task/WG-138-WG-140: improve logs for analytics to include application…

… and related DS system (#161)

* Improve logs for anlytics to include application and related DesignSafe project/systems

* Add unit tests

* Remove unused user1

* Use 'project_uuid' for uuids and 'project' for id

* Move analytics related logging to features endpoint

* Check query params for analytic information (due to WG-191)

Headers can't be used due to https://tacc-main.atlassian.net/browse/WG-191

* Add public view information to log statement

* Fix public-view check

* Update tests

* Fix linting

geoapi/models/project.py

@@ -29,6 +29,8 @@ class Project(Base):
     id = Column(Integer, primary_key=True)
     uuid = Column(UUID(as_uuid=True), default=uuid.uuid4, nullable=False)
     tenant_id = Column(String, nullable=False)
+    # Project system_id/system_path really not used except for analytics.
+    # This could be improved; see https://jira.tacc.utexas.edu/browse/WG-185
     system_id = Column(String, nullable=True)
     system_path = Column(String, nullable=True)
     system_file = Column(String, nullable=True)

geoapi/routes/projects.py

@@ -169,14 +169,15 @@ def get(self):
         uuid_subset = query.get("uuid")
 
         if uuid_subset:
-            logger.info("Get a subset of projects for user:{} projects:{}".format(u.username, uuid_subset))
+            logger.info(f"Getting a subset of projects for user:{u.username} project_uuid:{uuid_subset}")
+
             # Check each project and abort if user (authenticated or anonymous) can't access the project
             subset = [check_access_and_get_project(request.current_user, uuid=uuid, allow_public_use=True) for uuid in uuid_subset]
             return subset
         else:
             if is_anonymous(u):
                 abort(403, "Access denied")
-            logger.info("Get all projects for user:{}".format(u.username))
+            logger.info(f"Get all projects for user:{u.username}")
             return ProjectsService.list(db_session, u)
 
     @api.doc(id="createProject",
@@ -290,8 +291,23 @@ class ProjectFeaturesResource(Resource):
     @api.marshal_with(feature_collection_model, as_list=True)
     @project_permissions_allow_public
     def get(self, projectId: int):
-        logger.info("Get features of project:{} for user:{}".format(
-            projectId, request.current_user.username))
+        # Following log is for analytics, see https://confluence.tacc.utexas.edu/display/DES/Hazmapper+Logging
+        application = request.headers.get('X-Geoapi-Application')
+        if application is None:
+            #  Check if in query parameters due to https://tacc-main.atlassian.net/browse/WG-192 and WG-191 */
+            application = request.args.get('application')
+
+            if application is None:
+                application = "Unknown"
+
+        from geoapi.routes.public_projects import PublicProjectFeaturesResource
+        is_public_view = issubclass(self.__class__, PublicProjectFeaturesResource)
+
+        prj = ProjectsService.get(db_session, project_id=projectId, user=request.current_user)
+        logger.info(f"Get features of project for user:{request.current_user.username} application:{application}"
+                    f" public_view:{is_public_view} project_uuid:{prj.uuid} project:{prj.id} tapis_system_id:{prj.system_id} "
+                    f"tapis_system_path:{prj.system_path}")
+
         query = self.parser.parse_args()
         return ProjectsService.getFeatures(db_session, projectId, query)
 

geoapi/tests/api_tests/test_projects_routes.py

@@ -63,6 +63,16 @@ def test_get_projects_using_single_uuid(test_client, projects_fixture, projects_
     assert data[0]["deletable"] is True
 
 
+def test_get_projects_using_single_uuid_observable_project(test_client, observable_projects_fixture, user1):
+    resp = test_client.get('/projects/',
+                           query_string='uuid={}'.format(observable_projects_fixture.project.uuid),
+                           headers={'x-jwt-assertion-test': user1.jwt})
+    data = resp.get_json()
+    assert resp.status_code == 200
+    assert len(data) == 1
+    assert data[0]["uuid"] == str(observable_projects_fixture.project.uuid)
+
+
 def test_get_projects_using_single_uuid_that_is_wrong(test_client, user1):
     resp = test_client.get('/projects/',
                            query_string='uuid={}'.format(uuid.uuid4()),
@@ -262,20 +272,39 @@ def test_get_point_cloud(test_client, projects_fixture, point_cloud_fixture, use
     assert resp.status_code == 200
 
 
-def test_get_project_features_empty(test_client, projects_fixture, user1):
+def test_get_project_features_empty(test_client, projects_fixture, user1, caplog):
     resp = test_client.get(f'/projects/{projects_fixture.id}/features/',
                            headers={'x-jwt-assertion-test': user1.jwt})
     assert resp.status_code == 200
 
     data = resp.get_json()
     assert len(data['features']) == 0
+    log_statement_for_analytics = (f"Get features of project for user:{user1.username} application:Unknown public_view:False "
+                                   f"project_uuid:{projects_fixture.uuid} project:{projects_fixture.id} "
+                                   f"tapis_system_id:None tapis_system_path:None")
+    assert log_statement_for_analytics in caplog.text
 
 
-def test_get_project_features_empty_public_access(test_client, public_projects_fixture):
-    resp = test_client.get('/projects/{}/features/'.format(public_projects_fixture.id))
+def test_get_project_features_empty_public_access(test_client, public_projects_fixture, caplog):
+    resp = test_client.get('/public-projects/{}/features/'.format(public_projects_fixture.id))
     assert resp.status_code == 200
     data = resp.get_json()
     assert len(data['features']) == 0
+    log_statement_for_analytics = (f"Get features of project for user:Guest_Unknown application:Unknown public_view:True "
+                                   f"project_uuid:{public_projects_fixture.uuid} project:{public_projects_fixture.id} "
+                                   f"tapis_system_id:None tapis_system_path:None")
+    assert log_statement_for_analytics in caplog.text
+
+
+def test_get_project_features_analytics_with_query_params(test_client, public_projects_fixture, caplog):
+    # send analytics-related params to projects endpoint only (until we use headers again
+    # in https://tacc-main.atlassian.net/browse/WG-192)
+    query = {'application': 'hazmapper', 'guest_uuid': "1234"}
+    test_client.get('/public-projects/{}/features/'.format(public_projects_fixture.id), query_string=query)
+    log_statement_for_analytics = (f"Get features of project for user:Guest_1234 application:hazmapper public_view:True "
+                                   f"project_uuid:{public_projects_fixture.uuid} project:{public_projects_fixture.id} "
+                                   f"tapis_system_id:None tapis_system_path:None")
+    assert log_statement_for_analytics in caplog.text
 
 
 def test_get_project_features_single_feature(test_client, projects_fixture, feature_fixture, user1):

geoapi/tests/conftest.py

@@ -123,6 +123,12 @@ def observable_projects_fixture():
         path="/testPath",
         watch_content=True
     )
+
+    # Project system_id/system_path really not used except for analytics.
+    # This could be improved; see https://jira.tacc.utexas.edu/browse/WG-185
+    proj.system_id = obs.system_id
+    proj.system_path = obs.path
+
     obs.project = proj
     proj.users.append(u1)
     db_session.add(obs)

geoapi/utils/decorators.py

@@ -35,6 +35,9 @@ def wrapper(*args, **kwargs):
         except ValueError:
             # if not JWT information is provided in header, then this is a guest user
             guest_uuid = request.headers.get('X-Guest-UUID')
+            if guest_uuid is None:
+                #  Check if in query parameters due to https://tacc-main.atlassian.net/browse/WG-192 and WG-191 */
+                guest_uuid = request.args.get('guest_uuid')
             user = AnonymousUser(guest_unique_id=guest_uuid)
         if user is None:
             try: