def onMime(self, who, data):
''' Handles incoming MIME messages, such as those sent by
NotesBuddy containing embedded image data '''
from email.Parser import Parser
from StringIO import StringIO
print '<mime>%s' % who[0]
msg = Parser().parsestr(data)
html = StringIO() # combined text segments
images = {} # map of Content-ID:binary image
for part in msg.walk():
mt = part.get_content_maintype()
if mt == 'text':
html.write(part.get_payload(decode=True))
elif mt == 'image':
cid = part.get('Content-ID')
images[cid] = part.get_payload(decode=True)
print ' <text/html>:', html.getvalue()
html.close()
print ' <images>:', [k[1:-1] for k in images.keys()]
f = open("tmp.mime", 'w')
except Exception, e:
print >> sys.stderr, e
else:
f.write(data)
f.close()
# second, parse and display
msg = Parser().parsestr(data)
html = StringIO() # combined text segments
images = {} # map of Content-ID:binary image
for part in msg.walk():
mt = part.get_content_maintype()
if mt == 'text':
html.write(part.get_payload(decode=True))
elif mt == 'image':
cid = part.get('Content-ID')
images[cid] = part.get_payload(decode=True)
print ' <text/html>:', html.getvalue()
html.close()
print ' <images>:', [k[1:-1] for k in images.keys()]
def onSubject(self, who, subj):
print '<subject>%s: "%s"' % (who[0], subj)
def fetch(folder, username, password, server='imap.gmail.com', archive=True,
archive_folder=None, accept_only=None):
"""
Connects over IMAP to a mailbox and extracts attachments from all emails
folder
The IMAP folder name to parse for emails
server
The IMAP server URL
username
The username to connect to IMAP with
password
The username to connect to IMAP with
archive
Optionally move completed emails to a different folder (default: True)
archive_folder:
The folder to move completed mails to
(default: folder + "_archive")
accept_only (list):
One or more file type suffixes to extract from emails, from this list:
csv
excel
zip
gz
(Default: any from that list)
"""
mime_types = []
if accept_only is None:
accept_only = ('csv', 'excel', 'zip', 'gz')
if 'csv' in accept_only:
mime_types.extend(['text/csv', 'application/csv'])
if 'zip' in accept_only:
mime_types.extend([
'application/zip', 'multipart/x-zip',
'application/x-zip-compressed', 'application/octet-stream',
])
if 'gz' in accept_only:
mime_types.extend(['application/x-gzip', 'multipart/x-gzip'])
if 'excel' in accept_only:
mime_types.extend([
'vnd.openxmlformats-officedocument.spreadsheetml.sheet',
'application/vnd.ms-excel', 'application/octet-stream',
])
# default the IMAP archive folder for completed emails
if archive and archive_folder is None:
archive_folder = "{0}_archive".format(folder)
try:
# login to IMAP
M = imaplib.IMAP4_SSL(server)
M.login(username, password)
except socket.gaierror:
raise Exception("Bad credentials for IMAP server {0}".format(server))
files = []
to_archive = []
try:
# look at all mail in this IMAP folder
result, data = M.select(folder)
result, data = M.search(None, "ALL")
id_list = data[0].split()
for id in id_list:
result, data = M.fetch(id, "(RFC822)")
content = data[0][1]
if sys.version < '3':
msg = Parser().parsestr(content)
else:
msg = Parser().parsestr(content.decode('utf-8'))
# only look at if has attachment.
if msg.is_multipart():
for part in msg.walk():
if part.get_content_type() in mime_types:
# download attachment into temp file
tmpfile = tempfile.mkstemp(part.get_filename())
tmp_fhandle, tmp_fname = tmpfile
if sys.version < '3':
with os.fdopen(tmp_fhandle, "w") as f:
f.write(part.get_payload(decode=True))
else:
with os.fdopen(tmp_fhandle, "wb") as f:
f.write(part.get_payload(decode=True))
# decompress zip/gz into contained files
if _is_compressed(tmp_fname):
files.extend(_extract_csvs_from_zip(tmp_fname))
else:
files.append(tmp_fname)
# mark this email for archival
if archive is True:
to_archive.append(str(id))
# move mails from mailbox to archive_mbox
if len(to_archive) > 0:
message_set = ",".join(to_archive)
M.copy(message_set, archive_folder)
M.store(message_set, "+FLAGS", r"(\Deleted)")
M.expunge()
finally:
# close the mailbox and disconnect the IMAP connection
M.close()
M.logout()
return files
