fix: Exceptions include detail property for their value

[nicolassanmar]

Exceptions with detail but no message add that property to the exception value.

Fixes #2192

[sentrivana]

@nicolassanmar Thanks a ton for this PR! The change looks good, left a small comment. Could you please also add a testcase or two for this?

[nicolassanmar]

@nicolassanmar Thanks a ton for this PR! The change looks good, left a small comment. Could you please also add a testcase or two for this?

Sure, been busy this week, but I will update this when I can

[goharShoukat]

@nicolassanmar Thanks a ton for this PR! The change looks good, left a small comment. Could you please also add a testcase or two for this?

Sure, been busy this week, but I will update this when I can

@sentrivana I have added test to this PR.

For some reason, the new commit doesn't show up here and shows on a new branch.
#2294

[sentrivana]

Thanks @goharShoukat! I'll take a look sometime this week.

[sentrivana]

@nicolassanmar @goharShoukat Thank you both!

[goharShoukat]

Been following the changes you made Ivana. I had a question. The tests I wrote worked in Python 3.10. Is the attempt here to make the code backward compatible too?

This is my first every open-source contribution, albeit a very small one, I am still quite proud to make it. Looking forward to contributing in the future.

[sentrivana]

@goharShoukat It's a great first contribution! Congrats! :)

The change I made to the test had to do with the change I made to the behavior of the get_error_message function (this line specifically https://github.com/getsentry/sentry-python/pull/2193/files#diff-e9f88ca6368ce622839e078ed26c741a927752c01ceae64cd504d3ae86de6fd2R689). Before this PR, we were falling back to safe_str(exc_value) in case we didn't succeed in getting the message attribute, but in the previous version of this PR we would just return an empty string if we didn't find neither message nor detail, which would have been a breaking change from before. I then had to adjust the test to also check that the fallback is safe_str(exc_value) and not an empty string.

Hope that makes sense. Looking forward to more contributions from you in the future!