bpo-47182: Fix crash by named unicode characters after interpreter reinitialization...

Automerge-Triggered-By: GH:tiran

diff --git a/Lib/test/test_embed.py b/Lib/test/test_embed.py
index f0c88de68e89eb16d6586f2a1a00bef27fc007c3..7e5e4c144851ee3eed224916d6658b0cd0132594 100644 (file)
--- a/Lib/test/test_embed.py
+++ b/Lib/test/test_embed.py
@@ -343,6 +343,11 @@ class EmbeddingTests(EmbeddingTestsMixin, unittest.TestCase):
         out, err = self.run_embedded_interpreter("test_repeated_init_exec", code)
         self.assertEqual(out, 'Tests passed\n' * INIT_LOOPS)
 
+    def test_ucnhash_capi_reset(self):
+        # bpo-47182: unicodeobject.c:ucnhash_capi was not reset on shutdown.
+        code = "print('\\N{digit nine}')"
+        out, err = self.run_embedded_interpreter("test_repeated_init_exec", code)
+        self.assertEqual(out, '9\n' * INIT_LOOPS)
 
 class InitConfigTests(EmbeddingTestsMixin, unittest.TestCase):
     maxDiff = 4096

diff --git a/Misc/NEWS.d/next/Core and Builtins/2022-03-31-15-37-02.bpo-47182.e_4SsC.rst b/Misc/NEWS.d/next/Core and Builtins/2022-03-31-15-37-02.bpo-47182.e_4SsC.rst
new file mode 100644 (file)
index 0000000..08036bc
--- /dev/null
+++ b/Misc/NEWS.d/next/Core and Builtins/2022-03-31-15-37-02.bpo-47182.e_4SsC.rst
@@ -0,0 +1,2 @@
+Fix a crash when using a named unicode character like ``"\N{digit nine}"``
+after the main interpreter has been initialized a second time.

diff --git a/Objects/unicodeobject.c b/Objects/unicodeobject.c
index 5a1d2c063016724420898f557a96347868dfca95..2d4096397784f7f401639eddb19e25488467b4ec 100644 (file)
--- a/Objects/unicodeobject.c
+++ b/Objects/unicodeobject.c
@@ -16085,6 +16085,9 @@ _PyUnicode_Fini(PyInterpreterState *interp)
     if (_Py_IsMainInterpreter(interp)) {
         // _PyUnicode_ClearInterned() must be called before _PyUnicode_Fini()
         assert(interned == NULL);
+        // bpo-47182: force a unicodedata CAPI capsule re-import on
+        // subsequent initialization of main interpreter.
+        ucnhash_capi = NULL;
     }
 
     _PyUnicode_FiniEncodings(&state->fs_codec);