update naming and constants

Author: cchuong

src/acquisition/rvdss/constants.py

@@ -51,7 +51,7 @@
 # Regions are groups of provinces that are geographically close together. Some single provinces are reported as their own region (e.g. Québec, Ontario).
 REGIONS = ['atlantic','atl','at','province of québec','québec','qc','province of ontario','ontario','on',
             'prairies', 'pr', "british columbia",'bc',"territories",'terr',]
-NATION = ["canada","can",'ca',]
+NATION = ["canada","can",'ca']
 
 # Construct dashboard and data report URLS.
 DASHBOARD_BASE_URL = "https://health-infobase.canada.ca/src/data/respiratory-virus-detections/"

src/acquisition/rvdss/utils.py

@@ -100,15 +100,15 @@ def preprocess_table_columns(table):
     table.columns = [re.sub("canada","can",t) for t in table.columns]
     table.columns = [re.sub(r"\bcb\b","bc",t) for t in table.columns]
 
-    table.columns =[re.sub(r"h1n1 2009 |h1n12009|a_h1|ah1\b", "ah1n1pdm09", s)for s in table.columns]
+    table.columns =[re.sub(r"h1n1 2009 |h1n12009|a_h1|ah1\b|ah1pdm09", "ah1n1pdm09", s)for s in table.columns]
     table.columns =[re.sub(r"a_uns", "auns", s)for s in table.columns]
     table.columns =[re.sub(r"a_h3", "ah3", s)for s in table.columns]
 
     table.columns =[abbreviate_virus(col) for col in table.columns] # abbreviate viruses
     table.columns = [re.sub(r"flu a","flua",t) for t in table.columns]
     table.columns = [re.sub(r"flu b","flub",t) for t in table.columns]
     table.columns = [re.sub(r"flutest\b","flu test", col) for col in table.columns]
-    table.columns = [re.sub(r"other hpiv|other_hpiv","hpivother",t) for t in table.columns]
+    table.columns = [re.sub(r"other hpiv|other_hpiv|hpiv other|hpiv_other","hpivother",t) for t in table.columns]
 
     table.columns=[re.sub(r'bpositive','b_positive',c) for c in table.columns]
     table.columns=[re.sub(r'apositive','a_positive',c) for c in table.columns]