Python编程实现两个文件夹里文件的对比功能示例

Python编程实现两个文件夹里文件的对比功能示例

本文实例讲述了Python编程实现两个文件夹里文件的对比功能。分享给大家供大家参考，具体如下：

#-*-coding:utf-8-*-
#===============================================================================
# 目录对比工具(包含子目录 )，并列出
# 1、A比B多了哪些文件
# 2、B比A多了哪些文件
# 3、二者相同的文件:文件大小相同 VS 文件大小不同 (Size相同文件不打印:与Size不同文件显示未排序)
#===============================================================================
import os, time,difflib
AFILES = [] #EE
BFILES = [] #SVN
COMMON = [] #EE & SVN
def getPrettyTime(state):
  return time.strftime('%y-%m-%d %H:%M:%S', time.localtime(state.st_mtime))
# def getpathsize(dir): #获取文件大小的函数,未用上,仅供学习.故注释掉
#   size=0
#   for root, dirs, files in os.walk(dir):
#   #root:目录:str 如: C:CopySVNSystemObjectTopoProcedureBuilt-in
#   #dirs:目录名称:列表: 如 ['Parsers']
#   #files:名称:列表: 如 ['011D0961FB42416AA49D5E82945DE7E9.og',...]
#   #file:目录:str, 如 011D0961FB42416AA49D5E82945DE7E9.og
#     for file in files:
#path = os.path.join(root,file)
#size = os.path.getsize(path)
#   return size
def dirCompare(apath,bpath):
  afiles = []
  bfiles = []
  for root, dirs , files in os.walk(apath):
    for f in files:
      afiles.append(root + "\" + f)
  for root, dirs , files in os.walk(bpath):
    for f in files:
      bfiles.append(root + "\" + f)
      #sizeB = os.path.getsize(root + "\" + f) 此处定义的size无法在commonfiles进行比较. (A,B在各自的循环里面)
  # 去掉afiles中文件名的apath (拿A,B相同的路径文件名,做成集合,去找交集)
  apathlen = len(apath)
  aafiles = []
  for f in afiles:
    aafiles.append(f[apathlen:])
  # 去掉bfiles中文件名的bpath
  bpathlen = len(bpath)
  bbfiles = []
  for f in bfiles:
    bbfiles.append(f[bpathlen:])
  afiles = aafiles
  bfiles = bbfiles
  setA = set(afiles)
  setB = set(bfiles)
  #print('%$%'+str(len(setA)))
  #print('%%'+str(len(setB)))
  commonfiles = setA & setB # 处理共有文件
  #print ("===============File with different size in '", apath, "' and '", bpath, "'===============")
  #将结果输出到本地
  #with open(os.getcwd()+'diff.txt','w') as di:
    #di.write("===============File with different size in '", apath, "' and '", bpath, "'===============")
  for f in sorted(commonfiles):
    sA=os.path.getsize(apath + "\" + f)
    sB=os.path.getsize(bpath + "\" + f)
    if sA==sB: #共有文件的大小比较
      #pass #print (f + "tt" + getPrettyTime(os.stat(apath + "\" + f)) + "tt" + getPrettyTime(os.stat(bpath + "\" + f)))
      #以下代码是处理大小一致，但是内容可能不一致的情况
      #print("in sa=sb")
      #print(os.getcwd())
      saf=[]
      sbf=[]
      sAfile=open(apath + "\" + f)
      iter_f=iter(sAfile)
      for line in iter_f:
 saf.append(line)
      sAfile.close()
      sBfile=open(bpath + "\" + f)
      iter_fb=iter(sBfile)
      for line in iter_fb:
 sbf.append(line)
      sBfile.close()
      saf1=sorted(saf)
      sbf1=sorted(sbf)
      if(len(saf1)!=len(sbf1)):
 with open(os.getcwd()+'\comment_diff.txt','a') as fp:
   print(os.getcwd())
   fp.write(apath + "\" + f+" lines size not equal "+bpath + "\" + f+'n')
      else:
 for i in range(len(saf1)):
   #print("into pre")
   if(saf1[i]!=sbf1[i]):
     print('into commont')
     with open(os.getcwd()+'\comment_diff.txt','a') as fp1:
fp1.write(apath + "\" + f+" content not equal "+bpath + "\" + f+'n')
break
    else:
      with open (os.getcwd()+'\diff.txt','a') as di:
 di.write("File Name=%s  EEresource file size:%d  != SVN file size:%d" %(f,sA,sB)+'n')
      #print ("File Name=%s  EEresource file size:%d  != SVN file size:%d" %(f,sA,sB))
  # 处理仅出现在一个目录中的文件
  onlyFiles = setA ^ setB
  aonlyFiles = []
  bonlyFiles = []
  for of in onlyFiles:
    if of in afiles:
      aonlyFiles.append(of)
    elif of in bfiles:
      bonlyFiles.append(of)
  #print ("###################### EE resource onLY ###########################")
  #print ("#only files in ", apath)
  for of in sorted(aonlyFiles):
    with open (os.getcwd()+'\EEonly.txt','a') as ee:
      ee.write(of+'n')
    #print (of)
  #print ("*"*20+"SVN ONLY+"+"*"*20)
  #print ("#only files in ", bpath)
  for of in sorted(bonlyFiles):
    with open (os.getcwd()+'\svnonly.txt','a') as svn:
      svn.write(of+'n')
    #print (of)
if __name__ == '__main__':
  FolderEE = 'D:\search\bb\ObjectGroup - Copy\ObjectGroup\Built-in'
  FolderSVN = 'D:\search\bb\ObjectGroup\ObjectGroup\Built-in'
  dirCompare(FolderEE, FolderSVN)
  print("done!")

PS：这里再为大家推荐一款功能相似的在线工具供大家参考使用：

在线文本比较工具：
http://tools.jb51.net/aideddesign/txt_diff

更多关于Python相关内容感兴趣的读者可查看本站专题：《Python文件与目录 *** 作技巧汇总》、《Python文本文件 *** 作技巧汇总》、《Python数据结构与算法教程》、《Python函数使用技巧总结》、《Python字符串 *** 作技巧汇总》及《Python入门与进阶经典教程》

希望本文所述对大家Python程序设计有所帮助。