Browse Source

Attachment handling

master
Sven Slootweg 10 years ago
parent
commit
be1eb56f91
  1. 38
      parse

38
parse

@ -46,10 +46,21 @@ database = sqlite3.connect('emails.db')
cursor = database.cursor()
try:
# Try to create the table
# Create directories first
os.makedirs("attachments")
except OSError:
pass
try:
# Try to create emails table
cursor.execute("CREATE TABLE emails (`message_id`, `from`, `to`, `subject`, `date`, `body`, `html`, `hash`)")
except sqlite3.OperationalError:
# Table already exists
pass
try:
# Try to create attachments table
cursor.execute("CREATE TABLE attachments (`message_hash`, `filename`, `type`, `hash`, `size`)")
except sqlite3.OperationalError:
pass
file_list = glob.glob(options['pattern'])
@ -73,6 +84,8 @@ for email_file in file_list:
message_parts = [find_submessages(message)]
message_parts = flatten(message_parts)
attachment_list = []
for part in message_parts:
if part.get_filename() is None:
# Part of the message
@ -86,13 +99,32 @@ for email_file in file_list:
print "WARNING: Unknown message format encountered in %s, message may be incomplete." % sha1_hash
else:
# Attachment
print "Attachment found of type %s: %s" % (part.get_content_type(), part.get_filename())
attachment_data = part.get_payload(decode=True)
attachment_sha1 = hashlib.sha1(attachment_data).hexdigest()
attachment_filename = part.get_filename()
attachment_type = part.get_content_type()
attachment_extension = os.path.splitext(attachment_filename)[1][1:]
attachment_size = len(attachment_data)
attachment_destination = "attachments/%s.%s" % (attachment_sha1, attachment_extension)
#print "Attachment found of type %s: %s (%s)" % (attachment_type, attachment_filename, attachment_destination)
attachment_file = open(attachment_destination, "w")
attachment_file.write(attachment_data)
attachment_file.close()
attachment_list.append((attachment_filename, attachment_type, attachment_sha1, attachment_size))
timestamp = 0
new_row = (getheader(message['message-id']), getheader(message['from']), getheader(message['to']), getheader(subject), timestamp, textbody, htmlbody, sha1_hash)
cursor.execute("INSERT INTO emails VALUES (?, ?, ?, ?, ?, ?, ?, ?)", new_row)
print "Successfully parsed and inserted e-mail with SHA1 hash %s." % sha1_hash
if len(attachment_list) > 0:
for attachment in attachment_list:
new_row = (sha1_hash, attachment[0], attachment[1], attachment[2], attachment[3])
cursor.execute("INSERT INTO attachments VALUES (?, ?, ?, ?, ?)", new_row)
print "Successfully inserted %d attachment(s) for %s." % (len(attachment_list), sha1_hash)
database.commit()
print "Changes successfully committed to database, exiting..."
Loading…
Cancel
Save