def __get_mails(self, mails_info: dict, folder: Folder):
mails = mails_info.get('mailMap')
if mails is None or len(mails) == 0:
return
for k, mail in mails.items():
try:
mailid = mail.get('messageId')
msid = mail.get('msId')
newmail = mail.get('newMail')
eml = EML(self._clientid, self.task, self._userid, mailid, folder, self.task.apptype)
if newmail:
eml.state = 0
else:
eml.state = 1
eml.provider = mail.get('sender')
sendtime = mail.get('sendDate')
if sendtime is not None and sendtime != '':
eml.sendtime = datetime.datetime.fromtimestamp(sendtime // 1000)
eml.subject = mail.get('subject')
eml_info = self.__get_mail_streams(mailid, msid)
eml.io_stream = eml_info[0]
eml.stream_length = eml_info[1]
time.sleep(5) # 防止取邮件的频率过快
yield eml
except Exception:
self._logger.error(f'Get {mailid} email error, err:{traceback.format_exc()}')
continue
def _get_mails(self, folder: Folder) -> iter:
error_times = 0
current_page = 1
all_pagenum = 1000
mail_url = 'http://m0.mail.sina.com.cn/wa.php?a=list_mail'
headers = {
'Accept':
"*/*",
'Accept-Encoding':
"gzip, deflate",
'Accept-Language':
"zh-CN,zh;q=0.9,en;q=0.8",
'Cache-Control':
"no-cache",
'Content-Length':
"98",
'Content-type':
"application/x-www-form-urlencoded;charset=UTF-8",
'Cookie':
self.task.cookie,
'Host':
"m0.mail.sina.com.cn",
'Origin':
"http://m0.mail.sina.com.cn",
'Pragma':
"no-cache",
'Proxy-Connection':
"keep-alive",
'Referer':
"http://m0.mail.sina.com.cn/classic/index.php",
'User-Agent':
"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/71.0.3578.98 Safari/537.36"
}
while True:
try:
if current_page >= all_pagenum:
break
payload_data = {
'fid': '{}'.format(folder.folderid),
'order': 'htime',
'sorttype': 'desc',
'type': '0',
'pageno': '{}'.format(current_page),
'tag': '-1',
'webmail': '1'
}
response = requests.request("POST",
mail_url,
data=payload_data,
headers=headers)
mail_data_json = json.loads(response.text)
if not mail_data_json.get('result'):
break
if mail_data_json.get('errno') != 0:
break
mail_data = mail_data_json.get('data')
all_pagenum = mail_data.get('pagenum')
maillist = mail_data.get('maillist')
if len(maillist) == 0:
self._logger.info("No mail in mail list")
break
for mail_info in maillist:
mailid = mail_info[0]
eml = EML(self._clientid, self.task, self._userid, mailid,
folder, self.task.apptype)
eml.sendtime = datetime.datetime.fromtimestamp(
mail_info[4])
eml.subject = mail_info[3]
eml.provider = mail_info[1]
eml.owner = mail_info[2]
eml_info = self.__download_mails(folder.folderid, mailid)
eml.io_stream = eml_info[0]
eml.stream_length = eml_info[1]
time.sleep(5) # 防止取邮件的频率过快
yield eml
if current_page == all_pagenum:
break
else:
current_page += 1
except Exception:
if error_times >= 5:
break
self._logger.error(
f"Get mail or download mail error, err:{traceback.format_exc()}"
)
error_times += 1
continue
def parse_mail(self, message, fo: Folder):
try:
sj = json.loads(message)
respMsgs = sj["Body"]["ResponseMessages"]["Items"]
if not respMsgs:
yield
except:
return
for msg in respMsgs:
items = msg["RootFolder"]["Items"]
for item in items:
size = 0
rcvTime = None
if not item.__contains__('ItemId') or not item.__contains__(
'Subject') or not item.__contains__('IsRead'):
continue
if not item['ItemId'].__contains__('Id'):
continue
id = item["ItemId"]["Id"]
subject = item["Subject"]
isRead = item["IsRead"]
if item.__contains__('DateTimeReceived'):
# 2019-03-09T00:52:47+08:00
rcvTime = item["DateTimeReceived"]
rcvTime = datetime.datetime.strptime(
rcvTime,
'%Y-%m-%dT%H:%M:%S+08:00') + datetime.timedelta(
hours=8)
if item.__contains__('Size'):
size = item["Size"]
# sender
# if item.__contains__('IsDraft') and not item["IsDraft"]:
# if item.__contains__('Sender') and item['Sender'].__contains__('Mailbox'):
# if item["Sender"]["Mailbox"].__contains__('EmailAddress'):
# sender = item["Sender"]["Mailbox"]["EmailAddress"]
# elif item["Sender"]["Mailbox"].__contains__("Name"):
# sender = item["Sender"]["Mailbox"]["Name"]
# else:
# sender = "Unknow"
# elif item.__contains__("From") and item["From"].__contains__("Mailbox"):
# if item["From"]["Mailbox"].__contains__("EmailAddress"):
# sender = item["From"]["Mailbox"]["EmailAddress"]
# elif item["From"]["Mailbox"].__contains__("Name"):
# sender = item["From"]["Mailbox"]["Name"]
# else:
# sender = "Unknow"
# else:
# sender = "Unknow"
# else:
# sender = self._userid
m = EML(self._clientid, self.task, self._userid, id, fo,
self.task.apptype)
m.subject = subject
m.sendtime = rcvTime
m.provider = 'outlook.live.com'
if isRead:
m.state = 1
else:
m.state = 0
m.stream_length = size
yield m