%global pypi_name idc-index-data
%global mod_name  idc_index_data

Name:           python-%{pypi_name}
Version:        25.0.0
Release:        1%{?dist}
Summary:        Imaging Data Commons index used to query and download data

# Code and the bundled index are both MIT. The index is metadata about the
# Imaging Data Commons collections (identifiers, modalities, sizes, storage
# URLs and each collection's own licence), not image data.
License:        MIT
URL:            https://github.com/ImagingDataCommons/idc-index-data
Source0:        %{pypi_source %{mod_name} %{version}}
# The package finds its Parquet files through the wheel's RECORD file
# (importlib.metadata distribution().files), and Fedora's %%pyproject_install
# deletes RECORD, so files is None and the import itself fails. Fall back to
# the package directory. Upstream draft: docs/upstream/idc-index-data-lookup-without-record.md
Patch0:         python-idc-index-data-lookup-without-record.patch

BuildArch:      noarch

BuildRequires:  python3-devel
BuildRequires:  python3-pytest
BuildRequires:  python3-pandas
BuildRequires:  python3-pyarrow
BuildRequires:  python3-requests

%description
idc-index-data packages the index of the NCI Imaging Data Commons (IDC) as
Parquet files: one row per DICOM series with its collection, patient, study,
modality, size and storage location. The idc-index client queries it to find
and download public imaging data.

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

%description -n python3-%{pypi_name}
idc-index-data packages the index of the NCI Imaging Data Commons (IDC) as
Parquet files. The idc-index client queries it to find and download public
imaging data.

%prep
%autosetup -n %{mod_name}-%{version} -p1
# The build hook regenerates the index from Google BigQuery, but only when
# GCP_PROJECT is set; without it, it ships the Parquet files the sdist already
# carries. Drop the BigQuery/GCS client libraries from the build requirements
# (none is packaged in Fedora) so pyproject_buildrequires does not ask for them.
sed -i -e '/"db-dtypes",/d' -e '/"google-cloud-bigquery",/d' -e '/"google-cloud-storage",/d' pyproject.toml
! grep -n -e '"google-cloud' -e '"db-dtypes' pyproject.toml
# Version comes from git (hatch-vcs); the sdist has no .git.
test -f src/%{mod_name}/_version.py

%generate_buildrequires
export SETUPTOOLS_SCM_PRETEND_VERSION=%{version}
%pyproject_buildrequires

%build
unset GCP_PROJECT
export SETUPTOOLS_SCM_PRETEND_VERSION=%{version}
%pyproject_wheel

%install
%pyproject_install
%pyproject_save_files %{mod_name}

%check
%pyproject_check_import
# Upstream's package tests check the index files are bundled, readable and
# consistent with the shipped metadata. No network. The other test files cover
# the maintainers' index-generation scripts (scripts/), which are not shipped.
%pytest -v tests/test_package.py

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

%changelog
* Sun Oct 11 2026 Morgan Hough <morgan.hough@gmail.com> - 25.0.0-1
- Initial package, for the Imaging Data Commons client (idc-index)