python下载邮箱附件

  1 #!/usr/bin/env python3
  2 # -*- coding: utf-8 -*-
  3 import poplib
  4 import email
  5 from datetime import datetime, timezone, timedelta
  6 import time
  7 import os
  8 from email.parser import Parser
  9 from email.header import decode_header
 10 from email.utils import parseaddr
 11 
 12 # 输入邮件地址, 口令和POP3服务器地址:
 13 
 14 email = ‘‘
 15 password = ‘‘  # 此处密码是授权码,用于登录第三方邮件客户端（163邮箱需要授权码）
 16 pop3_server = ‘pop.126.com‘
 17 
 18 
 19 # 获取邮件自带的编码
 20 # mail_encode = decode_header(mail.get("Subject"))[0][1]
 21 # mail_title = decode_data(decode_header(mail.get("Subject"))[0][0], mail_encode)
 22 
 23 def decode_data(bytes, added_encode=None):
 24     """
 25     字节解码
 26     :param bytes:
 27     :return:
 28     """
 29 
 30     def _decode(bytes, encoding):
 31         try:
 32             return str(bytes, encoding=encoding)
 33         except Exception as e:
 34             return None
 35 
 36     encodes = [‘UTF-8‘, ‘GBK‘, ‘GB2312‘]
 37     if added_encode:
 38         encodes = [added_encode] + encodes
 39     for encoding in encodes:
 40         str_data = _decode(bytes, encoding)
 41         if str_data is not None:
 42             return str_data
 43     return None
 44 
 45 
 46 def decode_str(s):  # 字符编码转换
 47     value, charset = decode_header(s)[0]
 48     if charset:
 49         value = value.decode(charset)
 50     return value
 51 
 52 
 53 def get_att(msg):
 54     import email
 55     attachment_files = []
 56 
 57     for part in msg.walk():
 58         file_name = part.get_filename()  # 获取附件名称类型
 59         contType = part.get_content_type()
 60 
 61         if file_name:
 62             h = email.header.Header(file_name)
 63             dh = email.header.decode_header(h)  # 对附件名称进行解码
 64             filename = dh[0][0]
 65             if dh[0][1]:
 66                 filename = decode_str(str(filename, dh[0][1]))  # 将附件名称可读化
 67                 print(filename)
 68                 # filename = filename.encode("utf-8")
 69             data = part.get_payload(decode=True)  # 下载附件
 70             # 在指定目录下创建文件，注意二进制文件需要用wb模式打开
 71             att_file = open(‘G:\\3、hobby\\2、python\\电子发票整理\\下载文件\\‘ + filename, ‘wb‘)
 72             attachment_files.append(filename)
 73             att_file.write(data)  # 保存附件
 74             att_file.close()
 75     return attachment_files
 76 
 77 
 78 def parse_mail_time(mail_datetime):
 79     """
 80     邮件时间解析
 81     :param bytes:
 82     :return:
 83     """
 84     print(mail_datetime)
 85     print(len(mail_datetime))
 86     GMT_FORMAT = "%a, %d %b %Y %H:%M:%S"
 87     GMT_FORMAT2 = "%d %b %Y %H:%M:%S"
 88     index = mail_datetime.find(‘ +0‘)
 89     if index > 0:
 90         mail_datetime = mail_datetime[:index]  # 去掉+0800
 91     print(mail_datetime)
 92     formats = [GMT_FORMAT, GMT_FORMAT2]
 93     for ft in formats:
 94         try:
 95             mail_datetime = datetime.strptime(mail_datetime, ft)
 96             return mail_datetime
 97         except:
 98             pass
 99 
100     raise Exception("邮件时间格式解析错误")
101 
102 
103 # 连接到POP3服务器,有些邮箱服务器需要ssl加密，对于不需要加密的服务器可以使用poplib.POP3()
104 server = poplib.POP3(pop3_server)
105 server.set_debuglevel(1)
106 # 打印POP3服务器的欢迎文字:
107 print(server.getwelcome().decode(‘utf-8‘))
108 # 身份认证:
109 server.user(email)
110 server.pass_(password)
111 # 返回邮件数量和占用空间:
112 print(‘Messages: %s. Size: %s‘ % server.stat())
113 # list()返回所有邮件的编号:
114 resp, mails, octets = server.list()
115 # 可以查看返回的列表类似[b‘1 82923‘, b‘2 2184‘, ...]
116 print(mails)
117 index = len(mails)
118 
119 for i in range(index, 0, -1):
120     try:
121         # 倒序遍历邮件
122         resp, lines, octets = server.retr(i)
123         # lines存储了邮件的原始文本的每一行,
124         # 邮件的原始文本:
125         msg_content = b‘\r\n‘.join(lines).decode("utf8", "ignore")
126         # 解析邮件:
127         msg = Parser().parsestr(msg_content)
128         # 获取邮件时间
129 
130         date1 = parse_mail_time(msg.get("Date")[0:30])  # 格式化收件时间
131         print(date1)
132         date2 = date1.strftime("%Y%m%d")  # 邮件时间格式转换
133 
134         if (date2 < ‘20201231‘) | (date2 > ‘20200101‘):
135             f_list = get_att(msg)  # 获取附件
136             print("下载成功")
137     except:
138         pass
139     continue
140 
141 server.quit()