Catch KeyError from collector from missing TLE #116
Conversation
Codecov Report
@@ Coverage Diff @@
## main #116 +/- ##
==========================================
+ Coverage 89.58% 89.84% +0.25%
==========================================
Files 25 25
Lines 3611 3653 +42
==========================================
+ Hits 3235 3282 +47
+ Misses 376 371 -5
Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here. |
TAlonglong
requested review from
mraspaud and
pnuu
and removed request for
mraspaud
| 'uid': 'AVHRR_C_EUMP_20220901102203_51653_eps_o_amv_l2d.bin', | ||
| 'origin': '157.249.16.188:9062', | ||
| 'end_time': datetime(2022, 9, 1, 10, 25, 3)}) | ||
| assert "Found no TLE entry for 'METOP-B' to similate KeyError" in caplog.text |
There was a problem hiding this comment.
| assert "Found no TLE entry for 'METOP-B' to similate KeyError" in caplog.text | |
| assert "Found no TLE entry for 'METOP-B' to simulate" in caplog.text |
There was a problem hiding this comment.
Hmm, where does this exception text originate? I found "Found no TLE entry for '%s'" in https://github.com/pytroll/pyorbital/blob/main/pyorbital/tlefile.py#L200 but the rest of it I can't seem to find in either Pyorbital, Pytroll Schedule nor Pytroll Collectors.
There was a problem hiding this comment.
oh, how to explain
I just added some dummy text in the test to be sure I matched the test KeyError message and not the message from the actual code. If that make any sense. I struggled a bit before I got this right.
But sure, I can at least get the spelling right.
There was a problem hiding this comment.
Oh, right. Didn't grep the same file 😅
Now and then I get for some reason an exception as described in #30
What happens is then that the subscriber for this area stops and thus can not continue collecting new segments.
Other regions continue to work if there are more. Or the the geographic gatherer just continue to run.
One way to handle this, as proposed in this PR, is to catch the KeyError exception and then continue keeping the subscriber running. Therefore at the next run it will try again and it will still work.
I think this is better than just stop working. At least it tries again.
Another way is to do this is to make the whole geographic gatherer stop and then leave to the user how he or she will handle this. But this will affect all the regions in the gatherer.