Use normalize_summary for document creation (#1981)

* Use normalize_summary for document creation

* minor linter issue
* increase memory for functions

* Push normalize_summary after the LLM API call to avoid having devs to remember

* Not needed

Author: chiroptical

llm/backfill_summaries.py

@@ -21,7 +21,6 @@
 from firebase_admin import firestore
 from bill_on_document_created import get_categories_from_topics, CATEGORY_BY_TOPIC
 import csv
-from normalize_summaries import normalize_summary
 
 # Module constants
 FIREBASE_COLLECTION_PATH = "generalCourts/194/bills"
@@ -73,7 +72,7 @@ def make_bill_summary(bill_id, status, summary, topics):
             continue
         # Note: `normalize_summary` does some post-processing to clean up the summaries
         # As of 2025-10-21 this was necessary due to the LLM prompt
-        summary = normalize_summary(summary["summary"])
+        summary = summary["summary"]
         bill.reference.update({"summary": summary})
 
         # If the topics are already populated, just make a note of it

llm/llm_functions.py

@@ -63,6 +63,8 @@
 from prompts import *
 from tag_categories import *
 
+from normalize_summaries import normalize_summary
+
 GPT_MDOEL_VERSION = 'gpt-4o-mini'
 MAX_TOKEN_LIMIT = 128000
 
@@ -434,7 +436,7 @@ def get_summary_api_function(bill_id: str, bill_title: str, bill_text: str) -> d
     if status_code != 1: 
         return {'status': status_code, 'summary': ''}
     else: 
-        return {'status': status_code, 'summary': results.response}
+        return {'status': status_code, 'summary': normalize_summary(results.response)}
 
 def get_tags_api_function(bill_id: str, bill_title: str, bill_text: str) -> dict:
 
@@ -1042,4 +1044,4 @@ def large_docs(bill_details: BillDetails, query: str, llm: ChatOpenAI) -> str:
     except Exception as e: 
         print(e)
 
-    return response
+    return response

llm/main.py

@@ -22,10 +22,10 @@ def is_intersection(keys, required_keys):
 def set_openai_api_key():
     match os.environ.get("MAPLE_DEV"):
         case "prod":
-            if os.environ.get("OPENAI_PROD") != None:
+            if os.environ.get("OPENAI_PROD") is not None:
                 os.environ["OPENAI_API_KEY"] = os.environ["OPENAI_PROD"]
         case _:  # if "dev" or unspecified, use OPENAI_DEV
-            if os.environ.get("OPENAI_DEV") != None:
+            if os.environ.get("OPENAI_DEV") is not None:
                 os.environ["OPENAI_API_KEY"] = os.environ["OPENAI_DEV"]
 
 
@@ -81,6 +81,8 @@ def httpsflaskexample(req: https_fn.Request) -> https_fn.Response:
 
 @on_document_created(
     secrets=["OPENAI_DEV", "OPENAI_PROD"],
+    timeout_sec=300,
+    memory=options.MemoryOption.GB_1,
     document="generalCourts/{session_id}/bills/{bill_id}",
 )
 def add_summary_on_document_created(event: Event[DocumentSnapshot | None]) -> None: