#Lecture données caf
if 'CAF_aggregate_PDL.csv' not in os.listdir(dataDir + '/CAF/'):
filepath = dataDir + '/CAF/CAF.xlsx'
df1 = pd.read_excel(filepath, sheet_name='PPACOM2017')
df1 = df1.iloc[:,:].add_prefix('PPA_')
df2 = pd.read_excel(filepath, sheet_name='BASREVENUCOM2017')
df2 = df2.iloc[:,3:].add_prefix('BAS_')
df3 = pd.read_excel(filepath, sheet_name='ENFANTAGECOM2017')
df3 = df3.iloc[:,3:].add_prefix('ENF_')
df4 = pd.read_excel(filepath, sheet_name='EJCOM2017')
df4 = df4.iloc[:,3:].add_prefix('EJC_')
df5 = pd.read_excel(filepath, sheet_name='NIVCOMTOTAL2017')
df5 = df5.iloc[:,3:].add_prefix('NIV_')
df6 = pd.read_excel(filepath, sheet_name='RASCOM2017')
df6 = df6.iloc[:,3:].add_prefix('RAS_')
df7 = pd.read_excel(filepath, sheet_name='NIVCOMPARTPRESTA2017')
df7 = df7.iloc[:,3:].add_prefix('PRES_')
df8 = pd.read_excel(filepath, sheet_name='NIVCOMAAH2017')
df8 = df8.iloc[:,3:].add_prefix('PRES_')
#concatenation des colonnes et suppression des colonnes dupliquées, subset pdl
dfc = pd.concat([df1, df2, df3, df4, df5, df6, df7, df8], axis=1)
dfc = dfc.rename(columns={"PPA_Communes":"Communes","PPA_Numéro_département": "Numéro_département", "PPA_Codes_Insee": "Codes_Insee", "PPA_Numéro_département":"Numéro_département"})
dfc = dfc[dfc['Numéro_département'].isin([44,85,53,72,49])]
dfc.to_csv("./data/CAF_aggregate_PDL.csv")
dfc = pd.read_csv(dataDir + '/CAF/CAF_aggregate_PDL.csv")
print(len(dfc), " lignes de données caf")