file: email\_header_value_parser.py

line : encoded_part = part.fold(policy=policy)[:-1] # strip nl
modify to: encoded_part = part.fold(policy=policy)[:-2] # strip nl

because the nl is "\r\n"

python 3.7 same

Hi and thank you for opening this issue. 
What is the behaviour?
What is the test you used to reproduce this bug?

import smtplib
from email.message import EmailMessage
from email.utils import formataddr

server = smtplib.SMTP('smtp.xxx.com',port=25)
server.login('you@xxx.com', 'password')
msg = EmailMessage()

#if address username include Chinese or other multibytes language, it will be encode to utf8 and base64, like '?utf-8?B?6YKu5Lu25rWL6K+V?= <you@xxx.com>', but the raw smtp request data error, "From: xxx" and "To: xxx" fields has the error end char:"\r\r\r\r\n".
#From: ?utf-8?B?6YKu5Lu25rWL6K+V?= <you@xxx.com>\r\r\r\r\n

msg['From'] =formataddr(('中文','you@xxx.com'))
msg['To'] = formataddr(('中文姓名','to@yyy.com'))

msg['Subject'] =subject
msg.set_content('something')

server.send_message(msg)
server.quit()

This seems to be a duplicate of #34424. You'd have to upgrade to the latest 3.7 to get the fix.