Replace print-based diagnostics with the logging module#76
Open
kripnerl wants to merge 34 commits into
Open
Conversation
Merge phd-thesis work to the master See merge request kripner/pleque!141
Bump version to 0.0.10 and update release documentation See merge request kripner/pleque!142
- Remove four redundant in-method `mu_0` imports; module-level import covers them - Remove commented-out old __init__ signature left over from a prior refactor - Remove commented-out matplotlib debug code from arg_longest_monotonic_subsequence - Wire critical-point search grid (700×1200) to pleque.config.settings so users can override nr_grid/nz_grid via environment variables - Extract triplicated psi_xysq_func closure in equi_tools.py to a single _make_psi_grad_sq factory, eliminating the duplication across find_extremes, recognize_mg_axis, and recognize_x_points https://claude.ai/code/session_01FS9C1xD1y9tceTjGBvcyhn
- Remove three deprecated mis-named methods from Coordinates: incidence_angle_cos, impact_angle_cos, pol_projection_impact_angle_cos. All were wrappers around the correctly-named sin variants and had no callers anywhere in the codebase. - Delete pleque/io/_test_gfile_loader.py: dead code with no callers, superseded by the proper geqdsk.read() path, and contained Equilibrium monkey-patching. - Populate pleque/io/__init__.py with top-level re-exports (read_geqdsk, write_geqdsk, read_efithdf5, read_fiesta_equilibrium) so users can import directly from pleque.io. https://claude.ai/code/session_01FS9C1xD1y9tceTjGBvcyhn
…oup 5) __init__ is now a ~30-line orchestration that delegates to five private methods: - _load_spatial_data: loads psi grid, first wall, and spatial metadata from dataset - _load_profiles: loads 1D profile arrays (F, FFprime, pressure, pprime) - _setup_psi_spline: builds the 2D RectBivariateSpline for psi(R, Z) - _find_critical_points: finds x-points/o-points and recognizes plasma type - _setup_plasma_boundary: finds LCFS contour and strike points - _setup_1d_profiles: builds all UnivariateSpline profile interpolants No behavior change; all tests pass. Also included in this commit: - Remove three deprecated cos-misnamed methods from Coordinates that had no callers (incidence_angle_cos, impact_angle_cos, pol_projection_impact_angle_cos) https://claude.ai/code/session_01FS9C1xD1y9tceTjGBvcyhn
…2203 Refactor: clean up code structure (Groups 1 & 2)
- Replace flake8 and pylint with ruff + ty in pyproject.toml - Configure ruff with rules E,W,F,I,B,UP,RUF at 120-char line limit - Ignore N802/N803/N806: physics variable naming (R, Z, X, Y) is intentional and forms part of the public API - Add GitLab CI lint:ruff job after test:pytest - Add GitHub Actions CI with separate test, lint, and typecheck jobs (ty runs with continue-on-error until annotations are more complete) - Fix auto-fixable issues: unused imports, invalid escape sequences, import ordering, f-string modernisation, UP04x pyupgrade fixes - Fix bare except → except Exception in omas.py and plotting.py - Fix E741 ambiguous variable l → lis_len/arc_length in tools/fluxsurface - Fix B008 mutable default args: np.array/np.eye moved into function bodies - Fix B007 unused loop variable i → _i in tools.py arglis - Fix affine.py bug: np.zeros_like(transform) → np.zeros_like(self.transformation) - Change star import in fluxsurface.py to explicit import of deprecated - Use explicit re-export (X as X) in pleque/core/__init__.py - Add E402 noqa suppression in test_cdb.py (importorskip must come first) Memo – issues not fixed (would break public API or need larger refactor): - N802/N803/N806: physics coordinate names (R, Z, X, Y, B_tor, …) - C901/PLR0913/PLR0915: physics algorithms are inherently complex - E501 (18 remaining at 120 chars): all in long docstring prose - F841/RUF059: some unpacked variables are intentional - ANN*: type annotations largely absent; tracked separately https://claude.ai/code/session_01NoBcY348U4MKRmAgQHCx1w
Equilibrium must be imported last since equilibrium.py depends on Coordinates, FluxFunctions, Surface and SurfaceFunctions already being registered on the pleque.core module object. https://claude.ai/code/session_01NoBcY348U4MKRmAgQHCx1w
…-dependent Equilibrium must be imported last to avoid circular imports; isort cannot reorder these freely. https://claude.ai/code/session_01NoBcY348U4MKRmAgQHCx1w
Resolve conflicts with the develop refactor (PR #69 split of Equilibrium.__init__ into helpers and dead-code removal): - equilibrium.py: take develop's _setup_plasma_boundary() extraction; the inline boundary block my branch only lint-touched now lives in the dedicated helper. Re-applied the UP032 f-string fix there. - equi_tools.py: take develop's _make_psi_grad_sq() helper that replaces the duplicated nested psi_xysq_func. - _test_gfile_loader.py: accept develop's deletion (dead code, no references); my branch had only lint-touched it. - io/__init__.py: develop added public-API re-exports; sorted imports and added the module to ruff F401 per-file-ignores. ruff passes; full test suite green (775 passed, 37 skipped). https://claude.ai/code/session_01NoBcY348U4MKRmAgQHCx1w
Add ruff and ty linting; replace flake8/pylint; add CI workflows
- GitLab CI: run pytest with --junitxml and expose report.xml via artifacts.reports.junit so results appear in the GitLab test report UI - GitHub Actions: new workflow that installs Poetry, runs pytest with --junitxml, uploads report.xml via actions/upload-artifact, and publishes a check-run summary via dorny/test-reporter https://claude.ai/code/session_01SCWaAYTAKQdf4DpfiuczAM
Add pytest-cov to dev dependencies and run pytest with --cov=pleque --cov-report=xml:coverage.xml in both pipelines. - GitLab CI: exposes coverage.xml via artifacts.reports.coverage_report (Cobertura format) so GitLab renders line-level coverage diffs in MRs - GitHub Actions: uploads coverage.xml as artifact, publishes a coverage summary badge+table via irongut/CodeCoverageSummary, and posts a sticky comment on PRs with the coverage breakdown https://claude.ai/code/session_01SCWaAYTAKQdf4DpfiuczAM
…-ayr3rk ci: store pytest results as artifacts in GitLab and GitHub CI
…docs Coordinates.__init__ parsed the rich input syntax on every construction (~7 us even for a single point), which dominated single-point evaluations such as eq.B_R(R, Z) (~half of the 25 us call) and the field-line tracer RHS, which builds three Coordinates objects per ODE evaluation. Performance: - Add Coordinates.from_coords() classmethod as the canonical rich-syntax factory; Coordinates.__init__ now takes a fast path for the common case of 1-3 positional scalars/1-D arrays in the default coordinate type and emits a DeprecationWarning for direct rich-syntax construction (named coordinates, (N, dim) arrays, non-default coord_type). - Cache cocos_coefs() with lru_cache and return a read-only mapping (it is a pure function of the COCOS index and was rebuilt per init). - Move the _valid_coordinates* sets to class attributes. - Clean up the slow parser (module-level Iterable import, single np.atleast_1d conversions, set comparison instead of permutation scans; 3D named input is now also converted with np.atleast_1d). - Migrate internal call sites (Equilibrium.coordinates, Surface/ FluxSurface, metis.py, examples, tests) to the warning-free paths. Measured: Coordinates(eq, 0.7, 0.1) 7 -> 2.1 us; kwargs construction 7 -> 3.0 us; eq.B_R(0.7, 0.1) 25 -> 20.7 us. Bulk array calls are spline-bound and unchanged. Documentation: - The coordinate input syntax was fully documented only in Coordinates.__init__ and duplicated in docs/source/coordinates.rst, while the ~44 coordinate-accepting Equilibrium methods documented it not at all (empty :param: stubs). The syntax now lives once in COORDINATES_DOC/COORD_PARAMS_DOC and a new append_to_doc decorator injects the full text into Coordinates.from_coords and Equilibrium.coordinates, and a short cross-referencing note into all coordinate-accepting Equilibrium methods; the empty :param: stubs are removed and coordinates.rst embeds the canonical docstring via automethod instead of a copy. Tests: add fast/slow path parity tests and deprecation tests; full suite 777 passed. Sphinx build warning count drops from 10 to 8 (remaining ones pre-date this change). https://claude.ai/code/session_01KvV66UPfmyCtKH71MthmWo
…loading Extract magic numbers (grid resolutions, solver tolerances, search heuristics, spline parameters, plotting and IO writer defaults) from equilibrium.py, equi_tools.py, surfaces.py, field_line_tracers.py, plotting.py, geqdsk.py and omas.py into pleque.config.Settings, organized into per-topic sections (grid, splines, critical_points, lcfs, field_line_tracing, flux_surfaces, plotting, io). Function arguments still override settings per call; settings are now read at use time instead of being captured at import time. Settings load from the first existing of: $PLEQUE_CONFIG_FILE, ./pleque.toml, the per-user config dir (%APPDATA%\pleque on Windows, $XDG_CONFIG_HOME/pleque elsewhere), or a [tool.pleque] table in the nearest pyproject.toml, with PLEQUE_-prefixed environment variables taking precedence over files. The winning file is recorded on the settings object (config_source, config_files_consulted) and reload_settings() re-reads the configuration at runtime. The old flat fields moved into sections: npsi_grid -> flux_surfaces.n_psi, psin0 -> flux_surfaces.psi_n_min, nr_grid/nz_grid -> lcfs.search_grid_nr/nz (the None sentinel is gone; defaults are 700/1200 directly). Also fix lcfs_field_line ignoring its vect_no and xp_shift arguments (they were hardcoded in the track_plasma_boundary call). Add tests for defaults, precedence, source tracking, env vars and reload; document the feature in docs/source/configuration.rst (kept in sync by a test) and README. pydantic-settings gains the toml extra for Python 3.10 support. https://claude.ai/code/session_01HRFoBXP3uhqR5cZuURzdhS
…magic-numbers-7rq85x Centralize algorithm parameters in grouped Settings with config-file loading
- coordinates.py: remove unused itertools import, fix import sort order, remove redundant `object` base class, annotate mutable class-level sets/dict with ClassVar to satisfy RUF012 - equilibrium.py: remove unused `import pleque`, collapse duplicate import blocks (F811 redefinitions introduced by the previous commit), fix import sort order https://claude.ai/code/session_012QryKbWtWjQQrWmHdhe1pR
…tion-f2rrvs Optimize Coordinates construction and de-duplicate coordinate syntax docs
All modules now log through named loggers under the 'pleque' hierarchy. The package root installs a NullHandler (library best practice) and gains a pleque.set_log_level() convenience helper that attaches a single stream handler for interactive use. https://claude.ai/code/session_01T8Bv18R1iJkgxQ7e4rTAKc
- Initialisation hints (mg_axis, psi_lcfs, x_points, strike_points) can now also be provided as basedata variables; explicit arguments take precedence with a logged warning (idea from MR !63 / feature/fast_forward_init). - init_method is now functional: 'full' ignores all hints, 'hints' (default) uses them to assist the recognition, and 'fast_forward' takes the axis and X-point hints as final and skips the critical-point search (with fallback to 'hints' when the required hints are missing). Unknown values raise ValueError; read_geqdsk and EquilibriaTimeSlices.get_time_slice pass the option through. - The bare except around the initialisation now catches Exception, logs the traceback, and draws the (blocking) debug plot only when the new debug_plots setting is enabled (PLEQUE_DEBUG_PLOTS=1); the same gating applies to the diagnostic plot in find_extremes. Initialisation failures no longer open matplotlib windows in headless/CI environments. - First-wall resolution extracted into _resolve_first_wall with the synthetic rectangular wall construction moved to equi_tools.synthesize_rectangular_wall; the chosen wall source is logged. - find_extremes warns when no O-point candidate is found; F'=FF'/F is guarded against division by zero. - Class and constructor docstrings now describe the actual initialisation sequence, including where it deliberately differs from the thesis-chapter description (1D profile splines built after the critical-point search; LCFS found by marching squares + downhill refinement, not by field-line tracing). https://claude.ai/code/session_01T8Bv18R1iJkgxQ7e4rTAKc
- New tests/test_equilibrium_init.py: synthetic-wall generation, vacuum equilibria, profile-derivative vs direct-profile equivalence, hint arguments and basedata hints (incl. duplicity warning), init_method full/hints/fast_forward consistency on bundled cases, double-null X-points, spline-order override, logging assertions, and the failure path (logged exception, no debug plot by default, debug plot with PLEQUE_DEBUG_PLOTS=1). - The synthetic analytic equilibrium builder moved from tests/test_invert_equilibrium.py to pleque.tests.utils (synthetic_dataset/synthetic_test_wall) for reuse. - tests/test_equilibria.py::test_critical placeholder replaced with real assertions for limiter and X-point configurations. https://claude.ai/code/session_01T8Bv18R1iJkgxQ7e4rTAKc
Resolve conflict in pleque/core/coordinates.py: keep the Coordinates refactor from develop (PR #75) and re-apply the named logger, converting the warning print added by the refactor to logger.warning as well. https://claude.ai/code/session_01T8Bv18R1iJkgxQ7e4rTAKc
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
All modules now log through named loggers under the 'pleque' hierarchy.
The package root installs a NullHandler (library best practice) and gains
a pleque.set_log_level() convenience helper that attaches a single stream
handler for interactive use.
https://claude.ai/code/session_01T8Bv18R1iJkgxQ7e4rTAKc