Feature: Query authors and respondents #18

- Bugfix: Duplicated counts at aggregated levels (category, stream), due to the same message potentially being matched by multiple keywords.

fhir_zulip_nlp/fhir_zulip_nlp.py

@@ -410,7 +410,6 @@ def create_report_users(df: pd.DataFrame) -> Tuple[pd.DataFrame, pd.DataFrame, p
     for stream_id in streams:
         stream_name = stream_id_name_map[str(stream_id)]
         df_i = df[df['stream_id'] == stream_id]
-        # todo: Would they like the 0 totals as well? If so, rather than .unique(), should use `category_keywords`
         categories = df_i['category'].unique()
         for c in categories:
             df_i2 = df_i[df_i['category'] == c]
@@ -426,7 +425,6 @@ def create_report_users(df: pd.DataFrame) -> Tuple[pd.DataFrame, pd.DataFrame, p
                     author_id: int = list(
                         df_i4[df_i4['timestamp'] == author_timestamp]['sender_id'].to_dict().values())[0]
                     participant_roles[author_id] = 'author'
-
                     # Populate: user_participation
                     for user_id, role in participant_roles.items():
                         row = {
@@ -453,6 +451,7 @@ def create_report_users(df: pd.DataFrame) -> Tuple[pd.DataFrame, pd.DataFrame, p
         streams = list(df_i['stream'].unique())
         for s in streams:
             df_i2 = df_i[df_i['stream'] == s]
+            # TODO: Dedupe here?
             role_counts: Dict[str, int] = df_i2['user.role'].value_counts().to_dict()
             for role, count in role_counts.items():
                 row = {
@@ -467,6 +466,7 @@ def create_report_users(df: pd.DataFrame) -> Tuple[pd.DataFrame, pd.DataFrame, p
                 user_participation_stats.append(row)
             categories = list(df_i2['category'].unique())
             for c in categories:
+                # TODO: Dedupe here?
                 df_i3 = df_i2[df_i2['category'] == c]
                 role_counts: Dict[str, int] = df_i3['user.role'].value_counts().to_dict()
                 for role, count in role_counts.items():