Update Spreadsheet Retrieval using Metadata (#849)

* fix retrieve spreadsheet

* improved fallback mechanism

* get user friendly filename

data_store/controllers/ingest.py

@@ -28,6 +28,7 @@
     get_or_generate_submission_id,
 )
 from data_store.controllers.mappings import INGEST_MAPPINGS, DataMapping
+from data_store.controllers.retrieve_submission_file import get_custom_file_name
 from data_store.db import db
 from data_store.db.entities import Fund, Programme, ProgrammeJunction, Submission
 from data_store.db.queries import (
@@ -504,6 +505,7 @@ def save_submission_file_s3(excel_file: FileStorage, submission_id: str):
         metadata={
             "submission_id": submission_id,
             "filename": f"{fund_type}-{str(uuid)}.xlsx",
+            "download_filename": f"{get_custom_file_name(uuid)}.xlsx",
             "programme_name": programme_name,
         },
     )

data_store/controllers/retrieve_submission_file.py

@@ -1,3 +1,5 @@
+from typing import Optional
+
 from config import Config
 from data_store.aws import create_presigned_url, get_file_header
 from data_store.db.entities import Fund, Organisation, Programme, ProgrammeJunction, ReportingRound, Submission
@@ -52,7 +54,12 @@ def retrieve_submission_file(submission_id) -> str:
     return presigned_url
 
 
-def get_custom_file_name(submission_id: str) -> str:
+def get_custom_file_name(
+    submission_id: str,
+    fallback_date: Optional[str] = None,
+    fallback_fund_code: Optional[str] = None,
+    fallback_org_name: Optional[str] = None,
+) -> str:
     submission_info = (
         Programme.query.join(ProgrammeJunction)
         .join(Submission)
@@ -71,10 +78,16 @@ def get_custom_file_name(submission_id: str) -> str:
         .one_or_none()
     )
 
-    date = submission_info.submission_date.strftime("%Y-%m-%d")
-    start_date = submission_info.observation_period_start.strftime("%b%Y")
-    end_date = submission_info.observation_period_end.strftime("%b%Y")
+    if submission_info:
+        date = submission_info.submission_date.strftime("%Y-%m-%d")
+        start_date = submission_info.observation_period_start.strftime("%b%Y")
+        end_date = submission_info.observation_period_end.strftime("%b%Y")
 
-    file_name = f"{date}-{submission_info.fund_code}-{submission_info.organisation_name}-{start_date}-{end_date}"
+        file_name = f"{date}-{submission_info.fund_code}-{submission_info.organisation_name}-{start_date}-{end_date}"
+    else:
+        date = fallback_date
+        fund_code = fallback_fund_code
+        org_name = fallback_org_name
+        file_name = f"{date}-{fund_code}-{org_name}-{submission_id}"
 
     return file_name.replace(" ", "-").lower()

find/main/routes.py

@@ -195,12 +195,23 @@ def retrieve_spreadsheet(fund_code: str, submission_id: str):
         return abort(404)
 
     form = RetrieveForm()
-
+    file_metadata = file_header["metadata"]
+    file_name = file_metadata.get("download_filename")
+    programme_name = file_metadata.get("programme_name")
+    submission_date = file_header["last_modified"].strftime("%d %B %Y")
+    if not file_name:
+        custom_file_name = get_custom_file_name(
+            submission_id=submission_id,
+            fallback_date=submission_date,
+            fallback_fund_code=fund_code,
+            fallback_org_name=programme_name,
+        )
+        file_name = f"{custom_file_name}.xlsx"
     if form.validate_on_submit():
         presigned_url = create_presigned_url(
             bucket_name=Config.AWS_S3_BUCKET_SUCCESSFUL_FILES,
             file_key=object_name,
-            filename=f"{get_custom_file_name(submission_id)}.xlsx",
+            filename=file_name,
         )
 
         return redirect(presigned_url)