Protogen is used to generate a serialization/deserialization interface for the Arke Industries Interchange Protocol (AIIP). It is made available under the terms of the MIT license.
A protocol consists of RPC methods and objects. Every procedure has an ID which consists of a method and a category. A procedure takes a number of objects and returns a number of objects. What they are and how many is determined by the method signature.
An object is either a primitive type, or an ordered collection of objects (each of which is called a field). Objects have no means to identify themselves or their fields from inspecting their own representation. The Protocol defines what objects are used in what contexts.
protogen's serialization/deserialization model depends on a "stream" which provides the direct ability to read or write primitive types. The primitive types protogen requires are:
array{u,i}{8, 16, 24, 32, 64}f{32,64}string
array is an "unsized type". It consists of {N, val1, val2, ...valN}, where
N is the amount of elements (as a u16). They are homogeneous. Strings are
similar. More details later.
An example showing every feature of protogen is available in example.pg.
include is literal, textual inclusion, along the lines of the C preprocessor
(though without any of its other features).
There is a grammar in grammar.ebnf
protogen is primarily concerned with shuttling data across the wire, and as such does not add any padding to objects, and does not require any alignment. Each field of an object is laid out one after another, using only the exact amount of bytes required to store the field.
To run protogen, install Python and pip, run pip install -r requirements.txt, make, and then ./protogen path/to/definition.pg.