Parse a raw email

email_parse.py

"""
$ cat source.email | python email_parse.py
"""

import fileinput
import email
from email.Utils import parseaddr

def parse():
    raw_message = ''.join([line for line in fileinput.input()])
    message = email.message_from_string(raw_message)
    text_plain = None
    text_html = None

    for part in message.walk():
        if part.get_content_type() == 'text/plain' and text_plain is None:
            text_plain = part.get_payload()
        if part.get_content_type() == 'text/html' and text_html is None:
            text_html = part.get_payload()

    return {
      'to': parseaddr(message.get('To'))[1],
      'from': parseaddr(message.get('From'))[1],
      'delivered to': parseaddr(message.get('Delivered-To'))[1],
      'subject': message.get('Subject'),
      'text_plain': text_plain,
      'text_html': text_html,
    }

if __name__ == '__main__':
    print parse()

source.email

                                                                                                                                                                                                                                                               
MIME-Version: 1.0
Received: by 10.152.6.130 with HTTP; Wed, 4 Jul 2012 17:10:25 -0700 (PDT)
Date: Wed, 4 Jul 2012 20:10:25 -0400
Delivered-To: user1@gmail.com
Message-ID: <CAB2OXrM6oChuZdcWrRtf8btjadXtZfDxd3R+Pto6jLY6C3We3w@mail.gmail.com>
Subject: prueba
From: Favio <user1@gmail.com>
To: Favio <user1@gmail.com>
Content-Type: multipart/mixed; boundary=f46d043745a175980904c409f962

--f46d043745a175980904c409f962
Content-Type: multipart/alternative; boundary=f46d043745a175980204c409f960

--f46d043745a175980204c409f960
Content-Type: text/plain; charset=ISO-8859-1

probando

--f46d043745a175980204c409f960
Content-Type: text/html; charset=ISO-8859-1

probando

--f46d043745a175980204c409f960--
--f46d043745a175980904c409f962
Content-Type: text/plain; charset=US-ASCII; name="adj.txt"
Content-Disposition: attachment; filename="adj.txt"
Content-Transfer-Encoding: base64
X-Attachment-Id: f_h492x3111

YWRqCg==
--f46d043745a175980904c409f962
Content-Type: text/html; charset=US-ASCII; name="adj.html"
Content-Disposition: attachment; filename="adj.html"
Content-Transfer-Encoding: base64
X-Attachment-Id: f_h492wyx10

PGh0bWw+PGhlYWQ+PC9oZWFkPjxib2R5PjxwPmFkajwvcD48L2JvZHk+PC9odG1sPgo=
--f46d043745a175980904c409f962--