Merge pull request #5699 from ckan/5698-fix-missing-activities

smotornyuk · web-flow · commit 8eec3e27c320 · 2020-11-06T20:50:05.000+02:00
[#5698] Filter activities from user at the database level
diff --git a/changes/5699.bugfix b/changes/5699.bugfix
@@ -0,0 +1 @@
+Fix missing activities from UI when internal processes are run by ignored users
diff --git a/ckan/logic/action/get.py b/ckan/logic/action/get.py
@@ -53,23 +53,6 @@
 _text = sqlalchemy.text
 
 
-def _filter_activity_by_user(activity_list, users=[]):
-    '''
-    Return the given ``activity_list`` but with activities from the specified
-    users removed. The users parameters should be a list of ids.
-
-    A *new* filtered list is returned, the given ``activity_list`` itself is
-    not modified.
-    '''
-    if not len(users):
-        return activity_list
-    new_list = []
-    for activity in activity_list:
-        if activity.user_id not in users:
-            new_list.append(activity)
-    return new_list
-
-
 def _activity_stream_get_filtered_users():
     '''
     Get the list of users from the :ref:`ckan.hide_activity_from_users` config
@@ -2499,10 +2482,8 @@ def user_activity_list(context, data_dict):
     offset = data_dict.get('offset', 0)
     limit = data_dict['limit']  # defaulted, limited & made an int by schema
 
-    _activity_objects = model.activity.user_activity_list(
+    activity_objects = model.activity.user_activity_list(
         user.id, limit=limit, offset=offset)
-    activity_objects = _filter_activity_by_user(
-        _activity_objects, _activity_stream_get_filtered_users())
 
     return model_dictize.activity_list_dictize(
         activity_objects, context,
@@ -2553,13 +2534,10 @@ def package_activity_list(context, data_dict):
     offset = int(data_dict.get('offset', 0))
     limit = data_dict['limit']  # defaulted, limited & made an int by schema
 
-    _activity_objects = model.activity.package_activity_list(
-        package.id, limit=limit, offset=offset)
-    if not include_hidden_activity:
-        activity_objects = _filter_activity_by_user(
-            _activity_objects, _activity_stream_get_filtered_users())
-    else:
-        activity_objects = _activity_objects
+    activity_objects = model.activity.package_activity_list(
+        package.id, limit=limit, offset=offset,
+        include_hidden_activity=include_hidden_activity,
+    )
 
     return model_dictize.activity_list_dictize(
         activity_objects, context, include_data=data_dict['include_data'])
@@ -2608,13 +2586,10 @@ def group_activity_list(context, data_dict):
     group_show = logic.get_action('group_show')
     group_id = group_show(context, {'id': group_id})['id']
 
-    _activity_objects = model.activity.group_activity_list(
-        group_id, limit=limit, offset=offset)
-    if not include_hidden_activity:
-        activity_objects = _filter_activity_by_user(
-            _activity_objects, _activity_stream_get_filtered_users())
-    else:
-        activity_objects = _activity_objects
+    activity_objects = model.activity.group_activity_list(
+        group_id, limit=limit, offset=offset,
+        include_hidden_activity=include_hidden_activity,
+    )
 
     return model_dictize.activity_list_dictize(
         activity_objects, context,
@@ -2662,13 +2637,10 @@ def organization_activity_list(context, data_dict):
     org_show = logic.get_action('organization_show')
     org_id = org_show(context, {'id': org_id})['id']
 
-    _activity_objects = model.activity.organization_activity_list(
-        org_id, limit=limit, offset=offset)
-    if not include_hidden_activity:
-        activity_objects = _filter_activity_by_user(
-            _activity_objects, _activity_stream_get_filtered_users())
-    else:
-        activity_objects = _activity_objects
+    activity_objects = model.activity.organization_activity_list(
+        org_id, limit=limit, offset=offset,
+        include_hidden_activity=include_hidden_activity,
+    )
 
     return model_dictize.activity_list_dictize(
         activity_objects, context,
@@ -2698,12 +2670,9 @@ def recently_changed_packages_activity_list(context, data_dict):
     data_dict['include_data'] = False
     limit = data_dict['limit']  # defaulted, limited & made an int by schema
 
-    _activity_objects = \
+    activity_objects = \
         model.activity.recently_changed_packages_activity_list(
             limit=limit, offset=offset)
-    activity_objects = _filter_activity_by_user(
-        _activity_objects,
-        _activity_stream_get_filtered_users())
 
     return model_dictize.activity_list_dictize(
         activity_objects, context,
@@ -3216,13 +3185,11 @@ def dashboard_activity_list(context, data_dict):
 
     # FIXME: Filter out activities whose subject or object the user is not
     # authorized to read.
-    _activity_tuple_objects = model.activity.dashboard_activity_list(
+    activity_objects = model.activity.dashboard_activity_list(
         user_id, limit=limit, offset=offset)
 
-    activity_tuple_list = _filter_activity_by_user(
-        _activity_tuple_objects, _activity_stream_get_filtered_users())
     activity_dicts = model_dictize.activity_list_dictize(
-        activity_tuple_list, context)
+        activity_objects, context)
 
     # Mark the new (not yet seen by user) activities.
     strptime = datetime.datetime.strptime
diff --git a/ckan/model/activity.py b/ckan/model/activity.py
@@ -15,6 +15,7 @@
     text,
 )
 
+from ckan.common import config
 import ckan.model
 from ckan.model import meta
 from ckan.model import domain_object, types as _types
@@ -164,6 +165,9 @@ def user_activity_list(user_id, limit, offset):
 
     '''
     q = _user_activity_query(user_id, limit + offset)
+
+    q = _filter_activitites_from_users(q)
+
     return _activities_at_offset(q, limit, offset)
 
 
@@ -177,7 +181,8 @@ def _package_activity_query(package_id):
     return q
 
 
-def package_activity_list(package_id, limit, offset):
+def package_activity_list(
+        package_id, limit, offset, include_hidden_activity=False):
     '''Return the given dataset (package)'s public activity stream.
 
     Returns all activities about the given dataset, i.e. where the given
@@ -189,10 +194,14 @@ def package_activity_list(package_id, limit, offset):
 
     '''
     q = _package_activity_query(package_id)
+
+    if not include_hidden_activity:
+        q = _filter_activitites_from_users(q)
+
     return _activities_at_offset(q, limit, offset)
 
 
-def _group_activity_query(group_id):
+def _group_activity_query(group_id, include_hidden_activity=False):
     '''Return an SQLAlchemy query for all activities about group_id.
 
     Returns a query for all activities whose object is either the group itself
@@ -242,10 +251,13 @@ def _group_activity_query(group_id):
         )
     )
 
+    if not include_hidden_activity:
+        q = _filter_activitites_from_users(q)
+
     return q
 
 
-def _organization_activity_query(org_id):
+def _organization_activity_query(org_id, include_hidden_activity=False):
     '''Return an SQLAlchemy query for all activities about org_id.
 
     Returns a query for all activities whose object is either the org itself
@@ -278,11 +290,14 @@ def _organization_activity_query(org_id):
             model.Activity.object_id == org_id
         )
     )
+    if not include_hidden_activity:
+        q = _filter_activitites_from_users(q)
 
     return q
 
 
-def group_activity_list(group_id, limit, offset):
+def group_activity_list(group_id, limit, offset, include_hidden_activity=False):
+
     '''Return the given group's public activity stream.
 
     Returns activities where the given group or one of its datasets is the
@@ -293,11 +308,12 @@ def group_activity_list(group_id, limit, offset):
     etc.
 
     '''
-    q = _group_activity_query(group_id)
+    q = _group_activity_query(group_id, include_hidden_activity)
     return _activities_at_offset(q, limit, offset)
 
 
-def organization_activity_list(group_id, limit, offset):
+def organization_activity_list(
+        group_id, limit, offset, include_hidden_activity=False):
     '''Return the given org's public activity stream.
 
     Returns activities where the given org or one of its datasets is the
@@ -308,7 +324,7 @@ def organization_activity_list(group_id, limit, offset):
     etc.
 
     '''
-    q = _organization_activity_query(group_id)
+    q = _organization_activity_query(group_id, include_hidden_activity)
     return _activities_at_offset(q, limit, offset)
 
 
@@ -402,8 +418,12 @@ def dashboard_activity_list(user_id, limit, offset):
 
     '''
     q = _dashboard_activity_query(user_id, limit + offset)
+
+    q = _filter_activitites_from_users(q)
+
     return _activities_at_offset(q, limit, offset)
 
+
 def _changed_packages_activity_query():
     '''Return an SQLAlchemy query for all changed package activities.
 
@@ -425,4 +445,37 @@ def recently_changed_packages_activity_list(limit, offset):
 
     '''
     q = _changed_packages_activity_query()
+
+    q = _filter_activitites_from_users(q)
+
     return _activities_at_offset(q, limit, offset)
+
+
+def _filter_activitites_from_users(q):
+    '''
+    Adds a filter to an existing query object ot avoid activities from users
+    defined in :ref:`ckan.hide_activity_from_users` (defaults to the site user)
+    '''
+    users_to_avoid = _activity_stream_get_filtered_users()
+    if users_to_avoid:
+        q = q.filter(ckan.model.Activity.user_id.notin_(users_to_avoid))
+
+    return q
+
+
+def _activity_stream_get_filtered_users():
+    '''
+    Get the list of users from the :ref:`ckan.hide_activity_from_users` config
+    option and return a list of their ids. If the config is not specified,
+    returns the id of the site user.
+    '''
+    users = config.get('ckan.hide_activity_from_users')
+    if users:
+        users_list = users.split()
+    else:
+        from ckan.logic import get_action
+        context = {'ignore_auth': True}
+        site_user = get_action('get_site_user')(context)
+        users_list = [site_user.get('name')]
+
+    return ckan.model.User.user_ids_for_name_or_id(users_list)
diff --git a/ckan/tests/controllers/test_package.py b/ckan/tests/controllers/test_package.py
@@ -1864,7 +1864,8 @@ def test_redirect_when_given_id(self, app):
         assert response.headers['location'] == expected_url
 
     def test_redirect_also_with_activity_parameter(self, app):
-        dataset = factories.Dataset()
+        user = factories.User()
+        dataset = factories.Dataset(user=user)
         activity = activity_model.package_activity_list(
             dataset["id"], limit=1, offset=0
         )[0]
diff --git a/ckan/tests/logic/action/test_get.py b/ckan/tests/logic/action/test_get.py
@@ -3164,9 +3164,11 @@ def _create_bulk_package_activities(self, count):
         dataset = factories.Dataset()
         from ckan import model
 
+        user = factories.User()
+
         objs = [
             model.Activity(
-                user_id=None,
+                user_id=user['id'],
                 object_id=dataset["id"],
                 activity_type=None,
                 data=None,
@@ -3396,9 +3398,10 @@ def test_delete_org_by_updating_state(self):
         assert activities[0]["data"]["group"]["title"] == org["title"]
 
     def _create_bulk_user_activities(self, count):
-        user = factories.User()
         from ckan import model
 
+        user = factories.User()
+
         objs = [
             model.Activity(
                 user_id=user["id"],
@@ -3613,9 +3616,11 @@ def _create_bulk_group_activities(self, count):
         group = factories.Group()
         from ckan import model
 
+        user = factories.User()
+
         objs = [
             model.Activity(
-                user_id=None,
+                user_id=user['id'],
                 object_id=group["id"],
                 activity_type=None,
                 data=None,
@@ -3844,9 +3849,11 @@ def _create_bulk_org_activities(self, count):
         org = factories.Organization()
         from ckan import model
 
+        user = factories.User()
+
         objs = [
             model.Activity(
-                user_id=None,
+                user_id=user['id'],
                 object_id=org["id"],
                 activity_type=None,
                 data=None,
@@ -4013,9 +4020,11 @@ def test_delete_dataset(self):
     def _create_bulk_package_activities(self, count):
         from ckan import model
 
+        user = factories.User()
+
         objs = [
             model.Activity(
-                user_id=None,
+                user_id=user['id'],
                 object_id=None,
                 activity_type="new_package",
                 data=None,

Original file line number	Diff line number	Diff line change
`@@ -0,0 +1 @@`
	`1`	`+Fix missing activities from UI when internal processes are run by ignored users`