def main():
args = get_args()
parser = MailParser()
if args.file:
parser.parse_from_file(args.file)
elif args.string:
parser.parse_from_string(args.string)
if args.json:
j = json.loads(parser.parsed_mail_json)
safe_print(json.dumps(j, ensure_ascii=False, indent=4))
if args.body:
# safe_print(parser.body)
safe_print(parser.body)
if args.headers:
safe_print(parser.headers)
if args.to:
safe_print(parser.to_)
if args.from_:
safe_print(parser.from_)
if args.subject:
safe_print(parser.subject)
if args.defects:
for i in parser.defects_category:
safe_print(i)
if args.anomalies:
for i in parser.anomalies:
safe_print(i)
if args.senderip:
r = parser.get_server_ipaddress(args.senderip)
if r:
safe_print(r)
else:
safe_print("Not Found")
if args.attachments:
for i in parser.attachments_list:
safe_print(json.dumps(i, ensure_ascii=False, indent=4))
def main():
args = get_args()
parser = MailParser()
if args.file:
parser.parse_from_file(args.file)
elif args.string:
parser.parse_from_string(args.string)
if args.json:
j = json.loads(parser.parsed_mail_json)
print(json.dumps(j, ensure_ascii=False, indent=4).encode('utf-8'))
if args.body:
print(parser.body.encode('utf-8'))
if args.headers:
print(parser.headers.encode('utf-8'))
if args.to:
print(parser.to_.encode('utf-8'))
if args.from_:
print(parser.from_.encode('utf-8'))
if args.subject:
print(parser.subject.encode('utf-8'))
if args.defects:
for i in parser.defects_category:
print(i.encode('utf-8'))
if args.anomalies:
for i in parser.anomalies:
print(i.encode('utf-8'))
if args.senderip:
r = parser.get_server_ipaddress(args.senderip)
if r:
print(r.encode('utf-8'))
else:
print("Not Found")
if args.attachments:
for i in parser.attachments_list:
print(json.dumps(i, ensure_ascii=False, indent=4).encode('utf-8'))
class TestMailParser(unittest.TestCase):
def setUp(self):
# Init
self.parser = MailParser()
def test_ipaddress(self):
self.parser.parse_from_file(mail_test_2)
trust = "smtp.customers.net"
ip = "217.76.210.112"
result = self.parser.get_server_ipaddress(trust)
self.assertEqual(result, ip)
trust = ""
result = self.parser.get_server_ipaddress(trust)
self.assertEqual(result, None)
trust = " "
result = self.parser.get_server_ipaddress(trust)
self.assertEqual(result, None)
def test_fingerprints_body(self):
self.parser.parse_from_file(mail_test_1)
md5, sha1, sha256, sha512 = fingerprints(
self.parser.body.encode("utf-8"))
self.assertEqual(md5, "1bbdb7dcf511113bbc0c1b214aeac392")
self.assertEqual(sha1, "ce9e62b50fa4e2168278880b14460b905b24eb4b")
self.assertEqual(sha256, ("1e9b96e3f1bc74702f9703391e8ba0715b849"
"7127a7ff857013ab33385898574"))
self.assertEqual(sha512, ("ad858f7b5ec5549e55650fd13df7683e403489"
"77522995851fb6b625ac54744cf3a4bf652784"
"dba971ef99afeec4e6caf2fdd10be72eabb730"
"c312ffbe1c4de3"))
def test_malformed_mail(self):
self.parser.parse_from_file(mail_malformed_3)
defects_category = self.parser.defects_category
self.assertIn("StartBoundaryNotFoundDefect", defects_category)
self.assertIn("MultipartInvariantViolationDefect", defects_category)
def test_type_error(self):
self.parser.parse_from_file(mail_test_5)
self.assertEqual(len(self.parser.attachments_list), 5)
for i in self.parser.attachments_list:
self.assertIsInstance(i["filename"], six.text_type)
def test_valid_mail(self):
with self.assertRaises(InvalidMail):
self.parser.parse_from_string("fake mail")
def test_valid_date_mail(self):
self.parser.parse_from_file(mail_test_1),
self.assertIn("mail_without_date", self.parser.anomalies)
def test_parsing_know_values(self):
self.parser.parse_from_file(mail_test_2)
trust = "smtp.customers.net"
self.assertEqual(False, self.parser.has_defects)
raw = "217.76.210.112"
result = self.parser.get_server_ipaddress(trust)
self.assertEqual(raw, result)
raw = "<*****@*****.**>"
result = self.parser.message_id
self.assertEqual(raw, result)
raw = "mporcile@server_mail.it"
result = self.parser.to_
self.assertEqual(raw, result)
raw = "<*****@*****.**>"
result = self.parser.from_
self.assertEqual(raw, result)
raw = "Bollettino Meteorologico del 29/11/2015"
result = self.parser.subject
self.assertEqual(raw, result)
result = self.parser.has_defects
self.assertEqual(False, result)
result = len(self.parser.attachments_list)
self.assertEqual(3, result)
raw = "Sun, 29 Nov 2015 09:45:18 +0100"
raw_utc = datetime.datetime(2015, 11, 29, 8, 45, 18, 0).isoformat()
result = self.parser.date_mail.isoformat()
self.assertEqual(raw_utc, result)
def test_types(self):
self.parser.parse_from_file(mail_test_2)
trust = "smtp.customers.net"
self.assertEqual(False, self.parser.has_defects)
result = self.parser.parsed_mail_obj
self.assertIsInstance(result, dict)
self.assertNotIn("defects", result)
self.assertNotIn("anomalies", result)
self.assertIn("has_defects", result)
self.assertIn("has_anomalies", result)
result = self.parser.get_server_ipaddress(trust)
self.assertIsInstance(result, six.text_type)
result = self.parser.parsed_mail_json
self.assertIsInstance(result, six.text_type)
result = self.parser.headers
self.assertIsInstance(result, six.text_type)
result = self.parser.body
self.assertIsInstance(result, six.text_type)
result = self.parser.date_mail
self.assertIsInstance(result, datetime.datetime)
result = self.parser.from_
self.assertIsInstance(result, six.text_type)
result = self.parser.to_
self.assertIsInstance(result, six.text_type)
result = self.parser.subject
self.assertIsInstance(result, six.text_type)
result = self.parser.message_id
self.assertIsInstance(result, six.text_type)
result = self.parser.attachments_list
self.assertIsInstance(result, list)
result = self.parser.date_mail
self.assertIsInstance(result, datetime.datetime)
result = self.parser.defects
self.assertIsInstance(result, list)
result = self.parser.anomalies
self.assertIsInstance(result, list)
def test_defects_anomalies(self):
self.parser.parse_from_file(mail_malformed_1)
self.assertEqual(True, self.parser.has_defects)
self.assertEqual(1, len(self.parser.defects))
self.assertEqual(1, len(self.parser.defects_category))
self.assertIn("defects", self.parser.parsed_mail_obj)
self.assertIn("StartBoundaryNotFoundDefect",
self.parser.defects_category)
self.assertIsInstance(self.parser.parsed_mail_json, six.text_type)
result = len(self.parser.attachments_list)
self.assertEqual(1, result)
self.parser.parse_from_file(mail_test_1)
if six.PY2:
self.assertEqual(False, self.parser.has_defects)
self.assertNotIn("defects", self.parser.parsed_mail_obj)
elif six.PY3:
self.assertEqual(True, self.parser.has_defects)
self.assertEqual(1, len(self.parser.defects))
self.assertEqual(1, len(self.parser.defects_category))
self.assertIn("defects", self.parser.parsed_mail_obj)
self.assertIn(
"CloseBoundaryNotFoundDefect", self.parser.defects_category)
self.assertEqual(True, self.parser.has_anomalies)
self.assertEqual(2, len(self.parser.anomalies))
self.assertIn("anomalies", self.parser.parsed_mail_obj)
self.assertIn("has_anomalies", self.parser.parsed_mail_obj)
def test_defects_bug(self):
self.parser.parse_from_file(mail_malformed_2)
self.assertEqual(True, self.parser.has_defects)
self.assertEqual(1, len(self.parser.defects))
self.assertEqual(1, len(self.parser.defects_category))
self.assertIn("defects", self.parser.parsed_mail_obj)
self.assertIn("StartBoundaryNotFoundDefect",
self.parser.defects_category)
self.assertIsInstance(self.parser.parsed_mail_json, six.text_type)
result = len(self.parser.attachments_list)
self.assertEqual(0, result)
def test_add_content_type(self):
self.parser.parse_from_file(mail_test_3)
self.assertEqual(False, self.parser.has_defects)
result = self.parser.parsed_mail_obj
self.assertEqual(len(result["attachments"]), 1)
self.assertIsInstance(
result["attachments"][0]["mail_content_type"], six.text_type)
self.assertIsInstance(
result["attachments"][0]["payload"], six.text_type)
self.assertEqual(
result["attachments"][0]["content_transfer_encoding"],
"quoted-printable")
return rows
except Error as e:
#print(e)
logger.error(e)
finally:
cursor.close()
conn.close()
if __name__ == "__main__":
try:
content = sys.stdin.read()
parser = MailParser()
parser.parse_from_string(content)
except:
logger.error(traceback.format_exc())
sys.exit()
data = {}
sfrom = email.utils.parseaddr(parser.from_)
ssubject = parser.subject
if sfrom[0]:
sfrom = sfrom[0]
if '=?utf-8?' in sfrom or '=?UTF-8?' in sfrom:
sfrom = email.Header.decode_header(sfrom)[0][0]
else:
sfrom = sfrom[1]