fix: use submit_ids for flow submission counts

src/apps/answers/crud/answers.py

@@ -1005,20 +1005,25 @@ async def get_activity_and_flow_ids_by_source_subject(self, source_subject_id: u
 
     @staticmethod
     def _query_submissions_by_subject(subject_column: InstrumentedAttribute, subject_id: uuid.UUID) -> Query:
-        query: Query = select(
-            case(
+        query: Query = (
+            select(
+                func.count(func.distinct(AnswerSchema.submit_id)).label("submission_count"),
+                case(
+                    (
+                        AnswerSchema.flow_history_id.isnot(None),
+                        AnswerSchema.id_from_history_id(AnswerSchema.flow_history_id),
+                    ),
+                    else_=AnswerSchema.id_from_history_id(AnswerSchema.activity_history_id),
+                ).label("activity_id"),
                 (
-                    AnswerSchema.flow_history_id.isnot(None),
-                    AnswerSchema.id_from_history_id(AnswerSchema.flow_history_id),
-                ),
-                else_=AnswerSchema.id_from_history_id(AnswerSchema.activity_history_id),
-            ).label("activity_id"),
-            (
-                AnswerSchema.source_subject_id
-                if subject_column == AnswerSchema.target_subject_id
-                else AnswerSchema.target_subject_id
-            ).label("subject_id"),
-        ).where(subject_column == subject_id)
+                    AnswerSchema.source_subject_id
+                    if subject_column == AnswerSchema.target_subject_id
+                    else AnswerSchema.target_subject_id
+                ).label("subject_id"),
+            )
+            .where(subject_column == subject_id)
+            .group_by("activity_id", "subject_id")
+        )
         return query
 
     async def get_submissions_by_target_subject(self, target_subject_id: uuid.UUID) -> list[dict]:

src/apps/answers/service.py

@@ -1920,25 +1920,27 @@ async def get_target_subject_ids_by_respondent_and_activity_or_flow(
             respondent_subject_id, activity_or_flow_id
         )
 
-    async def get_activity_and_flow_ids_by_target_subject(self, target_subject_id: uuid.UUID) -> list[uuid.UUID]:
+    async def get_activity_and_flow_ids_by_target_subject(self, target_subject_id: uuid.UUID) -> list[dict]:
         """
-        Get a list of activity and flow IDs based on answers submitted for a target subject
+        Get a list of activity and flow IDs based on answers submitted for a target subject.
+        Excludes answers whose source subject was soft-deleted.
 
         The data returned is just a combined list of activity and flow IDs, without any
-        distinction between the two
+        distinction between the two.
         """
         results = await AnswersCRUD(self.answer_session).get_activity_and_flow_ids_by_target_subject(target_subject_id)
         existing_subject_ids = await self._filter_out_soft_deleted_subjects(results)
         activity_ids = [result["activity_id"] for result in results if result["subject_id"] in existing_subject_ids]
 
         return activity_ids
 
-    async def get_activity_and_flow_ids_by_source_subject(self, source_subject_id: uuid.UUID) -> list[uuid.UUID]:
+    async def get_activity_and_flow_ids_by_source_subject(self, source_subject_id: uuid.UUID) -> list[dict]:
         """
-        Get a list of activity and flow IDs based on answers submitted for a source subject
+        Get a list of activity and flow IDs based on answers submitted for a source subject.
+        Excludes answers whose target subject was soft-deleted.
 
         The data returned is just a combined list of activity and flow IDs, without any
-        distinction between the two
+        distinction between the two.
         """
         results = await AnswersCRUD(self.answer_session).get_activity_and_flow_ids_by_source_subject(source_subject_id)
         existing_subject_ids = await self._filter_out_soft_deleted_subjects(results)
@@ -1948,7 +1950,7 @@ async def get_activity_and_flow_ids_by_source_subject(self, source_subject_id: u
 
     async def _filter_out_soft_deleted_subjects(self, submissions: list[dict]) -> set[uuid.UUID]:
         """
-        Return subject_ids contained in given submissions array that correspond to soft-deleted subjects
+        Filter out submissions whose subject_id column corresponds to soft-deleted subjects
         """
         subject_ids = set([activityOrFlow["subject_id"] for activityOrFlow in submissions])
 
@@ -1970,23 +1972,23 @@ async def get_submissions_by_subject(self, subject_id: uuid.UUID) -> Submissions
 
         submissions_activity_count = SubmissionsActivityCountBySubject(subject_id=subject_id)
 
-        for activityOrFlow in submissions_target:
+        for activity_submissions in submissions_target:
             activity_counters = submissions_activity_count.activities.setdefault(
-                uuid.UUID(activityOrFlow["activity_id"]), SubmissionsSubjectCounters()
+                uuid.UUID(activity_submissions["activity_id"]), SubmissionsSubjectCounters()
             )
-            respondent_subject_id = activityOrFlow["subject_id"]
+            respondent_subject_id = activity_submissions["subject_id"]
             if respondent_subject_id in existing_subject_ids:
                 activity_counters.respondents.add(respondent_subject_id)
-                activity_counters.subject_submissions_count += 1
+                activity_counters.subject_submissions_count += activity_submissions["submission_count"]
 
-        for activityOrFlow in submissions_respondent:
+        for activity_submissions in submissions_respondent:
             activity_counters = submissions_activity_count.activities.setdefault(
-                uuid.UUID(activityOrFlow["activity_id"]), SubmissionsSubjectCounters()
+                uuid.UUID(activity_submissions["activity_id"]), SubmissionsSubjectCounters()
             )
-            target_subject_id = activityOrFlow["subject_id"]
+            target_subject_id = activity_submissions["subject_id"]
             if target_subject_id in existing_subject_ids:
                 activity_counters.subjects.add(target_subject_id)
-                activity_counters.respondent_submissions_count += 1
+                activity_counters.respondent_submissions_count += activity_submissions["submission_count"]
 
         return submissions_activity_count