sandialabs / LoQS

89%

push

github

Linting (except codepacks and parallelization/checkpointing files) (#129)

* Format loqs/ with black and fix mechanical flake8 issues

Get most of loqs/ clean under black and flake8, deferring the 7 pre-existing complexity violations and four files excluded from this pass.
- Reformat 40 files with black.
- Add raw-string prefixes to three PauliFrame docstrings containing literal LaTeX escape sequences.
- Strip trailing whitespace from the STIM circuit grammar docstring and remove an unused local variable in NumpyStatevectorQuantumState.
- Remove a duplicate pymatching import guard and a redundant nested import that already had access to the outer scope's binding via closure.
- Fix f-string operator spacing, and rewrite two Protocol stub methods to use a pass-bodied stub instead of an inline ..., so black won't keep re-collapsing them into a form flake8 flags.

* Extract helpers to fix remaining C901 complexity violations

Reduce the last two non-codepack flake8 complexity violations via helper-function extraction, with no behavior change beyond one explicitly-agreed edge case.
- Split NumpyStatevectorQuantumState.__init__ into _init_from_existing_state, _infer_num_subsystems, _resolve_dimensions, and _init_state_array; also removes a real duplication where dimensions were computed generically then immediately discarded whenever copy-constructing from another instance.
- Extract a shared _first_successful_rep helper used by both PyGSTiNoiseModel._get_gate_rep and _get_instrument_rep (try each candidate rep type, cache the first success, raise a combined error otherwise), and pull _get_instrument_rep's largest branch out into _build_outcome_dict_rep.
- Copy-constructing NumpyStatevectorQuantumState with an explicit d no longer raises on a length mismatch against the source's qubit count; that value was already silently discarded in every case, so this only removes a spurious check on an untested edge case.

* Fix mypy errors in DictNoiseModel/PyGSTiNois... (continued)

2984 of 3544 branches covered (84.2%)

Branch coverage included in aggregate %.

9602 of 10563 relevant lines covered (90.9%)

0.91 hits per line