    )
         assert res.status_code == 400
-        assert 'CEDAR metadata record' in res.json['errors'][0]['detail']
+
+    def test_switch_active_submission_with_cedar_record_succeeds(
+            self, app, user_one, project, url, payload, cedar_template):
+        from osf.models import CedarMetadataRecord
+        CedarMetadataRecord.objects.create(
+            guid=project.guids.first(),
+            template=cedar_template,
+            metadata={'title': 'Test'},
+            is_published=True,
+        )
+        with capture_notifications():
+            with mock_update_share():
+                with override_switch(features.COLLECTION_SUBMISSION_WITH_CEDAR, active=True):
+                    res = app.post_json_api(
+                        url,
+                        payload(guid=project._id),
+                        auth=user_one.auth,
+                    )
+        assert res.status_code == 201
+
+    def test_switch_inactive_submission_without_cedar_record_succeeds(
+            self, app, user_one, project, url, payload, cedar_template):
+        with capture_notifications():
+            with mock_update_share():
+                with override_switch(features.COLLECTION_SUBMISSION_WITH_CEDAR, active=False):
+                    res = app.post_json_api(url, payload(guid=project._id), auth=user_one.auth)
+        assert res.status_code == 201
+
+    def test_switch_active_update_does_not_alter_cedar_record(
+            self, app, user_one, project, url, payload, cedar_template, collection):
+        from osf.models import CedarMetadataRecord
+        original_metadata = {'title': 'Original'}
+        CedarMetadataRecord.objects.create(
+            guid=project.guids.first(),
+            template=cedar_template,
+            metadata=original_metadata,
+            is_published=True,
+        )
+        collection.status_choices = ['pending', 'approved']
+        collection.save()
+        with capture_notifications():
+            with mock_update_share():
+                with override_switch(features.COLLECTION_SUBMISSION_WITH_CEDAR, active=True):
+                    res = app.post_json_api(url, payload(guid=project._id, status='pending'), auth=user_one.auth)
+        assert res.status_code == 201
+
+        detail_url = f'/{API_BASE}collections/{collection._id}/collected_metadata/{project._id}/'
+        with override_switch(features.COLLECTION_SUBMISSION_WITH_CEDAR, active=True):
+            app.patch_json_api(detail_url, payload(status='approved'), auth=user_one.auth)
+
+        record = CedarMetadataRecord.objects.get(guid__in=project.guids.all(), template=cedar_template)
+        assert record.metadata == original_metadata
 
 
 class TestCollectedMetaSubjectFiltering(SubjectsFilterMixin):

osf/models/cedar_metadata.py

@@ -1,8 +1,19 @@
+from django.core.exceptions import ValidationError
 from django.db import models
+from jsonschema import validate as jsonschema_validate, ValidationError as JsonSchemaValidationError
 
 from osf.models.base import BaseModel, ObjectIDMixin
 from osf.utils.datetime_aware_jsonfield import DateTimeAwareJSONField
 
+# Fields generated by the CEDAR server on instance creation — frontend never sends them
+CEDAR_SERVER_GENERATED_FIELDS = frozenset({
+    '@id',
+    'pav:createdOn',
+    'pav:createdBy',
+    'pav:lastUpdatedOn',
+    'oslc:modifiedBy',
+})
+
 
 class CedarMetadataTemplate(ObjectIDMixin, BaseModel):
     schema_name = models.CharField(max_length=255, default=None)
@@ -47,6 +58,22 @@ def get_template_name(self):
     def get_template_version(self):
         return self.template.template_version
 
+    def clean(self):
+        if self.is_published:
+            schema = dict(self.template.template)
+            if 'required' in schema:
+                schema['required'] = [
+                    f for f in schema['required']
+                    if f not in CEDAR_SERVER_GENERATED_FIELDS
+                ]
+            metadata = {k: v for k, v in self.metadata.items() if v != {}}
+            try:
+                jsonschema_validate(metadata, schema)
+            except JsonSchemaValidationError as e:
+                raise ValidationError(
+                    f'CEDAR metadata does not validate against template "{self.template.schema_name}": {e.message}'
+                )
+
     def save(self, *args, **kwargs):
+        super().save(*args, **kwargs)
         self.guid.referent.update_search()
-        return super().save(*args, **kwargs)

osf/models/collection_submission.py

@@ -22,7 +22,6 @@
 
 logger = logging.getLogger(__name__)
 
-
 class CollectionSubmission(TaxonomizableMixin, BaseModel):
     primary_identifier_name = 'guid___id'
 

osf/models/provider.py

@@ -1,6 +1,5 @@
 import json
 import requests
-from jsonschema import validate as jsonschema_validate, ValidationError as JsonSchemaValidationError
 
 from django.apps import apps
 from django.contrib.postgres import fields
@@ -21,7 +20,6 @@
 from .brand import Brand
 from .citation import CitationStyle
 from .licenses import NodeLicense
-from .cedar_metadata import CedarMetadataRecord
 from .storage import ProviderAssetFile
 from .subject import Subject
 from osf.utils.datetime_aware_jsonfield import DateTimeAwareJSONField
@@ -208,6 +206,24 @@ def top_level_subjects(self):
     def readable_type(self):
         raise NotImplementedError
 
+    def validate_required_metadata(self, obj):
+        """
+        Raises ValidationError if obj does not have a published CedarMetadataRecord for
+        this provider's required_metadata_template.
+        Does nothing when required_metadata_template is not set.
+        """
+        if not self.required_metadata_template_id:
+            return
+        guid = obj.guids.first()
+        if guid is None or not guid.cedar_metadata_records.filter(
+            template_id=self.required_metadata_template_id,
+            is_published=True,
+        ).exists():
+            raise ValidationError(
+                f'Submitted object must have a published CEDAR metadata record for template '
+                f'"{self.required_metadata_template.schema_name}" to be submitted to this collection.'
+            )
+
     def get_asset_url(self, name):
         """ Helper that returns an associated ProviderAssetFile's url, or None
 
@@ -259,27 +275,6 @@ def setup_share_source(self, provider_home_page):
 
         self.save()
 
-    def validate_required_metadata(self, osf_obj):
-        if not self.required_metadata_template:
-            return
-
-        record = CedarMetadataRecord.objects.filter(
-            guid__in=osf_obj.guids.all(),
-            template=self.required_metadata_template,
-            is_published=True,
-        ).first()
-
-        if record is None:
-            raise ValidationError(
-                f'Object must have a published CEDAR metadata record for the required template '
-                f'"{self.required_metadata_template.schema_name}".'
-            )
-
-        try:
-            jsonschema_validate(record.metadata, self.required_metadata_template.template)
-        except JsonSchemaValidationError as e:
-            raise ValidationError(e.message)
-
 
 class CollectionProvider(AbstractProvider):
     DEFAULT_SUBSCRIPTIONS = [

osf_tests/test_collection_submission.py

@@ -833,7 +833,8 @@ def test_share_update_cedar_metadata_record(self, unmoderated_collection_submiss
             },
             'identifier': {}
         }
-        with mock.patch('api.share.utils.pls_send_trove_record'):
+        with mock.patch('api.share.utils.pls_send_trove_record'), \
+                mock.patch('api.share.utils.share_update_cedar_metadata_record'):
             record = CedarMetadataRecord.objects.create(
                 guid=unmoderated_collection_submission_public.guid,
                 template=cedar_template,

osf_tests/test_validate_required_metadata.py

@@ -84,20 +84,6 @@ def test_published_valid_record_passes(self, provider, cedar_template, preprint)
 
         provider.validate_required_metadata(preprint)
 
-    def test_published_invalid_record_raises(self, provider, cedar_template, preprint):
-        provider.required_metadata_template = cedar_template
-        provider.save()
-
-        CedarMetadataRecord.objects.create(
-            guid=preprint.guids.first(),
-            template=cedar_template,
-            metadata={'title': 123},
-            is_published=True,
-        )
-
-        with pytest.raises(ValidationError):
-            provider.validate_required_metadata(preprint)
-
     def test_record_for_wrong_template_raises(self, provider, cedar_template, preprint):
         provider.required_metadata_template = cedar_template
         provider.save()