pdo: Handle None cob_id correctly in __repr__

[stoicfeats]

fxes a crash when calling repr() on a pdomap that doesn't have a cob_id assigned yet.

before this, working in an interactive session would throw a typeerror if we evaluated uninitialized maps like node.rpdo[1].
the string formatter was trying to apply : X hex-formatting to a nonetype value.

this adds a quick inline fallback so it prints "unassigned" instead of throwing error on the console.

[stoicfeats]

@acolomb please have a look into this, thanks.

[codecov]

Codecov Report

❌ Patch coverage is 0% with 8 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
canopen/pdo/base.py	0.00%	7 Missing and 1 partial ⚠️

📢 Thoughts on this report? Let us know!

[acolomb]

Hi and thanks for your contribution. I've experienced the issue you mention several times. It's usually a sign of weird ordering in the API usage, because you need to assign a valid COB ID by some means first. But sure, the __repr__ case is very easy to hit without even trying any actual bus configuration.

We have several more places that can fail with the same error, mainly logging calls but also send_message() and friends. Ideally, we would forbid the cob_id member to ever be None (rather zero, which is also not valid). But that's a rather large change.

Would you mind checking the other places though, to fix other occurrences where the format strings can easily explode with cob_id = None?

[stoicfeats]

these were the places where cob_id = None could have cause the issues.

added explicit none-checks to subscribe, start, transmit, and remote_request to raise an immediate value error.

@acolomb

[stoicfeats]

@acolomb can you please have a look into this once ?

[acolomb]

Looks good in general, thank you!

I just broadened the checks a bit to cover more than just None values. Zero is not a valid COB ID for PDOs either, so checking for the thruth value of self.cob_id catches both cases.