add the case when there is no daily file stored

Author: Jingjing Tang

claims_hosp/delphi_claims_hosp/backfill.py

@@ -79,6 +79,8 @@ def merge_backfill_file(backfill_dir, backfill_merge_day, today,
         considerations
     """
     new_files = glob.glob(backfill_dir + "/claims_hosp_as_of_*")
+    if len(new_files) == 0: # if no any daily file is stored
+        return
 
     def get_date(file_link):
         # Keep the function here consistent with the backfill path in

claims_hosp/tests/backfill/.gitignore

@@ -47,10 +47,20 @@ def test_merge_backfill_file(self):
 
         today = datetime.today()
 
-        new_files = glob.glob(backfill_dir + "/claims_hosp*.parquet")
         fn = "claims_hosp_from_20200611_to_20200614.parquet"
         assert fn not in os.listdir(backfill_dir)
 
+        # Check when there is no daily file to merge.
+        today = datetime(2020, 6, 14)
+        merge_backfill_file(backfill_dir, today.weekday(), today, 
+                            test_mode=True, check_nd=8)
+        assert fn not in os.listdir(backfill_dir)
+        
+        # Generate backfill daily files     
+        for d in range(11, 15):
+            dropdate = datetime(2020, 6, d)        
+            store_backfill_file(DATA_FILEPATH, dropdate, backfill_dir)
+        
         # Check the when the merged file is not generated
         today = datetime(2020, 6, 14)
         merge_backfill_file(backfill_dir, today.weekday(), today, 
@@ -63,13 +73,18 @@ def test_merge_backfill_file(self):
         assert fn in os.listdir(backfill_dir)
 
         # Read daily file
+        new_files = glob.glob(backfill_dir + "/claims_hosp*.parquet")
         pdList = []        
         for file in new_files:
             df = pd.read_parquet(file, engine='pyarrow')
             issue_date = datetime.strptime(file[-16:-8], "%Y%m%d")
             df["issue_date"] = issue_date
             df["lag"] = [(issue_date - x).days for x in df["time_value"]]
             pdList.append(df)
+            os.remove(file)
+        new_files = glob.glob(backfill_dir + "/claims_hosp*.parquet")
+        assert len(new_files) == 1
+
         expected = pd.concat(pdList).sort_values(["time_value", "fips"])
 
         # Read the merged file