Add deallocator for str subclasses

Author: msimacek

graalpython/com.oracle.graal.python.cext/src/capi.c

@@ -103,6 +103,8 @@ typedef struct {
     PyObject *weakreflist;
 } PyPickleBufferObject;
 
+// defined in 'unicodeobject.c'
+void unicode_dealloc(PyObject *unicode);
 
 static void object_dealloc(PyObject *self) {
     Py_TYPE(self)->tp_free(self);

graalpython/com.oracle.graal.python.cext/src/capi.h

@@ -326,7 +326,7 @@ PY_TRUFFLE_TYPE(PyStaticMethod_Type, 	"staticmethod", 			&PyType_Type, sizeof(Py
 PY_TRUFFLE_TYPE(PySuper_Type, 			"super", 					&PyType_Type, sizeof(superobject)) \
 PY_TRUFFLE_TYPE(PyTraceBack_Type, 		"traceback", 				&PyType_Type, sizeof(PyTypeObject)) \
 PY_TRUFFLE_TYPE_GENERIC(PyTuple_Type, 	"tuple", 					&PyType_Type, sizeof(PyTupleObject) - sizeof(PyObject *), sizeof(PyObject *), PyTruffle_Tuple_Alloc, (destructor)PyTruffle_Tuple_Dealloc, 0, 0) \
-PY_TRUFFLE_TYPE(PyUnicode_Type, 		"str", 					    &PyType_Type, sizeof(PyUnicodeObject)) \
+PY_TRUFFLE_TYPE_GENERIC(PyUnicode_Type,	"str", 					    &PyType_Type, sizeof(PyUnicodeObject), 0, NULL, unicode_dealloc, PyObject_Del, 0) \
 /* NOTE: we use the same Python type (namely 'PBuiltinFunction') for 'wrapper_descriptor' as for 'method_descriptor'; so the flags must be the same! */ \
 PY_TRUFFLE_TYPE(PyWrapperDescr_Type, 	"wrapper_descriptor",       &PyType_Type, sizeof(PyWrapperDescrObject)) \
 PY_TRUFFLE_TYPE(PyZip_Type, 			"zip", 					    &PyType_Type, sizeof(zipobject)) \

graalpython/com.oracle.graal.python.cext/src/unicodeobject.c

@@ -119,7 +119,6 @@ extern "C" {
 #  define _PyUnicode_CHECK(op) PyUnicode_Check(op)
 #endif
 
-#if 0 // GraalPy change
 #define _PyUnicode_UTF8(op)                             \
     (_PyCompactUnicodeObject_CAST(op)->utf8)
 #define PyUnicode_UTF8(op)                              \
@@ -190,7 +189,6 @@ extern "C" {
     ((_PyUnicode_WSTR(op) &&                            \
       (!PyUnicode_IS_READY(op) ||                       \
        _PyUnicode_WSTR(op) != PyUnicode_DATA(op))))
-#endif // GraalPy change
 
 /* Generic helper macro to convert characters of different types.
    from_type and to_type have to be valid type names, begin and end
@@ -1715,8 +1713,8 @@ _PyUnicode_Ready(PyObject *unicode)
     return 0;
 }
 
-#if 0 // GraalPy change
-static void
+// GraalPy change: export
+PyAPI_FUNC(void)
 unicode_dealloc(PyObject *unicode)
 {
 #ifdef Py_DEBUG
@@ -1725,6 +1723,7 @@ unicode_dealloc(PyObject *unicode)
     }
 #endif
 
+#if 0 // GraalPy change
     switch (PyUnicode_CHECK_INTERNED(unicode)) {
     case SSTATE_NOT_INTERNED:
         break;
@@ -1753,6 +1752,7 @@ unicode_dealloc(PyObject *unicode)
     default:
         Py_UNREACHABLE();
     }
+#endif // GraalPy change
 
     if (_PyUnicode_HAS_WSTR_MEMORY(unicode)) {
         PyObject_Free(_PyUnicode_WSTR(unicode));
@@ -1767,6 +1767,7 @@ unicode_dealloc(PyObject *unicode)
     Py_TYPE(unicode)->tp_free(unicode);
 }
 
+#if 0 // GraalPy change
 #ifdef Py_DEBUG
 static int
 unicode_is_singleton(PyObject *unicode)
@@ -14466,24 +14467,6 @@ unicode_new_impl(PyTypeObject *type, PyObject *x, const char *encoding,
 PyAPI_FUNC(PyObject *) // GraalPy change: export for downcall
 unicode_subtype_new(PyTypeObject *type, PyObject *unicode)
 {
-    // GraalPy change: temporarily define struct access macros
-#define _PyUnicode_STATE(op)                            \
-    (_PyASCIIObject_CAST(op)->state)
-#define _PyUnicode_DATA_ANY(op)                         \
-    (_PyUnicodeObject_CAST(op)->data.any)
-#define _PyUnicode_LENGTH(op)                           \
-    (_PyASCIIObject_CAST(op)->length)
-#define _PyUnicode_HASH(op)                             \
-    (_PyASCIIObject_CAST(op)->hash)
-#define _PyUnicode_UTF8(op)                             \
-    (_PyCompactUnicodeObject_CAST(op)->utf8)
-#define _PyUnicode_UTF8_LENGTH(op)                      \
-    (_PyCompactUnicodeObject_CAST(op)->utf8_length)
-#define _PyUnicode_WSTR(op)                             \
-    (_PyASCIIObject_CAST(op)->wstr)
-#define _PyUnicode_WSTR_LENGTH(op)                      \
-    (_PyCompactUnicodeObject_CAST(op)->wstr_length)
-
     PyObject *self;
     Py_ssize_t length, char_size;
     int share_wstr, share_utf8;
@@ -14569,15 +14552,6 @@ unicode_subtype_new(PyTypeObject *type, PyObject *unicode)
 onError:
     Py_DECREF(self);
     return NULL;
-// GraalPy change
-#undef _PyUnicode_STATE
-#undef _PyUnicode_DATA_ANY
-#undef _PyUnicode_LENGTH
-#undef _PyUnicode_HASH
-#undef _PyUnicode_UTF8
-#undef _PyUnicode_UTF8_LENGTH
-#undef _PyUnicode_WSTR
-#undef _PyUnicode_WSTR_LENGTH
 }
 
 #if 0 // GraalPy change

graalpython/com.oracle.graal.python.test/src/tests/cpyext/test_unicode.py

@@ -224,7 +224,6 @@ def gen_intern_args():
     "UnicodeSubclass",
     '',
     struct_base='PyUnicodeObject base;',
-    tp_itemsize='sizeof(char)',
     tp_base='&PyUnicode_Type',
     tp_new='0',
     tp_alloc='0',

graalpython/com.oracle.graal.python/src/com/oracle/graal/python/builtins/modules/cext/PythonCextUnicodeBuiltins.java

@@ -1050,8 +1050,7 @@ static Object doNative(PythonAbstractNativeObject s,
             if (lib.isNull(utf8)) {
                 PBytes bytes = (PBytes) asUTF8String.execute(s, T_STRICT);
                 int len = bufferLib.getBufferLength(bytes);
-                // TODO leaked?
-                Object mem = allocateNode.alloc(len + 1);
+                Object mem = allocateNode.alloc(len + 1, true);
                 writeByteNode.writeByteArray(mem, bufferLib.getInternalByteArray(bytes), len, 0, 0);
                 writePointerNode.writeToObj(s, CFields.PyCompactUnicodeObject__utf8, mem);
                 writeLongNode.writeToObject(s, CFields.PyCompactUnicodeObject__utf8_length, len);