使用python实现正则匹配检索远端FTP目录下的文件

# coding=utf-8
#########################################################################
# File Name: reg_url.py
# Author: WangWeilong
# Company: Baidu
#########################################################################

import re
import sys
import os
from ftplib import FTP

dhccmd = "http://xxx/api/submit"

class DHC_FTP():
 def __init__(self, hostname, username="", passwd=""):
 self.hostname = hostname
 try:
 self.ftp = FTP(self.hostname)
 except:
 print "hostname error!"
 exit(-1)
 self.username = username
 self.passwd = passwd
 self.filelist = []
 self.reg_pattern = ""

 def getftpfilelist(self, path):
 self.ftp.login(self.username, self.passwd)
 self.ftp.cwd(path)
 self.filelist = self.ftp.nlst()

 def matchnames(self, regpattern):
 pattern = re.compile(regpattern)
 matchedfiles = []
 for file in self.filelist:
 match = pattern.search(file)
 if match:
 matchedfiles.append(match.string)
 return matchedfiles



if __name__ == "__main__":
 if len(sys.argv) != 4:
 print 'usage:python reg_url.py $ftpurldir $regular'
 else:
 #解析ftpurl字符串
 head = sys.argv[1].split("@")[1].split("/")[0]
 username = sys.argv[1].split("//")[1].split(":")[0]
 passwd = sys.argv[1].split("//")[1].split(":")[1].split("@")[0]
 pathdir = sys.argv[1].split(head)[1]

 dhc_ftp = DHC_FTP(head, username, passwd)
 dhc_ftp.getftpfilelist("./" + pathdir)

 #获取正则表达式
 regpattern = r'%s' % sys.argv[2]

 matchedfiles = dhc_ftp.matchnames(regpattern)
 for files in matchedfiles:
 ftpname = sys.argv[1] + "/" + files
 print ftpname