[3.12] gh-121188: Sanitize invalid XML characters in regrtest (GH-121195) (#121205)

gh-121188: Sanitize invalid XML characters in regrtest (GH-121195)

When creating the JUnit XML file, regrtest now escapes characters
which are invalid in XML, such as the chr(27) control character used
in ANSI escape sequences.
(cherry picked from commit af8c3d7a26)

Co-authored-by: Victor Stinner <vstinner@python.org>

Lib/test/test_regrtest.py

@@ -21,6 +21,8 @@ import sysconfig
 import tempfile
 import textwrap
 import unittest
+from xml.etree import ElementTree
+
 from test import support
 from test.support import os_helper
 from test.libregrtest import cmdline
@@ -2221,6 +2223,44 @@ class ArgsTestCase(BaseTestCase):
             self.check_executed_tests(output, testname, stats=1, parallel=True)
             self.assertNotIn('SPAM SPAM SPAM', output)
 
+    def test_xml(self):
+        code = textwrap.dedent(r"""
+            import unittest
+            from test import support
+
+            class VerboseTests(unittest.TestCase):
+                def test_failed(self):
+                    print("abc \x1b def")
+                    self.fail()
+        """)
+        testname = self.create_test(code=code)
+
+        # Run sequentially
+        filename = os_helper.TESTFN
+        self.addCleanup(os_helper.unlink, filename)
+
+        output = self.run_tests(testname, "--junit-xml", filename,
+                                exitcode=EXITCODE_BAD_TEST)
+        self.check_executed_tests(output, testname,
+                                  failed=testname,
+                                  stats=TestStats(1, 1, 0))
+
+        # Test generated XML
+        with open(filename, encoding="utf8") as fp:
+            content = fp.read()
+
+        testsuite = ElementTree.fromstring(content)
+        self.assertEqual(int(testsuite.get('tests')), 1)
+        self.assertEqual(int(testsuite.get('errors')), 0)
+        self.assertEqual(int(testsuite.get('failures')), 1)
+
+        testcase = testsuite[0][0]
+        self.assertEqual(testcase.get('status'), 'run')
+        self.assertEqual(testcase.get('result'), 'completed')
+        self.assertGreater(float(testcase.get('time')), 0)
+        for out in testcase.iter('system-out'):
+            self.assertEqual(out.text, r"abc \x1b def")
+
 
 class TestUtils(unittest.TestCase):
     def test_format_duration(self):
@@ -2403,6 +2443,25 @@ class TestUtils(unittest.TestCase):
             self.assertTrue(match_test(test_chdir))
             self.assertFalse(match_test(test_copy))
 
+    def test_sanitize_xml(self):
+        sanitize_xml = utils.sanitize_xml
+
+        # escape invalid XML characters
+        self.assertEqual(sanitize_xml('abc \x1b\x1f def'),
+                         r'abc \x1b\x1f def')
+        self.assertEqual(sanitize_xml('nul:\x00, bell:\x07'),
+                         r'nul:\x00, bell:\x07')
+        self.assertEqual(sanitize_xml('surrogate:\uDC80'),
+                         r'surrogate:\udc80')
+        self.assertEqual(sanitize_xml('illegal \uFFFE and \uFFFF'),
+                         r'illegal \ufffe and \uffff')
+
+        # no escape for valid XML characters
+        self.assertEqual(sanitize_xml('a\n\tb'),
+                         'a\n\tb')
+        self.assertEqual(sanitize_xml('valid t\xe9xt \u20ac'),
+                         'valid t\xe9xt \u20ac')
+
 
 if __name__ == '__main__':
     unittest.main()