When we fail to encrypt a message, we log a message with a complete traceback.
The goal is to ensure that when diagnosing it later, we have complete
information.
We want to avoid deserialising message contents, because Python's email module
might produce different representation than the MUA sending original message.
The result would be a transformed message, which could mean broken message in
certain conditions.
- ContentManager sets default Content-Type even if it was missing in the
original message.
- Make sure that when Content-Type is missing, copying parameters doesn't
raise an error.
- Add a unit-test to check that.
Introduce modules:
- lacre.transport - for actual delivery via SMTP
- lacre.smime - to take care of S/MIME stuff
Implement lacre.transport.SendFrom class that does a almost exactly the same
thing as the original send_msg function, but without using global variable to
store original message sender.
In the daemon, specify policy as SMTPUTF8. That sets the deafult message type
to EmailMessage.
EmailMessage class is richer, including support for Content Managers, giving
it the capability to properly handle textual data and its encodings.
Also: add another contract test.
- Use MIMEPart instead of Message when encrypting in PGP/MIME mode.
- Wrap text/plain messages in MIMEPart, instead of manipulating payloads
manually.
- Add a test for wrapping.
- Clean up PGP/MIME flow by using API instead of explicit/manual generation of
headers.
- Fix E2E test configuration for PGP/MIME case.
- Add first lacre.core unit tests.
- Add another Contract Test.
When GnuPG refuses to encrypt a message (e.g. when key has expired), record
information about the failure and send to logs, then deliver cleartext. This
way we won't bounce email that could be delivered without encryption.
Also: add more E2E tests.
smtplib.SMTP expects ASCII-only message bodies when message body is provided
as a 'str'. If we pass a 'bytes', we need to choose encoding earlier and we
do this by calling 'as_bytes' on messages with SMTP policy, which takes care
of formatting the body properly.
As a result, ISO-8859-x messages are converted to Quoted Printable and UTF-8
messages are Base64-encoded.
Testing this behaviour is tricky, because we use the same SMTP client to send
test data. For this reason, test code has become a bit ugly, but it does
exactly what we need.
