message.h

Tries to downcast this message to a generated message type.

Returns nullptr if this class is not an instance of T. This works even if RTTI is disabled.

This also has the effect of creating a strong reference to T that will prevent the linker from stripping it out at link time. This can be important if you are using a DynamicMessageFactory that delegates to the generated factory.

Call this function to ensure that this message's reflection is linked into the binary:

google::protobuf::LinkMessageReflection<FooMessage>();

This will ensure that the following lookup will succeed:

DescriptorPool::generated_pool()->FindMessageTypeByName("FooMessage");

As a side-effect, it will also guarantee that anything else from the same .proto file will also be available for lookup in the generated pool.

This function does not actually register the message, so it does not need to be called before the lookup. However it does need to occur in a function that cannot be stripped from the binary (ie. it must be reachable from main).

Best practice is to call this function as close as possible to where the reflection is actually needed. This function is very cheap to call, so you should not need to worry about its runtime overhead except in the tightest of loops (on x86-64 it compiles into two "mov" instructions).

Construct a new instance of the same type.

Ownership is passed to the caller. (This is also defined in MessageLite, but is defined again here for return-type covariance.)

Construct a new instance on the arena.

Ownership is passed to the caller if arena is a nullptr. Default implementation allows for API compatibility during the Arena transition.

Make this message into a copy of the given message.

The given message must have the same descriptor, but need not necessarily be the same class. By default this is just implemented as "Clear(); MergeFrom(from);".

Merge the fields from the given message into this message.

Singular fields will be overwritten, if specified in from, except for embedded messages which will be merged. Repeated fields will be concatenated. The given message must be of the same type as this message (i.e. the exact same class).

Verifies that IsInitialized() returns true.

GOOGLE_CHECK-fails otherwise, with a nice error message.

Slowly build a list of all required fields that are not set.

This is much, much slower than IsInitialized() as it is implemented purely via reflection. Generally, you should not call this unless you have already determined that an error exists by calling IsInitialized().

Clears all unknown fields from this message and all embedded messages.

Normally, if unknown tag numbers are encountered when parsing a message, the tag and value are stored in the message's UnknownFieldSet and then written back out when the message is serialized. This allows servers which simply route messages to other servers to pass through messages that have new field definitions which they don't yet know about. However, this behavior can have security implications. To avoid it, call this method after parsing.

See Reflection::GetUnknownFields() for more on unknown fields.

Computes (an estimate of) the total number of bytes currently used for storing the message in memory.

The default implementation calls the Reflection object's SpaceUsed() method.

SpaceUsed() is noticeably slower than ByteSize(), as it is implemented using reflection (rather than the generated code implementation for ByteSize()). Like ByteSize(), its CPU time is linear in the number of fields defined for the proto.

Clear all fields of the message and set them to their default values.

Clear() avoids freeing memory, assuming that any memory allocated to hold parts of the message will be needed again to hold the next message. If you actually want to free the memory used by a Message, you must delete it.

Returns whether all required fields have been set.

Note that required fields no longer exist starting in proto3.

If |other| is the exact same class as this, calls MergeFrom().

Otherwise, results are undefined (probably crash).

Computes the serialized size of the message.

This recursively calls ByteSizeLong() on all embedded messages.

ByteSizeLong() is generally linear in the number of fields defined for the proto.

Fast path when conditions match (ie.

non-deterministic) uint8* _InternalSerialize(uint8* ptr) const;

Get a non-owning pointer to a Descriptor for this message's type.

This describes what fields the message contains, the types of those fields, etc. This object remains property of the Message.

Get a non-owning pointer to the Reflection interface for this Message, which can be used to read and modify the fields of the Message dynamically (in other words, without knowing the message type at compile time).

This object remains property of the Message.

Get the UnknownFieldSet for the message.

This contains fields which were seen when the Message was parsed but were not recognized according to the Message's definition.

Get a mutable pointer to the UnknownFieldSet for the message.

This contains fields which were seen when the Message was parsed but were not recognized according to the Message's definition.

Check if the oneof is set.

Returns true if any field in oneof is set, false otherwise.

Removes the last element of a repeated field.

We don't provide a way to remove any element other than the last because it invites inefficient use, such as O(n^2) filtering loops that should have been O(n). If you want to remove an element other than the last, the best way to do it is to re-arrange the elements (using Swap()) so that the one you want removed is at the end, then call RemoveLast().

Removes the last element of a repeated message field, and returns the pointer to the caller.

Caller takes ownership of the returned pointer.

List all fields of the message which are currently set, except for unknown fields, but including extension known to the parser (i.e.

compiled in). Singular fields will only be listed if HasField(field) would return true and repeated fields will only be listed if FieldSize(field) would return non-zero. Fields (both normal fields and extension fields) will be listed ordered by field number. Use Reflection::GetUnknownFields() or message.unknown_fields() to also get access to fields/extensions unknown to the parser.

GetEnumValue() returns an enum field's value as an integer rather than an EnumValueDescriptor*.

If the integer value does not correspond to a known value descriptor, a new value descriptor is created. (Such a value will only be present when the new unknown-enum-value semantics are enabled for a message.)

Get a string value without copying, if possible.

GetString() necessarily returns a copy of the string. This can be inefficient when the std::string is already stored in a std::string object in the underlying message. GetStringReference() will return a reference to the underlying std::string in this case. Otherwise, it will copy the string into *scratch and return that.

Note: It is perfectly reasonable and useful to write code like:

str = reflection->GetStringReference(message, field, &str);

This line would ensure that only one copy of the string is made regardless of the field's underlying representation. When initializing a newly-constructed string, though, it's just as fast and more readable to use code like:

std::string str = reflection->GetString(message, field);

Set an enum field's value with an integer rather than EnumValueDescriptor.

For proto3 this is just setting the enum field to the value specified, for proto2 it's more complicated. If value is a known enum value the field is set as usual. If the value is unknown then it is added to the unknown field set. Note this matches the behavior of parsing unknown enum values. If multiple calls with unknown values happen than they are all added to the unknown field set in order of the calls.

Get a mutable pointer to a field with a message type.

If a MessageFactory is provided, it will be used to construct instances of the sub-message; otherwise, the default factory is used. If the field is an extension that does not live in the same pool as the containing message's descriptor (e.g. it lives in an overlay pool), then a MessageFactory must be provided. If you have no idea what that meant, then you probably don't need to worry about it (don't provide a MessageFactory). WARNING: If the FieldDescriptor is for a compiled-in extension, then factory->GetPrototype(field->message_type()) MUST return an instance of the compiled-in class for this type, NOT DynamicMessage.

Replaces the message specified by 'field' with the already-allocated object sub_message, passing ownership to the message.

If the field contained a message, that message is deleted. If sub_message is nullptr, the field is cleared.

Similar to SetAllocatedMessage, but omits all internal safety and ownership checks.

This method should only be used when the objects are on the same arena or paired with a call to UnsafeArenaReleaseMessage.

Releases the message specified by 'field' and returns the pointer, ReleaseMessage() will return the message the message object if it exists.

Otherwise, it may or may not return nullptr. In any case, if the return value is non-null, the caller takes ownership of the pointer. If the field existed (HasField() is true), then the returned pointer will be the same as the pointer returned by MutableMessage(). This function has the same effect as ClearField().

Similar to ReleaseMessage, but omits all internal safety and ownership checks.

This method should only be used when the objects are on the same arena or paired with a call to UnsafeArenaSetAllocatedMessage.

GetRepeatedEnumValue() returns an enum field's value as an integer rather than an EnumValueDescriptor*.

If the integer value does not correspond to a known value descriptor, a new value descriptor is created. (Such a value will only be present when the new unknown-enum-value semantics are enabled for a message.)

Set an enum field's value with an integer rather than EnumValueDescriptor.

For proto3 this is just setting the enum field to the value specified, for proto2 it's more complicated. If value is a known enum value the field is set as usual. If the value is unknown then it is added to the unknown field set. Note this matches the behavior of parsing unknown enum values. If multiple calls with unknown values happen than they are all added to the unknown field set in order of the calls.

Add an integer value to a repeated enum field rather than EnumValueDescriptor.

For proto3 this is just setting the enum field to the value specified, for proto2 it's more complicated. If value is a known enum value the field is set as usual. If the value is unknown then it is added to the unknown field set. Note this matches the behavior of parsing unknown enum values. If multiple calls with unknown values happen than they are all added to the unknown field set in order of the calls.

Get a RepeatedFieldRef object that can be used to read the underlying repeated field.

The type parameter T must be set according to the field's cpp type. The following table shows the mapping from cpp type to acceptable T.

field->cpp_type()      T
CPPTYPE_INT32        int32
CPPTYPE_UINT32       uint32
CPPTYPE_INT64        int64
CPPTYPE_UINT64       uint64
CPPTYPE_DOUBLE       double
CPPTYPE_FLOAT        float
CPPTYPE_BOOL         bool
CPPTYPE_ENUM         generated enum type or int32
CPPTYPE_STRING       std::string
CPPTYPE_MESSAGE      generated message type or google::protobuf::Message

A RepeatedFieldRef object can be copied and the resulted object will point to the same repeated field in the same message. The object can be used as long as the message is not destroyed.

Note that to use this method users need to include the header file "reflection.h" (which defines the RepeatedFieldRef class templates).

DEPRECATED.

Please use GetRepeatedFieldRef().

for T = Cord and all protobuf scalar types except enums.

DEPRECATED.

Please use GetMutableRepeatedFieldRef().

for T = Cord and all protobuf scalar types except enums.

DEPRECATED.

Please use GetRepeatedFieldRef().

for T = std::string, google::protobuf::internal::StringPieceField

google::protobuf::Message & descendants.

DEPRECATED.

Please use GetMutableRepeatedFieldRef().

for T = std::string, google::protobuf::internal::StringPieceField

google::protobuf::Message & descendants.

Try to find an extension of this message type by fully-qualified field name.

Returns nullptr if no extension is known for this name or number.

Try to find an extension of this message type by field number.

Returns nullptr if no extension is known for this name or number.

Does this message support storing arbitrary integer values in enum fields? If |true|, GetEnumValue/SetEnumValue and associated repeated-field versions take arbitrary integer values, and the legacy GetEnum() getter will dynamically create an EnumValueDescriptor for any integer value without one.

If |false|, setting an unknown enum value via the integer-based setters results in undefined behavior (in practice, GOOGLE_DCHECK-fails).

Generic code that uses reflection to handle messages with enum fields should check this flag before using the integer-based setter, and either downgrade to a compatible value or use the UnknownFieldSet if not. For example:

int new_value = GetValueFromApplicationLogic();
if (reflection->SupportsUnknownEnumValues()) {
  reflection->SetEnumValue(message, field, new_value);
} else {
  if (field_descriptor->enum_type()->
          FindValueByNumber(new_value) != nullptr) {
    reflection->SetEnumValue(message, field, new_value);
  } else if (emit_unknown_enum_values) {
    reflection->MutableUnknownFields(message)->AddVarint(
        field->number(), new_value);
  } else {
    // convert value to a compatible/default value.
    new_value = CompatibleDowngrade(new_value);
    reflection->SetEnumValue(message, field, new_value);
  }
}

Returns the MessageFactory associated with this message.

This can be useful for determining if a message is a generated message or not, for example:

if (message->GetReflection()->GetMessageFactory() ==
    google::protobuf::MessageFactory::generated_factory()) {
  // This is a generated message.
}

It can also be used to create more messages of this type, though Message::New() is an easier way to accomplish this.

Given a Descriptor, gets or constructs the default (prototype) Message of that type.

You can then call that message's New() method to construct a mutable message of that type.

Calling this method twice with the same Descriptor returns the same object. The returned object remains property of the factory. Also, any objects created by calling the prototype's New() method share some data with the prototype, so these must be destroyed before the MessageFactory is destroyed.

The given descriptor must outlive the returned message, and hence must outlive the MessageFactory.

Some implementations do not support all types. GetPrototype() will return nullptr if the descriptor passed in is not supported.

This method may or may not be thread-safe depending on the implementation. Each implementation should document its own degree thread-safety.

Gets a MessageFactory which supports all generated, compiled-in messages.

In other words, for any compiled-in type FooMessage, the following is true:

MessageFactory::generated_factory()->GetPrototype(
  FooMessage::descriptor()) == FooMessage::default_instance()

This factory supports all types which are found in DescriptorPool::generated_pool(). If given a descriptor from any other pool, GetPrototype() will return nullptr. (You can also check if a descriptor is for a generated message by checking if descriptor->file()->pool() == DescriptorPool::generated_pool().)

This factory is 100% thread-safe; calling GetPrototype() does not modify any shared data.

This factory is a singleton. The caller must not delete the object.

For internal use only: Registers a .proto file at static initialization time, to be placed in generated_factory.

The first time GetPrototype() is called with a descriptor from this file, |register_messages| will be called, with the file name as the parameter. It must call InternalRegisterGeneratedMessage() (below) to register each message type in the file. This strange mechanism is necessary because descriptors are built lazily, so we can't register types by their descriptor until we know that the descriptor exists. |filename| must be a permanent string.

For internal use only: Registers a message type.

Called only by the functions which are registered with InternalRegisterGeneratedFile(), above.