From e0b69e18f1105f40be63b00d478d60c5ce2e7ec2 Mon Sep 17 00:00:00 2001
From: Yorick Barbanneau <ephase@xieme-art.org>
Date: Tue, 19 Apr 2022 22:40:35 +0200
Subject: [PATCH] Process towns DataFrame

Add area colum from statistics
---
 create_db.py | 20 +++++++++++++++++---
 1 file changed, 17 insertions(+), 3 deletions(-)

diff --git a/create_db.py b/create_db.py
index 08a6973..6a7e641 100755
--- a/create_db.py
+++ b/create_db.py
@@ -111,6 +111,7 @@ def import_statistics_csv(raw_file):
     """
 
     logger.info('import town from {}'.format(raw_file))
+    stats_convert= lambda x: x if len(str(x)) == 5 else f'0{x}'
     stats = pd.read_csv(raw_file,
         usecols=["CODGEO","SUPERF","P18_POP","P13_POP","P08_POP","D99_POP",
             "NAIS1318","NAIS0813","NAIS9908","NAIS9099","NAIS8290","DECE1318",
@@ -120,7 +121,9 @@ def import_statistics_csv(raw_file):
             "P13_RP","P08_RP","D99_RP","D90_RP","D82_RP", "P18_RSECOCC",
             "P13_RSECOCC","P08_RSECOCC","D99_RSECOCC","D90_RSECOCC",
             "D82_RSECOCC"],
-            sep=';')
+            sep=';',
+            converters={'CODGEO':stats_convert}
+            )
     return stats
 
 def get_single_date(attr):
@@ -241,8 +244,7 @@ if __name__ == '__main__':
     ## create statistics dataframes
     # 
     # We need to first iterate on statistics
-    if args.verbose or args.debug:
-        t.start()
+    t.start('Process_Statistics')
         
     c_stats = pd.DataFrame(columns = ['com','id_indicateur','date_debut',
         'date_fin','valeur']
@@ -290,6 +292,18 @@ if __name__ == '__main__':
                 temp['valeur'].append(value)
 
     t.stop()
+
+    t.start('Process_town')
+    print(statistics[['SUPERF','CODGEO']])
+    towns = pd.merge(towns, 
+            statistics[['CODGEO', 'SUPERF']], 
+            left_on=['COM'],
+            right_on=['CODGEO'],
+            how = 'left'
+    )[['COM','NCC','LIBELLE', 'DEP', 'SUPERF']]
+    t.stop()
+    logger.debug(towns)
+
     t.get_total_time()
 
     sys.exit()