Message type "PrettyList" has no field named "driver_activity_v1_2". Available Fields(except extensions): ['entities']

Question

I want to handle messages like this in protobuf:

Python code:

request_proto = PrettyRequest()
Parse(body, request_proto)

I have an error:

Message type "some_path.PrettyList" has no field named "driver_activity_v1_2". Available Fields(except extensions): ['entities']

How to fix protobuf structure?

Json-body example

{
    "my_services": {
        "service1": {
            "entities": {
                "driver_id":[1001, 1003]
            }
        },
        "serviceN": {
            "entities": {
                "driver_id":[1, 2]
            }
        }
    }
}

My proto file:

message PrettyRequest {
    PrettyList my_services = 1;
}

message PrettyList {
    map<string, EntitiesDict> entities = 1;
}

message EntitiesDict {
    map<string, RepeatedValue> entities = 1;
}

message RepeatedValue {
  repeated Value val = 1;
}

message Value {
  // ValueType is referenced by the metadata types, FeatureInfo and EntityInfo.
  // The enum values do not have to match the oneof val field ids, but they should.
  // In JSON "*_val" field can be omitted
  oneof val {
    bytes bytes_val = 1;
    string string_val = 2;
    int32 int32_val = 3;
  }
}

Answer 1

There are at least 2 ways that you can achieve this.

Generally (!) using JSON as a template for a Protobuf schema is challenging because JSON is weakly typed whereas Protobuf messages are strongly-typed.

Protobuf

syntax="prpto3";

message X {
  map<string,Y> my_services = 1;
}
message Y {
  repeated int32 driver_id = 1;
}

x = foo_pb2.X(
    my_services={
        "service1":foo_pb2.Y(
            driver_id=[1001,1003]
        ),
        "serviceN":foo_pb2.Y(
            driver_id=[1,2]
        )
    }
)
json_string = json_format.MessageToJson(x,preserving_proto_field_name=True)
print(json_string)

NOTE I've used X and Y as placeholders.

Yields:

{
  "my_services": {
    "serviceN": {
      "driver_id": [
        1,
        2
      ]
    },
    "service1": {
      "driver_id": [
        1001,
        1003
      ]
    }
  }
}

NOTE This matches your JSON sample.

Well-known Type 'Value'

Google Well-known Types include Value which can be used to represent arbitrary (any type) JSON objects.

from google.protobuf import json_format
from google.protobuf.struct_pb2 import Value

v = Value()

j = '''{
    "my_services": {
        "service1": {
            "entities": {
                "driver_id":[1001, 1003]
            }
        },
        "serviceN": {
            "entities": {
                "driver_id":[1, 2]
            }
        }
    }
}'''

# print(j)

json_format.Parse(j,v)
# print(v)

json_string = json_format.MessageToJson(v)
print(json_string)

Yields:

{
  "my_services": {
    "serviceN": {
      "entities": {
        "driver_id": [
          1.0,
          2.0
        ]
      }
    },
    "service1": {
      "entities": {
        "driver_id": [
          1001.0,
          1003.0
        ]
      }
    }
  }
}

NOTE This matches your JSON sample.

The main challenges with using Value are that (a) the Protobuf message structure is complicated; (b) the Protobuf message structure accepts any JSON not just messages from types inferred from your JSON sample.