import sys
import nltk

args = sys.argv
dir='label_feature/'

if(len(args)<2):
        print("Usage: python3 %s train/test celltype/antigen" % args[0])

for i in range(10):
	fin=open(dir+'c10_'+args[1]+str(i+1)+'_human_'+args[2]+'_feature.txt')	
	fout=open(dir+'c10_'+args[1]+str(i+1)+'_human_'+args[2]+'_feature_normalized.txt','w')	
	puncts='+_-.;()'
	for text in fin:
		text = text.lower()
		for sym in puncts:
			text = text.replace(sym, '')
		fout.write("%s" % text)
	fout.close()