gh-101819: Adapt _io types to heap types, batch 1 (GH-101949)

Adapt StringIO, TextIOWrapper, FileIO, Buffered*, and BytesIO types.

Automerge-Triggered-By: GH:erlend-aasland

Author: erlend-aasland

Lib/test/test_io.py

@@ -1042,6 +1042,95 @@ def close(self):
         support.gc_collect()
         self.assertIsNone(wr(), wr)
 
+@support.cpython_only
+class TestIOCTypes(unittest.TestCase):
+    def setUp(self):
+        _io = import_helper.import_module("_io")
+        self.types = [
+            _io.BufferedRWPair,
+            _io.BufferedRandom,
+            _io.BufferedReader,
+            _io.BufferedWriter,
+            _io.BytesIO,
+            _io.FileIO,
+            _io.IncrementalNewlineDecoder,
+            _io.StringIO,
+            _io.TextIOWrapper,
+            _io._BufferedIOBase,
+            _io._BytesIOBuffer,
+            _io._IOBase,
+            _io._RawIOBase,
+            _io._TextIOBase,
+        ]
+        if sys.platform == "win32":
+            self.types.append(_io._WindowsConsoleIO)
+        self._io = _io
+
+    def test_immutable_types(self):
+        for tp in self.types:
+            with self.subTest(tp=tp):
+                with self.assertRaisesRegex(TypeError, "immutable"):
+                    tp.foo = "bar"
+
+    def test_class_hierarchy(self):
+        def check_subs(types, base):
+            for tp in types:
+                with self.subTest(tp=tp, base=base):
+                    self.assertTrue(issubclass(tp, base))
+
+        def recursive_check(d):
+            for k, v in d.items():
+                if isinstance(v, dict):
+                    recursive_check(v)
+                elif isinstance(v, set):
+                    check_subs(v, k)
+                else:
+                    self.fail("corrupt test dataset")
+
+        _io = self._io
+        hierarchy = {
+            _io._IOBase: {
+                _io._BufferedIOBase: {
+                    _io.BufferedRWPair,
+                    _io.BufferedRandom,
+                    _io.BufferedReader,
+                    _io.BufferedWriter,
+                    _io.BytesIO,
+                },
+                _io._RawIOBase: {
+                    _io.FileIO,
+                },
+                _io._TextIOBase: {
+                    _io.StringIO,
+                    _io.TextIOWrapper,
+                },
+            },
+        }
+        if sys.platform == "win32":
+            hierarchy[_io._IOBase][_io._RawIOBase].add(_io._WindowsConsoleIO)
+
+        recursive_check(hierarchy)
+
+    def test_subclassing(self):
+        _io = self._io
+        dataset = {k: True for k in self.types}
+        dataset[_io._BytesIOBuffer] = False
+
+        for tp, is_basetype in dataset.items():
+            with self.subTest(tp=tp, is_basetype=is_basetype):
+                name = f"{tp.__name__}_subclass"
+                bases = (tp,)
+                if is_basetype:
+                    _ = type(name, bases, {})
+                else:
+                    msg = "not an acceptable base type"
+                    with self.assertRaisesRegex(TypeError, msg):
+                        _ = type(name, bases, {})
+
+    def test_disallow_instantiation(self):
+        _io = self._io
+        support.check_disallow_instantiation(self, _io._BytesIOBuffer)
+
 class PyIOTest(IOTest):
     pass
 
@@ -4671,7 +4760,7 @@ def load_tests(loader, tests, pattern):
              CIncrementalNewlineDecoderTest, PyIncrementalNewlineDecoderTest,
              CTextIOWrapperTest, PyTextIOWrapperTest,
              CMiscIOTest, PyMiscIOTest,
-             CSignalsTest, PySignalsTest,
+             CSignalsTest, PySignalsTest, TestIOCTypes,
              )
 
     # Put the namespaces of the IO module we are testing and some useful mock

Modules/_io/_iomodule.c

@@ -10,7 +10,6 @@
 #define PY_SSIZE_T_CLEAN
 #include "Python.h"
 #include "_iomodule.h"
-#include "pycore_moduleobject.h"  // _PyModule_GetState()
 #include "pycore_pystate.h"       // _PyInterpreterState_GET()
 
 #ifdef HAVE_SYS_TYPES_H
@@ -315,8 +314,9 @@ _io_open_impl(PyObject *module, PyObject *file, const char *mode,
     }
 
     /* Create the Raw file stream */
+    _PyIO_State *state = get_io_state(module);
     {
-        PyObject *RawIO_class = (PyObject *)&PyFileIO_Type;
+        PyObject *RawIO_class = (PyObject *)state->PyFileIO_Type;
 #ifdef MS_WINDOWS
         const PyConfig *config = _Py_GetConfig();
         if (!config->legacy_windows_stdio && _PyIO_get_console_type(path_or_fd) != '\0') {
@@ -390,12 +390,15 @@ _io_open_impl(PyObject *module, PyObject *file, const char *mode,
     {
         PyObject *Buffered_class;
 
-        if (updating)
-            Buffered_class = (PyObject *)&PyBufferedRandom_Type;
-        else if (creating || writing || appending)
-            Buffered_class = (PyObject *)&PyBufferedWriter_Type;
-        else if (reading)
-            Buffered_class = (PyObject *)&PyBufferedReader_Type;
+        if (updating) {
+            Buffered_class = (PyObject *)state->PyBufferedRandom_Type;
+        }
+        else if (creating || writing || appending) {
+            Buffered_class = (PyObject *)state->PyBufferedWriter_Type;
+        }
+        else if (reading) {
+            Buffered_class = (PyObject *)state->PyBufferedReader_Type;
+        }
         else {
             PyErr_Format(PyExc_ValueError,
                          "unknown mode: '%s'", mode);
@@ -417,7 +420,7 @@ _io_open_impl(PyObject *module, PyObject *file, const char *mode,
     }
 
     /* wraps into a TextIOWrapper */
-    wrapper = PyObject_CallFunction((PyObject *)&PyTextIOWrapper_Type,
+    wrapper = PyObject_CallFunction((PyObject *)state->PyTextIOWrapper_Type,
                                     "OsssO",
                                     buffer,
                                     encoding, errors, newline,
@@ -558,14 +561,6 @@ PyNumber_AsOff_t(PyObject *item, PyObject *err)
     return result;
 }
 
-static inline _PyIO_State*
-get_io_state(PyObject *module)
-{
-    void *state = _PyModule_GetState(module);
-    assert(state != NULL);
-    return (_PyIO_State *)state;
-}
-
 _PyIO_State *
 _PyIO_get_module_state(void)
 {
@@ -587,6 +582,15 @@ iomodule_traverse(PyObject *mod, visitproc visit, void *arg) {
         return 0;
     Py_VISIT(state->locale_module);
     Py_VISIT(state->unsupported_operation);
+
+    Py_VISIT(state->PyBufferedRWPair_Type);
+    Py_VISIT(state->PyBufferedRandom_Type);
+    Py_VISIT(state->PyBufferedReader_Type);
+    Py_VISIT(state->PyBufferedWriter_Type);
+    Py_VISIT(state->PyBytesIO_Type);
+    Py_VISIT(state->PyFileIO_Type);
+    Py_VISIT(state->PyStringIO_Type);
+    Py_VISIT(state->PyTextIOWrapper_Type);
     return 0;
 }
 
@@ -599,6 +603,15 @@ iomodule_clear(PyObject *mod) {
     if (state->locale_module != NULL)
         Py_CLEAR(state->locale_module);
     Py_CLEAR(state->unsupported_operation);
+
+    Py_CLEAR(state->PyBufferedRWPair_Type);
+    Py_CLEAR(state->PyBufferedRandom_Type);
+    Py_CLEAR(state->PyBufferedReader_Type);
+    Py_CLEAR(state->PyBufferedWriter_Type);
+    Py_CLEAR(state->PyBytesIO_Type);
+    Py_CLEAR(state->PyFileIO_Type);
+    Py_CLEAR(state->PyStringIO_Type);
+    Py_CLEAR(state->PyTextIOWrapper_Type);
     return 0;
 }
 
@@ -612,7 +625,9 @@ iomodule_free(PyObject *mod) {
  * Module definition
  */
 
+#define clinic_state() (get_io_state(module))
 #include "clinic/_iomodule.c.h"
+#undef clinic_state
 
 static PyMethodDef module_methods[] = {
     _IO_OPEN_METHODDEF
@@ -644,23 +659,11 @@ static PyTypeObject* static_types[] = {
     &PyRawIOBase_Type,
     &PyTextIOBase_Type,
 
-    // PyBufferedIOBase_Type(PyIOBase_Type) subclasses
-    &PyBytesIO_Type,
-    &PyBufferedReader_Type,
-    &PyBufferedWriter_Type,
-    &PyBufferedRWPair_Type,
-    &PyBufferedRandom_Type,
-
     // PyRawIOBase_Type(PyIOBase_Type) subclasses
-    &PyFileIO_Type,
     &_PyBytesIOBuffer_Type,
 #ifdef MS_WINDOWS
     &PyWindowsConsoleIO_Type,
 #endif
-
-    // PyTextIOBase_Type(PyIOBase_Type) subclasses
-    &PyStringIO_Type,
-    &PyTextIOWrapper_Type,
 };
 
 
@@ -673,6 +676,17 @@ _PyIO_Fini(void)
     }
 }
 
+#define ADD_TYPE(module, type, spec, base)                               \
+do {                                                                     \
+    type = (PyTypeObject *)PyType_FromModuleAndSpec(module, spec,        \
+                                                    (PyObject *)base);   \
+    if (type == NULL) {                                                  \
+        goto fail;                                                       \
+    }                                                                    \
+    if (PyModule_AddType(module, type) < 0) {                            \
+        goto fail;                                                       \
+    }                                                                    \
+} while (0)
 
 PyMODINIT_FUNC
 PyInit__io(void)
@@ -705,17 +719,9 @@ PyInit__io(void)
     }
 
     // Set type base classes
-    PyFileIO_Type.tp_base = &PyRawIOBase_Type;
-    PyBytesIO_Type.tp_base = &PyBufferedIOBase_Type;
-    PyStringIO_Type.tp_base = &PyTextIOBase_Type;
 #ifdef MS_WINDOWS
     PyWindowsConsoleIO_Type.tp_base = &PyRawIOBase_Type;
 #endif
-    PyBufferedReader_Type.tp_base = &PyBufferedIOBase_Type;
-    PyBufferedWriter_Type.tp_base = &PyBufferedIOBase_Type;
-    PyBufferedRWPair_Type.tp_base = &PyBufferedIOBase_Type;
-    PyBufferedRandom_Type.tp_base = &PyBufferedIOBase_Type;
-    PyTextIOWrapper_Type.tp_base = &PyTextIOBase_Type;
 
     // Add types
     for (size_t i=0; i < Py_ARRAY_LENGTH(static_types); i++) {
@@ -725,6 +731,25 @@ PyInit__io(void)
         }
     }
 
+    // PyBufferedIOBase_Type(PyIOBase_Type) subclasses
+    ADD_TYPE(m, state->PyBytesIO_Type, &bytesio_spec, &PyBufferedIOBase_Type);
+    ADD_TYPE(m, state->PyBufferedWriter_Type, &bufferedwriter_spec,
+             &PyBufferedIOBase_Type);
+    ADD_TYPE(m, state->PyBufferedReader_Type, &bufferedreader_spec,
+             &PyBufferedIOBase_Type);
+    ADD_TYPE(m, state->PyBufferedRWPair_Type, &bufferedrwpair_spec,
+             &PyBufferedIOBase_Type);
+    ADD_TYPE(m, state->PyBufferedRandom_Type, &bufferedrandom_spec,
+             &PyBufferedIOBase_Type);
+
+    // PyRawIOBase_Type(PyIOBase_Type) subclasses
+    ADD_TYPE(m, state->PyFileIO_Type, &fileio_spec, &PyRawIOBase_Type);
+
+    // PyTextIOBase_Type(PyIOBase_Type) subclasses
+    ADD_TYPE(m, state->PyStringIO_Type, &stringio_spec, &PyTextIOBase_Type);
+    ADD_TYPE(m, state->PyTextIOWrapper_Type, &textiowrapper_spec,
+             &PyTextIOBase_Type);
+
     state->initialized = 1;
 
     return m;

Modules/_io/_iomodule.h

@@ -4,23 +4,28 @@
 
 #include "exports.h"
 
+#include "pycore_moduleobject.h"  // _PyModule_GetState()
+#include "structmember.h"
+
 /* ABCs */
 extern PyTypeObject PyIOBase_Type;
 extern PyTypeObject PyRawIOBase_Type;
 extern PyTypeObject PyBufferedIOBase_Type;
 extern PyTypeObject PyTextIOBase_Type;
 
 /* Concrete classes */
-extern PyTypeObject PyFileIO_Type;
-extern PyTypeObject PyBytesIO_Type;
-extern PyTypeObject PyStringIO_Type;
-extern PyTypeObject PyBufferedReader_Type;
-extern PyTypeObject PyBufferedWriter_Type;
-extern PyTypeObject PyBufferedRWPair_Type;
-extern PyTypeObject PyBufferedRandom_Type;
-extern PyTypeObject PyTextIOWrapper_Type;
 extern PyTypeObject PyIncrementalNewlineDecoder_Type;
 
+/* Type specs */
+extern PyType_Spec bufferedrandom_spec;
+extern PyType_Spec bufferedreader_spec;
+extern PyType_Spec bufferedrwpair_spec;
+extern PyType_Spec bufferedwriter_spec;
+extern PyType_Spec bytesio_spec;
+extern PyType_Spec fileio_spec;
+extern PyType_Spec stringio_spec;
+extern PyType_Spec textiowrapper_spec;
+
 #ifdef MS_WINDOWS
 extern PyTypeObject PyWindowsConsoleIO_Type;
 #endif /* MS_WINDOWS */
@@ -140,11 +145,37 @@ typedef struct {
     PyObject *locale_module;
 
     PyObject *unsupported_operation;
+
+    /* Types */
+    PyTypeObject *PyBufferedRWPair_Type;
+    PyTypeObject *PyBufferedRandom_Type;
+    PyTypeObject *PyBufferedReader_Type;
+    PyTypeObject *PyBufferedWriter_Type;
+    PyTypeObject *PyBytesIO_Type;
+    PyTypeObject *PyFileIO_Type;
+    PyTypeObject *PyStringIO_Type;
+    PyTypeObject *PyTextIOWrapper_Type;
 } _PyIO_State;
 
 #define IO_MOD_STATE(mod) ((_PyIO_State *)PyModule_GetState(mod))
 #define IO_STATE() _PyIO_get_module_state()
 
+static inline _PyIO_State *
+get_io_state(PyObject *module)
+{
+    void *state = _PyModule_GetState(module);
+    assert(state != NULL);
+    return (_PyIO_State *)state;
+}
+
+static inline _PyIO_State *
+find_io_state_by_def(PyTypeObject *type)
+{
+    PyObject *mod = PyType_GetModuleByDef(type, &_PyIO_Module);
+    assert(mod != NULL);
+    return get_io_state(mod);
+}
+
 extern _PyIO_State *_PyIO_get_module_state(void);
 
 #ifdef MS_WINDOWS