Resolve "données cirm-cfbp 7 juin 2021 : entrées non conforme"

Closes #45 (closed)

softwares/Florilege/scripts/preprocess-cirm-cfbp.py

@@ -18,12 +18,13 @@ strain_index = int(args.strain_index)
 habitat_indexes = map(int,args.habitat_index.split(','))
 
 cirm_data = pd.read_excel(args.input, dtype='object')
+cirm_data.replace(to_replace=[r"\\n|\\r", "\n|\r"], value=[" "," "], regex=True, inplace=True)
+cirm_data.replace(to_replace=[r'["“”]'], value=[""], regex=True, inplace=True)
 
 taxa_dict = set()
 for i in range(len(cirm_data)) :
   taxon = str(cirm_data.iloc[i, taxa_index])
   strain = str(cirm_data.iloc[i, strain_index])
-  taxon = taxon.replace('"','')
   taxa_dict.add(taxon)
   taxa_dict.add(taxon + " CFBP " + strain)
   taxa_dict.add(taxon + " CFBP" + strain)