python 比較兩文件夾的內容,具有通用性。
#-*-coding:utf-8-*-
#===============================================================================
# 目錄對比工具(包含子目錄 )，並列出
# 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:與Size不同文件顯示未排序)
# 4、可通過修改，比較文件名且包含文件格式或者只比較文件名但不包括格式後綴。
#===============================================================================
import os, time,difflib,sys
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): #獲取文件大小的函數,未用上,僅供學習.故註釋掉
# size=0
# for root, dirs, files in os.walk(dir):
# #root:目錄:str 如: C:\CopySVN\SystemObject\TopoProcedure\Built-in\
# #dirs:目錄名稱:列表: 如 [‘Parsers‘]
# #files:名稱:列表: 如 [‘011D0961FB42416AA49D5E82945DE7E9.og‘,...]
# #file:目錄: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):
print apath,‘All files numbers:‘,len(files)
for f in files:
#比較文件名不含格式後綴
#afiles.append(root + f[0:-4])
#比較文件名含格式後綴
afiles.append(root + f)
for root, dirs , files in os.walk(bpath):
print bpath,‘All files numbers:‘,len(files)
for f in files:
#比較文件名不含格式後綴
#bfiles.append(root + f[0:-4])
#比較文件名含格式後綴
bfiles.append(root + f)
#sizeB = os.path.getsize(root + "/" + f) 此處定義的size無法在commonfiles進行比較. (A,B在各自的循環裏面)
# 去掉afiles中文件名的apath (拿A,B相同的路徑\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 # 處理共有文件
#print ("===============File with different size in ‘", apath, "‘ and ‘", bpath, "‘===============")
#將結果輸出到本地
#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: #共有文件的大小比較
#pass #print (f + "\t\t" + getPrettyTime(os.stat(apath + "/" + f)) + "\t\t" + getPrettyTime(os.stat(bpath + "/" + f)))
#以下代碼是處理大小一致，但是內容可能不一致的情況
#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))
# 處理僅出現在一個目錄中的文件
onlyFiles = setA ^ setB
aonlyFiles = []
bonlyFiles = []
for of in onlyFiles:
if of in afiles:
aonlyFiles.append(of)
elif of in bfiles:
bonlyFiles.append(of)
print apath,‘only files numbers:‘,len(aonlyFiles)
print bpath,‘only files numbers:‘,len(bonlyFiles)
#print ("###################### EE resource ONLY ###########################")
#print ("#only files in ", apath)
if os.path.exists(os.getcwd()+‘/Aonly.txt‘):
os.remove(os.getcwd()+‘/Aonly.txt‘)
if os.path.exists(os.getcwd()+‘/Bonly.txt‘):
os.remove(os.getcwd()+‘/Bonly.txt‘)
for of in sorted(aonlyFiles):
with open (os.getcwd()+‘/Aonly.txt‘,‘a‘) as a:
a.write(of+‘\n‘)
#print (of)
#print ("*"*20+"SVN ONLY+"+"*"*20)
#print ("#only files in ", bpath)
for of in sorted(bonlyFiles):
with open (os.getcwd()+‘/Bonly.txt‘,‘a‘) as b:
b.write(of+‘\n‘)
#print (of)
if __name__ == ‘__main__‘:
FolderEE = sys.argv[1]
FolderSVN = sys.argv[2]
dirCompare(FolderEE, FolderSVN)
print("done!")
ps：本文參考 http://www.cnblogs.com/luo-mao/p/5872532.html 貓兒爹，經修改符合自身使用，感謝作者。
python 比較兩文件夾的內容,具有通用性。