fix: Added discord summary type!

To query the database for thread summaries, it wasn't possible to do with just checking the null values for thread in metadata.
TogetherCrew · Feb 12, 2024 · 8d820d2 · 8d820d2
1 parent 72b01a5
commit 8d820d2
Showing 1 changed file with 6 additions and 10 deletions.
diff --git a/dags/hivemind_etl_helpers/src/db/discord/summary/summary_utils.py b/dags/hivemind_etl_helpers/src/db/discord/summary/summary_utils.py
@@ -33,8 +33,9 @@ def transform_thread_summary_to_document(
             "date": summary_date,
             "thread": thread_name,
             "channel": thread_channel,
+            "type": "thread",
         },
-        excluded_embed_metadata_keys=["date", "thread", "channel"],
+        excluded_embed_metadata_keys=["date", "thread", "channel", "type"],
     )
 
     return thread_summary_document
@@ -65,11 +66,8 @@ def transform_channel_summary_to_document(
 
     channel_summary_document = Document(
         text=channel_summary,
-        metadata={
-            "date": summary_date,
-            "channel": channel_name,
-        },
-        excluded_embed_metadata_keys=["date", "thread", "channel"],
+        metadata={"date": summary_date, "channel": channel_name, "type": "channel"},
+        excluded_embed_metadata_keys=["date", "thread", "channel", "type"],
     )
 
     return channel_summary_document
@@ -99,10 +97,8 @@ def transform_daily_summary_to_document(
         summary = daily_summary[date]
         doc = Document(
             text=summary,
-            metadata={
-                "date": date,
-            },
-            excluded_embed_metadata_keys=["date", "thread", "channel"],
+            metadata={"date": date, "type": "day"},
+            excluded_embed_metadata_keys=["date", "thread", "channel", "type"],
         )
         daily_summary_documents.append(doc)