'''function：將excel文件轉換為textauthor：Nstockdate：2018/3/1'''import pandas as pdimport reimport codecs #將excel轉化為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()) #去除記錄行首的數字和空格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)) #行數 texts = [re.sub(r'(/d)+(/s)+','',lines[num]) for num in range(len(lines)) ]  texts = list(set(texts)) #去重如果要保留重復記錄注釋該行  line_num = len(texts)#  for num in range(line_num):   #查看轉化后的文本#  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')