Python实现SVN的目录周期性备份实例

# -*- coding: utf-8 -*-
#!/usr/bin/python
from hashlib import md5
import os 
def calMD5(str):
  m = md5()
  m.update(str)
  return m.hexdigest()
def calMD5ForFile(file):
  statinfo = os.stat(file)
  if int(statinfo.st_size) / (1024*1024) >= 1000:
#     print("File size > 1000, move to big file...")
    return calMD5ForBigFile(file)
  m = md5()
  f = open(file, 'rb')
  m.update(f.read())
  f.close()
  return m.hexdigest()
def calMD5ForFolder(dir, MD5File):
  outfile = open(MD5File,'w')
  for root, subdirs, files in os.walk(dir):
    for file in files:
      filefullpath = os.path.join(root, file)
      """print filefullpath"""
      filerelpath = os.path.relpath(filefullpath, dir)
      md5 = calMD5ForFile(filefullpath)
      outfile.write(filerelpath+' '+md5+"\n")
  outfile.close()
def calMD5ForBigFile(file):
  m = md5()
  f = open(file, 'rb')
  buffer = 8192  # why is 8192 | 8192 is fast than 2048
  while 1:
    chunk = f.read(buffer)
    if not chunk : break
    m.update(chunk)
  f.close()
  return m.hexdigest()
if __name__=="__main__":
  print(calMD5ForFile("e:/test/target/a/b/rabbit.txt"))

# -*- coding: utf-8 -*-
#!/usr/bin/python
import os,shutil
from md5_caculate import calMD5ForFile
# 拷贝源目录到目标目录
def copyDir(srcDir, dstDir):
  if srcDir in dstDir: # 源目录包含在目标目录，则直接返回
    return
  if not os.path.isdir(srcDir):
    print(srcDir, "路径指定的源目录不存在！")
    return;
  if not os.path.exists(dstDir): # 目标目录不存在时则创建
    os.makedirs(dstDir)
  for fileOrDirName in os.listdir(srcDir): # 源目录下的所有文件（包括文件和目录） TODO BUG：如果srcDir为一个空目录？
#     fileOrDirPath = srcDir + "/" + fileOrDirName
    fileOrDirPath = os.path.join(srcDir, fileOrDirName)
    if os.path.isfile(fileOrDirPath): # 如果当前是一个子文件，则直接复制文件
      copyFile(fileOrDirPath, dstDir)
    if os.path.isdir(fileOrDirPath): # 如果当前是一个子目录，则递归复制目录
      copyDir(fileOrDirPath, os.path.join(dstDir, fileOrDirName))
# 拷贝源文件到目标目录
def copyFile(srcFile, dstDir):
  if not os.path.isfile(srcFile):
    print(srcFile, "路径指定的源文件不存在！")
    return
  fileName = os.path.basename(srcFile)
  dstFile = os.path.join(dstDir, fileName)
  if os.path.isfile(dstFile): # 有同名的目标文件，则检查MD5值是否相同，如果不同才Copy
    if calMD5ForFile(srcFile) != calMD5ForFile(dstFile):
      try:
        shutil.copy(srcFile, dstDir)
      except PermissionError:
        print("PermissionError occurs: ", srcFile)
  else:
    shutil.copy(srcFile, dstDir)
if __name__=="__main__":
  copyDir("e:/test/src", "e:/test/target")

# -*- coding: utf-8 -*-
#!/usr/bin/python
import os, time
from file_util import copyDir
# -------------------------------------------------------------
# 定时备份源目录到目标目录
# version = 1.0
# author = Will
# -------------------------------------------------------------
# 定时备份源目录到目标目录，sleepIntervalSeconds为备份时间间隔秒数
def backupDir(srcDir, dstDir, sleepIntervalSeconds):
  if not os.path.isdir(srcDir):
    print("请指定要备份源目录并确保目录存在！")
    return;
  while True:
    print("备份：", srcDir, "，到：", dstDir)
    copyDir(srcDir, dstDir)
    print("开始休眠", sleepIntervalSeconds, "秒...")
    time.sleep(sleepIntervalSeconds)
if __name__=="__main__":
  backupDir("d:/Documents/workspace/workspace/griddle", "e:/backup/griddle", 20)