python_文本文件里面随机抽取若干行，写入新的文本文件里面

天一直阴沉沉的，真想早点下班啊。。。。。。

 

#encoding=utf-8
import random
from sets import Set
   
def copyfile(srcfile, dstfile, linenum):
     """
         get linenum different lines out from srcfile at random
         and write them into dstfile
     """
     result = []
     ret = False
     try:
         srcfd = open(srcfile,'r')
     except IOError:
         print 'srcfile doesnot exist!'
         return ret
     try:
         dstfd = open(dstfile,'w')
     except IOError:
         print 'dstfile doesnot exist!'
         return ret
     srclines = srcfd.readlines()
     srclen = len(srclines)
     while len(Set(result)) < int(linenum):
         s = random.randint(0,srclen-1)
         result.append(srclines[s])
     for content in Set(result):
         dstfd.write(content)
     srcfd.close()
     dstfd.close()
     ret = True
     return ret
  
if __name__ == "__main__":
     srcpath = raw_input('input srcfile path')
     dstpath = raw_input('input dstfile path')
     linenum = raw_input('input linenum')
     print copyfile(srcpath,dstpath,linenum)

 

 hadoop快点跑吧。。。。。。