Merge branch 'release/25.02.0'

CHANGELOG

@@ -2,9 +2,12 @@
 
 We follow the CalVer (https://calver.org/) versioning scheme: YY.MINOR.MICRO.
 
-25.01.0 (2025-01-21)
+25.02.0 (2025-01-27)
 ====================
+- DOI Versioning - BE Release
 
+25.01.0 (2025-01-21)
+====================
 - Institutional Access Curator - BE Release
 
 24.11.0 (2024-12-11)

addons/base/views.py

@@ -310,7 +310,8 @@ def decrypt_and_decode_jwt_payload():
 
 
 def get_authenticated_resource(resource_id):
-    resource = AbstractNode.load(resource_id) or Preprint.load(resource_id)
+    resource, _ = Guid.load_referent(resource_id)
+
     if not resource:
         raise HTTPError(http_status.HTTP_404_NOT_FOUND, message='Resource not found.')
 
@@ -390,8 +391,7 @@ def get_auth(auth, **kwargs):
     waterbutler_data = decrypt_and_decode_jwt_payload()
 
     # Authenticate the resource based on the node_id and handle potential draft nodes
-    resource = get_authenticated_resource(waterbutler_data['nid'])
-
+    resource = get_authenticated_resource(waterbutler_data.get('nid'))
     # Authenticate the user using cookie or Oauth if possible
     authenticate_user_if_needed(auth, waterbutler_data, resource)
 
@@ -529,9 +529,9 @@ def create_waterbutler_log(payload, **kwargs):
             auth = payload['auth']
             # Don't log download actions
             if payload['action'] in DOWNLOAD_ACTIONS:
-                guid = Guid.load(payload['metadata'].get('nid'))
-                if guid:
-                    node = guid.referent
+                guid_id = payload['metadata'].get('nid')
+
+                node, _ = Guid.load_referent(guid_id)
                 return {'status': 'success'}
 
             user = OSFUser.load(auth['id'])
@@ -890,7 +890,7 @@ def addon_view_or_download_file(auth, path, provider, **kwargs):
     extras.pop('_', None)  # Clean up our url params a bit
     action = extras.get('action', 'view')
     guid = kwargs.get('guid')
-    guid_target = getattr(Guid.load(guid), 'referent', None)
+    guid_target, _ = Guid.load_referent(guid)
     target = guid_target or kwargs.get('node') or kwargs['project']
 
     provider_safe = markupsafe.escape(provider)

addons/osfstorage/decorators.py

@@ -33,8 +33,8 @@ def load_guid_as_target(func):
 
     @functools.wraps(func)
     def wrapped(*args, **kwargs):
-        guid = kwargs.get('guid')
-        target = getattr(Guid.load(guid), 'referent', None)
+        target, _ = Guid.load_referent(kwargs.get('guid'))
+
         if not target:
             raise HTTPError(
                 http_status.HTTP_404_NOT_FOUND,
@@ -88,7 +88,9 @@ def wrapped(payload, *args, **kwargs):
             user = OSFUser.load(payload['user'])
             # Waterbutler is sending back ['node'] under the destination payload - WB should change to target
             target = payload['destination'].get('target') or payload['destination'].get('node')
-            dest_target = Guid.load(target).referent
+            dest_target, _ = Guid.load_referent(target)
+            if not dest_target:
+                raise HTTPError(http_status.HTTP_404_NOT_FOUND)
             source = OsfStorageFileNode.get(payload['source'], kwargs['target'])
             dest_parent = OsfStorageFolder.get(payload['destination']['parent'], dest_target)
 

addons/osfstorage/listeners.py

@@ -24,7 +24,7 @@ def checkin_files_task(node_id, user_id):
         Guid = apps.get_model('osf.Guid')
         OSFUser = apps.get_model('osf.OSFUser')
 
-        node = Guid.load(node_id).referent
+        node, _ = Guid.load_referent(node_id)
         assert isinstance(node, (AbstractNode, Preprint))
         user = OSFUser.load(user_id)
 

addons/osfstorage/utils.py

@@ -9,7 +9,7 @@
 from framework.celery_tasks import app
 from framework.postcommit_tasks.handlers import enqueue_postcommit_task
 from framework.sessions import get_session
-from osf.models import BaseFileNode, Guid
+from osf.models import BaseFileNode, Guid, Preprint
 
 from addons.osfstorage import settings
 
@@ -21,7 +21,7 @@ def enqueue_update_analytics(node, file, version_idx, action='download'):
 
 @app.task(max_retries=5, default_retry_delay=60)
 def update_analytics_async(node_id, file_id, version_idx, session_key=None, action='download'):
-    node = Guid.load(node_id).referent
+    node, _ = Guid.load_referent(node_id)
     file = BaseFileNode.load(file_id)
     update_analytics(node, file, version_idx, session_key, action)
 
@@ -43,7 +43,7 @@ def update_analytics(node, file, version_idx, session_key, action='download'):
     node_info = {
         'contributors': contributors
     }
-    resource = node.guids.first()
+    resource = node.get_guid() if isinstance(node, Preprint) else node.guids.first()
 
     update_counter(resource, file, version=None, action=action, node_info=node_info, session_key=session_key)
     update_counter(resource, file, version_idx, action, node_info=node_info, session_key=session_key)

addons/osfstorage/views.py

@@ -187,9 +187,12 @@ def osfstorage_get_metadata(file_node, **kwargs):
 @decorators.autoload_filenode(must_be='folder')
 def osfstorage_get_children(file_node, **kwargs):
     from django.contrib.contenttypes.models import ContentType
+    from osf.models.preprint import Preprint
     user_id = request.args.get('user_id')
     user_content_type_id = ContentType.objects.get_for_model(OSFUser).id
     user_pk = OSFUser.objects.filter(guids___id=user_id, guids___id__isnull=False).values_list('pk', flat=True).first()
+    guid_id = file_node.target.get_guid().id if isinstance(file_node.target, Preprint) else file_node.target.guids.first().id,
+
     with connection.cursor() as cursor:
         # Read the documentation on FileVersion's fields before reading this code
         cursor.execute("""
@@ -281,7 +284,7 @@ def osfstorage_get_children(file_node, **kwargs):
             AND (NOT F.type IN ('osf.trashedfilenode', 'osf.trashedfile', 'osf.trashedfolder'))
         """, [
             user_content_type_id,
-            file_node.target.guids.first().id,
+            guid_id,
             user_pk,
             user_pk,
             user_id,

admin/preprints/urls.py

@@ -9,24 +9,19 @@
     re_path(r'^known_spam$', views.PreprintKnownSpamList.as_view(), name='known-spam'),
     re_path(r'^known_ham$', views.PreprintKnownHamList.as_view(), name='known-ham'),
     re_path(r'^withdrawal_requests$', views.PreprintWithdrawalRequestList.as_view(), name='withdrawal-requests'),
-    re_path(r'^(?P<guid>[a-z0-9]+)/$', views.PreprintView.as_view(), name='preprint'),
-    re_path(r'^(?P<guid>[a-z0-9]+)/change_provider/$', views.PreprintProviderChangeView.as_view(), name='preprint-provider'),
-    re_path(r'^(?P<guid>[a-z0-9]+)/machine_state/$', views.PreprintMachineStateView.as_view(), name='preprint-machine-state'),
-    re_path(r'^(?P<guid>[a-z0-9]+)/reindex_share_preprint/$', views.PreprintReindexShare.as_view(),
-        name='reindex-share-preprint'),
-    re_path(r'^(?P<guid>[a-z0-9]+)/remove_user/(?P<user_id>[a-z0-9]+)/$', views.PreprintRemoveContributorView.as_view(),
-        name='remove-user'),
-    re_path(r'^(?P<guid>[a-z0-9]+)/make_private/$', views.PreprintMakePrivate.as_view(), name='make-private'),
-    re_path(r'^(?P<guid>[a-z0-9]+)/make_public/$', views.PreprintMakePublic.as_view(), name='make-public'),
-    re_path(r'^(?P<guid>[a-z0-9]+)/remove/$', views.PreprintDeleteView.as_view(), name='remove'),
-    re_path(r'^(?P<guid>[a-z0-9]+)/restore/$', views.PreprintDeleteView.as_view(), name='restore'),
-    re_path(r'^(?P<guid>[a-z0-9]+)/confirm_unflag/$', views.PreprintConfirmUnflagView.as_view(), name='confirm-unflag'),
-    re_path(r'^(?P<guid>[a-z0-9]+)/confirm_spam/$', views.PreprintConfirmSpamView.as_view(), name='confirm-spam'),
-    re_path(r'^(?P<guid>[a-z0-9]+)/confirm_ham/$', views.PreprintConfirmHamView.as_view(), name='confirm-ham'),
-    re_path(r'^(?P<guid>[a-z0-9]+)/reindex_elastic_preprint/$', views.PreprintReindexElastic.as_view(),
-        name='reindex-elastic-preprint'),
-    re_path(r'^(?P<guid>[a-z0-9]+)/approve_withdrawal/$', views.PreprintApproveWithdrawalRequest.as_view(),
-        name='approve-withdrawal'),
-    re_path(r'^(?P<guid>[a-z0-9]+)/reject_withdrawal/$', views.PreprintRejectWithdrawalRequest.as_view(),
-        name='reject-withdrawal'),
+    re_path(r'^(?P<guid>\w+)/$', views.PreprintView.as_view(), name='preprint'),
+    re_path(r'^(?P<guid>\w+)/change_provider/$', views.PreprintProviderChangeView.as_view(), name='preprint-provider'),
+    re_path(r'^(?P<guid>\w+)/machine_state/$', views.PreprintMachineStateView.as_view(), name='preprint-machine-state'),
+    re_path(r'^(?P<guid>\w+)/reindex_share_preprint/$', views.PreprintReindexShare.as_view(), name='reindex-share-preprint'),
+    re_path(r'^(?P<guid>\w+)/remove_user/(?P<user_id>[a-z0-9]+)/$', views.PreprintRemoveContributorView.as_view(), name='remove-user'),
+    re_path(r'^(?P<guid>\w+)/make_private/$', views.PreprintMakePrivate.as_view(), name='make-private'),
+    re_path(r'^(?P<guid>\w+)/make_public/$', views.PreprintMakePublic.as_view(), name='make-public'),
+    re_path(r'^(?P<guid>\w+)/remove/$', views.PreprintDeleteView.as_view(), name='remove'),
+    re_path(r'^(?P<guid>\w+)/restore/$', views.PreprintDeleteView.as_view(), name='restore'),
+    re_path(r'^(?P<guid>\w+)/confirm_unflag/$', views.PreprintConfirmUnflagView.as_view(), name='confirm-unflag'),
+    re_path(r'^(?P<guid>\w+)/confirm_spam/$', views.PreprintConfirmSpamView.as_view(), name='confirm-spam'),
+    re_path(r'^(?P<guid>\w+)/confirm_ham/$', views.PreprintConfirmHamView.as_view(), name='confirm-ham'),
+    re_path(r'^(?P<guid>\w+)/reindex_elastic_preprint/$', views.PreprintReindexElastic.as_view(), name='reindex-elastic-preprint'),
+    re_path(r'^(?P<guid>\w+)/approve_withdrawal/$', views.PreprintApproveWithdrawalRequest.as_view(), name='approve-withdrawal'),
+    re_path(r'^(?P<guid>\w+)/reject_withdrawal/$', views.PreprintRejectWithdrawalRequest.as_view(), name='reject-withdrawal'),
 ]

admin/preprints/views.py

@@ -48,7 +48,7 @@
 class PreprintMixin(PermissionRequiredMixin):
 
     def get_object(self):
-        preprint = Preprint.objects.get(guids___id=self.kwargs['guid'])
+        preprint = Preprint.load(self.kwargs['guid'])
         # Django template does not like attributes with underscores for some reason
         preprint.guid = preprint._id
         return preprint
@@ -324,9 +324,10 @@ class WithdrawalRequestMixin(PermissionRequiredMixin):
     permission_required = 'osf.change_preprintrequest'
 
     def get_object(self):
+        target = Preprint.load(self.kwargs['guid'])
         return PreprintRequest.objects.filter(
             request_type='withdrawal',
-            target__guids___id=self.kwargs['guid'],
+            target=target,
         ).first()
 
     def get_success_url(self):

api/actions/views.py

@@ -6,7 +6,7 @@
 from api.actions.permissions import ReviewActionPermission
 from api.actions.serializers import NodeRequestActionSerializer, ReviewActionSerializer, PreprintRequestActionSerializer
 from api.base.exceptions import Conflict
-from api.base.filters import ListFilterMixin
+from api.base.filters import TargetFilterMixin
 from api.base.views import JSONAPIBaseView
 from api.base.parsers import (
     JSONAPIMultipleRelationshipsParser,
@@ -110,7 +110,7 @@ def get_object(self):
         return action
 
 
-class ReviewActionListCreate(JSONAPIBaseView, generics.ListCreateAPIView, ListFilterMixin):
+class ReviewActionListCreate(JSONAPIBaseView, generics.ListCreateAPIView, TargetFilterMixin):
     """List of review actions viewable by this user
 
     Actions represent state changes and/or comments on a reviewable object (e.g. a preprint)

api/base/filters.py

@@ -19,8 +19,9 @@
 from rest_framework import serializers as ser
 from rest_framework.filters import OrderingFilter
 from osf.models import Subject, Preprint
-from osf.models.base import GuidMixin
+from osf.models.base import GuidMixin, Guid
 from functools import cmp_to_key
+from framework import sentry
 
 def lowercase(lower):
     if hasattr(lower, '__call__'):
@@ -568,24 +569,102 @@ def get_serializer_method(self, field_name):
 
 
 class PreprintFilterMixin(ListFilterMixin):
-    """View mixin that uses ListFilterMixin, adding postprocessing for preprint querying
+    """View mixin for many preprint listing views. It inherits from ListFilterMixin and customize postprocessing for
+    preprint querying by provider, subjects and versioned `_id`.
 
-       Subclasses must define `get_default_queryset()`.
+    Note: Subclasses must define `get_default_queryset()`.
     """
+
+    @staticmethod
+    def postprocess_versioned_guid_id_query_param(operation):
+        """Handle query parameters when filtering on `_id` for preprint which are now versioned.
+
+        Note: preprint achieves versioning by using versioned guid, and thus no long has guid or guid._id for every
+        version. Must convert `guid___id__in=` look-up to `pk__in` look-up.
+        """
+        preprint_pk_list = []
+        for val in operation['value']:
+            referent, version = Guid.load_referent(val)
+            if referent is None:
+                continue
+            preprint_pk_list.append(referent.id)
+        # Override the operation to filter `id__in=preprint_pk_list`
+        operation['source_field_name'] = 'id__in'
+        operation['value'] = preprint_pk_list
+        operation['op'] = 'eq'
+
     def postprocess_query_param(self, key, field_name, operation):
         if field_name == 'provider':
             operation['source_field_name'] = 'provider___id'
 
         if field_name == 'id':
-            operation['source_field_name'] = 'guids___id'
+            PreprintFilterMixin.postprocess_versioned_guid_id_query_param(operation)
 
         if field_name == 'subjects':
             self.postprocess_subject_query_param(operation)
 
-    def preprints_queryset(self, base_queryset, auth_user, allow_contribs=True, public_only=False):
-        return Preprint.objects.can_view(
+    def preprints_queryset(self, base_queryset, auth_user, allow_contribs=True, public_only=False, latest_only=False):
+        preprints = Preprint.objects.can_view(
             base_queryset=base_queryset,
             user=auth_user,
             allow_contribs=allow_contribs,
             public_only=public_only,
         )
+        if latest_only:
+            preprints = preprints.filter(guids__isnull=False)
+        return preprints
+
+
+class TargetFilterMixin(ListFilterMixin):
+    """View mixin for multi-content-type list views (e.g. `ReviewActionListCreate`). It inherits from `ListFilterMixin`
+     and customizes the postprocessing of the `target` field when target is a preprint.
+
+    Note: Subclasses must define `get_default_queryset()`.
+    """
+
+    @staticmethod
+    def postprocess_preprint_as_target_query_param(operation, target_pk):
+        """When target is a preprint, which must be versioned, the traditional non-versioned `guid___id==_id`
+        look-up no longer works. Must convert it to PK look-up `target__id==pk`.
+        """
+        # Override the operation to filter `target__id==pk`
+        operation['source_field_name'] = 'target__id'
+        operation['value'] = target_pk
+        operation['op'] = 'eq'
+
+    def postprocess_query_param(self, key, field_name, operation):
+        """Handles a special case when filtering on `target` when `target` is a Preprint.
+        """
+        if field_name == 'target':
+            referent, version = Guid.load_referent(operation['value'])
+            if referent:
+                if version:
+                    TargetFilterMixin.postprocess_preprint_as_target_query_param(operation, referent.id)
+                else:
+                    super().postprocess_query_param(key, field_name, operation)
+            else:
+                sentry.log_message(f'Target object invalid or not found: [target={operation['value']}]')
+                return
+        else:
+            super().postprocess_query_param(key, field_name, operation)
+
+
+class PreprintAsTargetFilterMixin(TargetFilterMixin):
+    """View mixin for preprint related list views (e.g. `PreprintProviderWithdrawRequestList` and `PreprintActionList`).
+    It inherits from `TargetFilterMixin` and customizes postprocessing the `target` field for preprint.
+
+    Note: Subclasses must define `get_default_queryset()`.
+    """
+
+    def postprocess_query_param(self, key, field_name, operation):
+        """Handles a special case when filtering on `target`.
+        """
+        if field_name == 'target':
+            referent, version = Guid.load_referent(operation['value'])
+            # A valid preprint must have referent and version
+            if not referent or not version:
+                sentry.log_message(f'Preprint invalid or note found: [target={operation['value']}]')
+                return
+            TargetFilterMixin.postprocess_preprint_as_target_query_param(operation, referent.id)
+        else:
+            super().postprocess_query_param(key, field_name, operation)

api/base/pagination.py

@@ -238,8 +238,7 @@ def get_paginated_response(self, data):
 class PreprintContributorPagination(NodeContributorPagination):
 
     def get_resource(self, kwargs):
-        resource_id = kwargs.get('preprint_id')
-        return Preprint.load(resource_id)
+        return Preprint.load(kwargs.get('preprint_id'))
 
 
 class DraftRegistrationContributorPagination(NodeContributorPagination):

api/base/utils.py

@@ -16,7 +16,7 @@
 from framework.auth import Auth
 from framework.auth.cas import CasResponse
 from framework.auth.oauth_scopes import ComposedScopes, normalize_scopes
-from osf.models.base import GuidMixin
+from osf.models.base import GuidMixin, VersionedGuidMixin
 from osf.utils.requests import check_select_for_update
 from website import settings as website_settings
 from website import util as website_util  # noqa
@@ -103,8 +103,14 @@ def get_object_or_error(model_or_qs, query_or_pk=None, request=None, display_nam
             raise NotFound
 
     elif isinstance(query_or_pk, str):
-        # they passed a 5-char guid as a string
-        if issubclass(model_cls, GuidMixin):
+        # If the class is a subclass of `VersionedGuidMixin`, get obj directly from model using `.load()`. The naming
+        # for `query_or_pk` no longer matches the actual case. It is neither a query nor a pk, but a guid str.
+        if issubclass(model_cls, VersionedGuidMixin):
+            obj = model_cls.load(query_or_pk, select_for_update=select_for_update)
+        # If the class is a subclass of `GuidMixin` (except for `VersionedGuidMixin`), turn it into a query dictionary.
+        # The naming for `query_or_pk` no longer matches the actual case either. It is neither a query nor a pk, but a
+        # 5-char guid str. We should be able to use the `.load()` the same way as in the `VersionedGuidMixin` case.
+        elif issubclass(model_cls, GuidMixin):
             # if it's a subclass of GuidMixin we know it's primary_identifier_name
             query = {'guids___id': query_or_pk}
         else:

api/chronos/views.py

@@ -91,8 +91,9 @@ def get_serializer_class(self):
 
     def get_default_queryset(self):
         user = get_user_auth(self.request).user
-        preprint_contributors = Preprint.load(self.kwargs['preprint_id'])._contributors
-        queryset = ChronosSubmission.objects.filter(preprint__guids___id=self.kwargs['preprint_id'])
+        preprint = Preprint.load(self.kwargs['preprint_id'])
+        preprint_contributors = preprint._contributors
+        queryset = ChronosSubmission.objects.filter(preprint=preprint)
 
         # Get the list of stale submissions and queue a task to update them
         update_list_id = queryset.filter(