Down integrate from Google internal.

11 files changed, 161 insertions, 92 deletions

diff --git a/python/google/protobuf/pyext/descriptor.cc b/python/google/protobuf/pyext/descriptor.cc
index 07550706..23557538 100644
--- a/python/google/protobuf/pyext/descriptor.cc
+++ b/python/google/protobuf/pyext/descriptor.cc
@@ -200,8 +200,8 @@ static PyObject* GetOrBuildOptions(const DescriptorClass *descriptor) {
   // read-only instance.
   const Message& options(descriptor->options());
   const Descriptor *message_type = options.GetDescriptor();
-  PyObject* message_class(cdescriptor_pool::GetMessageClass(
-      pool, message_type));
+  CMessageClass* message_class(
+      cdescriptor_pool::GetMessageClass(pool, message_type));
   if (message_class == NULL) {
     // The Options message was not found in the current DescriptorPool.
     // In this case, there cannot be extensions to these options, and we can
@@ -215,7 +215,8 @@ static PyObject* GetOrBuildOptions(const DescriptorClass *descriptor) {
                  message_type->full_name().c_str());
     return NULL;
   }
-  ScopedPyObjectPtr value(PyEval_CallObject(message_class, NULL));
+  ScopedPyObjectPtr value(
+      PyEval_CallObject(message_class->AsPyObject(), NULL));
   if (value == NULL) {
     return NULL;
   }
@@ -433,11 +434,11 @@ static PyObject* GetConcreteClass(PyBaseDescriptor* self, void *closure) {
   // which contains this descriptor.
   // This might not be the one you expect! For example the returned object does
   // not know about extensions defined in a custom pool.
-  PyObject* concrete_class(cdescriptor_pool::GetMessageClass(
+  CMessageClass* concrete_class(cdescriptor_pool::GetMessageClass(
       GetDescriptorPool_FromPool(_GetDescriptor(self)->file()->pool()),
       _GetDescriptor(self)));
   Py_XINCREF(concrete_class);
-  return concrete_class;
+  return concrete_class->AsPyObject();
 }
 
 static PyObject* GetFieldsByName(PyBaseDescriptor* self, void *closure) {
diff --git a/python/google/protobuf/pyext/descriptor_pool.cc b/python/google/protobuf/pyext/descriptor_pool.cc
index 0bc76bc9..1faff96b 100644
--- a/python/google/protobuf/pyext/descriptor_pool.cc
+++ b/python/google/protobuf/pyext/descriptor_pool.cc
@@ -190,8 +190,8 @@ PyObject* FindMessageByName(PyDescriptorPool* self, PyObject* arg) {
 
 // Add a message class to our database.
 int RegisterMessageClass(PyDescriptorPool* self,
-                         const Descriptor *message_descriptor,
-                         PyObject *message_class) {
+                         const Descriptor* message_descriptor,
+                         CMessageClass* message_class) {
   Py_INCREF(message_class);
   typedef PyDescriptorPool::ClassesByMessageMap::iterator iterator;
   std::pair<iterator, bool> ret = self->classes_by_descriptor->insert(
@@ -205,8 +205,8 @@ int RegisterMessageClass(PyDescriptorPool* self,
 }
 
 // Retrieve the message class added to our database.
-PyObject *GetMessageClass(PyDescriptorPool* self,
-                          const Descriptor *message_descriptor) {
+CMessageClass* GetMessageClass(PyDescriptorPool* self,
+                               const Descriptor* message_descriptor) {
   typedef PyDescriptorPool::ClassesByMessageMap::iterator iterator;
   iterator ret = self->classes_by_descriptor->find(message_descriptor);
   if (ret == self->classes_by_descriptor->end()) {
diff --git a/python/google/protobuf/pyext/descriptor_pool.h b/python/google/protobuf/pyext/descriptor_pool.h
index 16bc910c..2a42c112 100644
--- a/python/google/protobuf/pyext/descriptor_pool.h
+++ b/python/google/protobuf/pyext/descriptor_pool.h
@@ -42,6 +42,9 @@ class MessageFactory;
 
 namespace python {
 
+// The (meta) type of all Messages classes.
+struct CMessageClass;
+
 // Wraps operations to the global DescriptorPool which contains information
 // about all messages and fields.
 //
@@ -78,7 +81,7 @@ typedef struct PyDescriptorPool {
   //
   // Descriptor pointers stored here are owned by the DescriptorPool above.
   // Python references to classes are owned by this PyDescriptorPool.
-  typedef hash_map<const Descriptor*, PyObject*> ClassesByMessageMap;
+  typedef hash_map<const Descriptor*, CMessageClass*> ClassesByMessageMap;
   ClassesByMessageMap* classes_by_descriptor;
 
   // Cache the options for any kind of descriptor.
@@ -101,14 +104,14 @@ const Descriptor* FindMessageTypeByName(PyDescriptorPool* self,
 // On error, returns -1 with a Python exception set.
 int RegisterMessageClass(PyDescriptorPool* self,
                          const Descriptor* message_descriptor,
-                         PyObject* message_class);
+                         CMessageClass* message_class);
 
 // Retrieves the Python class registered with the given message descriptor.
 //
 // Returns a *borrowed* reference if found, otherwise returns NULL with an
 // exception set.
-PyObject* GetMessageClass(PyDescriptorPool* self,
-                          const Descriptor* message_descriptor);
+CMessageClass* GetMessageClass(PyDescriptorPool* self,
+                               const Descriptor* message_descriptor);
 
 // The functions below are also exposed as methods of the DescriptorPool type.
 
diff --git a/python/google/protobuf/pyext/extension_dict.cc b/python/google/protobuf/pyext/extension_dict.cc
index 555bd293..21bbb8c2 100644
--- a/python/google/protobuf/pyext/extension_dict.cc
+++ b/python/google/protobuf/pyext/extension_dict.cc
@@ -130,7 +130,7 @@ PyObject* subscript(ExtensionDict* self, PyObject* key) {
 
   if (descriptor->label() == FieldDescriptor::LABEL_REPEATED) {
     if (descriptor->cpp_type() == FieldDescriptor::CPPTYPE_MESSAGE) {
-      PyObject *message_class = cdescriptor_pool::GetMessageClass(
+      CMessageClass* message_class = cdescriptor_pool::GetMessageClass(
           cmessage::GetDescriptorPoolForMessage(self->parent),
           descriptor->message_type());
       if (message_class == NULL) {
@@ -239,6 +239,21 @@ PyObject* _FindExtensionByName(ExtensionDict* self, PyObject* name) {
   }
 }
 
+PyObject* _FindExtensionByNumber(ExtensionDict* self, PyObject* number) {
+  ScopedPyObjectPtr extensions_by_number(PyObject_GetAttrString(
+      reinterpret_cast<PyObject*>(self->parent), "_extensions_by_number"));
+  if (extensions_by_number == NULL) {
+    return NULL;
+  }
+  PyObject* result = PyDict_GetItem(extensions_by_number.get(), number);
+  if (result == NULL) {
+    Py_RETURN_NONE;
+  } else {
+    Py_INCREF(result);
+    return result;
+  }
+}
+
 ExtensionDict* NewExtensionDict(CMessage *parent) {
   ExtensionDict* self = reinterpret_cast<ExtensionDict*>(
       PyType_GenericAlloc(&ExtensionDict_Type, 0));
@@ -271,6 +286,8 @@ static PyMethodDef Methods[] = {
   EDMETHOD(HasExtension, METH_O, "Checks if the object has an extension."),
   EDMETHOD(_FindExtensionByName, METH_O,
            "Finds an extension by name."),
+  EDMETHOD(_FindExtensionByNumber, METH_O,
+           "Finds an extension by field number."),
   { NULL, NULL }
 };
 
diff --git a/python/google/protobuf/pyext/extension_dict.h b/python/google/protobuf/pyext/extension_dict.h
index 1e7f6f7b..049d2e45 100644
--- a/python/google/protobuf/pyext/extension_dict.h
+++ b/python/google/protobuf/pyext/extension_dict.h
@@ -123,6 +123,12 @@ PyObject* ClearExtension(ExtensionDict* self,
 // Returns a new reference.
 PyObject* _FindExtensionByName(ExtensionDict* self, PyObject* name);
 
+// Gets an extension from the dict given the extension field number as
+// opposed to descriptor.
+//
+// Returns a new reference.
+PyObject* _FindExtensionByNumber(ExtensionDict* self, PyObject* number);
+
 }  // namespace extension_dict
 }  // namespace python
 }  // namespace protobuf
diff --git a/python/google/protobuf/pyext/map_container.cc b/python/google/protobuf/pyext/map_container.cc
index df9138a4..e022406d 100644
--- a/python/google/protobuf/pyext/map_container.cc
+++ b/python/google/protobuf/pyext/map_container.cc
@@ -32,6 +32,11 @@
 
 #include <google/protobuf/pyext/map_container.h>
 
+#include <memory>
+#ifndef _SHARED_PTR_H
+#include <google/protobuf/stubs/shared_ptr.h>
+#endif
+
 #include <google/protobuf/stubs/logging.h>
 #include <google/protobuf/stubs/common.h>
 #include <google/protobuf/stubs/scoped_ptr.h>
@@ -70,7 +75,7 @@ class MapReflectionFriend {
 struct MapIterator {
   PyObject_HEAD;
 
-  scoped_ptr< ::google::protobuf::MapIterator> iter;
+  google::protobuf::scoped_ptr< ::google::protobuf::MapIterator> iter;
 
   // A pointer back to the container, so we can notice changes to the version.
   // We own a ref on this.
@@ -610,8 +615,7 @@ static PyObject* GetCMessage(MessageMapContainer* self, Message* message) {
   PyObject* ret = PyDict_GetItem(self->message_dict, key.get());
 
   if (ret == NULL) {
-    CMessage* cmsg = cmessage::NewEmptyMessage(self->subclass_init,
-                                               message->GetDescriptor());
+    CMessage* cmsg = cmessage::NewEmptyMessage(self->message_class);
     ret = reinterpret_cast<PyObject*>(cmsg);
 
     if (cmsg == NULL) {
@@ -634,7 +638,7 @@ static PyObject* GetCMessage(MessageMapContainer* self, Message* message) {
 
 PyObject* NewMessageMapContainer(
     CMessage* parent, const google::protobuf::FieldDescriptor* parent_field_descriptor,
-    PyObject* concrete_class) {
+    CMessageClass* message_class) {
   if (!CheckFieldBelongsToMessage(parent_field_descriptor, parent->message)) {
     return NULL;
   }
@@ -669,8 +673,8 @@ PyObject* NewMessageMapContainer(
                         "Could not allocate message dict.");
   }
 
-  Py_INCREF(concrete_class);
-  self->subclass_init = concrete_class;
+  Py_INCREF(message_class);
+  self->message_class = message_class;
 
   if (self->key_field_descriptor == NULL ||
       self->value_field_descriptor == NULL) {
@@ -763,6 +767,7 @@ static void MessageMapDealloc(PyObject* _self) {
   MessageMapContainer* self = GetMessageMap(_self);
   self->owner.reset();
   Py_DECREF(self->message_dict);
+  Py_DECREF(self->message_class);
   Py_TYPE(_self)->tp_free(_self);
 }
 
diff --git a/python/google/protobuf/pyext/map_container.h b/python/google/protobuf/pyext/map_container.h
index 27ee6dbd..b11dfa34 100644
--- a/python/google/protobuf/pyext/map_container.h
+++ b/python/google/protobuf/pyext/map_container.h
@@ -55,6 +55,7 @@ using internal::shared_ptr;
 namespace python {
 
 struct CMessage;
+struct CMessageClass;
 
 // This struct is used directly for ScalarMap, and is the base class of
 // MessageMapContainer, which is used for MessageMap.
@@ -104,8 +105,8 @@ struct MapContainer {
 };
 
 struct MessageMapContainer : public MapContainer {
-  // A callable that is used to create new child messages.
-  PyObject* subclass_init;
+  // The type used to create new child messages.
+  CMessageClass* message_class;
 
   // A dict mapping Message* -> CMessage.
   PyObject* message_dict;
@@ -132,7 +133,7 @@ extern PyObject* NewScalarMapContainer(
 // field descriptor.
 extern PyObject* NewMessageMapContainer(
     CMessage* parent, const FieldDescriptor* parent_field_descriptor,
-    PyObject* concrete_class);
+    CMessageClass* message_class);
 
 }  // namespace python
 }  // namespace protobuf
diff --git a/python/google/protobuf/pyext/message.cc b/python/google/protobuf/pyext/message.cc
index 6d7b2b0f..83c151ff 100644
--- a/python/google/protobuf/pyext/message.cc
+++ b/python/google/protobuf/pyext/message.cc
@@ -98,31 +98,6 @@ static PyObject* PythonMessage_class;
 static PyObject* kEmptyWeakref;
 static PyObject* WKT_classes = NULL;
 
-// Defines the Metaclass of all Message classes.
-// It allows us to cache some C++ pointers in the class object itself, they are
-// faster to extract than from the type's dictionary.
-
-struct PyMessageMeta {
-  // This is how CPython subclasses C structures: the base structure must be
-  // the first member of the object.
-  PyHeapTypeObject super;
-
-  // C++ descriptor of this message.
-  const Descriptor* message_descriptor;
-
-  // Owned reference, used to keep the pointer above alive.
-  PyObject* py_message_descriptor;
-
-  // The Python DescriptorPool used to create the class. It is needed to resolve
-  // fields descriptors, including extensions fields; its C++ MessageFactory is
-  // used to instantiate submessages.
-  // This can be different from DESCRIPTOR.file.pool, in the case of a custom
-  // DescriptorPool which defines new extensions.
-  // We own the reference, because it's important to keep the descriptors and
-  // factory alive.
-  PyDescriptorPool* py_descriptor_pool;
-};
-
 namespace message_meta {
 
 static int InsertEmptyWeakref(PyTypeObject* base);
@@ -173,10 +148,6 @@ static int AddDescriptors(PyObject* cls, const Descriptor* descriptor) {
   }
 
   // For each enum set cls.<enum name> = EnumTypeWrapper(<enum descriptor>).
-  //
-  // The enum descriptor we get from
-  // <messagedescriptor>.enum_types_by_name[name]
-  // which was built previously.
   for (int i = 0; i < descriptor->enum_type_count(); ++i) {
     const EnumDescriptor* enum_descriptor = descriptor->enum_type(i);
     ScopedPyObjectPtr enum_type(
@@ -309,7 +280,7 @@ static PyObject* New(PyTypeObject* type,
   if (result == NULL) {
     return NULL;
   }
-  PyMessageMeta* newtype = reinterpret_cast<PyMessageMeta*>(result.get());
+  CMessageClass* newtype = reinterpret_cast<CMessageClass*>(result.get());
 
   // Insert the empty weakref into the base classes.
   if (InsertEmptyWeakref(
@@ -338,7 +309,7 @@ static PyObject* New(PyTypeObject* type,
 
   // Add the message to the DescriptorPool.
   if (cdescriptor_pool::RegisterMessageClass(newtype->py_descriptor_pool,
-                                             descriptor, result.get()) < 0) {
+                                             descriptor, newtype) < 0) {
     return NULL;
   }
 
@@ -349,7 +320,7 @@ static PyObject* New(PyTypeObject* type,
   return result.release();
 }
 
-static void Dealloc(PyMessageMeta *self) {
+static void Dealloc(CMessageClass *self) {
   Py_DECREF(self->py_message_descriptor);
   Py_DECREF(self->py_descriptor_pool);
   Py_TYPE(self)->tp_free(reinterpret_cast<PyObject*>(self));
@@ -378,10 +349,10 @@ static int InsertEmptyWeakref(PyTypeObject *base_type) {
 
 }  // namespace message_meta
 
-PyTypeObject PyMessageMeta_Type = {
+PyTypeObject CMessageClass_Type = {
   PyVarObject_HEAD_INIT(&PyType_Type, 0)
   FULL_MODULE_NAME ".MessageMeta",     // tp_name
-  sizeof(PyMessageMeta),               // tp_basicsize
+  sizeof(CMessageClass),               // tp_basicsize
   0,                                   // tp_itemsize
   (destructor)message_meta::Dealloc,   // tp_dealloc
   0,                                   // tp_print
@@ -419,16 +390,16 @@ PyTypeObject PyMessageMeta_Type = {
   message_meta::New,                   // tp_new
 };
 
-static PyMessageMeta* CheckMessageClass(PyTypeObject* cls) {
-  if (!PyObject_TypeCheck(cls, &PyMessageMeta_Type)) {
+static CMessageClass* CheckMessageClass(PyTypeObject* cls) {
+  if (!PyObject_TypeCheck(cls, &CMessageClass_Type)) {
     PyErr_Format(PyExc_TypeError, "Class %s is not a Message", cls->tp_name);
     return NULL;
   }
-  return reinterpret_cast<PyMessageMeta*>(cls);
+  return reinterpret_cast<CMessageClass*>(cls);
 }
 
 static const Descriptor* GetMessageDescriptor(PyTypeObject* cls) {
-  PyMessageMeta* type = CheckMessageClass(cls);
+  CMessageClass* type = CheckMessageClass(cls);
   if (type == NULL) {
     return NULL;
   }
@@ -783,9 +754,9 @@ namespace cmessage {
 
 PyDescriptorPool* GetDescriptorPoolForMessage(CMessage* message) {
   // No need to check the type: the type of instances of CMessage is always
-  // an instance of PyMessageMeta. Let's prove it with a debug-only check.
+  // an instance of CMessageClass. Let's prove it with a debug-only check.
   GOOGLE_DCHECK(PyObject_TypeCheck(message, &CMessage_Type));
-  return reinterpret_cast<PyMessageMeta*>(Py_TYPE(message))->py_descriptor_pool;
+  return reinterpret_cast<CMessageClass*>(Py_TYPE(message))->py_descriptor_pool;
 }
 
 MessageFactory* GetFactoryForMessage(CMessage* message) {
@@ -1090,6 +1061,10 @@ int InitAttributes(CMessage* self, PyObject* kwargs) {
                    PyString_AsString(name));
       return -1;
     }
+    if (value == Py_None) {
+      // field=None is the same as no field at all.
+      continue;
+    }
     if (descriptor->is_map()) {
       ScopedPyObjectPtr map(GetAttr(self, name));
       const FieldDescriptor* value_descriptor =
@@ -1220,9 +1195,9 @@ int InitAttributes(CMessage* self, PyObject* kwargs) {
 
 // Allocates an incomplete Python Message: the caller must fill self->message,
 // self->owner and eventually self->parent.
-CMessage* NewEmptyMessage(PyObject* type, const Descriptor *descriptor) {
+CMessage* NewEmptyMessage(CMessageClass* type) {
   CMessage* self = reinterpret_cast<CMessage*>(
-      PyType_GenericAlloc(reinterpret_cast<PyTypeObject*>(type), 0));
+      PyType_GenericAlloc(&type->super.ht_type, 0));
   if (self == NULL) {
     return NULL;
   }
@@ -1242,7 +1217,7 @@ CMessage* NewEmptyMessage(PyObject* type, const Descriptor *descriptor) {
 // Creates a new C++ message and takes ownership.
 static PyObject* New(PyTypeObject* cls,
                      PyObject* unused_args, PyObject* unused_kwargs) {
-  PyMessageMeta* type = CheckMessageClass(cls);
+  CMessageClass* type = CheckMessageClass(cls);
   if (type == NULL) {
     return NULL;
   }
@@ -1258,8 +1233,7 @@ static PyObject* New(PyTypeObject* cls,
     return NULL;
   }
 
-  CMessage* self = NewEmptyMessage(reinterpret_cast<PyObject*>(type),
-                                   message_descriptor);
+  CMessage* self = NewEmptyMessage(type);
   if (self == NULL) {
     return NULL;
   }
@@ -2023,10 +1997,34 @@ static PyObject* RegisterExtension(PyObject* cls,
     PyErr_SetString(PyExc_TypeError, "no extensions_by_number on class");
     return NULL;
   }
+
   ScopedPyObjectPtr number(PyObject_GetAttrString(extension_handle, "number"));
   if (number == NULL) {
     return NULL;
   }
+
+  // If the extension was already registered by number, check that it is the
+  // same.
+  existing_extension = PyDict_GetItem(extensions_by_number.get(), number.get());
+  if (existing_extension != NULL) {
+    const FieldDescriptor* existing_extension_descriptor =
+        GetExtensionDescriptor(existing_extension);
+    if (existing_extension_descriptor != descriptor) {
+      const Descriptor* msg_desc = GetMessageDescriptor(
+          reinterpret_cast<PyTypeObject*>(cls));
+      PyErr_Format(
+          PyExc_ValueError,
+          "Extensions \"%s\" and \"%s\" both try to extend message type "
+          "\"%s\" with field number %ld.",
+          existing_extension_descriptor->full_name().c_str(),
+          descriptor->full_name().c_str(),
+          msg_desc->full_name().c_str(),
+          PyInt_AsLong(number.get()));
+      return NULL;
+    }
+    // Nothing else to do.
+    Py_RETURN_NONE;
+  }
   if (PyDict_SetItem(extensions_by_number.get(), number.get(),
                      extension_handle) < 0) {
     return NULL;
@@ -2166,6 +2164,12 @@ static PyObject* ListFields(CMessage* self) {
   return all_fields.release();
 }
 
+static PyObject* DiscardUnknownFields(CMessage* self) {
+  AssureWritable(self);
+  self->message->DiscardUnknownFields();
+  Py_RETURN_NONE;
+}
+
 PyObject* FindInitializationErrors(CMessage* self) {
   Message* message = self->message;
   vector<string> errors;
@@ -2309,14 +2313,13 @@ PyObject* InternalGetSubMessage(
   const Message& sub_message = reflection->GetMessage(
       *self->message, field_descriptor, pool->message_factory);
 
-  PyObject *message_class = cdescriptor_pool::GetMessageClass(
+  CMessageClass* message_class = cdescriptor_pool::GetMessageClass(
       pool, field_descriptor->message_type());
   if (message_class == NULL) {
     return NULL;
   }
 
-  CMessage* cmsg = cmessage::NewEmptyMessage(message_class,
-                                             sub_message.GetDescriptor());
+  CMessage* cmsg = cmessage::NewEmptyMessage(message_class);
   if (cmsg == NULL) {
     return NULL;
   }
@@ -2585,6 +2588,8 @@ static PyMethodDef Methods[] = {
     "Clears a message field." },
   { "CopyFrom", (PyCFunction)CopyFrom, METH_O,
     "Copies a protocol message into the current message." },
+  { "DiscardUnknownFields", (PyCFunction)DiscardUnknownFields, METH_NOARGS,
+    "Discards the unknown fields." },
   { "FindInitializationErrors", (PyCFunction)FindInitializationErrors,
     METH_NOARGS,
     "Finds unset required fields." },
@@ -2654,7 +2659,7 @@ PyObject* GetAttr(CMessage* self, PyObject* name) {
     const Descriptor* entry_type = field_descriptor->message_type();
     const FieldDescriptor* value_type = entry_type->FindFieldByName("value");
     if (value_type->cpp_type() == FieldDescriptor::CPPTYPE_MESSAGE) {
-      PyObject* value_class = cdescriptor_pool::GetMessageClass(
+      CMessageClass* value_class = cdescriptor_pool::GetMessageClass(
           GetDescriptorPoolForMessage(self), value_type->message_type());
       if (value_class == NULL) {
         return NULL;
@@ -2677,7 +2682,7 @@ PyObject* GetAttr(CMessage* self, PyObject* name) {
   if (field_descriptor->label() == FieldDescriptor::LABEL_REPEATED) {
     PyObject* py_container = NULL;
     if (field_descriptor->cpp_type() == FieldDescriptor::CPPTYPE_MESSAGE) {
-      PyObject *message_class = cdescriptor_pool::GetMessageClass(
+      CMessageClass* message_class = cdescriptor_pool::GetMessageClass(
           GetDescriptorPoolForMessage(self), field_descriptor->message_type());
       if (message_class == NULL) {
         return NULL;
@@ -2749,7 +2754,7 @@ int SetAttr(CMessage* self, PyObject* name, PyObject* value) {
 }  // namespace cmessage
 
 PyTypeObject CMessage_Type = {
-  PyVarObject_HEAD_INIT(&PyMessageMeta_Type, 0)
+  PyVarObject_HEAD_INIT(&CMessageClass_Type, 0)
   FULL_MODULE_NAME ".CMessage",        // tp_name
   sizeof(CMessage),                    // tp_basicsize
   0,                                   //  tp_itemsize
@@ -2864,12 +2869,12 @@ bool InitProto2MessageModule(PyObject *m) {
   // Initialize constants defined in this file.
   InitGlobals();
 
-  PyMessageMeta_Type.tp_base = &PyType_Type;
-  if (PyType_Ready(&PyMessageMeta_Type) < 0) {
+  CMessageClass_Type.tp_base = &PyType_Type;
+  if (PyType_Ready(&CMessageClass_Type) < 0) {
     return false;
   }
   PyModule_AddObject(m, "MessageMeta",
-                     reinterpret_cast<PyObject*>(&PyMessageMeta_Type));
+                     reinterpret_cast<PyObject*>(&CMessageClass_Type));
 
   if (PyType_Ready(&CMessage_Type) < 0) {
     return false;
@@ -3077,9 +3082,10 @@ bool InitProto2MessageModule(PyObject *m) {
 }  // namespace protobuf
 
 static PyMethodDef ModuleMethods[] = {
-    {"SetAllowOversizeProtos",
-     (PyCFunction)google::protobuf::python::cmessage::SetAllowOversizeProtos,
-     METH_O, "Enable/disable oversize proto parsing."},
+  {"SetAllowOversizeProtos",
+    (PyCFunction)google::protobuf::python::cmessage::SetAllowOversizeProtos,
+    METH_O, "Enable/disable oversize proto parsing."},
+  { NULL, NULL}
 };
 
 #if PY_MAJOR_VERSION >= 3
diff --git a/python/google/protobuf/pyext/message.h b/python/google/protobuf/pyext/message.h
index c2b62649..9dce198f 100644
--- a/python/google/protobuf/pyext/message.h
+++ b/python/google/protobuf/pyext/message.h
@@ -116,12 +116,43 @@ typedef struct CMessage {
 
 extern PyTypeObject CMessage_Type;
 
+
+// The (meta) type of all Messages classes.
+// It allows us to cache some C++ pointers in the class object itself, they are
+// faster to extract than from the type's dictionary.
+
+struct CMessageClass {
+  // This is how CPython subclasses C structures: the base structure must be
+  // the first member of the object.
+  PyHeapTypeObject super;
+
+  // C++ descriptor of this message.
+  const Descriptor* message_descriptor;
+
+  // Owned reference, used to keep the pointer above alive.
+  PyObject* py_message_descriptor;
+
+  // The Python DescriptorPool used to create the class. It is needed to resolve
+  // fields descriptors, including extensions fields; its C++ MessageFactory is
+  // used to instantiate submessages.
+  // This can be different from DESCRIPTOR.file.pool, in the case of a custom
+  // DescriptorPool which defines new extensions.
+  // We own the reference, because it's important to keep the descriptors and
+  // factory alive.
+  PyDescriptorPool* py_descriptor_pool;
+
+  PyObject* AsPyObject() {
+    return reinterpret_cast<PyObject*>(this);
+  }
+};
+
+
 namespace cmessage {
 
 // Internal function to create a new empty Message Python object, but with empty
 // pointers to the C++ objects.
 // The caller must fill self->message, self->owner and eventually self->parent.
-CMessage* NewEmptyMessage(PyObject* type, const Descriptor* descriptor);
+CMessage* NewEmptyMessage(CMessageClass* type);
 
 // Release a submessage from its proto tree, making it a new top-level messgae.
 // A new message will be created if this is a read-only default instance.
diff --git a/python/google/protobuf/pyext/repeated_composite_container.cc b/python/google/protobuf/pyext/repeated_composite_container.cc
index b01123b4..4f339e77 100644
--- a/python/google/protobuf/pyext/repeated_composite_container.cc
+++ b/python/google/protobuf/pyext/repeated_composite_container.cc
@@ -107,8 +107,7 @@ static int UpdateChildMessages(RepeatedCompositeContainer* self) {
   for (Py_ssize_t i = child_length; i < message_length; ++i) {
     const Message& sub_message = reflection->GetRepeatedMessage(
         *(self->message), self->parent_field_descriptor, i);
-    CMessage* cmsg = cmessage::NewEmptyMessage(self->subclass_init,
-                                               sub_message.GetDescriptor());
+    CMessage* cmsg = cmessage::NewEmptyMessage(self->child_message_class);
     ScopedPyObjectPtr py_cmsg(reinterpret_cast<PyObject*>(cmsg));
     if (cmsg == NULL) {
       return -1;
@@ -140,8 +139,7 @@ static PyObject* AddToAttached(RepeatedCompositeContainer* self,
   Message* sub_message =
       message->GetReflection()->AddMessage(message,
                                            self->parent_field_descriptor);
-  CMessage* cmsg = cmessage::NewEmptyMessage(self->subclass_init,
-                                             sub_message->GetDescriptor());
+  CMessage* cmsg = cmessage::NewEmptyMessage(self->child_message_class);
   if (cmsg == NULL)
     return NULL;
 
@@ -168,7 +166,7 @@ static PyObject* AddToReleased(RepeatedCompositeContainer* self,
 
   // Create a new Message detached from the rest.
   PyObject* py_cmsg = PyEval_CallObjectWithKeywords(
-      self->subclass_init, NULL, kwargs);
+      self->child_message_class->AsPyObject(), NULL, kwargs);
   if (py_cmsg == NULL)
     return NULL;
 
@@ -506,7 +504,7 @@ int SetOwner(RepeatedCompositeContainer* self,
 PyObject *NewContainer(
     CMessage* parent,
     const FieldDescriptor* parent_field_descriptor,
-    PyObject *concrete_class) {
+    CMessageClass* concrete_class) {
   if (!CheckFieldBelongsToMessage(parent_field_descriptor, parent->message)) {
     return NULL;
   }
@@ -523,7 +521,7 @@ PyObject *NewContainer(
   self->parent_field_descriptor = parent_field_descriptor;
   self->owner = parent->owner;
   Py_INCREF(concrete_class);
-  self->subclass_init = concrete_class;
+  self->child_message_class = concrete_class;
   self->child_messages = PyList_New(0);
 
   return reinterpret_cast<PyObject*>(self);
@@ -531,7 +529,7 @@ PyObject *NewContainer(
 
 static void Dealloc(RepeatedCompositeContainer* self) {
   Py_CLEAR(self->child_messages);
-  Py_CLEAR(self->subclass_init);
+  Py_CLEAR(self->child_message_class);
   // TODO(tibell): Do we need to call delete on these objects to make
   // sure their destructors are called?
   self->owner.reset();
diff --git a/python/google/protobuf/pyext/repeated_composite_container.h b/python/google/protobuf/pyext/repeated_composite_container.h
index 442ce7e3..25463037 100644
--- a/python/google/protobuf/pyext/repeated_composite_container.h
+++ b/python/google/protobuf/pyext/repeated_composite_container.h
@@ -58,6 +58,7 @@ using internal::shared_ptr;
 namespace python {
 
 struct CMessage;
+struct CMessageClass;
 
 // A RepeatedCompositeContainer can be in one of two states: attached
 // or released.
@@ -94,8 +95,8 @@ typedef struct RepeatedCompositeContainer {
   // calling Clear() or ClearField() on the parent.
   Message* message;
 
-  // A callable that is used to create new child messages.
-  PyObject* subclass_init;
+  // The type used to create new child messages.
+  CMessageClass* child_message_class;
 
   // A list of child messages.
   PyObject* child_messages;
@@ -110,7 +111,7 @@ namespace repeated_composite_container {
 PyObject *NewContainer(
     CMessage* parent,
     const FieldDescriptor* parent_field_descriptor,
-    PyObject *concrete_class);
+    CMessageClass *child_message_class);
 
 // Appends a new CMessage to the container and returns it.  The
 // CMessage is initialized using the content of kwargs.