Fix get_source_script

.gitignore

@@ -1,7 +1,3 @@
 /chardet-*.tar.gz
 /chardet-*/
 /chardet-4.0.0-filtered.tar.xz
-/chardet-5.0.0-filtered.tar.xz
-/chardet-5.1.0-filtered.tar.xz
-/chardet-5.2.0-filtered.tar.xz
-/chardet-5.2.0-filtered.tar.zst

changelog

@@ -1,9 +1,3 @@
-* Mon Jun 13 2022 Python Maint <python-maint@redhat.com> - 4.0.0-7
-- Rebuilt for Python 3.11
-
-* Mon Jun 13 2022 Python Maint <python-maint@redhat.com> - 4.0.0-6
-- Bootstrap for Python 3.11
-
 * Fri Jan 21 2022 Fedora Release Engineering <releng@fedoraproject.org> - 4.0.0-5
 - Rebuilt for https://fedoraproject.org/wiki/Fedora_36_Mass_Rebuild
 

chardetect.1

@@ -19,11 +19,5 @@ File whose encoding we would like to determine.
 .B \-h\fR, \fB\-\-help
 show a help message and exit
 .TP
-.B \-\-minimal
-Print only the encoding to standard output
-.TP
-.B \-l\fR, \fB\-\-legacy
-Rename legacy encodings to more modern ones.
-.TP
 .B \-\-version
 show program\(cqs version number and exit

get_source

@@ -26,23 +26,9 @@ ARCHIVE="$(find . -mindepth 1 -maxdepth 1 -type f -name '*.tar.gz' -print -quit)
 echo "--> Extracting: $(basename "${ARCHIVE}")" 1>&2
 tar -xzf "${ARCHIVE}"
 echo '--> Removing tests due to licensing issues' 1>&2
-TARDIR="$(basename "${ARCHIVE}" '.tar.gz')"
-MTIME="$(stat -c '%Y' "${TARDIR}")"
-rm -rvf "${TARDIR}/tests/"
-# Restore the original mtime even though we modified the base directory by
-# removing the tests.
-touch -d @"${MTIME}" "${TARDIR}"
-FILTERED="$(basename "${ARCHIVE}" .tar.gz)-filtered.tar.zst"
+rm -rvf */tests/
+FILTERED="$(basename "${ARCHIVE}" .tar.gz)-filtered.tar.xz"
 echo "--> Re-archiving: ${FILTERED}" 1>&2
-# https://www.gnu.org/software/tar/manual/html_section/Reproducibility.html
-TZ=UTC LC_ALL=C tar \
-    --create --verbose \
-    --sort=name \
-    --format=posix \
-    --numeric-owner --owner=0 --group=0 \
-    --mode=go+u,go-w \
-    --pax-option='delete=atime,delete=ctime' \
-    "${TARDIR}/" |
-  zstdmt --ultra -22 > "${FILTERED}"
+tar -c "${FILTERED}" */ | xz -9e > "${FILTERED}"
 mv -v "${FILTERED}" "${OUTDIR}"
 echo 'Done.' 1>&2

python-chardet.rpmlintrc

@@ -1,11 +0,0 @@
-# Not a real spelling error:
-addFilter(r" spelling-error \('([Pp]rober)s?',")
-# This Source cannot be a URL because we must filter out certain files.
-addFilter(r" invalid-url Source[0-9]+: chardet-.*-filtered\.tar\.zst")
-# Bogus; it is a perfectly fine zstandard file.
-addFilter(r" inconsistent-file-extension chardet-.*-filtered\.tar\.zst$")
-# The -doc subpackage was removed without an equivalent replacement.
-addFilter(r" obsolete-not-provided python-chardet-doc$")
-# Update LGPLv2.1 license text and notices for remote-only FSF
-# https://github.com/chardet/chardet/pull/307
-addFilter(r" incorrect-fsf-address ")

python-chardet.spec

@@ -1,11 +1,23 @@
+# Sphinx-generated HTML documentation is not suitable for packaging; see
+# https://bugzilla.redhat.com/show_bug.cgi?id=2006555 for discussion.
+#
+# We can generate PDF documentation as a substitute.
+%bcond_without doc_pdf
+
 Name:           python-chardet
-Version:        5.2.0
+Version:        4.0.0
 Release:        %autorelease
 Summary:        Python character encoding detector
 
-# The entire source, after tests/ are filtered, is LGPL-2.0-or-later. See the
-# comment above Source0 for more details on tests/.
-License:        LGPL-2.0-or-later
+# The entire source, after tests/ are filtered, is LGPLv2+. See the comment
+# above Source0 for more details on tests/.
+#
+# See also:
+#   clarify Trove license classifier
+#   https://github.com/chardet/chardet/issues/162
+#   Clarify LGPL version in License trove classifier
+#   https://github.com/chardet/chardet/pull/255
+License:        LGPLv2+
 URL:            https://github.com/chardet/chardet
 # A filtered source tarball, obtained by (see Source1):
 #
@@ -16,16 +28,25 @@ URL:            https://github.com/chardet/chardet
 #
 #   problematic licensing of /tests?
 #   https://github.com/chardet/chardet/issues/231
-Source0:        chardet-%{version}-filtered.tar.zst
+Source0:        chardet-%{version}-filtered.tar.xz
 Source1:        get_source
 # Hand-written for Fedora in groff_man(7) format based on --help output
 Source2:        chardetect.1
 
-BuildSystem:            pyproject
-BuildOption(install):   -l chardet
-
 BuildArch:      noarch
 
+BuildRequires:  python3-devel
+
+%if %{with doc_pdf}
+BuildRequires:  make
+BuildRequires:  python3dist(sphinx)
+BuildRequires:  python3dist(sphinx-rtd-theme)
+BuildRequires:  python3-sphinx-latex
+BuildRequires:  latexmk
+BuildRequires:  tex-xetex-bin
+BuildRequires:  /usr/bin/xindy
+%endif
+
 %global common_description %{expand:
 Chardet: The Universal Character Encoding Detector
 
@@ -53,26 +74,65 @@ disabled.}
 %package -n python3-chardet
 Summary:        %{summary}
 
-# Removed in F41:
-Obsoletes:      python-chardet-doc < 5.2.0-12
-
 %description -n python3-chardet
 %{common_description}
 
 
-%install -a
+%package        doc
+Summary:        Documentation for python-chardet
+
+%description    doc
+%{common_description}
+
+
+%prep
+%autosetup -p1 -n chardet-%{version}
+
+# Remove useless shebangs
+# https://github.com/chardet/chardet/commit/1e94b33329
+grep -lr "^#\!/usr/bin/env python" chardet/ | xargs sed -i "1d"
+
+# Since pdflatex cannot handle Unicode inputs in general:
+echo "latex_engine = 'xelatex'" >> docs/conf.py
+
+
+%generate_buildrequires
+%pyproject_buildrequires
+
+
+%build
+%pyproject_wheel
+
+%if %{with doc_pdf}
+PYTHONPATH="${PWD}" %make_build -C docs latex SPHINXOPTS='%{?_smp_mflags}'
+%make_build -C docs/_build/latex LATEXMKOPTS='-quiet'
+%endif
+
+
+%install
+%pyproject_install
+%pyproject_save_files chardet
 install -t '%{buildroot}%{_mandir}/man1' -D -p -m 0644 '%{SOURCE2}'
 
 
+%check
 # We cannot run the upstream tests because they would require data files with
 # problematic license status.
+%pyproject_check_import
 
 
 %files -n python3-chardet -f %{pyproject_files}
-%doc README.rst
 %{_bindir}/chardetect
 %{_mandir}/man1/chardetect.1*
 
 
+%files doc
+%license LICENSE
+%doc README.rst
+%if %{with doc_pdf}
+%doc docs/_build/latex/chardet.pdf
+%endif
+
+
 %changelog
 %autochangelog

sources

@@ -1 +1 @@
-SHA512 (chardet-5.2.0-filtered.tar.zst) = 410b2d62c54781f9b848ee78c659d4607bbb7f94f6117d7fa7cd5f82a50dfaf10d6e4c6dfaefa9b9a721b846c167aa12eae029debbaa66b0d830875c8ce5749c
+SHA512 (chardet-4.0.0-filtered.tar.xz) = 4c41d5f9311e6796a3270b5ecd63abc51e52962c97a7599e8756e82ba97998306e2e841e14da96441a63c446fa9142534bac7154579822b55882e621f3bbe2e0