%global pypi_name mapca

Name:           python-%{pypi_name}
Version:        0.0.8
Release:        1%{?dist}
Summary:        Moving Average PCA for fMRI dimensionality estimation

# mapca/mapca.py (and the GIFT algorithm it ports) carries a "GPL version 2
# or (at your option) any later version" header; the repository LICENSE is the
# GPL-2.0 text. The pyproject classifier says LGPL, which is stale.
License:        GPL-2.0-or-later
URL:            https://github.com/ME-ICA/mapca
Source0:        %{pypi_source %{pypi_name}}

BuildArch:      noarch
BuildRequires:  python3-devel
BuildRequires:  python3-pytest

%global _description %{expand:
MAPCA (Moving Average Principal Components Analysis) estimates the number
of independent components in fMRI data using the stationary Gaussian
moving-average model of the GIFT toolbox, and performs the PCA with that
dimensionality. It is the component-count step in tedana.}

%description %_description

%package -n python3-%{pypi_name}
Summary:        %{summary}

%description -n python3-%{pypi_name} %_description

%prep
%autosetup -n %{pypi_name}-%{version}

%generate_buildrequires
%pyproject_buildrequires

%build
%pyproject_wheel

%install
%pyproject_install
%pyproject_save_files -l mapca

%check
# test_integration.py fetches its reference arrays from OSF; the rest run offline.
%pytest mapca/tests/test_utils.py mapca/tests/test_mapca.py
# Five time courses mixed into noise: every criterion must find exactly five.
cd %{_tmppath}
%{py3_test_envvars} %{python3} - <<'PY'
import numpy as np, nibabel as nib
from mapca.mapca import MovingAveragePCA
rng = np.random.default_rng(1)
nx, ny, nz, nt, k = 12, 12, 8, 120, 5
t = np.arange(nt)
src = np.stack([np.sin(2*np.pi*(i+2)*t/nt) + 0.3*rng.standard_normal(nt)
                for i in range(k)])
maps = rng.standard_normal((nx*ny*nz, k))
data = (maps @ src + 0.5*rng.standard_normal((nx*ny*nz, nt))).reshape(nx, ny, nz, nt)
img = nib.Nifti1Image(data.astype('float32'), np.eye(4))
mask = nib.Nifti1Image(np.ones((nx, ny, nz), 'uint8'), np.eye(4))
for crit in ('aic', 'kic', 'mdl'):
    u = MovingAveragePCA(criterion=crit).fit_transform(img, mask)
    print(crit, u.shape)
    assert u.shape == (nx, ny, nz, k), (crit, u.shape)
PY

%files -n python3-%{pypi_name} -f %{pyproject_files}
%license LICENSE
%doc README.md

%changelog
* Fri Oct 16 2026 Morgan Hough <morgan.hough@gmail.com> - 0.0.8-1
- New package: required by tedana 26