johanjoensson / impurityModel / 40071905347

95%

Build Type

push

github

Committed by johanjoensson

Commit Message

chore(lint): make ruff, black and mypy clean across the package

`make check` was red on master after merging nio-iter1-findings: 138 ruff findings
(125 of them pre-existing under ruff 0.15/0.17), one file black would reformat and
40 mypy errors (37 pre-existing). Now `ruff check .`, `black --check .` and
`mypy -p impurityModel` all pass, under ruff 0.15.17 and 0.17.2 alike.

Mostly mechanical: import sorting, unused imports and variables, `dict.fromkeys`,
raw regex patterns in `pytest.raises(match=...)`, docstring lines over 120 columns,
`Optional` dataclass fields spelled `X | None`. The non-mechanical ones:

- `test_block_lanczos_perf.py` reported a `W` buffer that `block_lanczos` stopped
  returning (undefined name in an opt-in benchmark); the row is dropped.
- `gf_solvers.block_Green_sparse` reads `len(alphas)` after a loop that may leave it
  `None`; the block count is now taken once, as 0 in that case.
- `memory_estimate` spells its NaN checks with `math.isnan` instead of `x != x`.
- `plot_spectra` carried a dead `plotted` list; `lie_algebra._residual_r_factor` a
  dead product.
- Two test lambdas closing over a per-iteration list now bind it as a default (B023).
- `PLR0917` (ruff >= 0.17's too-many-positional-arguments) joins `PLR0913` in the
  ignore list; `BlockLanczosCore` and `rspt2spectra` join mypy's no-stubs stanza.

Checked: ruff, black, mypy green; `pytest --co` collects 3346 tests with no errors.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

Coverage Stats

39 of 42 new or added lines in 18 files covered. (92.86%)

238 existing lines in 22 files now uncovered.

16032 of 16925 relevant lines covered (94.72%)

0.95 hits per line