#coding:utf-8

 from hashlib import md5

 import os,time,sys

 import xlrd

 import os

 md5dic = {}

 #读取Excel表中的md5值

 def readXls(fname):

     xls = xlrd.open_workbook(fname)

     sheet_name = xls.sheet_names()

     shxrange = range(xls.nsheets)

     try:

         sh = xls.sheet_by_name(sheet_name[0])

     except:

         print "no sheet existed"

     #获取行数

     nrows = sh.nrows

     #获取列数

     ncols = sh.ncols

     #print "nrows %d, ncols %d" % (nrows,ncols)

     #获取第一行第一列数据

     cell_value = sh.cell_value(1,1)

     #print cell_value

     row_list = []

     #获取各行数据

     for i in range(1,nrows):

         row_data = sh.row_values(i)[0]

         #print row_data

         row_list.append(row_data)

     md5list = list(set(row_list))

     return md5list

 def calMD5(str):

   m = md5()

   m.update(str)

   return m.hexdigest() 

 #获取指定文件的md5值

 def calMD5ForFile(file):

   m = md5()

   a_file = open(file, 'rb')

   m.update(a_file.read())

   a_file.close()

   return m.hexdigest()

 #获取指定目录下文件的md5值保存到字典中

 def calMD5ForFolder(dir):

     for root, subdirs, files in os.walk(dir):

         for file in files:

             filefullpath = os.path.join(root,file)

             filerelpath = os.path.relpath(filefullpath,dir)

             md5 = calMD5ForFile(filefullpath)

             md5dic[filefullpath] = md5

     return md5dic

 #删除与excel表中md5值相同的文件

 def delFile(filemd5,md5dic):

     for filefullpath in md5dic.keys():

         if md5dic[filefullpath] == filemd5:

             os.remove(filefullpath)

             print filemd5,filefullpath

 if __name__ == "__main__":

     if len(sys.argv) < 3:

         print "Please input the file path."

         sys.exit()

     xlsdir = sys.argv[1]

     sourceDir = sys.argv[2]

     if os.path.exists(sourceDir)==False:

         print "path not exists"

     if os.path.isdir(sourceDir)==False:

         print "not a directory"

     md5dic = calMD5ForFolder(sourceDir)

     md5list = readXls(xlsdir)

     for i in md5list:

         delFile(i,md5dic)