From 7f6d5f61df3d053b7cc392f03b12f059fb97a4a3 Mon Sep 17 00:00:00 2001
From: Charalampos Stratakis <cstratak@redhat.com>
Date: Thu, 16 Feb 2023 05:24:25 +0100
Subject: [PATCH] Remove the isoschematron submodule

Resolves: rhbz#2157708
---
 .gitignore           |  1 +
 get-lxml-source.sh   | 29 +++++++++++++++++++++++++++++
 python3.11-lxml.spec | 20 +++++++++++++++++---
 sources              |  2 +-
 4 files changed, 48 insertions(+), 4 deletions(-)
 create mode 100755 get-lxml-source.sh

diff --git a/.gitignore b/.gitignore
index f3c0026..d2660f9 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,2 +1,3 @@
 /lxml-4.9.1.tar.gz
 /lxml-4.9.2.tar.gz
+/lxml-4.9.2-no-isoschematron.tar.gz
diff --git a/get-lxml-source.sh b/get-lxml-source.sh
new file mode 100755
index 0000000..9d79f3a
--- /dev/null
+++ b/get-lxml-source.sh
@@ -0,0 +1,29 @@
+#! /bin/bash -ex
+
+# Download a release of lxml (if missing) and remove the isoschematron module from it
+
+version=$1
+
+if [ -z "${version}" ]; then
+    echo "Usage: $0 VERSION" >& 2
+    echo "" >& 2
+    echo "example: $0 4.9.2" >& 2
+    exit 1
+fi
+
+versionedname=lxml-${version}
+orig_archive=${versionedname}.tar.gz
+new_archive=${versionedname}-no-isoschematron.tar.gz
+
+if [ ! -e ${orig_archive} ]; then
+    wget -N https://files.pythonhosted.org/packages/source/l/lxml/${orig_archive}
+fi
+
+deleted_module=lxml-${version}/src/lxml/isoschematron/
+deleted_test=lxml-${version}/src/lxml/tests/test_isoschematron.py
+
+# tar --delete does not operate on compressed archives, so do
+# gz decompression explicitly
+gzip --decompress ${orig_archive}
+tar -v --delete -f ${orig_archive//.gz} {$deleted_module,$deleted_test}
+gzip -cf ${orig_archive//.gz} > ${new_archive}
diff --git a/python3.11-lxml.spec b/python3.11-lxml.spec
index 2fc2154..e3d2ccc 100644
--- a/python3.11-lxml.spec
+++ b/python3.11-lxml.spec
@@ -3,7 +3,7 @@
 
 Name:           python%{python3_pkgversion}-lxml
 Version:        4.9.2
-Release:        2%{?dist}
+Release:        3%{?dist}
 Summary:        XML processing library combining libxml2/libxslt with the ElementTree API
 
 # The lxml project is licensed under BSD-3-Clause
@@ -12,7 +12,12 @@ Summary:        XML processing library combining libxml2/libxslt with the Elemen
 # .xsl schematron files are under the MIT license
 License:        BSD and MIT
 URL:            https://github.com/lxml/lxml
-Source:         %{pypi_source lxml}
+# We use the get-lxml-source.sh script to generate the tarball
+# without the isoschematron submodule as it contains a problematic
+# license.
+# See: https://gitlab.com/fedora/legal/fedora-license-data/-/issues/154
+Source0:         lxml-%{version}-no-isoschematron.tar.gz
+Source1:         get-lxml-source.sh
 
 BuildRequires:  gcc
 BuildRequires:  libxml2-devel
@@ -34,6 +39,12 @@ XML Schema, XSLT, C14N and much more.
 %prep
 %autosetup -n lxml-%{version} -p1
 
+# Remove isoschematron module due to problematic license
+sed -i "s/, 'lxml.isoschematron'//" setup.py
+# Remove the doctests for it (the documentation is not shipped)
+# The command [d]eletes all lines from the first pattern to the second
+sed -Ei '/^Schematron$/,/^\(Pre-ISO-Schematron\)$/d' doc/validation.txt
+
 # Remove pregenerated Cython C sources
 # We need to do this after %%pyproject_buildrequires because setup.py errors
 # without Cython and without the .c files.
@@ -54,11 +65,14 @@ cp -a build/lib.%{python3_platform}-*/* src/
 
 %files -n python%{python3_pkgversion}-lxml
 %license doc/licenses/BSD.txt doc/licenses/elementtree.txt
-%doc README.rst src/lxml/isoschematron/resources/xsl/iso-schematron-xslt1/readme.txt
+%doc README.rst
 %{python3_sitearch}/lxml/
 %{python3_sitearch}/lxml-*.egg-info/
 
 %changelog
+* Thu Feb 16 2023 Charalampos Stratakis <cstratak@redhat.com> - 4.9.2-3
+- Remove the isoschematron submodule
+
 * Fri Feb 10 2023 Charalampos Stratakis <cstratak@redhat.com> - 4.9.2-2
 - Remove the extras subpackages
 
diff --git a/sources b/sources
index ff93acf..2599e86 100644
--- a/sources
+++ b/sources
@@ -1 +1 @@
-SHA512 (lxml-4.9.2.tar.gz) = 4597784c96219145fef30607f257097ad312264a6796e65b3dc74463853cd29f55379b6a903b401c1120750126b57f58dd9fc459dd453e57030883fc76b87ea3
+SHA512 (lxml-4.9.2-no-isoschematron.tar.gz) = 3ee80967dfa69c840a27ad99546727346f8bb058ed64bf93d2aa8959fb3000e78c1227f4b8709d25aaadeb9a4ccdf97a03eb53a1171de8feda93fbeaa15a8b39