def test_that_as_dict_works(email_with_attachments: bytes):
message = ParsedEmail.from_bytes(email_with_attachments)
attachments = list(message.get_attachments())
expected = {
'from': '*****@*****.**',
'to': ['*****@*****.**'],
'subject': 'An email with attachments',
'body': 'This is the message body\n',
'attachments': [
{
'filename': 'a-text-file.txt',
'content-type': 'text/plain',
'body': b64encode(attachments[0].body).decode(),
'base64': True,
},
{
'filename': 'a-pdf-file.pdf',
'content-type': 'application/pdf',
'body': b64encode(attachments[1].body).decode(),
'base64': True,
},
{
'filename': 'an-excel-file.xlsx',
'content-type': 'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet',
'body': b64encode(attachments[2].body).decode(),
'base64': True,
}
]
}
assert message.as_dict() == expected
def test_that_email_message_can_read_an_email_without_attachments(base_email: bytes):
message = ParsedEmail.from_bytes(base_email)
assert message.from_addr == '*****@*****.**'
assert message.to_addr == ('*****@*****.**',)
assert message.subject == 'Example Email'
assert message.body == 'This is an example email\n*This is an example email*\nThis is an example email\n'
def test_that_lambda_handler_wires_everything_correctly(
base_email: bytes,
mock_sns_client: botocore.client.BaseClient,
set_environment_variable: Callable[[str, str], None]):
sns_arn = 'I am an ARM'
bucket = 'bucket-bucket-bucket'
prefix = 'tank-recipient'
message_id = 'id1234'
set_environment_variable('TRANSFORMER_ARN', sns_arn)
set_environment_variable('EMAIL_BUCKET_NAME', bucket)
create_s3_object(bucket, f'{prefix}/{message_id}', base_email)
event = {
'Records': [
{
'ses': {
'mail': {'messageId': message_id},
'receipt': {
'recipients': [f'{prefix}@foo.com'],
},
},
},
],
}
lambda_handler(event, None)
mock_sns_client.publish.assert_called_once_with(
TargetArn=sns_arn,
Message=json.dumps({'default': ParsedEmail.from_bytes(base_email).as_json()}),
MessageStructure='json')
def test_that_attachments_are_extracted(email_with_attachments: bytes):
message = ParsedEmail.from_bytes(email_with_attachments)
attachments = list(message.get_attachments())
assert message.has_attachments
assert len(attachments) == 3
assert attachments[0].filename == 'a-text-file.txt'
assert attachments[1].filename == 'a-pdf-file.pdf'
assert attachments[2].filename == 'an-excel-file.xlsx'
def lambda_handler(event: dict, _context):
_configure_logging()
transformer_arn = os.environ['TRANSFORMER_ARN']
bucket_name = os.environ['EMAIL_BUCKET_NAME']
message_id = event['Records'][0]['ses']['mail']['messageId']
# key prefix is the first part of the email address (i.e. 'jdoe' from '*****@*****.**')
key_prefix = event['Records'][0]['ses']['receipt']['recipients'][0].split('@')[0]
raw_email = get_raw_email(message_id, bucket_name, key_prefix)
message = ParsedEmail.from_bytes(raw_email)
publish_to_sns(message.as_json(), transformer_arn)
def test_that_attachments_get_decoded(email_with_attachments: bytes):
message = ParsedEmail.from_bytes(email_with_attachments)
text_file = list(message.get_attachments())[0]
pdf_file = list(message.get_attachments())[1]
excel_file = list(message.get_attachments())[2]
assert text_file.filename == 'a-text-file.txt'
assert text_file.body == b'This is a text file\n'
assert pdf_file.filename == 'a-pdf-file.pdf'
assert pdf_file.body.startswith(b'%PDF-1.4\n%\xc7\xec\x8f\xa2\n5')
assert excel_file.filename == 'an-excel-file.xlsx'
assert excel_file.body.startswith(b'PK\x03\x04\x14\x00\x00\x00\x08\x00\xc5\xa5RN')
def test_to_addr_contains_all_email_recipients(two_recipient_email: bytes):
message = ParsedEmail.from_bytes(two_recipient_email)
assert message.to_addr == ('*****@*****.**', '*****@*****.**')
def test_that_html_emails_are_converted_to_plain_text(html_email: bytes):
message = ParsedEmail.from_bytes(html_email)
assert message.body == 'This is an example email\n\n **This is an example email**\n\nThis is an example email'