udata 10.4.2.dev35427__py2.py3-none-any.whl → 10.4.2.dev35441__py2.py3-none-any.whl

udata/core/user/activities.py

@@ -30,11 +30,6 @@ class DiscussActivity(object):
     badge_type = "warning"
 
 
-class UserStarredOrganization(FollowActivity, OrgRelatedActivity, Activity):
-    key = "organization:followed"
-    label = _("followed an organization")
-
-
 class UserFollowedUser(FollowActivity, Activity):
     key = "user:followed"
     label = _("followed a user")

udata/migrations/2025-05-22-purge-duplicate-activities.py

@@ -0,0 +1,101 @@
+"""
+This migration updates Topic.featured to False when it is None.
+"""
+
+import logging
+from datetime import datetime, timedelta
+
+from mongoengine.connection import get_db
+
+from udata.core.dataset.activities import UserCreatedDataset, UserDeletedDataset, UserUpdatedDataset
+from udata.core.organization.activities import UserUpdatedOrganization
+from udata.core.reuse.activities import UserCreatedReuse, UserDeletedReuse, UserUpdatedReuse
+from udata.core.user.activities import (
+    UserDiscussedDataset,
+    UserDiscussedReuse,
+    UserFollowedDataset,
+    UserFollowedOrganization,
+    UserFollowedReuse,
+)
+
+log = logging.getLogger(__name__)
+
+
+def migrate(db):
+    # Remove legacy fields (`as_organization`, `kwargs`) from old activities
+    result = get_db().activity.update_many({}, {"$unset": {"as_organization": ""}})
+    log.info(
+        f"Legacy field `as_organization` removed from {result.modified_count} activity objects"
+    )
+
+    result = get_db().activity.update_many({}, {"$unset": {"kwargs": ""}})
+    log.info(f"Legacy field `kwargs` removed from {result.modified_count} activity objects")
+
+    # Clean duplicate activities in case of discussion or following
+    # - remove the "updated" activity on the discussed/followed object
+    # - remove the activity on the organization
+    # The heuristic is to look for specific activities by the same actor on the targeted object
+    # within a -1 +1 second timespan
+    for action_related_activity, object_updated_activity in [
+        (UserDiscussedDataset, UserUpdatedDataset),
+        (UserDiscussedReuse, UserUpdatedReuse),
+        (UserFollowedDataset, UserUpdatedDataset),
+        (UserFollowedReuse, UserUpdatedReuse),
+    ]:
+        org_activity_count = 0
+        object_activity_count = 0
+        activities = (
+            action_related_activity.objects()
+            .no_dereference()  # We use no_dereference in query to prevent DBref DoesNotExist errors
+            .no_cache()
+            .timeout(False)
+        )
+        log.info(
+            f"{datetime.utcnow()}: Processing {activities.count()} {action_related_activity} activities..."
+        )
+        for act in activities:
+            object_activity_count += object_updated_activity.objects(
+                actor=act.actor.id,
+                related_to=act.related_to.id,
+                created_at__gte=act.created_at - timedelta(seconds=1),
+                created_at__lte=act.created_at + timedelta(seconds=1),
+            ).delete()
+            if act.organization:
+                org_activity_count += UserUpdatedOrganization.objects(
+                    actor=act.actor.id,
+                    related_to=act.organization,
+                    created_at__gte=act.created_at - timedelta(seconds=1),
+                    created_at__lte=act.created_at + timedelta(seconds=1),
+                ).delete()
+        log.info(
+            f"{datetime.utcnow()}: Deleted {object_activity_count} {object_updated_activity} and {org_activity_count} UserUpdatedOrganization activities"
+        )
+
+    # Clean duplicated UserUpdatedOrganization activities on organization for any object related activity
+    for object_related_activity in [
+        UserCreatedDataset,
+        UserUpdatedDataset,
+        UserDeletedDataset,
+        UserCreatedReuse,
+        UserUpdatedReuse,
+        UserDeletedReuse,
+        UserFollowedOrganization,
+    ]:
+        count = 0
+        activities = (
+            object_related_activity.objects(organization__exists=True)
+            .no_dereference()  # We use no_dereference in query to prevent DBref DoesNotExist errors
+            .no_cache()
+            .timeout(False)
+        )
+        log.info(
+            f"{datetime.utcnow()}: Processing {activities.count()} {object_related_activity} activities..."
+        )
+        for act in activities:
+            count += UserUpdatedOrganization.objects(
+                actor=act.actor.id,
+                related_to=act.organization,
+                created_at__gte=act.created_at - timedelta(seconds=1),
+                created_at__lte=act.created_at + timedelta(seconds=1),
+            ).delete()
+        log.info(f"{datetime.utcnow()}: Deleted {count} UserUpdatedOrganization activities")

{udata-10.4.2.dev35427.dist-info → udata-10.4.2.dev35441.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: udata
-Version: 10.4.2.dev35427
+Version: 10.4.2.dev35441
 Summary: Open data portal
 Home-page: https://github.com/opendatateam/udata
 Author: Opendata Team
@@ -144,6 +144,7 @@ It is collectively taken care of by members of the
 -  Add test for removing last contact point [#3322](https://github.com/opendatateam/udata/pull/3322) 
 - Add activities to dataservices and resources, add Auditable class to refactor improve code [#3308](https://github.com/opendatateam/udata/pull/3308)
 - Store activities for private objects [#3328](https://github.com/opendatateam/udata/pull/3328)
+- :warning: Add migration to clean duplicate activities [#3327](https://github.com/opendatateam/udata/pull/3327)
 - Do not crash if file doesn't exists during resource deletion [#3323](https://github.com/opendatateam/udata/pull/3323)
 - Show user domain in suggest [#3324](https://github.com/opendatateam/udata/pull/3324)
 - Keep the existing frequency if not found during harvesting [#3330](https://github.com/opendatateam/udata/pull/3330)

{udata-10.4.2.dev35427.dist-info → udata-10.4.2.dev35441.dist-info}/RECORD

@@ -237,7 +237,7 @@ udata/core/topic/models.py,sha256=X-jScC_mMNdZp0hQ_SD-NBHsIPS8aYqpq99x6l4dKz4,20
 udata/core/topic/parsers.py,sha256=ugkBd-w8TewInqowNF2w36UPwKYMYluK4U-grkAu-wg,2411
 udata/core/topic/permissions.py,sha256=RtFPPlxuU_Bv7ip6LDO4AoPrKFnIOEs9cCMXaSSmEdk,118
 udata/core/user/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-udata/core/user/activities.py,sha256=RAotXwFERt6TfLHiNfV13e1AgKGhlGwzlPm-ihj1sHk,3493
+udata/core/user/activities.py,sha256=d91G45XEgoxYUJetkSjVolSM-nqmNSuM5b-6geayGE4,3338
 udata/core/user/api.py,sha256=TO5iolYhb7wWSqBwrSHQOSYR-r_8Jrn4LtMF5BiQELk,13473
 udata/core/user/api_fields.py,sha256=S4wONsS5p_Ys31HdBxS_hBreCFHEegjfagg5aprwkNA,6026
 udata/core/user/apiv2.py,sha256=4eNsvJjb4ChJQFrXtVbkOtAvXEcQcQpZf8vkEbriXRA,1125
@@ -365,6 +365,7 @@ udata/migrations/2024-12-05-contact-point-is-now-a-list.py,sha256=il2qSFhOTq-Yhq
 udata/migrations/2025-01-05-dataservices-fields-changes.py,sha256=HlqHg3sG3rk3sYVrOwAlXMNhTmTKd1YT82P-gXOqmZM,4647
 udata/migrations/2025-03-20-save-quality-for-datasets.py,sha256=FPTfGVByXSHr18V4RFlktC7t-H-5rgEcZQMTRpMrGqo,607
 udata/migrations/2025-04-24-topic-featured-default-false.py,sha256=t4OyhehtPQiosKtAo9SWRWm3ObGUHvqD_roXQ34vIH4,369
+udata/migrations/2025-05-22-purge-duplicate-activities.py,sha256=iemQwEWtbxedtpkeTFIeoNP7WcY7ntnDrjUClPzzWfg,4212
 udata/migrations/__init__.py,sha256=RBCBDaTlLjuMs_Qzwji6Z6T4r7FCGXhESKoxQbT5qAA,11221
 udata/models/__init__.py,sha256=txbZwa-lRG3mq99eQ9E5YcFWiNUdjDVSyJJvlqUMFfs,1413
 udata/mongo/__init__.py,sha256=y4Rv-kq3o_kcEulcNpePLzocXPBNpx3Jd82G-VZPaMc,1421
@@ -727,9 +728,9 @@ udata/translations/pt/LC_MESSAGES/udata.mo,sha256=ViV14tUmjSydHS0TWG_mFikKQfyUaT
 udata/translations/pt/LC_MESSAGES/udata.po,sha256=rzAD_MVoV54TmN3w1ECz3H2Ru5pM7hWMVH03SkY28Q8,47250
 udata/translations/sr/LC_MESSAGES/udata.mo,sha256=EHX1_D-Uglj38832G7BrA0QC5IuY3p8dKqi9T0DgPmE,29169
 udata/translations/sr/LC_MESSAGES/udata.po,sha256=3PMnbVhKVJh6Q8ABi1ZTZ8Dcf-sMjngLJZqLbonJoec,54225
-udata-10.4.2.dev35427.dist-info/LICENSE,sha256=V8j_M8nAz8PvAOZQocyRDX7keai8UJ9skgmnwqETmdY,34520
-udata-10.4.2.dev35427.dist-info/METADATA,sha256=2m-J49idkyg2Qj4O0OP16KkloSjW2AHzQknkvRmOqso,147154
-udata-10.4.2.dev35427.dist-info/WHEEL,sha256=Kh9pAotZVRFj97E15yTA4iADqXdQfIVTHcNaZTjxeGM,110
-udata-10.4.2.dev35427.dist-info/entry_points.txt,sha256=ETvkR4r6G1duBsh_V_fGWENQy17GTFuobi95MYBAl1A,498
-udata-10.4.2.dev35427.dist-info/top_level.txt,sha256=39OCg-VWFWOq4gCKnjKNu-s3OwFlZIu_dVH8Gl6ndHw,12
-udata-10.4.2.dev35427.dist-info/RECORD,,
+udata-10.4.2.dev35441.dist-info/LICENSE,sha256=V8j_M8nAz8PvAOZQocyRDX7keai8UJ9skgmnwqETmdY,34520
+udata-10.4.2.dev35441.dist-info/METADATA,sha256=kKLW5dRSdxUIexOLVuiXwKXqEGbH9k1LBzUH4j8gMOQ,147267
+udata-10.4.2.dev35441.dist-info/WHEEL,sha256=Kh9pAotZVRFj97E15yTA4iADqXdQfIVTHcNaZTjxeGM,110
+udata-10.4.2.dev35441.dist-info/entry_points.txt,sha256=ETvkR4r6G1duBsh_V_fGWENQy17GTFuobi95MYBAl1A,498
+udata-10.4.2.dev35441.dist-info/top_level.txt,sha256=39OCg-VWFWOq4gCKnjKNu-s3OwFlZIu_dVH8Gl6ndHw,12
+udata-10.4.2.dev35441.dist-info/RECORD,,