Prune fails when OOPSes are zero in size

[Haw Loeung]

Hi,

We've ran into a bug where the prune job would abort when it encounters a zero sized OOPS:

| oops_tools@neem:/srv/oops.canonical.com/python-oops-datedir-repo$ cd $OOPS_DATEDIR && bin/prune --repo /srv/oops.canonical.com/oops-amqp/launchpad/staging --projectgroup launchpad-project
| File: <open file '/srv/oops.canonical.com/oops-amqp/launchpad/staging/2013-12-01/OOPS-75141b99c882a0f6a00906c3da62dcce', mode 'rb' at 0xf15ec00>
|
| Traceback (most recent call last):
| File "bin/prune", line 26, in <module>
| oops_datedir_repo.prune.main()
| File "/srv/oops.canonical.com/python-oops-datedir-repo/oops_datedir_repo/prune.py", line 154, in main
| repo.prune_unreferenced(prune_from, prune_until, references)
| File "/srv/oops.canonical.com/python-oops-datedir-repo/oops_datedir_repo/repository.py", line 289, in prune_unreferenced
| report = serializer.read(report_file)
| File "/srv/oops.canonical.com/python-oops-datedir-repo/oops_datedir_repo/serializer.py", line 57, in read
| raise IOError("Empty OOPS Report")
| IOError: Empty OOPS Report

| -rw-r--r-- 1 oops_tools oops_tools 0 Dec 1 22:33
| /srv/oops.canonical.com/oops-amqp/launchpad/staging/2013-12-01/OOPS-75141b99c882a0f6a00906c3da62dcce
|

I think what should happen is that it should skip or remove these OOPSes and continue pruning the rest.

Regards,

Haw