'''
function：将excel文件转换为text
author：Nstock
date：2018/3/1
'''
import pandas as pd
import re
import codecs
 
#将excel转化为txt文件
def exceltotxt(excel_dir, txt_dir): 
 with codecs.open(txt_dir, 'w', 'utf-8') as f:
 neg=pd.read_excel(excel_dir, header=None, index=None)
 f.write(neg.to_string())
 
#去除记录行首的数字和空格
def del_linehead_number_speace(orig_txt_dir,saveas_txt_dir):
 with open(orig_txt_dir,'r+') as f, open(saveas_txt_dir,'r+') as fw:
 lines = f.readlines()
 print(len(lines)) #行数
 texts = [re.sub(r'(\d)+(\s)+','',lines[num]) for num in range(len(lines)) ]
 
 texts = list(set(texts)) #去重如果要保留重复记录注释该行
 
 line_num = len(texts)
#  for num in range(line_num):   #查看转化后的文本
#  print(texts[num])
 fw.writelines(texts)
 
exceltotxt('./data/neg.xls', './data/neg_temp.txt') 
del_linehead_number_speace('./data/neg_temp.txt','./data/neg.txt')