sklearn实战-乳腺癌细胞数据挖掘

# -*- coding: utf-8 -*-

"""

Created on Mon Sep 19 09:36:48 2016

@author: Administrator

"""

import os

from os.path import join, getsize  

import shutil,xlrd

excelFilename="test.xlsx"

sheetName="Sheet1"

data = xlrd.open_workbook(excelFilename)

table = data.sheets()[0]

#总pdf列表

totalpdfs_list=table.col_values(0)

pdf_test="20160607_2.pdf"

#存储大于2M的pdf文件

list_bigFileSize=[]

#存储大于2M的pdf文件列表

def get_list_bigFileSize():  

   for file in totalpdfs_list:

      size= getsize(file)

      if size>2048000:

          list_bigFileSize.append(file)

   return list_bigFileSize  

#移动函数，目录里不匹配文件移入unmatching_file文件夹

def RemoveFile():

    dir="大于2M的pdf"

    for file in list_bigFileSize:

        shutil.move(file,dir)

list_bigFileSize=get_list_bigFileSize()

RemoveFile()

https://study.163.com/course/introduction.htm?courseId=1005214003&utm_campaign=commission&utm_source=cp-400000000398149&utm_medium=share