Add python check (#955)

* Add PythonCheck (#423)

This adds a Python specific check that performs 2 tasks:

1) Verify the egg-info is folder (setuptools based) and not file based
   (distutils style)

2) Check for Python specific forbidden dirs and output various warnings and
   errors about them.

* PythonCheck: remove src-in-package check

* PythonCheck: extend test cases to /usr/lib64

* PythonCheck: ignore tests in -test{,s} packages.

* Update rpmlint/descriptions/PythonCheck.toml

Co-authored-by: Miro Hrončok <miro@hroncok.cz>

* PythonCheck: update to the latest codebase

* Rename test file to be able to use get_tested_package

* Improve PythonCheck error descrpitions

Co-authored-by: Miro Hrončok <miro@hroncok.cz>

* PythonCheck: juse endswith with tuple instead of or

* PythonCheck: Use pathlib.Path instead of os.path

* PythonCheck: recommend dist-info instead of egg-info

* PythonCheck: compile re at module level

* fixup! PythonCheck: Use pathlib.Path instead of os.path

* PythonCheck: Update test binary packages

These binary packages are generated from the following source spec:
https://build.opensuse.org/package/show/devel:openSUSE:Factory:rpmlint:tests/pythoncheck

* PythonCheck: Convert python-tests-in-package to Error

* PythonCheck: Do not warn on python modules named doc

* PythonCheck: Improve python-tests-in-package descriptions

Co-authored-by: Miro Hrončok <miro@hroncok.cz>

* PythonCheck: Improve python-doc-in-package descriptions

Co-authored-by: Miro Hrončok <miro@hroncok.cz>

* PythonCheck: Improve python-egg-info-distutils-style descriptions

Co-authored-by: Miro Hrončok <miro@hroncok.cz>

Co-authored-by: Johannes Grassler <johannes.grassler@suse.com>
Co-authored-by: Johannes Grassler <jgr-github@btw23.de>
Co-authored-by: Miro Hrončok <miro@hroncok.cz>

rpmlint/checks/PythonCheck.py

@@ -0,0 +1,76 @@
+from pathlib import Path
+import re
+
+from rpmlint.checks.AbstractCheck import AbstractFilesCheck
+
+# Warning messages
+WARNS = {
+    'doc': 'python-doc-in-package',
+}
+
+# Error messages
+ERRS = {
+    'egg-distutils': 'python-egg-info-distutils-style',
+    'tests': 'python-tests-in-site-packages',
+    'doc': 'python-doc-in-site-packages',
+    'src': 'python-src-in-site-packages',
+    'tests-package': 'python-tests-in-package',
+}
+
+SITELIB_RE = '/usr/lib[^/]*/python[^/]*/site-packages'
+
+# Paths that shouldn't be in any packages, ever, because they clobber global
+# name space.
+ERR_PATHS = [
+    (re.compile(f'{SITELIB_RE}/tests?$'), 'tests'),
+    (re.compile(f'{SITELIB_RE}/docs?$'), 'doc'),
+    (re.compile(f'{SITELIB_RE}/src$'), 'src'),
+    (re.compile(f'{SITELIB_RE}/[^/]+/tests?$'), 'tests-package'),
+]
+
+# Paths that shouldn't be in any packages, but might need to be under
+# sufficiently special circumstances.
+WARN_PATHS = [
+    (re.compile(f'{SITELIB_RE}/[^/]+/docs?$'), 'doc'),
+]
+
+
+class PythonCheck(AbstractFilesCheck):
+    def __init__(self, config, output):
+        super().__init__(config, output, r'.*')
+
+    def check_file(self, pkg, filename):
+        egg_info_re = re.compile('.*egg-info$')
+
+        if egg_info_re.match(filename):
+            self.check_egginfo(pkg, filename)
+
+        for path_re, key in WARN_PATHS:
+            if path_re.match(filename):
+                if key == 'doc':
+                    # Check for __init__.py file inside doc, maybe this is a
+                    # module, not documentation
+                    module_file = f'{filename}/__init__.py'
+                    if module_file in pkg.files.keys():
+                        continue
+                self.output.add_info('W', pkg, WARNS[key], filename)
+
+        for path_re, key in ERR_PATHS:
+            if path_re.match(filename):
+                if key == 'tests-package':
+                    # Ignore "-test" and "-tests" packages since these are
+                    # supposed to contain tests.
+                    if pkg.name.endswith(('test', 'tests')):
+                        continue
+                self.output.add_info('E', pkg, ERRS[key], filename)
+
+    def check_egginfo(self, pkg, filename):
+        """
+        Check type of egg-info metadata and check Requires against egg-info
+        metadata if applicable.
+        """
+
+        filepath = Path(pkg.dir_name() or '/', filename.lstrip('/'))
+        # Check for (deprecated) distutils style metadata.
+        if filepath.is_file():
+            self.output.add_info('E', pkg, ERRS['egg-distutils'], filename)

rpmlint/configdefaults.toml

@@ -20,6 +20,7 @@ Checks = [
     "MixedOwnershipCheck",
     "PkgConfigCheck",
     "PostCheck",
+    "PythonCheck",
     "SignatureCheck",
     "SourceCheck",
     "SpecCheck",

rpmlint/descriptions/PythonCheck.toml

@@ -0,0 +1,20 @@
+python-tests-in-package="""
+test/ or tests/ directory in Python package directory. Tests in %{python_sitelib}/<pkgname> should be packaged in a separate test or tests subpackage or not packaged."""
+python-doc-in-package="""
+doc/ or docs/ directory in Python package directory. Documentation should go into %{docdir}, not %{python_sitelib}/<pkgname>"""
+python-egg-info-distutils-style="""
+The Python package's egg-info is a distutils style file.
+Please update to dist-info standardized core metadata.
+"""
+python-tests-in-site-packages="""
+test/ or tests/ directory in %{python_sitelib}. This should never happen since
+this is a global name space not owned by any particular package.
+"""
+python-doc-in-site-packages="""
+doc/ or docs directory installed to %{python_sitelib}. This should never happen
+since this is a global name space not owned by any particular package.
+"""
+python-src-in-site-packages="""
+src/ directory installed to %{python_sitelib}. This should never happen
+since this is a global name space not owned by any particular package.
+"""

test/test_lint.py

@@ -48,6 +48,7 @@ basic_tests = [
     'MixedOwnershipCheck',
     'PkgConfigCheck',
     'PostCheck',
+    'PythonCheck',
     'SignatureCheck',
     'SourceCheck',
     'SpecCheck',

test/test_python.py

@@ -0,0 +1,125 @@
+import pytest
+from rpmlint.checks.PythonCheck import PythonCheck
+from rpmlint.filter import Filter
+
+from Testing import CONFIG, get_tested_package
+
+
+@pytest.fixture(scope='function', autouse=True)
+def pythoncheck():
+    CONFIG.info = True
+    output = Filter(CONFIG)
+    test = PythonCheck(CONFIG, output)
+    return output, test
+
+
+@pytest.mark.parametrize('package', ['binary/pythoncheck-python-doc-in-package'])
+def test_python_doc_in_package(tmpdir, package, pythoncheck):
+    output, test = pythoncheck
+    test.check(get_tested_package(package, tmpdir))
+    out = output.print_results(output.results)
+    assert 'W: python-doc-in-package /usr/lib/python2.7/site-packages/python-mypackage/doc' in out
+    assert 'W: python-doc-in-package /usr/lib/python2.7/site-packages/python-mypackage/docs' in out
+    assert 'W: python-doc-in-package /usr/lib/python3.10/site-packages/python-mypackage/doc' in out
+    assert 'W: python-doc-in-package /usr/lib/python3.10/site-packages/python-mypackage/docs' in out
+    assert 'W: python-doc-in-package /usr/lib64/python2.7/site-packages/python-mypackage/doc' in out
+    assert 'W: python-doc-in-package /usr/lib64/python2.7/site-packages/python-mypackage/docs' in out
+    assert 'W: python-doc-in-package /usr/lib64/python3.10/site-packages/python-mypackage/doc' in out
+    assert 'W: python-doc-in-package /usr/lib64/python3.10/site-packages/python-mypackage/docs' in out
+
+
+@pytest.mark.parametrize('package', ['binary/pythoncheck-python-doc-module-in-package'])
+def test_python_doc_module_in_package(tmpdir, package, pythoncheck):
+    output, test = pythoncheck
+    test.check(get_tested_package(package, tmpdir))
+    out = output.print_results(output.results)
+    assert 'W: python-doc-in-package /usr/lib/python2.7/site-packages/python-mypackage/doc' not in out
+    assert 'W: python-doc-in-package /usr/lib/python2.7/site-packages/python-mypackage/docs' not in out
+    assert 'W: python-doc-in-package /usr/lib/python3.10/site-packages/python-mypackage/doc' not in out
+    assert 'W: python-doc-in-package /usr/lib/python3.10/site-packages/python-mypackage/docs' not in out
+    assert 'W: python-doc-in-package /usr/lib64/python2.7/site-packages/python-mypackage/doc' not in out
+    assert 'W: python-doc-in-package /usr/lib64/python2.7/site-packages/python-mypackage/docs' not in out
+    assert 'W: python-doc-in-package /usr/lib64/python3.10/site-packages/python-mypackage/doc' not in out
+    assert 'W: python-doc-in-package /usr/lib64/python3.10/site-packages/python-mypackage/docs' not in out
+
+
+@pytest.mark.parametrize('package', ['binary/pythoncheck-python-tests-in-package2'])
+def test_python_tests_in_package(tmpdir, package, pythoncheck):
+    output, test = pythoncheck
+    test.check(get_tested_package(package, tmpdir))
+    out = output.print_results(output.results)
+    assert 'E: python-tests-in-package /usr/lib/python2.7/site-packages/python-mypackage/test' in out
+    assert 'E: python-tests-in-package /usr/lib/python2.7/site-packages/python-mypackage/tests' in out
+    assert 'E: python-tests-in-package /usr/lib/python3.10/site-packages/python-mypackage/test' in out
+    assert 'E: python-tests-in-package /usr/lib/python3.10/site-packages/python-mypackage/tests' in out
+    assert 'E: python-tests-in-package /usr/lib64/python2.7/site-packages/python-mypackage/test' in out
+    assert 'E: python-tests-in-package /usr/lib64/python2.7/site-packages/python-mypackage/tests' in out
+    assert 'E: python-tests-in-package /usr/lib64/python3.10/site-packages/python-mypackage/test' in out
+    assert 'E: python-tests-in-package /usr/lib64/python3.10/site-packages/python-mypackage/tests' in out
+
+
+@pytest.mark.parametrize('package', ['binary/pythoncheck-python-tests-in-package-test'])
+def test_python_tests_in_test_package(tmpdir, package, pythoncheck):
+    output, test = pythoncheck
+    test.check(get_tested_package(package, tmpdir))
+    out = output.print_results(output.results)
+    assert 'E: python-tests-in-package /usr/lib/python2.7/site-packages/python-mypackage/test' not in out
+    assert 'E: python-tests-in-package /usr/lib/python2.7/site-packages/python-mypackage/tests' not in out
+    assert 'E: python-tests-in-package /usr/lib/python3.10/site-packages/python-mypackage/test' not in out
+    assert 'E: python-tests-in-package /usr/lib/python3.10/site-packages/python-mypackage/tests' not in out
+    assert 'E: python-tests-in-package /usr/lib64/python2.7/site-packages/python-mypackage/test' not in out
+    assert 'E: python-tests-in-package /usr/lib64/python2.7/site-packages/python-mypackage/tests' not in out
+    assert 'E: python-tests-in-package /usr/lib64/python3.10/site-packages/python-mypackage/test' not in out
+    assert 'E: python-tests-in-package /usr/lib64/python3.10/site-packages/python-mypackage/tests' not in out
+
+
+@pytest.mark.parametrize('package', ['binary/pythoncheck-python-egg-info-distutils-style'])
+def test_python_distutils_egg_info(tmpdir, package, pythoncheck):
+    output, test = pythoncheck
+    test.check(get_tested_package(package, tmpdir))
+    out = output.print_results(output.results)
+    assert 'E: python-egg-info-distutils-style /usr/lib/python2.7/site-packages/mydistutilspackage.egg-info' in out
+    assert 'E: python-egg-info-distutils-style /usr/lib/python3.10/site-packages/mydistutilspackage.egg-info' in out
+    assert 'E: python-egg-info-distutils-style /usr/lib64/python2.7/site-packages/mydistutilspackage.egg-info' in out
+    assert 'E: python-egg-info-distutils-style /usr/lib64/python3.10/site-packages/mydistutilspackage.egg-info' in out
+
+
+@pytest.mark.parametrize('package', ['binary/pythoncheck-python-doc-in-site-packages'])
+def test_python_doc_in_site_packages(tmpdir, package, pythoncheck):
+    output, test = pythoncheck
+    test.check(get_tested_package(package, tmpdir))
+    out = output.print_results(output.results)
+    assert 'E: python-doc-in-site-packages /usr/lib/python2.7/site-packages/doc' in out
+    assert 'E: python-doc-in-site-packages /usr/lib/python2.7/site-packages/docs' in out
+    assert 'E: python-doc-in-site-packages /usr/lib/python3.10/site-packages/doc' in out
+    assert 'E: python-doc-in-site-packages /usr/lib/python3.10/site-packages/docs' in out
+    assert 'E: python-doc-in-site-packages /usr/lib64/python2.7/site-packages/doc' in out
+    assert 'E: python-doc-in-site-packages /usr/lib64/python2.7/site-packages/docs' in out
+    assert 'E: python-doc-in-site-packages /usr/lib64/python3.10/site-packages/doc' in out
+    assert 'E: python-doc-in-site-packages /usr/lib64/python3.10/site-packages/docs' in out
+
+
+@pytest.mark.parametrize('package', ['binary/pythoncheck-python-src-in-site-packages'])
+def test_python_src_in_site_packages(tmpdir, package, pythoncheck):
+    output, test = pythoncheck
+    test.check(get_tested_package(package, tmpdir))
+    out = output.print_results(output.results)
+    assert 'E: python-src-in-site-packages /usr/lib/python2.7/site-packages/src' in out
+    assert 'E: python-src-in-site-packages /usr/lib/python3.10/site-packages/src' in out
+    assert 'E: python-src-in-site-packages /usr/lib64/python2.7/site-packages/src' in out
+    assert 'E: python-src-in-site-packages /usr/lib64/python3.10/site-packages/src' in out
+
+
+@pytest.mark.parametrize('package', ['binary/pythoncheck-python-tests-in-site-packages'])
+def test_python_tests_in_site_packages(tmpdir, package, pythoncheck):
+    output, test = pythoncheck
+    test.check(get_tested_package(package, tmpdir))
+    out = output.print_results(output.results)
+    assert 'E: python-tests-in-site-packages /usr/lib/python2.7/site-packages/test' in out
+    assert 'E: python-tests-in-site-packages /usr/lib/python2.7/site-packages/tests' in out
+    assert 'E: python-tests-in-site-packages /usr/lib/python3.10/site-packages/test' in out
+    assert 'E: python-tests-in-site-packages /usr/lib/python3.10/site-packages/tests' in out
+    assert 'E: python-tests-in-site-packages /usr/lib64/python2.7/site-packages/test' in out
+    assert 'E: python-tests-in-site-packages /usr/lib64/python2.7/site-packages/tests' in out
+    assert 'E: python-tests-in-site-packages /usr/lib64/python3.10/site-packages/test' in out
+    assert 'E: python-tests-in-site-packages /usr/lib64/python3.10/site-packages/tests' in out