dep-protobuf/python/repeated.c
Runze Wang 78a09ada5c Fix RepeatedContainer_pop to raise IndexError for out-of-bounds indices.
The previous implementation of RepeatedContainer_pop would clamp the index to the last element if it was out of bounds. This change removes the index clamping, ensuring that pop() raises an IndexError when given an index that is out of range, matching the other 2 pyproto implementation as well as standard Python list behavior. Tests are added to confirm this corrected behavior.

This is the 5th bug listed in https://github.com/protocolbuffers/protobuf/issues/26596

PiperOrigin-RevId: 902737839
2026-04-20 11:30:37 -07:00

1015 lines
36 KiB
C

// Protocol Buffers - Google's data interchange format
// Copyright 2023 Google LLC. All rights reserved.
//
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file or at
// https://developers.google.com/open-source/licenses/bsd
#include "python/repeated.h"
#include "python/convert.h"
#include "python/message.h"
#include "python/protobuf.h"
#include "upb/port/def.inc"
static PyObject* PyUpb_RepeatedCompositeContainer_Append(PyObject* _self,
PyObject* value);
static PyObject* PyUpb_RepeatedScalarContainer_Append(PyObject* _self,
PyObject* value);
// Wrapper for a repeated field.
typedef struct {
// clang-format off
PyObject_HEAD
PyObject* arena;
// clang-format on
// The field descriptor (PyObject*).
// The low bit indicates whether the container is reified (see ptr below).
// - low bit set: repeated field is a stub (no underlying data).
// - low bit clear: repeated field is reified (points to upb_Array).
uintptr_t field;
union {
PyObject* parent; // stub: owning pointer to parent message.
upb_Array* arr; // reified: the data for this array.
} ptr;
} PyUpb_RepeatedContainer;
static bool PyUpb_RepeatedContainer_IsStub(PyUpb_RepeatedContainer* self) {
return self->field & 1;
}
static PyObject* PyUpb_RepeatedContainer_GetFieldDescriptor(
PyUpb_RepeatedContainer* self) {
return (PyObject*)(self->field & ~(uintptr_t)1);
}
static const upb_FieldDef* PyUpb_RepeatedContainer_GetField(
PyUpb_RepeatedContainer* self) {
return PyUpb_FieldDescriptor_GetDef(
PyUpb_RepeatedContainer_GetFieldDescriptor(self));
}
// If the repeated field is reified, returns it. Otherwise, returns NULL.
// If NULL is returned, the object is empty and has no underlying data.
static upb_Array* PyUpb_RepeatedContainer_GetIfReified(
PyUpb_RepeatedContainer* self) {
return PyUpb_RepeatedContainer_IsStub(self) ? NULL : self->ptr.arr;
}
upb_Array* PyUpb_RepeatedContainer_Reify(PyObject* _self, upb_Array* arr,
PyUpb_WeakMap* subobj_map,
intptr_t iter) {
PyUpb_RepeatedContainer* self = (PyUpb_RepeatedContainer*)_self;
assert(PyUpb_RepeatedContainer_IsStub(self));
const upb_FieldDef* f = PyUpb_RepeatedContainer_GetField(self);
if (!arr) {
upb_Arena* arena = PyUpb_Arena_Get(self->arena);
arr = upb_Array_New(arena, upb_FieldDef_CType(f));
}
if (subobj_map) {
PyUpb_WeakMap_DeleteIter(subobj_map, &iter);
} else {
PyUpb_Message_SetConcreteSubobj(self->ptr.parent, f,
(upb_MessageValue){.array_val = arr});
}
PyUpb_ObjCache_Add(arr, &self->ob_base);
Py_DECREF(self->ptr.parent);
self->ptr.arr = arr; // Overwrites self->ptr.parent.
self->field &= ~(uintptr_t)1;
assert(!PyUpb_RepeatedContainer_IsStub(self));
return arr;
}
upb_Array* PyUpb_RepeatedContainer_EnsureReified(PyObject* _self) {
PyUpb_RepeatedContainer* self = (PyUpb_RepeatedContainer*)_self;
upb_Array* arr = PyUpb_RepeatedContainer_GetIfReified(self);
if (arr) return arr; // Already writable.
return PyUpb_RepeatedContainer_Reify((PyObject*)self, NULL, NULL, 0);
}
static void PyUpb_RepeatedContainer_Dealloc(PyObject* _self) {
PyUpb_RepeatedContainer* self = (PyUpb_RepeatedContainer*)_self;
Py_DECREF(self->arena);
if (PyUpb_RepeatedContainer_IsStub(self)) {
PyUpb_Message_CacheDelete(self->ptr.parent,
PyUpb_RepeatedContainer_GetField(self));
Py_DECREF(self->ptr.parent);
} else {
PyUpb_ObjCache_Delete(self->ptr.arr);
}
Py_DECREF(PyUpb_RepeatedContainer_GetFieldDescriptor(self));
PyUpb_Dealloc(self);
}
static PyTypeObject* PyUpb_RepeatedContainer_GetClass(const upb_FieldDef* f) {
assert(upb_FieldDef_IsRepeated(f) && !upb_FieldDef_IsMap(f));
PyUpb_ModuleState* state = PyUpb_ModuleState_Get();
return upb_FieldDef_IsSubMessage(f) ? state->repeated_composite_container_type
: state->repeated_scalar_container_type;
}
static Py_ssize_t PyUpb_RepeatedContainer_Length(PyObject* self) {
upb_Array* arr =
PyUpb_RepeatedContainer_GetIfReified((PyUpb_RepeatedContainer*)self);
return arr ? upb_Array_Size(arr) : 0;
}
PyObject* PyUpb_RepeatedContainer_NewStub(PyObject* parent,
const upb_FieldDef* f,
PyObject* arena) {
// We only create stubs when the parent is reified, by convention. However
// this is not an invariant: the parent could become reified at any time.
assert(PyUpb_Message_GetIfReified(parent) == NULL);
PyTypeObject* cls = PyUpb_RepeatedContainer_GetClass(f);
PyUpb_RepeatedContainer* repeated = (void*)PyType_GenericAlloc(cls, 0);
repeated->arena = arena;
repeated->field = (uintptr_t)PyUpb_FieldDescriptor_Get(f) | 1;
repeated->ptr.parent = parent;
Py_INCREF(arena);
Py_INCREF(parent);
return &repeated->ob_base;
}
PyObject* PyUpb_RepeatedContainer_GetOrCreateWrapper(upb_Array* arr,
const upb_FieldDef* f,
PyObject* arena) {
PyObject* ret = PyUpb_ObjCache_Get(arr);
if (ret) return ret;
PyTypeObject* cls = PyUpb_RepeatedContainer_GetClass(f);
PyUpb_RepeatedContainer* repeated = (void*)PyType_GenericAlloc(cls, 0);
repeated->arena = arena;
repeated->field = (uintptr_t)PyUpb_FieldDescriptor_Get(f);
repeated->ptr.arr = arr;
ret = &repeated->ob_base;
Py_INCREF(arena);
PyUpb_ObjCache_Add(arr, ret);
return ret;
}
static PyObject* PyUpb_RepeatedContainer_MergeFrom(PyObject* _self,
PyObject* args);
PyObject* PyUpb_RepeatedContainer_DeepCopy(PyObject* _self, PyObject* value) {
PyUpb_RepeatedContainer* self = (PyUpb_RepeatedContainer*)_self;
PyUpb_RepeatedContainer* clone =
(void*)PyType_GenericAlloc(Py_TYPE(_self), 0);
if (clone == NULL) return NULL;
const upb_FieldDef* f = PyUpb_RepeatedContainer_GetField(self);
clone->arena = PyUpb_Arena_New();
clone->field = (uintptr_t)PyUpb_FieldDescriptor_Get(f);
clone->ptr.arr =
upb_Array_New(PyUpb_Arena_Get(clone->arena), upb_FieldDef_CType(f));
PyUpb_ObjCache_Add(clone->ptr.arr, (PyObject*)clone);
PyObject* result = PyUpb_RepeatedContainer_MergeFrom((PyObject*)clone, _self);
if (!result) {
Py_DECREF(clone);
return NULL;
}
Py_DECREF(result);
return (PyObject*)clone;
}
PyObject* PyUpb_RepeatedContainer_Extend(PyObject* _self, PyObject* value) {
PyUpb_RepeatedContainer* self = (PyUpb_RepeatedContainer*)_self;
upb_Array* arr = PyUpb_RepeatedContainer_EnsureReified(_self);
size_t start_size = upb_Array_Size(arr);
PyObject* it = PyObject_GetIter(value);
if (!it) {
PyErr_SetString(PyExc_TypeError, "Value must be iterable");
return NULL;
}
const upb_FieldDef* f = PyUpb_RepeatedContainer_GetField(self);
bool submsg = upb_FieldDef_IsSubMessage(f);
PyObject* e;
if (submsg) {
while ((e = PyIter_Next(it))) {
PyObject* ret = PyUpb_RepeatedCompositeContainer_Append(_self, e);
Py_XDECREF(ret);
Py_DECREF(e);
}
} else {
upb_Arena* arena = PyUpb_Arena_Get(self->arena);
Py_ssize_t size = PyObject_Size(value);
if (size < 0) {
// Some iterables may not have len. Size() will return -1 and
// set an error in such cases.
PyErr_Clear();
} else {
upb_Array_Reserve(arr, start_size + size, arena);
}
while ((e = PyIter_Next(it))) {
upb_MessageValue msgval;
if (!PyUpb_PyToUpb(e, f, &msgval, arena)) {
assert(PyErr_Occurred());
Py_DECREF(e);
break;
}
upb_Array_Append(arr, msgval, arena);
Py_DECREF(e);
}
}
Py_DECREF(it);
if (PyErr_Occurred()) {
upb_Array_Resize(arr, start_size, NULL);
return NULL;
}
Py_RETURN_NONE;
}
static PyObject* PyUpb_RepeatedContainer_Item(PyObject* _self,
Py_ssize_t index) {
PyUpb_RepeatedContainer* self = (PyUpb_RepeatedContainer*)_self;
upb_Array* arr = PyUpb_RepeatedContainer_GetIfReified(self);
Py_ssize_t size = arr ? upb_Array_Size(arr) : 0;
if (index < 0 || index >= size) {
PyErr_Format(PyExc_IndexError, "list index (%zd) out of range", index);
return NULL;
}
const upb_FieldDef* f = PyUpb_RepeatedContainer_GetField(self);
return PyUpb_UpbToPy(upb_Array_Get(arr, index), f, self->arena);
}
PyObject* PyUpb_RepeatedContainer_ToList(PyObject* _self) {
PyUpb_RepeatedContainer* self = (PyUpb_RepeatedContainer*)_self;
upb_Array* arr = PyUpb_RepeatedContainer_GetIfReified(self);
if (!arr) return PyList_New(0);
const upb_FieldDef* f = PyUpb_RepeatedContainer_GetField(self);
size_t n = upb_Array_Size(arr);
PyObject* list = PyList_New(n);
for (size_t i = 0; i < n; i++) {
PyObject* val = PyUpb_UpbToPy(upb_Array_Get(arr, i), f, self->arena);
if (!val) {
Py_DECREF(list);
return NULL;
}
PyList_SetItem(list, i, val);
}
return list;
}
static PyObject* PyUpb_RepeatedContainer_Repr(PyObject* _self) {
PyObject* list = PyUpb_RepeatedContainer_ToList(_self);
if (!list) return NULL;
assert(!PyErr_Occurred());
PyObject* repr = PyObject_Repr(list);
Py_DECREF(list);
return repr;
}
static PyObject* PyUpb_RepeatedContainer_RichCompare(PyObject* _self,
PyObject* _other,
int opid) {
if (opid != Py_EQ && opid != Py_NE) {
Py_INCREF(Py_NotImplemented);
return Py_NotImplemented;
}
PyObject* list1 = PyUpb_RepeatedContainer_ToList(_self);
PyObject* list2 = _other;
PyObject* del = NULL;
if (PyObject_TypeCheck(_other, _self->ob_type)) {
del = list2 = PyUpb_RepeatedContainer_ToList(_other);
}
PyObject* ret = PyObject_RichCompare(list1, list2, opid);
Py_DECREF(list1);
Py_XDECREF(del);
return ret;
}
static PyObject* PyUpb_RepeatedContainer_Subscript(PyObject* _self,
PyObject* key) {
PyUpb_RepeatedContainer* self = (PyUpb_RepeatedContainer*)_self;
upb_Array* arr = PyUpb_RepeatedContainer_GetIfReified(self);
Py_ssize_t size = arr ? upb_Array_Size(arr) : 0;
Py_ssize_t idx, count, step;
if (!PyUpb_IndexToRange(key, size, &idx, &count, &step)) return NULL;
const upb_FieldDef* f = PyUpb_RepeatedContainer_GetField(self);
if (step == 0) {
return PyUpb_UpbToPy(upb_Array_Get(arr, idx), f, self->arena);
} else {
PyObject* list = PyList_New(count);
for (Py_ssize_t i = 0; i < count; i++, idx += step) {
upb_MessageValue msgval = upb_Array_Get(self->ptr.arr, idx);
PyObject* item = PyUpb_UpbToPy(msgval, f, self->arena);
if (!item) {
Py_DECREF(list);
return NULL;
}
PyList_SetItem(list, i, item);
}
return list;
}
}
static int PyUpb_RepeatedContainer_SetSubscript(
PyUpb_RepeatedContainer* self, upb_Array* arr, const upb_FieldDef* f,
Py_ssize_t idx, Py_ssize_t count, Py_ssize_t step, PyObject* value) {
upb_Arena* arena = PyUpb_Arena_Get(self->arena);
if (upb_FieldDef_IsSubMessage(f)) {
PyErr_SetString(PyExc_TypeError, "does not support assignment");
return -1;
}
if (step == 0) {
// Set single value.
upb_MessageValue msgval;
if (!PyUpb_PyToUpb(value, f, &msgval, arena)) return -1;
upb_Array_Set(arr, idx, msgval);
return 0;
}
// Set range.
PyObject* seq =
PySequence_Fast(value, "must assign iterable to extended slice");
PyObject* item = NULL;
int ret = -1;
if (!seq) goto err;
Py_ssize_t seq_size = PySequence_Size(seq);
if (seq_size != count) {
if (step == 1) {
// We must shift the tail elements (either right or left).
size_t tail = upb_Array_Size(arr) - (idx + count);
upb_Array_Resize(arr, idx + seq_size + tail, arena);
upb_Array_Move(arr, idx + seq_size, idx + count, tail);
count = seq_size;
} else {
PyErr_Format(PyExc_ValueError,
"attempt to assign sequence of %zd to extended slice "
"of size %zd",
seq_size, count);
goto err;
}
}
for (Py_ssize_t i = 0; i < count; i++, idx += step) {
upb_MessageValue msgval;
item = PySequence_GetItem(seq, i);
if (!item) goto err;
// XXX: if this fails we can leave the list partially mutated.
if (!PyUpb_PyToUpb(item, f, &msgval, arena)) goto err;
Py_DECREF(item);
item = NULL;
upb_Array_Set(arr, idx, msgval);
}
ret = 0;
err:
Py_XDECREF(seq);
Py_XDECREF(item);
return ret;
}
static int PyUpb_RepeatedContainer_DeleteSubscript(upb_Array* arr,
Py_ssize_t idx,
Py_ssize_t count,
Py_ssize_t step) {
// Normalize direction: deletion is order-independent.
Py_ssize_t start = idx;
if (step < 0) {
Py_ssize_t end = start + step * (count - 1);
start = end;
step = -step;
}
size_t dst = start;
size_t src;
if (step > 1) {
// Move elements between steps:
//
// src
// |
// |------X---X---X---X------------------------------|
// |
// dst <-------- tail -------------->
src = start + 1;
for (Py_ssize_t i = 1; i < count; i++, dst += step - 1, src += step) {
upb_Array_Move(arr, dst, src, step);
}
} else {
src = start + count;
}
// Move tail.
size_t tail = upb_Array_Size(arr) - src;
size_t new_size = dst + tail;
assert(new_size == upb_Array_Size(arr) - count);
upb_Array_Move(arr, dst, src, tail);
upb_Array_Resize(arr, new_size, NULL);
return 0;
}
static int PyUpb_RepeatedContainer_AssignSubscript(PyObject* _self,
PyObject* key,
PyObject* value) {
PyUpb_RepeatedContainer* self = (PyUpb_RepeatedContainer*)_self;
const upb_FieldDef* f = PyUpb_RepeatedContainer_GetField(self);
upb_Array* arr = PyUpb_RepeatedContainer_EnsureReified(_self);
Py_ssize_t size = arr ? upb_Array_Size(arr) : 0;
Py_ssize_t idx, count, step;
if (!PyUpb_IndexToRange(key, size, &idx, &count, &step)) return -1;
if (value) {
return PyUpb_RepeatedContainer_SetSubscript(self, arr, f, idx, count, step,
value);
} else {
return PyUpb_RepeatedContainer_DeleteSubscript(arr, idx, count, step);
}
}
static PyObject* PyUpb_RepeatedContainer_Pop(PyObject* _self, PyObject* args) {
PyUpb_RepeatedContainer* self = (PyUpb_RepeatedContainer*)_self;
Py_ssize_t index = -1;
if (!PyArg_ParseTuple(args, "|n", &index)) return NULL;
upb_Array* arr = PyUpb_RepeatedContainer_EnsureReified(_self);
size_t size = upb_Array_Size(arr);
if (index < 0) index += size;
#if UPB_FUTURE_REMOVE_POP_CLAMP
#else
if (index >= size) index = size - 1;
#endif // UPB_FUTURE_REMOVE_POP_CLAMP
PyObject* ret = PyUpb_RepeatedContainer_Item(_self, index);
if (!ret) return NULL;
upb_Array_Delete(self->ptr.arr, index, 1);
return ret;
}
static PyObject* PyUpb_RepeatedContainer_Remove(PyObject* _self,
PyObject* value) {
upb_Array* arr = PyUpb_RepeatedContainer_EnsureReified(_self);
Py_ssize_t match_index = -1;
Py_ssize_t n = PyUpb_RepeatedContainer_Length(_self);
for (Py_ssize_t i = 0; i < n; ++i) {
PyObject* elem = PyUpb_RepeatedContainer_Item(_self, i);
if (!elem) return NULL;
int eq = PyObject_RichCompareBool(elem, value, Py_EQ);
Py_DECREF(elem);
if (eq) {
match_index = i;
break;
}
}
if (match_index == -1) {
PyErr_SetString(PyExc_ValueError, "remove(x): x not in container");
return NULL;
}
if (PyUpb_RepeatedContainer_DeleteSubscript(arr, match_index, 1, 1) < 0) {
return NULL;
}
Py_RETURN_NONE;
}
// A helper function used only for Sort().
static bool PyUpb_RepeatedContainer_Assign(PyObject* _self, PyObject* list) {
PyUpb_RepeatedContainer* self = (PyUpb_RepeatedContainer*)_self;
const upb_FieldDef* f = PyUpb_RepeatedContainer_GetField(self);
upb_Array* arr = PyUpb_RepeatedContainer_EnsureReified(_self);
Py_ssize_t size = PyList_Size(list);
bool submsg = upb_FieldDef_IsSubMessage(f);
upb_Arena* arena = PyUpb_Arena_Get(self->arena);
for (Py_ssize_t i = 0; i < size; ++i) {
PyObject* obj = PyList_GetItem(list, i);
upb_MessageValue msgval;
if (submsg) {
msgval.msg_val = PyUpb_Message_GetIfReified(obj);
assert(msgval.msg_val);
} else {
if (!PyUpb_PyToUpb(obj, f, &msgval, arena)) return false;
}
upb_Array_Set(arr, i, msgval);
}
return true;
}
static PyObject* PyUpb_RepeatedContainer_Sort(PyObject* pself, PyObject* args,
PyObject* kwds) {
// Support the old sort_function argument for backwards
// compatibility.
if (kwds != NULL) {
PyObject* sort_func = PyDict_GetItemString(kwds, "sort_function");
if (sort_func != NULL) {
// Must set before deleting as sort_func is a borrowed reference
// and kwds might be the only thing keeping it alive.
if (PyDict_SetItemString(kwds, "cmp", sort_func) == -1) return NULL;
if (PyDict_DelItemString(kwds, "sort_function") == -1) return NULL;
}
}
if (PyUpb_RepeatedContainer_Length(pself) == 0) Py_RETURN_NONE;
PyObject* ret = NULL;
PyObject* full_slice = NULL;
PyObject* list = NULL;
PyObject* m = NULL;
PyObject* res = NULL;
if ((full_slice = PySlice_New(NULL, NULL, NULL)) &&
(list = PyUpb_RepeatedContainer_Subscript(pself, full_slice)) &&
(m = PyObject_GetAttrString(list, "sort")) &&
(res = PyObject_Call(m, args, kwds)) &&
PyUpb_RepeatedContainer_Assign(pself, list)) {
Py_INCREF(Py_None);
ret = Py_None;
}
Py_XDECREF(full_slice);
Py_XDECREF(list);
Py_XDECREF(m);
Py_XDECREF(res);
return ret;
}
static PyObject* PyUpb_RepeatedContainer_Reverse(PyObject* _self) {
upb_Array* arr = PyUpb_RepeatedContainer_EnsureReified(_self);
size_t n = upb_Array_Size(arr);
size_t half = n / 2; // Rounds down.
for (size_t i = 0; i < half; i++) {
size_t i2 = n - i - 1;
upb_MessageValue val1 = upb_Array_Get(arr, i);
upb_MessageValue val2 = upb_Array_Get(arr, i2);
upb_Array_Set(arr, i, val2);
upb_Array_Set(arr, i2, val1);
}
Py_RETURN_NONE;
}
static PyObject* PyUpb_RepeatedContainer_Clear(PyObject* _self) {
PyUpb_RepeatedContainer* self = (PyUpb_RepeatedContainer*)_self;
Py_ssize_t size = PyUpb_RepeatedContainer_Length(_self);
if (size > 0) {
upb_Array_Delete(self->ptr.arr, 0, size);
}
Py_RETURN_NONE;
}
static PyObject* PyUpb_RepeatedContainer_MergeFrom(PyObject* _self,
PyObject* args) {
return PyUpb_RepeatedContainer_Extend(_self, args);
}
// -----------------------------------------------------------------------------
// RepeatedCompositeContainer
// -----------------------------------------------------------------------------
static PyObject* PyUpb_RepeatedCompositeContainer_AppendNew(PyObject* _self) {
PyUpb_RepeatedContainer* self = (PyUpb_RepeatedContainer*)_self;
upb_Array* arr = PyUpb_RepeatedContainer_EnsureReified(_self);
if (!arr) return NULL;
const upb_FieldDef* f = PyUpb_RepeatedContainer_GetField(self);
upb_Arena* arena = PyUpb_Arena_Get(self->arena);
const upb_MessageDef* m = upb_FieldDef_MessageSubDef(f);
const upb_MiniTable* layout = upb_MessageDef_MiniTable(m);
upb_Message* msg = upb_Message_New(layout, arena);
upb_MessageValue msgval = {.msg_val = msg};
upb_Array_Append(arr, msgval, arena);
return PyUpb_Message_Get(msg, m, self->arena);
}
PyObject* PyUpb_RepeatedCompositeContainer_Add(PyObject* _self, PyObject* args,
PyObject* kwargs) {
PyUpb_RepeatedContainer* self = (PyUpb_RepeatedContainer*)_self;
PyObject* py_msg = PyUpb_RepeatedCompositeContainer_AppendNew(_self);
if (!py_msg) return NULL;
if (PyUpb_Message_InitAttributes(py_msg, args, kwargs) < 0) {
Py_DECREF(py_msg);
upb_Array_Delete(self->ptr.arr, upb_Array_Size(self->ptr.arr) - 1, 1);
return NULL;
}
return py_msg;
}
static PyObject* PyUpb_RepeatedCompositeContainer_Append(PyObject* _self,
PyObject* value) {
if (!PyUpb_Message_Verify(value)) return NULL;
PyObject* py_msg = PyUpb_RepeatedCompositeContainer_AppendNew(_self);
if (!py_msg) return NULL;
PyObject* none = PyUpb_Message_MergeFrom(py_msg, value);
if (!none) {
Py_DECREF(py_msg);
return NULL;
}
Py_DECREF(none);
return py_msg;
}
static PyObject* PyUpb_RepeatedContainer_Insert(PyObject* _self,
PyObject* args) {
PyUpb_RepeatedContainer* self = (PyUpb_RepeatedContainer*)_self;
Py_ssize_t index;
PyObject* value;
if (!PyArg_ParseTuple(args, "nO", &index, &value)) return NULL;
upb_Array* arr = PyUpb_RepeatedContainer_EnsureReified(_self);
if (!arr) return NULL;
// Normalize index.
Py_ssize_t size = upb_Array_Size(arr);
if (index < 0) index += size;
if (index < 0) index = 0;
if (index > size) index = size;
const upb_FieldDef* f = PyUpb_RepeatedContainer_GetField(self);
upb_MessageValue msgval;
upb_Arena* arena = PyUpb_Arena_Get(self->arena);
if (upb_FieldDef_IsSubMessage(f)) {
// Create message.
const upb_MessageDef* m = upb_FieldDef_MessageSubDef(f);
const upb_MiniTable* layout = upb_MessageDef_MiniTable(m);
upb_Message* msg = upb_Message_New(layout, arena);
PyObject* py_msg = PyUpb_Message_Get(msg, m, self->arena);
PyObject* ret = PyUpb_Message_MergeFrom(py_msg, value);
Py_DECREF(py_msg);
if (!ret) return NULL;
Py_DECREF(ret);
msgval.msg_val = msg;
} else {
if (!PyUpb_PyToUpb(value, f, &msgval, arena)) return NULL;
}
upb_Array_Insert(arr, index, 1, arena);
upb_Array_Set(arr, index, msgval);
Py_RETURN_NONE;
}
static PyMethodDef PyUpb_RepeatedCompositeContainer_Methods[] = {
{"__deepcopy__", PyUpb_RepeatedContainer_DeepCopy, METH_VARARGS,
"Makes a deep copy of the class."},
{"add", (PyCFunction)PyUpb_RepeatedCompositeContainer_Add,
METH_VARARGS | METH_KEYWORDS, "Adds an object to the repeated container."},
{"append", PyUpb_RepeatedCompositeContainer_Append, METH_O,
"Appends a message to the end of the repeated container."},
{"insert", PyUpb_RepeatedContainer_Insert, METH_VARARGS,
"Inserts a message before the specified index."},
{"extend", PyUpb_RepeatedContainer_Extend, METH_O,
"Adds objects to the repeated container."},
{"pop", PyUpb_RepeatedContainer_Pop, METH_VARARGS,
"Removes an object from the repeated container and returns it."},
{"remove", PyUpb_RepeatedContainer_Remove, METH_O,
"Removes an object from the repeated container."},
{"sort", (PyCFunction)PyUpb_RepeatedContainer_Sort,
METH_VARARGS | METH_KEYWORDS, "Sorts the repeated container."},
{"reverse", (PyCFunction)PyUpb_RepeatedContainer_Reverse, METH_NOARGS,
"Reverses elements order of the repeated container."},
{"clear", (PyCFunction)PyUpb_RepeatedContainer_Clear, METH_NOARGS,
"Clears repeated container."},
{"MergeFrom", PyUpb_RepeatedContainer_MergeFrom, METH_O,
"Adds objects to the repeated container."},
{NULL, NULL}};
static PyType_Slot PyUpb_RepeatedCompositeContainer_Slots[] = {
{Py_tp_dealloc, PyUpb_RepeatedContainer_Dealloc},
{Py_tp_methods, PyUpb_RepeatedCompositeContainer_Methods},
{Py_sq_length, PyUpb_RepeatedContainer_Length},
{Py_sq_item, PyUpb_RepeatedContainer_Item},
{Py_mp_length, PyUpb_RepeatedContainer_Length},
{Py_tp_repr, PyUpb_RepeatedContainer_Repr},
{Py_mp_subscript, PyUpb_RepeatedContainer_Subscript},
{Py_mp_ass_subscript, PyUpb_RepeatedContainer_AssignSubscript},
{Py_tp_new, PyUpb_Forbidden_New},
{Py_tp_richcompare, PyUpb_RepeatedContainer_RichCompare},
{Py_tp_hash, PyObject_HashNotImplemented},
{0, NULL}};
static PyType_Spec PyUpb_RepeatedCompositeContainer_Spec = {
PYUPB_MODULE_NAME ".RepeatedCompositeContainer",
sizeof(PyUpb_RepeatedContainer),
0, // tp_itemsize
Py_TPFLAGS_DEFAULT,
PyUpb_RepeatedCompositeContainer_Slots,
};
// -----------------------------------------------------------------------------
// RepeatedScalarContainer
// -----------------------------------------------------------------------------
static PyObject* PyUpb_RepeatedScalarContainer_Append(PyObject* _self,
PyObject* value) {
PyUpb_RepeatedContainer* self = (PyUpb_RepeatedContainer*)_self;
upb_Array* arr = PyUpb_RepeatedContainer_EnsureReified(_self);
upb_Arena* arena = PyUpb_Arena_Get(self->arena);
const upb_FieldDef* f = PyUpb_RepeatedContainer_GetField(self);
upb_MessageValue msgval;
if (!PyUpb_PyToUpb(value, f, &msgval, arena)) {
return NULL;
}
upb_Array_Append(arr, msgval, arena);
Py_RETURN_NONE;
}
static int PyUpb_RepeatedScalarContainer_AssignItem(PyObject* _self,
Py_ssize_t index,
PyObject* item) {
PyUpb_RepeatedContainer* self = (PyUpb_RepeatedContainer*)_self;
upb_Array* arr = PyUpb_RepeatedContainer_GetIfReified(self);
Py_ssize_t size = arr ? upb_Array_Size(arr) : 0;
if (index < 0 || index >= size) {
PyErr_Format(PyExc_IndexError, "list index (%zd) out of range", index);
return -1;
}
const upb_FieldDef* f = PyUpb_RepeatedContainer_GetField(self);
upb_MessageValue msgval;
upb_Arena* arena = PyUpb_Arena_Get(self->arena);
if (!PyUpb_PyToUpb(item, f, &msgval, arena)) {
return -1;
}
upb_Array_Set(self->ptr.arr, index, msgval);
return 0;
}
static PyObject* PyUpb_RepeatedScalarContainer_Reduce(PyObject* unused_self,
PyObject* unused_other) {
PyObject* pickle_module = PyImport_ImportModule("pickle");
if (!pickle_module) return NULL;
PyObject* pickle_error = PyObject_GetAttrString(pickle_module, "PickleError");
Py_DECREF(pickle_module);
if (!pickle_error) return NULL;
PyErr_Format(pickle_error,
"can't pickle repeated message fields, convert to list first");
Py_DECREF(pickle_error);
return NULL;
}
char* GetDefaultDTypeStr(upb_CType cpp_type) {
switch (cpp_type) {
case kUpb_CType_Float:
return "f4";
case kUpb_CType_Int32:
return "i4";
case kUpb_CType_Int64:
return "i8";
case kUpb_CType_UInt32:
return "u4";
case kUpb_CType_UInt64:
return "u8";
case kUpb_CType_Double:
return "f8";
case kUpb_CType_Bool:
return "?"; // Boolean is '?' per official docs.
case kUpb_CType_Enum:
return "i4";
case kUpb_CType_String:
case kUpb_CType_Bytes:
case kUpb_CType_Message:
return NULL;
}
return NULL;
}
PyObject* CreateArrayFromView(PyObject* _self, PyObject* np_module) {
PyUpb_RepeatedContainer* self = (PyUpb_RepeatedContainer*)(_self);
upb_Array* arr = PyUpb_RepeatedContainer_GetIfReified(self);
size_t size = arr ? upb_Array_Size(arr) : 0;
const upb_FieldDef* f = PyUpb_RepeatedContainer_GetField(self);
Py_ssize_t out_buffer_size_bytes;
const char* out_dtype = GetDefaultDTypeStr(upb_FieldDef_CType(f));
switch (upb_FieldDef_CType(f)) {
case kUpb_CType_Float: {
out_buffer_size_bytes = sizeof(float) * size;
break;
}
case kUpb_CType_Int32: {
out_buffer_size_bytes = 4 * size;
break;
}
case kUpb_CType_Int64: {
out_buffer_size_bytes = 8 * size;
break;
}
case kUpb_CType_UInt32: {
out_buffer_size_bytes = 4 * size;
break;
}
case kUpb_CType_UInt64: {
out_buffer_size_bytes = 8 * size;
break;
}
case kUpb_CType_Double: {
out_buffer_size_bytes = sizeof(double) * size;
break;
}
case kUpb_CType_Bool: {
out_buffer_size_bytes = sizeof(bool) * size;
break;
}
case kUpb_CType_Enum: {
out_buffer_size_bytes = 4 * size;
break;
}
case kUpb_CType_Message:
case kUpb_CType_Bytes:
case kUpb_CType_String: {
PyErr_Format(PyExc_SystemError,
"Code should never reach here: cpp type "
"should not be message nor string in CreateArrayFromView.");
return NULL;
}
}
if (out_buffer_size_bytes == 0) {
return PyObject_CallMethod(np_module, "empty", "is", 0, out_dtype);
}
const void* out_ptr = upb_Array_DataPtr(arr);
PyObject* view = PyMemoryView_FromMemory(
(char*)(const char*)out_ptr, out_buffer_size_bytes, 0x100 /*PyBUF_READ*/);
PyObject* return_value =
PyObject_CallMethod(np_module, "frombuffer", "Os", view, out_dtype);
Py_DECREF(view);
return return_value;
}
// CPython equivalent of nparray.astype(dtype_requested, copy=copy).
PyObject* NpArrayAsType(PyObject* nparray, PyObject* dtype_requested,
bool copy) {
PyObject* astype_args = Py_BuildValue("(O)", dtype_requested);
PyObject* keywords = PyDict_New();
PyDict_SetItemString(keywords, "copy", copy ? Py_True : Py_False);
PyObject* np_array_astype = PyObject_GetAttrString(nparray, "astype");
PyObject* return_value =
PyObject_Call(np_array_astype, astype_args, keywords);
Py_DECREF(np_array_astype);
Py_DECREF(keywords);
Py_DECREF(astype_args);
return return_value;
}
PyObject* ConstructArrayByIteration(PyObject* _self, PyObject* np_module) {
PyUpb_RepeatedContainer* self = (PyUpb_RepeatedContainer*)_self;
upb_Array* arr = PyUpb_RepeatedContainer_GetIfReified(self);
Py_ssize_t array_length = PyUpb_RepeatedContainer_Length(_self);
const upb_FieldDef* f = PyUpb_RepeatedContainer_GetField(self);
PyObject* nparray =
PyObject_CallMethod(np_module, "empty", "is", array_length, "O");
PyObject* arena = self->arena;
for (Py_ssize_t i = 0; i < array_length; ++i) {
upb_MessageValue msgval = upb_Array_Get(arr, i);
PyObject* item = PyUpb_UpbToPy(msgval, f, arena);
PySequence_SetItem(nparray, i, item);
Py_DECREF(item);
}
return nparray;
}
static PyObject* PyUpb_RepeatedScalarContainer_AsNpArray(PyObject* _self,
PyObject* args,
PyObject* kwargs) {
static const char* kwlist[] = {"dtype", "copy", NULL};
PyObject* dtype_requested = NULL;
PyObject* copy = NULL;
PyObject* return_value = NULL;
PyObject* default_dtype = NULL;
PyObject* nparray = NULL;
if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|OO", (char**)kwlist,
&dtype_requested, &copy)) {
return NULL;
}
PyObject* np_module = PyImport_ImportModule("numpy");
if (np_module == NULL) {
PyErr_Format(PyExc_ImportError, "Unable to import numpy.");
return NULL;
}
PyUpb_RepeatedContainer* self = (PyUpb_RepeatedContainer*)(_self);
const upb_FieldDef* f = PyUpb_RepeatedContainer_GetField(self);
upb_CType c_type = upb_FieldDef_CType(f);
// string repeated
if (c_type == kUpb_CType_String || c_type == kUpb_CType_Bytes) {
nparray = ConstructArrayByIteration(_self, np_module);
if (dtype_requested == NULL) {
if (c_type == kUpb_CType_String) {
dtype_requested = PyObject_CallMethod(np_module, "dtype", "s", "str");
} else {
dtype_requested = PyObject_CallMethod(np_module, "dtype", "s", "S");
}
default_dtype = dtype_requested;
}
return_value = NpArrayAsType(nparray, dtype_requested, false);
goto ret;
}
// non-string scalar repeated
if (dtype_requested == NULL) {
char* returned_dtype = GetDefaultDTypeStr(c_type);
if (returned_dtype == NULL) {
PyErr_SetString(PyExc_SystemError,
"Implementation error: Unhandled case in AsNpArray.");
return NULL;
}
if (PyUpb_RepeatedContainer_Length(_self) == 0) {
return_value =
PyObject_CallMethod(np_module, "empty", "is", 0, returned_dtype);
goto ret;
}
dtype_requested =
PyObject_CallMethod(np_module, "dtype", "s", returned_dtype);
// For ref-counting.
default_dtype = dtype_requested;
}
nparray = CreateArrayFromView(_self, np_module);
if (nparray == NULL) {
goto ret;
}
return_value = NpArrayAsType(nparray, dtype_requested, true);
ret:
Py_XDECREF(default_dtype);
Py_XDECREF(nparray);
Py_DECREF(np_module);
assert(!PyErr_Occurred());
return return_value;
}
static PyMethodDef PyUpb_RepeatedScalarContainer_Methods[] = {
{"__deepcopy__", PyUpb_RepeatedContainer_DeepCopy, METH_VARARGS,
"Makes a deep copy of the class."},
{"__array__", (PyCFunction)PyUpb_RepeatedScalarContainer_AsNpArray,
METH_VARARGS | METH_KEYWORDS, "Returns a np.array."},
{"__reduce__", PyUpb_RepeatedScalarContainer_Reduce, METH_NOARGS,
"Outputs picklable representation of the repeated field."},
{"append", PyUpb_RepeatedScalarContainer_Append, METH_O,
"Appends an object to the repeated container."},
{"extend", PyUpb_RepeatedContainer_Extend, METH_O,
"Appends objects to the repeated container."},
{"insert", PyUpb_RepeatedContainer_Insert, METH_VARARGS,
"Inserts an object at the specified position in the container."},
{"pop", PyUpb_RepeatedContainer_Pop, METH_VARARGS,
"Removes an object from the repeated container and returns it."},
{"remove", PyUpb_RepeatedContainer_Remove, METH_O,
"Removes an object from the repeated container."},
{"sort", (PyCFunction)PyUpb_RepeatedContainer_Sort,
METH_VARARGS | METH_KEYWORDS, "Sorts the repeated container."},
{"reverse", (PyCFunction)PyUpb_RepeatedContainer_Reverse, METH_NOARGS,
"Reverses elements order of the repeated container."},
{"clear", (PyCFunction)PyUpb_RepeatedContainer_Clear, METH_NOARGS,
"Clears repeated container."},
{"MergeFrom", PyUpb_RepeatedContainer_MergeFrom, METH_O,
"Merges a repeated container into the current container."},
{NULL, NULL}};
static PyType_Slot PyUpb_RepeatedScalarContainer_Slots[] = {
{Py_tp_dealloc, PyUpb_RepeatedContainer_Dealloc},
{Py_tp_methods, PyUpb_RepeatedScalarContainer_Methods},
{Py_tp_new, PyUpb_Forbidden_New},
{Py_tp_repr, PyUpb_RepeatedContainer_Repr},
{Py_sq_length, PyUpb_RepeatedContainer_Length},
{Py_sq_item, PyUpb_RepeatedContainer_Item},
{Py_sq_ass_item, PyUpb_RepeatedScalarContainer_AssignItem},
{Py_mp_length, PyUpb_RepeatedContainer_Length},
{Py_mp_subscript, PyUpb_RepeatedContainer_Subscript},
{Py_mp_ass_subscript, PyUpb_RepeatedContainer_AssignSubscript},
{Py_tp_richcompare, PyUpb_RepeatedContainer_RichCompare},
{Py_tp_hash, PyObject_HashNotImplemented},
{0, NULL}};
static PyType_Spec PyUpb_RepeatedScalarContainer_Spec = {
PYUPB_MODULE_NAME ".RepeatedScalarContainer",
sizeof(PyUpb_RepeatedContainer),
0, // tp_itemsize
Py_TPFLAGS_DEFAULT,
PyUpb_RepeatedScalarContainer_Slots,
};
// -----------------------------------------------------------------------------
// Top Level
// -----------------------------------------------------------------------------
static bool PyUpb_Repeated_RegisterAsSequence(PyUpb_ModuleState* state) {
PyObject* collections = NULL;
PyObject* seq = NULL;
PyObject* ret1 = NULL;
PyObject* ret2 = NULL;
PyTypeObject* type1 = state->repeated_scalar_container_type;
PyTypeObject* type2 = state->repeated_composite_container_type;
bool ok = (collections = PyImport_ImportModule("collections.abc")) &&
(seq = PyObject_GetAttrString(collections, "MutableSequence")) &&
(ret1 = PyObject_CallMethod(seq, "register", "O", type1)) &&
(ret2 = PyObject_CallMethod(seq, "register", "O", type2));
Py_XDECREF(collections);
Py_XDECREF(seq);
Py_XDECREF(ret1);
Py_XDECREF(ret2);
return ok;
}
bool PyUpb_Repeated_Init(PyObject* m) {
PyUpb_ModuleState* state = PyUpb_ModuleState_GetFromModule(m);
state->repeated_composite_container_type =
PyUpb_AddClass(m, &PyUpb_RepeatedCompositeContainer_Spec);
state->repeated_scalar_container_type =
PyUpb_AddClass(m, &PyUpb_RepeatedScalarContainer_Spec);
return state->repeated_composite_container_type &&
state->repeated_scalar_container_type &&
PyUpb_Repeated_RegisterAsSequence(state);
}
#include "upb/port/undef.inc"