bug loading MPL style from the configuration

[1ucian0]

While testing mthree with rc1, I detected a small typo in qiskit.visualization.circuit.qcstyle.load_style introduced in #11536:

https://github.com/Qiskit-Extensions/mthree/actions/runs/7828157591/job/21357435372?pr=196

TypeError: expected str, bytes or os.PathLike object, not list

locally also is reproducible:

from qiskit import QuantumCircuit

qc = QuantumCircuit(6)
qc.reset(range(6))
qc.h(3)
qc.measure_all()
qc.draw('mpl')

[qiskit-bot]

One or more of the the following people are requested to review this:

• @enavarro51
• @Qiskit/terra-core

[jakelishman]

This looks right thanks, but can we add a regression test so it doesn't happen again?

[coveralls]

Pull Request Test Coverage Report for Build 7916495351

Warning: This coverage report may be inaccurate.

This pull request's base commit is no longer the HEAD commit of its target branch. This means it includes changes from outside the original pull request, including, potentially, unrelated coverage changes.

• For more information on this, see Tracking coverage changes with pull request builds.
• To avoid this issue with future PRs, see these Recommended CI Configurations.
• For a quick fix, rebase this PR at GitHub. Your next report should be accurate.

Details

• 0 of 2 (100.0%) changed or added relevant lines in 2 files are covered.
• 175 unchanged lines in 22 files lost coverage.
• Overall coverage increased (+0.08%) to 89.289%

---

Files with Coverage Reduction	New Missed Lines	%
crates/qasm2/src/lex.rs	1	91.94%
qiskit/synthesis/clifford/clifford_decompose_ag.py	1	98.72%
qiskit/synthesis/clifford/clifford_decompose_full.py	1	87.5%
qiskit/synthesis/linear/cnot_synth.py	1	98.0%
qiskit/synthesis/clifford/clifford_decompose_greedy.py	2	98.88%
qiskit/synthesis/linear/linear_depth_lnn.py	2	98.33%
qiskit/synthesis/stabilizer/stabilizer_decompose.py	2	96.83%
qiskit/synthesis/unitary/qsd.py	2	97.52%
qiskit/synthesis/clifford/clifford_decompose_bm.py	3	97.83%
qiskit/synthesis/cnotdihedral/cnotdihedral_decompose_general.py	3	96.05%

Totals
Change from base Build 7904240889:	0.08%
Covered Lines:	58901
Relevant Lines:	65967

---

💛 - Coveralls

[jakelishman]

that's a pretty questionable decision by GitHub to allow a commit in an entirely separate organisation (let alone repo) to close this PR.

[1ucian0]

can we add a regression test so it doesn't happen again?

Added in dddca99

[jakelishman]

Thanks for this Luciano. The spirit of the test looks good, though I had a minor comment. Let me know if you've not got time to respond and I can just do it if you prefer.

[jakelishman]

This particular form causes the test to generate a warning (we don't forbid UserWarning in CI) unless mthree is installed, which drops the quantum-light.json file into ~/.qiskit.

We maybe want to make the test resilient against that - perhaps use a temporary directory as the style_path that includes a valid style file we can use?

[1ucian0]

true. Done in ae5f93c I ended up with a kinda complicated solution. Any ideas how to improve it?

[jakelishman]

I think what you had was fine in principle - the interface of the circuit drawer is kind of missing a couple of things that would have made it a bit easier to do, so I think everything was always going to look a little messy. I tweaked it over to use TemporaryDirectory instead and to add a test of a failed lookup in 05eb769, but mostly just to expand the coverage more than anything else.

[jakelishman]

Thanks for this Luciano.

I think your test before was generally ok, but I've tweaked the test here to avoid using another internal function (load_style) and needing to do a direct JSON dump, in favour of copying over a known-valid style into a temporary directory and using that. I also expanded the test to test the behaviour if we're searching for a file that doesn't exist.

It's perhaps a bit cleaner to use a TemporaryDirectory directly instead of of a NamedTemporaryFile because it's technically a directory that we actually want to test, and NamedTemporaryFile has a few complications between POSIX and Windows around the rules for re-opening it, and stuff like that.