refactor: make safe_lxml an ordinary folder in openedx/core/lib (#25689)

cms/celery.py

@@ -8,7 +8,7 @@ Taken from: https://celery.readthedocs.org/en/latest/django/first-steps-with-dja
 import os
 
 # Patch the xml libs before anything else.
-from safe_lxml import defuse_xml_libs
+from openedx.core.lib.safe_lxml import defuse_xml_libs
 
 defuse_xml_libs()
 

cms/conftest.py

@@ -16,7 +16,7 @@ import pytest
 from openedx.core.pytest_hooks import DeferPlugin
 
 # Patch the xml libs before anything else.
-from safe_lxml import defuse_xml_libs  # isort:skip  # lint-amnesty, pylint: disable=wrong-import-order
+from openedx.core.lib.safe_lxml import defuse_xml_libs  # isort:skip
 defuse_xml_libs()
 
 

cms/wsgi.py

@@ -12,7 +12,7 @@ isort:skip_file
 """
 
 # Patch the xml libs before anything else.
-from safe_lxml import defuse_xml_libs
+from openedx.core.lib.safe_lxml import defuse_xml_libs
 defuse_xml_libs()
 
 import os  # lint-amnesty, pylint: disable=wrong-import-order, wrong-import-position

common/lib/conftest.py

@@ -5,7 +5,7 @@
 
 import pytest
 
-from safe_lxml import defuse_xml_libs
+from openedx.core.lib.safe_lxml import defuse_xml_libs
 
 # This import is needed for pytest plugin configuration, so please avoid deleting this during refactoring
 from openedx.core.pytest_hooks import pytest_configure  # pylint: disable=unused-import

common/lib/safe_lxml/setup.py

@@ -1,16 +0,0 @@
-"""
-Setup.py for safe_lxml.
-"""
-
-
-from setuptools import setup
-
-setup(
-    name="safe_lxml",
-    version="1.0",
-    packages=["safe_lxml"],
-    install_requires=[
-        "lxml",
-        "defusedxml"
-    ],
-)

common/test/conftest.py

@@ -1,4 +1,6 @@
 """Code run by pylint before running any tests."""
-from safe_lxml import defuse_xml_libs
+
+# Patch the xml libs before anything else.
+from openedx.core.lib.safe_lxml import defuse_xml_libs
 
 defuse_xml_libs()

docs/guides/conf.py

@@ -21,7 +21,6 @@ root = Path('../..').abspath()
 sys.path.insert(0, root)
 sys.path.append(root / "docs/guides")
 sys.path.append(root / "common/lib/capa")
-sys.path.append(root / "common/lib/safe_lxml")
 sys.path.append(root / "common/lib/xmodule")
 
 # Use a settings module that allows all LMS and Studio code to be imported
@@ -223,7 +222,6 @@ autodoc_mock_imports = [
 modules = {
     'cms': 'cms',
     'common/lib/capa/capa': 'common/lib/capa',
-    'common/lib/safe_lxml/safe_lxml': 'common/lib/safe_lxml',
     'common/lib/xmodule/xmodule': 'common/lib/xmodule',
     'lms': 'lms',
     'openedx': 'openedx',

docs/guides/docstrings/common_lib.rst

@@ -9,5 +9,4 @@ out from edx-platform into separate packages at some point.
     :maxdepth: 2
 
     common/lib/capa/modules
-    common/lib/safe_lxml/modules
     common/lib/xmodule/modules

lms/celery.py

@@ -8,7 +8,7 @@ Taken from: https://celery.readthedocs.org/en/latest/django/first-steps-with-dja
 import os
 
 # Patch the xml libs before anything else.
-from safe_lxml import defuse_xml_libs
+from openedx.core.lib.safe_lxml import defuse_xml_libs
 
 defuse_xml_libs()
 

lms/wsgi.py

@@ -9,7 +9,7 @@ It exposes a module-level variable named ``application``. Django's
 """
 
 # Patch the xml libs
-from safe_lxml import defuse_xml_libs
+from openedx.core.lib.safe_lxml import defuse_xml_libs
 defuse_xml_libs()
 
 import os  # lint-amnesty, pylint: disable=wrong-import-order, wrong-import-position

lms/wsgi_apache_lms.py

@@ -6,7 +6,7 @@ It exposes a module-level variable named ``application``.
 """
 
 # Patch the xml libs before anything else.
-from safe_lxml import defuse_xml_libs
+from openedx.core.lib.safe_lxml import defuse_xml_libs
 defuse_xml_libs()
 
 import os  # lint-amnesty, pylint: disable=wrong-import-order, wrong-import-position

manage.py

@@ -17,7 +17,7 @@ from openedx.core.lib.logsettings import log_python_warnings
 log_python_warnings()
 
 # Patch the xml libs before anything else.
-from safe_lxml import defuse_xml_libs
+from openedx.core.lib.safe_lxml import defuse_xml_libs  # isort:skip
 defuse_xml_libs()
 
 import importlib

openedx/core/lib/safe_lxml/conftest.py

@@ -0,0 +1,7 @@
+"""
+Code run by pytest before running any tests in the safe_lxml directory.
+"""
+from openedx.core.lib.safe_lxml import defuse_xml_libs
+
+
+defuse_xml_libs()

openedx/core/lib/safe_lxml/etree.py

@@ -5,6 +5,8 @@ It overrides some unsafe functions from lxml.etree with safer versions from defu
 It also includes a safer XMLParser.
 
 For processing xml always prefer this over using lxml.etree directly.
+
+isort:skip_file
 """
 
 # Names are imported into this module so that it can be a stand-in for

openedx/core/lib/safe_lxml/tests.py

@@ -1,4 +1,12 @@
-"""Test that we have defused XML."""
+"""
+Test that we have defused XML.
+
+For these tests, the defusing will happen in one or more of the `conftest.py`
+files that runs at pytest startup calls `defuse_xml_libs()`.
+
+In production, the defusing happens when the LMS or Studio `wsgi.py` files
+call `defuse_xml_libs()`.
+"""
 
 
 import defusedxml

openedx/core/lib/xblock_builtin/xblock_discussion/tests.py

@@ -14,7 +14,7 @@ from xblock.fields import NO_CACHE_VALUE, UNIQUE_ID, ScopeIds
 from xblock.runtime import Runtime
 
 from openedx.core.lib.xblock_builtin.xblock_discussion.xblock_discussion import DiscussionXBlock
-from safe_lxml import etree  # lint-amnesty, pylint: disable=wrong-import-order
+from openedx.core.lib.safe_lxml import etree
 
 
 def attribute_pair_repr(self):

requirements/edx/base.txt

@@ -22,8 +22,6 @@
     # via -r requirements/edx/local.in
 -e git+https://github.com/edx/RateXBlock.git@2.0.1#egg=rate-xblock
     # via -r requirements/edx/github.in
--e common/lib/safe_lxml
-    # via -r requirements/edx/local.in
 -e common/lib/sandbox-packages
     # via -r requirements/edx/local.in
 -e openedx/core/lib/xblock_builtin/xblock_discussion

requirements/edx/development.txt

@@ -22,8 +22,6 @@
     # via -r requirements/edx/testing.txt
 -e git+https://github.com/edx/RateXBlock.git@2.0.1#egg=rate-xblock
     # via -r requirements/edx/testing.txt
--e common/lib/safe_lxml
-    # via -r requirements/edx/testing.txt
 -e common/lib/sandbox-packages
     # via -r requirements/edx/testing.txt
 -e openedx/core/lib/xblock_builtin/xblock_discussion

requirements/edx/local.in

@@ -1,7 +1,6 @@
 # Python libraries to install that are local to the edx-platform repo
 -e .
 -e common/lib/capa
--e common/lib/safe_lxml
 -e common/lib/sandbox-packages
 -e common/lib/xmodule
 

requirements/edx/testing.txt

@@ -22,8 +22,6 @@
     # via -r requirements/edx/base.txt
 -e git+https://github.com/edx/RateXBlock.git@2.0.1#egg=rate-xblock
     # via -r requirements/edx/base.txt
--e common/lib/safe_lxml
-    # via -r requirements/edx/base.txt
 -e common/lib/sandbox-packages
     # via -r requirements/edx/base.txt
 -e openedx/core/lib/xblock_builtin/xblock_discussion

safe_lxml/__init__.py

@@ -0,0 +1,27 @@
+"""
+Temporary import path shim module.
+
+Previously, the safe_lxml package was housed in common/lib/safe_lxml.
+It was installed as its own Python project, so instead of its import path
+being, as one would expect:
+
+    import common.lib.safe_lxml.safe_lxml
+
+it was instead just:
+
+    import safe_lxml
+
+To increase the sanity of edx-platform and simplify its tooling, we are
+moving the safe_lxml package to openedx/core/lib (in tihs same repo) and
+changing its import path to:
+
+    import openedx.core.lib.safe_lxml
+
+In order to maintain backwards-compatibility with code using the
+old import path for one release, we expose this compatibility module.
+
+Jira ticket (public, but requires account): https://openedx.atlassian.net/browse/BOM-2583
+Target removal for this shim module: by Olive.
+"""
+
+from openedx.core.lib.safe_lxml import *  # pylint: disable=unused-wildcard-import,wrong-import-order

scripts/verify-dunder-init.sh

@@ -38,7 +38,7 @@ exclude+='|^common/test/data/?.*$'
 #  * common/lib/xmodule -> EXCLUDE from check.
 #  * common/lib/xmodule/xmodule/modulestore -> INCLUDE in check.
 exclude+='|^common/lib$'
-exclude+='|^common/lib/(capa|safe_lxml|sandbox-packages|xmodule)$'
+exclude+='|^common/lib/(capa|sandbox-packages|xmodule)$'
 
 # Docs, scripts.
 exclude+='|^docs/.*$'

setup.cfg

@@ -47,4 +47,3 @@ multi_line_output=3
 skip=
     envs
     migrations
-    common/lib/safe_lxml/safe_lxml/etree.py