Google Protocol Buffers, how to handle multiple me

问题:

Is it possible to get the Type of the serialized Protocol Buffer message?

I have this example

option java_outer_classname="ProtoUser";

message User {
    required int32  id = 1; 
    required string name = 2;
    required string firstname = 3;
    required string lastname = 4;
    required string ssn= 5; 
}

message Address {
    required int32 id = 1;
    required string country = 2 [default = "US"];; 
    optional string state = 3;
    optional string city = 4;
    optional string street = 5;
    optional string zip = 6;
}

In Java I have this code

Address addr = ProtoUser.Address.newBuilder().setCity("Weston").setCountry("USA").setId(1).setState("FL").setStreet("123 Lakeshore").setZip("90210")
            .build();

    User user = ProtoUser.User.newBuilder().setId(1).setFirstname("Luis").setLastname("Atencio").setName("luisat").setSsn("555-555-5555").build();

if(....){
    FileOutputStream output = new FileOutputStream("out1.ser");
    user.writeTo(output);
    output.close();
}else{
    FileOutputStream output = new FileOutputStream("out1.ser");
    addr.writeTo(output);
    output.close();
}

Now, can I determine if the file contains a Address or a User? What is the common way to handle multiple Message-Types? How can I determine which Message-Type I have received?

回答1:

We cannot determine if the file contains a Address or a User. Because there is not type information encoded in the data.

To handle multiple Message-Types, you can use meta data like:

• Extension of the filename
• Headers in HTTP
• Specific frame header in frame base stream protocol
• ...

回答2:

Either you:

1. Include the information whether this is an address or a name in the file-ending (if files only contain one type at a time)
2. Explicitly specify the type of the packet in the serialized form, i.e. add a field required int32 type and deduce the type from that field. (if both types are included in one file at a time)
3. Devise a specific outer message format that explicitly includes this information and wraps the protocol buffer.

Whichever is appropriate, if you multiplex them onto one channel - which you do by choosing the same file-ending - you have to demultiplex them again when you receive them.