mbox_misc.py

import mailbox
import email.utils
from mboxparse import convert_to_json

filename = "metadatatest2525.mbox"


#the following returns a list containing message objects.  Each message objects is composed of the components listed on line 20.
mbox = mailbox.mbox(filename)

#Get the first message in the mail box to test the below functions out with
message = mbox[0]

#Each message has the following attributes
#print message.keys()
#results in the following
#['X-GM-THRID', 'X-Gmail-Labels', 'MIME-Version', 'Received', 'Bcc', 'Date', 'Delivered-To', 'Message-ID', 'Subject', 'From', 'To', 'Cc', 'Content-Type']



#example looping with python
#loop through the messages and print the subject of each message

'''for message in mbox:
    print message['subject']
'''





#this function parses the body of the message (not sure if you need this) and prints/returns it
def print_message_body(msg):
	payload = msg.get_payload()

	if msg.is_multipart():
		div = ''
		for subMsg in payload:
			print div
			print_message_body(subMsg)
			div = '------------------------------'
	else:
		print msg.get_content_type()
		#only prints the first 200 items
		print payload[:200]
		return payload



#This function prints the meta data of each message
def print_message_metadata(message):
	for key,value in message.items():
		print '\n'
		print '{}: '.format(key)
		print '\t{}: '.format(value)

#Print the metadata of the message
#print_message_metadata(message)

#Print the message body
#print_message_body(message)


#CONVERT MBOX FILE TO JSON FILE
#-------------
#uses the function from the other script
convert_to_json(filename, 'output_test_1.json')

#READ in JSON file
import json
with open('output_test_1.json', 'rU') as f:
	data = json.load(f)
#print in first message
print data[0]
#returns the following:
{u'X-Gmail-Labels': u'Sent,Important', u'Delivered-To': u'[email protected]', u'From': u'John McCarthy <[email protected]>', u'MIME-Version': u'1.0', u'Received': u'by 10.114.78.7 with HTTP; Thu, 10 Dec 2015 08:00:08 -0800 (PST)', u'Cc': [u'JohnMcCarthy<[email protected]>'], u'X-GM-THRID': u'1520186905955984162', u'Bcc': [u'MikeMaelzer<[email protected]>'], u'To': [u'HeatherMcCarthy<[email protected]>', u'JohnMcCarthy<[email protected]>', u'SandraMcCarthy<[email protected]>', u'NeilJaffe<[email protected]>', u'MaryJaffe<[email protected]>'], u'parts': [{u'content': u'can someone "reply all" to this please (except the mysterious BCC person,\r\nwho can\'t)\r\n', u'contentType': u'text/plain'}, {u'content': u'can someone &quot;reply all&quot; to this please (except the mysterious BCC person, who can&#39;t)\xa0\n', u'contentType': u'text/html'}], u'Date': u'Thu, 10 Dec 2015 11:00:08 -0500', u'Message-ID': u'<CAFVywbLvJ-tJ34vB7u+XUPYpOqnj6TiC6O+sJohY35GeGVBNeQ@mail.gmail.com>', u'Content-Type': u'multipart/alternative; boundary=001a11c37a8a0ab77e05268d4da6', u'Subject': u'1-to-5 outbound (1 CC) (1 BCC)'}