Add more edge cases, fix incorrect time_values

Author: cchuong

src/acquisition/rvdss/pull_historic.py

@@ -276,6 +276,9 @@ def create_percent_positive_detection_table(table,modified_date,start_year, flu=
     table.columns=[re.sub("tests",virus+"_tests",c) for c in table.columns]
     table.columns =[re.sub(" ","_",col) for col in table.columns]
 
+    # make sure weeks 1-9 have leading 0 when creating epiweek
+    table['week'] = pd.to_numeric(table['week'],downcast="integer")
+    
     table=table.rename(columns={'week':"epiweek"})
     table['epiweek'] = [get_report_date(week, start_year,epi=True) for week in table['epiweek']]
 
@@ -322,10 +325,13 @@ def fix_edge_cases(table,season,caption,current_week):
         #  In week 11 of the 2022-2023 season, in the positive hmpv table,
         # a date is written as 022-09-03, instead of 2022-09-03
          table.loc[table['week'] == 35, 'week end'] = "2022-09-03"
-    elif season[0] == '2016' and current_week == 32 and "influenza" in caption.text.lower():
-        #  In week 11 of the 2022-2023 season, in the positive hmpv table,
-        # a date is written as 022-09-03, instead of 2022-09-03
-         table.loc[table['week'] == 32, 'week end'] = "2017-08-12"
+    elif season[0] == '2016':
+        if current_week == 32 and "influenza" in caption.text.lower():
+            #  In week 11 of the 2022-2023 season, in the positive hmpv table,
+            # a date is written as 022-09-03, instead of 2022-09-03
+            table.loc[table['week'] == 32, 'week end'] = "2017-08-12"
+        if "detections" not in caption.text.lower():
+            table.loc[table['week'] == 40, 'week end'] = "2016-10-08"
     elif season[0] == '2021' and "parainfluenza" in caption.text.lower():
         #  In multiple weeks of the 2021-2022 season, in the positive hpiv table,
         # the date for week 12 is 2022-03-19, instead of 2022-03-26
@@ -338,6 +344,8 @@ def fix_edge_cases(table,season,caption,current_week):
             table.loc[table['week'] == 36, 'week end'] = "2013-09-07"
         elif current_week == 35:
             table.loc[table['week'] == 35, 'week end'] = "2013-08-31"
+        elif current_week == 38:
+            table.loc[table['week'] == 38, 'week end'] = "2013-09-21"
     return(table)
 
 def fetch_one_season_from_report(url):

tests/acquisition/rvdss/test_pull_historic.py

@@ -49,7 +49,9 @@
     pd.DataFrame([{"week":32,"week end":"2017-08-17"}]),
     pd.DataFrame([{"week":12,"week end":"2022-03-19"}]),
     pd.DataFrame([{"week":35,"week end":"2012-08-31"}]),
-    pd.DataFrame({"week":[35,36],"week end":["2012-08-31","2012-09-07"]})]
+    pd.DataFrame({"week":[35,36],"week end":["2012-08-31","2012-09-07"]}),
+    pd.DataFrame([{"week":40,"week end":"2016-10-09"}]),
+    pd.DataFrame([{"week":38,"week end":"2013-09-22"}])]
 
 expected_edge_case_tables=[
     pd.DataFrame(columns=['week', 'week end', 'canada tests', 'entero/rhino%', 'at tests',
@@ -73,7 +75,9 @@
     pd.DataFrame([{"week":32,"week end":"2017-08-12"}]),
     pd.DataFrame([{"week":12,"week end":"2022-03-26"}]),
     pd.DataFrame([{"week":35,"week end":"2013-08-31"}]),
-    pd.DataFrame({"week":[35,36],"week end":["2013-08-31","2013-09-07"]})]
+    pd.DataFrame({"week":[35,36],"week end":["2013-08-31","2013-09-07"]}),
+    pd.DataFrame([{"week":40,"week end":"2016-10-08"}]),
+    pd.DataFrame([{"week":38,"week end":"2013-09-21"}])]
 
 example_edge_case_captions=[
     [t for t in captions if "Entero" in t.text][0],
@@ -85,14 +89,16 @@
     [t for t in captions if "Influenza" in t.text][0],
     [t for t in captions if "Para" in t.text][0],
     [t for t in captions if "RSV" in t.text][0],
+    [t for t in captions if "Influenza" in t.text][0],
+    [t for t in captions if "Influenza" in t.text][0],
     [t for t in captions if "Influenza" in t.text][0]]
 
 example_edge_case_seasons=[["2017","2018"],["2017","2018"],["2017","2018"],
                            ["2015","2016"],["2022","2023"],["2021","2022"],
                            ["2016","2017"],["2021","2022"], ["2013","2014"],
-                           ["2013","2014"]]
+                           ["2013","2014"],["2016","2017"],["2013","2014"]]
 
-example_edge_case_weeks=[35,35,47,41,11,10,32,14,35,36]
+example_edge_case_weeks=[35,35,47,41,11,10,32,14,35,36,10,38]
 
 class TestPullHistoric():