[rifatkinali]

Fixes #228.

Problem

A dict produced by asdict() of a truncated type 6, 8 or 24 message carries None for the fields that were not present in the payload. The create() dispatchers of these types call int() directly on dac/fid (6, 8), wmo (8 DAC 1 FID 21) and mmsi/partno (24), so

>>> encode_dict(decode(b"!AIVDM,1,1,,A,6B?1,0*5C").asdict())
TypeError: int() argument must be a string, a bytes-like object or a real number, not 'NoneType'

The generic Payload.create() already treats None as "use the default" for every other field; only the dispatch step was inconsistent.

Change

• _dispatch_int(kwargs, key, default=0): reads an integer dispatch field and treats None like a missing key.
• Used in MessageType6.create, MessageType8.create, MessageType8Dac1Fid21.create (wmo) and MessageType24.create (mmsi, partno). No behaviour change for present values; partno=None selects Part A like a missing partno would.
• Tests: truncated round trip for all three types (decode → asdict → encode_dict → decode, type and MMSI preserved) and explicit None dispatch fields. Both tests fail on main before the fix.

Verification

• pytest tests/ on main + this change: 544 passed, 1 skipped; the two remaining failures (test_tcp_server, test_tcp_stream.test_invalid_endpoint) are Windows-only (SIGALRM, 0.0.0.0 bind) and fail identically without this change.
• flake8 and mypy --strict ./pyais clean.
• Against the fuzzing corpus behind Malformed NMEA input escapes the documented exception hierarchy and ends stream iteration (IndexError / UnicodeDecodeError / ValueError / TypeError) #227/encode_dict(decode(raw).asdict()) raises TypeError for truncated type 6/8/24 messages #228: with fix: various low-severity bugs found by Claude Mythos 5.1 #230 plus this change all 18 minimal reproducers are handled and all 158 real-sentence seeds still decode.

🤖 Generated with Claude Code

[M0r13n]

LGTM