def close(self):
try:
self.connection.close()
return True
except:
Logger.error(traceback.format_exc())
return False
def get_new_mail_ids(self):
try:
# 连接到POP3服务器:
self.pop3_session = poplib.POP3(self.pop3_server)
# 身份认证:
self.pop3_session.user(self.r_account)
self.pop3_session.pass_(self.r_password)
# 获取邮件唯一编号
resp, mails, octets = self.pop3_session.uidl()
# 获取邮件唯一编号和在服务器中的索引
mail_dict = dict()
for mail_idx_id in mails:
twice = mail_idx_id.split()
mail_idx = int(twice[0])
mail_id = twice[1]
mail_dict[mail_idx] = mail_id
# 过虑已读邮件
for mail_idx, mail_id in mail_dict.items():
if self.recorder.exist(mail_id):
del mail_dict[mail_idx]
# 返回结果
return mail_dict
except:
Logger.error(traceback.format_exc())
return None
def __init__(self, file_name, stay_seconds):
try:
self.file_name = file_name
self.stay_seconds = stay_seconds
self.dict = dict()
self.load_from_disk()
self.update = False
except:
Logger.error(traceback.format_exc())
def put(self, a_key):
try:
if isinstance(a_key, str):
a_key = a_key.decode('utf-8')
self.dict[a_key] = time.time()
self.update = True
return True
except:
Logger.error(traceback.format_exc())
return False
def filter(self, mail_info):
try:
if not self.__is_leave(mail_info):
return False
if self.__is_discard(mail_info):
return False
return True
except:
Logger.error(traceback.format_exc())
return True
def load_from_disk(self):
try:
with open(self.file_name, 'ab+') as content_file:
content_file.seek(0, 0)
content = content_file.read()
if content:
self.dict = json.loads(content)
return True
except:
Logger.error(traceback.format_exc())
return False
def put(self, a_key):
try:
language = 'INSERT OR REPLACE INTO mails(key, time) VALUES("%s", "%s")' % (
a_key, datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S'))
cursor = self.connection.execute(language)
cursor.execute(language)
self.connection.commit()
return True
except:
Logger.error(traceback.format_exc())
return False
def flush_to_disk(self):
try:
if not self.update:
return True
self.update = False
content = json.dumps(self.dict, ensure_ascii=False)
with open(self.file_name, 'w') as content_file:
content_file.write(content)
return True
except:
Logger.error(traceback.format_exc())
return False
def load(self, xml_file):
try:
root = xml.etree.ElementTree.parse(xml_file).getroot()
for child in root:
if child.tag == 'leave':
self.__parse(child, leave=True)
if child.tag == 'discard':
self.__parse(child, leave=False)
return True
except:
Logger.error(traceback.format_exc())
return False
def clear_dead(self):
try:
now = time.time()
for key, value in self.dict.items():
stay = now - value
if stay > self.stay_seconds:
del self.dict[key]
self.update = True
Logger.info('del self.dict[%s]' % key)
return True
except:
Logger.error(traceback.format_exc())
return False
def guess_charset(cls, root):
try:
# 先从root对象获取编码:
charset = root.get_charset()
if charset is None:
# 如果获取不到，再从Content-Type字段获取:
content_type = root.get('Content-Type', '').lower()
pos = content_type.find('charset=')
if pos >= 0:
charset = content_type[pos + 8:].strip()
return charset
except:
Logger.error(traceback.format_exc())
return None
def handle(self):
try:
# 获取邮件唯一编号
mail_dict = self.get_new_mail_ids()
# 依次处理所有邮件
for mail_idx, mail_id in mail_dict.items():
self.handle_one(mail_idx, mail_id)
# recorder内存刷到磁盘
Logger.info("close sqlite")
self.recorder.close()
except:
Logger.error(traceback.format_exc())
def __parse(self, e, leave):
for item in e:
rule = Rule(leave)
for child in item:
if child.tag == 'from':
rule.from_address = child.text
if child.tag == 'to':
rule.to_address = child.text
if child.tag == 'cc':
rule.cc_address = child.text
if child.tag == 'subject':
rule.subject = child.text
if child.tag == 'content':
rule.content = child.text
if leave:
self.__leave.append(rule)
else:
self.__discard.append(rule)
Logger.info(str(rule))
def handle_one(self, mail_idx, mail_id):
try:
Logger.info("get mail[%s]" % mail_idx)
resp, lines, octets = self.pop3_session.retr(mail_idx)
msg_content = '\r\n'.join(lines)
root = Parser().parsestr(msg_content)
# 记录邮件
self.recorder.put(mail_id)
# 解析邮件
mail_info = self.parse_mail(root)
if not mail_info:
return
# 过滤邮件
msg = "idx[{}], id[{}], from[{}], to[{}], cc[{}], subject[{}]". \
format(mail_idx,
mail_id,
mail_info.from_address,
mail_info.to_address,
mail_info.cc_address,
mail_info.subject)
if not self.filters.filter(mail_info):
Logger.report("filter: " + msg)
return
# 发送邮件
Logger.report("send: " + msg)
if self.send_flag:
self.send_mime(root)
else:
Logger.report(mail_info.content)
except:
Logger.error(traceback.format_exc())
def loop_once(r_account, r_password, pop3_server,
s_account, s_password, smpt_server,
recorder_file, filter_file,
send_flag, target_account,
idx):
"""
:type r_account: str
:type r_password: str
:type pop3_server: str
:type s_account: str
:type s_password: str
:type recorder_file: str
:type filter_file: str
:type smpt_server: str
:type send_flag: bool
:type target_account: str
:type idx: int
:return:
"""
try:
# 打印标志
Logger.info("-----------------------%s--------------------------" % idx)
# 初始化已读邮件记录
recorder = DBDict(recorder_file)
# 初始化过滤策略
filters = Filter()
if not filters.load(filter_file):
Logger.error("init mail filter error")
return False
# 初始化邮件 proxy
proxy = MailFilterProxy(r_account, r_password, pop3_server,
s_account, s_password, smpt_server,
recorder, filters,
send_flag, target_account)
proxy.handle()
except:
Logger.error(traceback.format_exc())
def send_mime(self, root):
"""
:type root: MIMEBase
"""
try:
# 显示原始信息
for key in root.keys():
raw_header = 'old_key[%s] ====> [%s]' % (key, root.get(key))
Logger.info(raw_header)
# 仅保留以下 Header
left_header = ('from',
'boundary',
'content-type',
'mime-version',
'subject',
'date',
'message-id',
'content-transfer-encoding')
for key in root.keys():
little = key.lower()
if little not in left_header:
del root[key]
Logger.info("delete key[%s]" % key)
root['to'] = self.target_account
# 打印新key
# for key in root.keys():
# raw_header = 'new_key[%s] ====> [%s]' % (key, root.get(key))
# Logger.info(raw_header)
# 发送邮件
server = smtplib.SMTP(self.smpt_server, 25)
server.login(self.s_account, self.s_password)
server.sendmail(self.s_account, [self.target_account], root.as_string())
server.quit()
except:
Logger.error(traceback.format_exc())
def __init__(self, file_name):
try:
self.file_name = file_name
self.connection = sqlite3.connect(file_name)
except:
Logger.error(traceback.format_exc())
def parse_mail(self, root):
try:
mail_info = MailInfo()
# from
value = root.get('From', '')
values = self.parse_address(value)
if values:
for (header, address) in values:
name = self.decode_str(header)
mail_info.from_name.add(name)
mail_info.from_address.add(address)
# to
value = root.get('To', '')
values = self.parse_address(value)
if values:
for (header, address) in values:
name = self.decode_str(header)
mail_info.to_name.add(name)
mail_info.to_address.add(address)
# cc
value = root.get('Cc', '')
values = self.parse_address(value)
if values:
for (header, address) in values:
name = self.decode_str(header)
mail_info.cc_name.add(name)
mail_info.cc_address.add(address)
# subject
value = root.get('Subject', '')
if value:
mail_info.subject = self.decode_str(value)
# child MIMEMultipart
if root.is_multipart():
parts = root.get_payload()
for n, part in enumerate(parts):
sub_mail_info = self.parse_mail(part)
mail_info.child_mail.append(sub_mail_info)
else:
content_type = root.get_content_type()
if content_type == 'text/plain' or content_type == 'text/html':
# 纯文本或HTML内容:
content = root.get_payload(decode=True)
# 要检测文本编码:
charset = self.guess_charset(root)
if charset:
content = content.decode(charset)
mail_info.content = content
# 调试用
# Logger.report(mail_info.content)
else:
# 不是文本,作为附件处理
pass
return mail_info
except:
Logger.error(traceback.format_exc())
return None
def __main__():
# utf-8
reload(sys)
sys.setdefaultencoding('utf-8')
# 日志目录
log_target = "logs"
# 邮件过滤策
filter_file = "config/filter.xml"
# 收件账号
r_account = raw_input("receive mail account: ")
# 收件密码
r_password = getpass.getpass('receive mail password: '******'send mail password: '******'receive mail password: '******'send mail password: '******'Y' and input_flag != 'N':
print >> sys.stderr, "Invalid send_flag values, it must be Y or N"
return False
send_flag = True if input_flag == 'Y' else False
print >> sys.stdout, "program is running in background, please check the running log!"
# 初始化日志
Logger.init(LogEnv.develop, log_target, "result", max_file_count=10)
Logger.info("program is starting......")
# 非调试状态时在后台启动
try:
pid = os.fork()
if pid > 0:
Logger.info("#1 parent exit")
os._exit(0)
except:
Logger.error(traceback.format_exc())
try:
pid = os.fork()
if pid > 0:
Logger.info("#2 parent exit")
Logger.info("pid[%s] is running..." % pid)
os._exit(0)
except:
Logger.error(traceback.format_exc())
# 邮件记录文件
self_dir = os.path.dirname(os.path.abspath(__file__))
recorder_dir = os.path.join(self_dir, 'data')
if not os.path.exists(recorder_dir):
os.mkdir(recorder_dir)
recorder_file = os.path.join(self_dir, 'data', 'transfer.db')
# 循环接收邮件
idx = 0
while True:
try:
loop_once(r_account, r_password, pop3_server,
s_account, s_password, smpt_server,
recorder_file, filter_file,
send_flag, target_account,
idx)
idx += 1
# 非调试状态时循环转发邮件
time.sleep(60)
# 调试用
# break
except:
Logger.error(traceback.format_exc())
def __del__(self):
try:
self.clear_dead()
self.flush_to_disk()
except:
Logger.error(traceback.format_exc())