Name:           python-zimscraperlib
Version:        5.5.1
Release:        1%{?dist}
Summary:        Shared code for openZIM scrapers that build ZIM archives

# zimscraperlib is GPL-3.0-or-later. The sdist ships two JavaScript files
# for offline replay: wombat.js (AGPL-3.0-or-later, fetched by upstream's
# build from @webrecorder/wombat 3.11.0) and wombatSetup.js (built by
# upstream with rollup from javascript/, GPL-3.0-or-later, bundling uri-js,
# BSD-2-Clause).
License:        GPL-3.0-or-later AND AGPL-3.0-or-later AND BSD-2-Clause
URL:            https://github.com/openzim/python-scraperlib
# The PyPI sdist, not the git tag: it carries the generated rules and the
# JavaScript statics, so the build hook downloads nothing.
Source0:        %{pypi_source zimscraperlib}
# Port to python-libzim 4.0: add_illustration() takes (content, width),
# search results are SearchResult objects, camelCase search methods and
# keyword arguments became snake_case, Archive._get_entry_by_id() is gone.
Patch0:         zimscraperlib-5.5.1-libzim4.patch

BuildArch:      noarch
BuildRequires:  python3-devel
BuildRequires:  python3-pytest
BuildRequires:  python3-pytest-mock
# The build hook runs "python rules/generate_rules.py".
BuildRequires:  python-unversioned-command
# python-libzim needs libzim >= 9.4, which Fedora 43 lacks.
BuildRequires:  python3-libzim >= 4
# GIF optimization runs gifsicle.
BuildRequires:  gifsicle

%global _description %{expand:
zimscraperlib is the common library behind openZIM's scrapers, the tools
that turn websites, YouTube channels, TED talks and other sources into ZIM
archives for offline reading in Kiwix. It wraps libzim's creator, downloads
and re-encodes media, optimizes and resizes images, rewrites HTML, CSS and
JavaScript URLs for offline replay (with wombat.js), detects languages and
builds full-text indexes from HTML and PDF.}

%description %{_description}

%package -n     python3-zimscraperlib
Summary:        %{summary}
Provides:       bundled(js-wombat) = 3.11.0
Provides:       bundled(js-uri-js) = 4.4.1
Requires:       gifsicle

%description -n python3-zimscraperlib %{_description}

%prep
%autosetup -p1 -n zimscraperlib-%{version}
# Upstream pins tightly for its scrapers' reproducibility:
# - only Python 3.14 (f45 and rawhide have 3.15; the tests decide);
# - libzim < 4.0 (Patch0 ports the two affected calls to python-libzim 4.0);
# - pymupdf-lite >= 2 is the PyMuPDF module minus its 2.0 layout engine;
#   the code uses only open(), metadata and get_text(), which Fedora's
#   PyMuPDF 1.x has;
# - types-xxhash is a typing stub, not needed at run time;
# - exact jinja2 and PyYAML versions for the build.
sed -i \
    -e 's/^requires-python = ">=3.14,<3.15"/requires-python = ">=3.14"/' \
    -e 's/"libzim>=3.8.0,<4.0"/"libzim>=3.8.0"/' \
    -e 's/"pymupdf-lite>=2.0,<3.0"/"pymupdf"/' \
    -e '/"types-xxhash/d' \
    -e 's/"jinja2==3.1.6"/"jinja2"/' \
    -e 's/"PyYAML==6.0.3"/"PyYAML"/' \
    pyproject.toml
grep -q '"pymupdf",' pyproject.toml
! grep -q 'types-xxhash\|<3.15\|==3.1.6' pyproject.toml

%generate_buildrequires
%pyproject_buildrequires

%build
%pyproject_wheel

%install
%pyproject_install
%pyproject_save_files -l zimscraperlib

%check
%pyproject_check_import
# Deselected: tests that fetch URLs or sample ZIMs without the "slow" marker
# (slow tests are skipped by default), and one that compares a PNG's exact
# byte size, which depends on the zlib and Pillow build.
deselect=(
  tests/download/test_download.py::test_first_block_download_default_session
  tests/download/test_download.py::test_filehandler
  "tests/image/test_illustration.py::test_get_zim_illustration[png_url]"
  "tests/image/test_illustration.py::test_get_zim_illustration[svg_url]"
  tests/image/test_image.py::test_remove_png_transparency
  tests/inputs/test_inputs.py::test_remote
  tests/inputs/test_inputs.py::test_remote_dest
  tests/inputs/test_inputs.py::test_remote_indir
  tests/zim/test_archive.py::test_metadata
  tests/zim/test_archive.py::test_entry_by_id
  tests/zim/test_archive.py::test_get_item
  tests/zim/test_archive.py::test_suggestions
  tests/zim/test_archive.py::test_suggestions_end_index
  tests/zim/test_archive.py::test_search_no_fti
  tests/zim/test_archive.py::test_counters
  tests/zim/test_archive.py::test_get_tags
  tests/zim/test_zim_creator.py::test_urlitem_html
  tests/zim/test_zim_creator.py::test_urlitem_binary
  tests/zim/test_zim_creator.py::test_filelikeprovider_nosize
  tests/zim/test_zim_creator.py::test_urlprovider
  tests/zim/test_zim_creator.py::test_urlprovider_nolength
)
%pytest ${deselect[@]/#/--deselect=}

%files -n python3-zimscraperlib -f %{pyproject_files}
%doc README.md CHANGELOG.md
%{_bindir}/fix_ogvjs_dist

%changelog
* Sun Nov 01 2026 Morgan Hough <morgan.hough@gmail.com> - 5.5.1-1
- Initial package