lightprotobuf 1.0b3

A light full Python3 Protocol Buffers implementation

lightprotobuf

Introduction

lightprotobuf is a full Python 3 implementation of the Protocol Buffers as described by Google.

Documentation

The main class to use is lightprotobuf.Message

It must be the base class of every messages you define. The rest follows the .proto design:

This message:

enum FooEnum {
   FIELD = 5;
   FIELD2 = 6;
}

message FooMsg {
   required int32 foo_field = 1;
   optional FooEnum foo_enum = 2;
   enum BarEnum {
      FIELD = 5;
      BAR = 6;
   }
   required BarEnum bar_enum = 3;
   repeated string foo_rep = 4;
   repeated int32 foo_pak = 5 [packed=true];
}

message BarMsg {
   required FooMsg.BarEnum bar_enum = 1;
}

Is translated in python to:

from enum import IntEnum
from lightprotobuf import *
class FooEnum(IntEnum):
     FIELD = 5
     FIELD2 = 6

class FooMsg(Message):
     class BarEnum(IntEnum):
             FIELD = 5
             BAR = 6
     foo_field = Field(1, Int32, Field.REQUIRED, **{})
     bar_enum = Field(3, BarEnum, Field.REQUIRED, **{})
     foo_enum = Field(2, FooEnum, Field.OPTIONAL, **{})

class BarMsg(Message):
     bar_enum = Field(1, FooMsg.BarEnum, Field.REQUIRED, **{})

As you can see, the fields follow this template

<name> = Field(<tag number>, <type>, Field.<REQUIRED|OPTIONEL|REPEATED>, **{<options as a dict (optional)>}

Enums are python’s enum.IntEnum

Nested types are real python nested types referenced just like in .proto

API

The fields are actually transformed as attributes via descriptors. So you can access fields easily:

m = FooMsg()
m.foo_field = 5
m.foo_field # returns 5
m.foo_enum = 5 # Error because it expects a FooEnum object
m.foo_enum = FooEnum.FIELD # OK
m.bar_enum = FooMsg.BarEnum.BAR # OK

Repeated fields atc like a list:

m.foo_rep = ["a string", "another"] # OK
li = m.foo_rep # Get a reference to the list
li.append("a string") # OK, append the string
li.append(b'a bytes') # TypeError because there is a check to avoid mistakes

Note : packed fields are able to decode either data packed either multiple occurence of the field e.g. the test case:

class Repeated(Message):
        r = Field(1, Int32, Field.REPEATED, packed="True")
nb = [1,150,1,2,3,150]
b = io.BytesIO(b'\x0C\x08\x01\x08\x96\x01\x0A\x05\x01\x02\x03\x96\x01')
m = Repeated.from_stream(b)
self.assertEqual(list(m.r), nb)
# 0C (12) bytes following
# 08 = 1 << 3 | 0
# varints etc.
# 0A = 1 << 3 | 2
# 05 bytes following
# concatened varints etc.

To encode a message, lightprotobuf uses stream objects : each DataType has a to_stream and from_stream class method. Just to call it from a message to encode/decode a message:

import io
s = io.BytesIO()
Message.to_stream(s, m)
s.getvalue() # b'\x06\x08\x05\x10\x05\x18\x06'

m = FooMsg()
s = io.BytesIO(b'\x06\x08\x05\x10\x05\x18\x06')
m = Message.from_stream(s)

_Note_ : if required field is missing, it raises a FieldNotOptional exception

Release Notes

1.0.b3

• WARNING : module moved at top-level. Use import lightprotobuf rather than from lightprotobuf import lightprotobuf

• Add support for repeated fields, packed and not packed

1.0.b2

• Remove DESCRIPTION.rst because duplicate of README.rst