""" Axis-role resolution: metadata first, names only as a fallback. """ import os import numpy as np import pytest from dispnc.conventions import normalize from dispnc.coords import resolve_variable_axes from dispnc.io import open_source from conftest import EXAMPLES, open_sample, sample_files def roles_of(ds, varname, unstructured=None): axes = resolve_variable_axes(ds, varname, unstructured) return {a.dim: (a.role, a.source) for a in axes} def test_xios_resolves_from_axis_attribute(xios_ds): ds, _ = normalize(xios_ds) got = roles_of(ds, 'ps_avg') assert got['lat'] == ('Y', 'axis-attr') assert got['lon'] == ('X', 'axis-attr') assert got['time_counter'] == ('T', 'axis-attr') def test_soildepth_resolves_by_positive_not_by_fake_standard_name(xios_ds): """ XIOS writes soildepth:standard_name = "Soil mid-layer depth". The role must come from `positive`, and the bogus name must not appear as the source. """ xios_ds['probe'] = (('soildepth', 'lat'), np.zeros((3, 3))) ds, _ = normalize(xios_ds) axis = resolve_variable_axes(ds, 'probe').by_dim('soildepth') assert axis.role == 'Z' assert axis.source in ('axis-attr', 'positive') assert axis.positive == 'down' # the sentence XIOS put in standard_name must not be what decided the role assert 'standard_name' not in ds['soildepth'].attrs def test_pem_resolves_after_normalization(pem_ds): ds, _ = normalize(pem_ds) got = roles_of(ds, 'tsurf') assert got['latitude'][0] == 'Y' assert got['longitude'][0] == 'X' assert got['Time'][0] == 'T' def test_staggered_rlonu_gets_a_role(lmdz_ds): """ rlonu/rlatv match no name table in the original code and rendered as meaningless axes. After normalization they resolve as real longitudes. """ ds, _ = normalize(lmdz_ds) axis = resolve_variable_axes(ds, 'ucov').by_dim('rlonu') assert axis.role == 'X' assert axis.units == 'degrees_east' def test_meaningless_dimensions_are_S(): import xarray as xr ds = xr.Dataset({'v': (('nslope', 'index'), np.zeros((2, 3)))}) got = roles_of(ds, 'v') assert got['nslope'][0] == 'S' assert got['index'][0] == 'S' def test_duplicate_roles_are_broken_by_confidence(): """ Two dimensions must never both be X: the less confident one is demoted. """ import xarray as xr ds = xr.Dataset( {'v': (('lon', 'rlonu'), np.zeros((3, 4)))}, coords={'lon': ('lon', [0.0, 1, 2], {'axis': 'X'}), 'rlonu': ('rlonu', [0.0, 1, 2, 3], {})}, ) axes = resolve_variable_axes(ds, 'v') assert [a.role for a in axes].count('X') == 1 assert axes.by_dim('lon').role == 'X' assert axes.by_dim('rlonu').role == 'S' def test_cyclic_longitude_detected(xios_ds): """ XIOS longitudes run -180..168.75 in 11.25 degree steps: they wrap the globe even though the last point is not 180. """ import xarray as xr lon = np.arange(-180, 180, 11.25) ds = xr.Dataset({'v': (('lon',), np.zeros(lon.size))}, coords={'lon': ('lon', lon, {'axis': 'X', 'units': 'degrees_east'})}) assert resolve_variable_axes(ds, 'v').x.is_cyclic def test_non_cyclic_longitude_subset(): import xarray as xr lon = np.arange(0, 90, 10.0) ds = xr.Dataset({'v': (('lon',), np.zeros(lon.size))}, coords={'lon': ('lon', lon, {'axis': 'X', 'units': 'degrees_east'})}) assert not resolve_variable_axes(ds, 'v').x.is_cyclic # --- the real files ----------------------------------------------------------- # # Two kinds of test read sample_files/. The table below names files and # variables, and each case skips when its own file or variable is not in the # sample set of the day. The sweep after it names nothing: it asserts what must # hold of any file the tool can be pointed at, so it keeps working when the # samples are replaced. Prefer adding to the sweep; the table is for the # specific quirks worth pinning down. REAL_CASES = [ ('diagevo.nc', 'tsurf', {'Time': 'T', 'latitude': 'Y', 'longitude': 'X'}), ('diagfi.nc', 'phisinit', {'latitude': 'Y', 'longitude': 'X'}), ('xoutyearly4pem_y1.nc', 'ps_avg', {'time_counter': 'T', 'lat': 'Y', 'lon': 'X'}), ('xoutyearly4pem_y1.nc', 'tsoil_avg', {'time_counter': 'T', 'soildepth': 'Z', 'lat': 'Y', 'lon': 'X'}), ('start.nc', 'ucov', {'Time': 'T', 'altitude': 'Z', 'latitude': 'Y', 'rlonu': 'X'}), ('startfi.nc', 'tsurf', {'physical_points': 'U'}), ('startevo.nc', 'tsoil', {'Time': 'T', 'nslope': 'S', 'subsurface_layers': 'Z', 'physical_points': 'U'}), ] @pytest.mark.parametrize('filename,varname,expected', REAL_CASES) def test_real_files_resolve_every_dimension(filename, varname, expected): with open_sample(filename, varname) as (ds, report): axes = resolve_variable_axes(ds, varname, report.unstructured) got = {a.dim: a.role for a in axes} assert got == expected def test_ucov_x_axis_is_the_staggered_one(): """ start.nc's ucov lives on rlonu, not on longitude. Resolving it as X is what makes the plot meaningful. """ with open_sample('start.nc', 'ucov') as (ds, report): axes = resolve_variable_axes(ds, 'ucov', report.unstructured) assert axes.x.dim == 'rlonu' assert axes.x.units == 'degrees_east' # 185.625 degrees: past 180, which is why the radian test needs a 2*pi # bound rather than the pi bound the original code used assert axes.x.values.max() == pytest.approx(185.625, abs=1e-3) # One case per sample file, whatever they happen to be, and a readable skip # rather than pytest's "empty parameter set" when there are none. SWEEP = sample_files() or [pytest.param( None, id='none', marks=pytest.mark.skip(reason=f"no .nc files in {EXAMPLES}; see README.md"))] @pytest.mark.parametrize('path', SWEEP, ids=os.path.basename) def test_every_variable_of_every_sample_resolves(path): """ Whatever the sample set holds, every variable of every file must resolve to exactly one axis per dimension, with no role claimed twice - a file that resolved `lat` and `latitude` both as Y would plot nonsense. This is the test that survives the samples being replaced. """ src = open_source(path) try: ds, report = normalize(src.ds) for varname in ds.variables: axes = resolve_variable_axes(ds, varname, report.unstructured) dims = list(ds[varname].dims) assert [a.dim for a in axes] == dims, f"{varname}: axes do not match dims" claimed = [a.role for a in axes if a.role != 'S'] assert len(claimed) == len(set(claimed)), \ f"{varname}: role claimed twice ({claimed})" finally: src.close()