Fix TypeError in report_collect with _collect_report_last_write

[blueyed]

_collect_report_last_write might be None, when pytest_collection was
not called before. Not sure if this indicates another problem, but it
can be reproduced with testing/test_collection.py::TestCollector::()::test_getcustomfile_roundtrip.

Fixes #4329

[nicoddemus]

Thanks @blueyed, you are on 🔥

but it
can be reproduced with testing/test_collection.py::TestCollector::()::test_getcustomfile_roundtrip

What do you mean? This test passes for me, and the output looks OK as well.

Oh just saw the issue, and I can replicate using the command posted there.

But I worry that this was passing before, because we might introduce a regression by mistake in the future.

[blueyed]

But I worry that this was passing before, because we might introduce a regression by mistake in the future.

When doing this initially I assumed that both hooks were called in sequence always, which apparently is not the case here.
This fix is just was I would have done without that assumption, so I think it is good - but the issue itself of both hooks not being called always might be worth investigating..

[nicoddemus]

This fix is just was I would have done without that assumption, so I think it is good

Fair enough. 👍

[blueyed]

#4331 contains a test for this, but fails due to the capture plugin.

[blueyed]

Included in #4331.

[codecov]

Codecov Report

Merging #4330 into master will decrease coverage by 2.12%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #4330      +/-   ##
==========================================
- Coverage   95.84%   93.72%   -2.13%     
==========================================
  Files         111      111              
  Lines       24880    24875       -5     
  Branches     2430     2427       -3     
==========================================
- Hits        23847    23314     -533     
- Misses        736     1230     +494     
- Partials      297      331      +34

Flag	Coverage Δ
#docs	?
#doctesting	?
#linting	?
#linux	93.72% <100%> (-1.92%)	⬇️
#nobyte	?
#numpy	?
#pexpect	?
#py27	91.64% <100%> (-2.38%)	⬇️
#py34	92.02% <100%> (-0.07%)	⬇️
#py35	?
#py36	92% <100%> (-1.9%)	⬇️
#py37	?
#trial	?
#windows	?
#xdist	?

Impacted Files	Coverage Δ
src/_pytest/terminal.py	91.32% <100%> (-1.71%)	⬇️
testing/test_pdb.py	42.4% <0%> (-52.95%)	⬇️
src/_pytest/debugging.py	51.38% <0%> (-32.64%)	⬇️
testing/python/approx.py	79.84% <0%> (-19.77%)	⬇️
src/_pytest/unittest.py	75.14% <0%> (-19.21%)	⬇️
testing/test_unittest.py	85.32% <0%> (-13.77%)	⬇️
src/_pytest/reports.py	86.9% <0%> (-10.72%)	⬇️
src/_pytest/python_api.py	87.12% <0%> (-10.31%)	⬇️
testing/test_pathlib.py	91.17% <0%> (-8.83%)	⬇️
src/_pytest/assertion/util.py	92.92% <0%> (-5.19%)	⬇️
... and 22 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 64762d2...5470533. Read the comment docs.

[blueyed]

Going to merge this already, since #4331 (which implicitly tests this) might take longer.