Skip to content

Utilities for working with Protobuf & gRPC in Python, e.g; compiling Python 3 Dataclasses from compiled proto-message-python code and casting between the two, quickly initializing large/nested Protobuf object, simplifying gRPC service interfaces etc.

License

Notifications You must be signed in to change notification settings

ccpgames/protoplasm

Repository files navigation

Protoplasm

Utilities for working with Protobuf & gRPC in Python, e.g; compiling Python 3 Dataclasses from compiled proto-message-python code and casting between the two, quickly initializing large/nested Protobuf object, simplifying gRPC service interfaces etc.

IMPORTANT

Update this README file after moving from CCP's internal code repo to Github.

Protoplasm 4

  • Unify the unary functionality of Protoplasm 2 with the streaming functionality of Protoplasm 3
    • The two turn out to be completely incompatible and API shattering
    • Protoplasm 4 must incorporate BOTH functionalities wile being backwards compatible enough for both Protoplasm 2 and 3 projects to be able to migrate to 4
    • The key here is detecting the stream keyword in protos that denote streaming input and/or output
  • Add piled up functionality/utility/QoL improvements/bugs that's been on The List™ for a while
    • Cast to/from base64 encoded strings
    • Utilize the __all__ directive to isolate import * side effects
    • Integrate the Neobuf Builder CLI (from various other projects) into Protoplasm and generalize it
    • Address the "None is default value" issue
    • Explore the pros/cons of making non-existing Message/Object attributes return Empty or EmptyDict to simplify nested attribute fetching...?
  • Refactor and restructure the package properly
    • Separate the 4 main roles of the package logically
      1. Cross-piling *.proto to *_pb2.py and *_pb2_grpc.py
      2. Cross-piling *_pb2.py to *_dc.py Neobuf Dataclasses
      3. Generating *_api.py interfaces
      4. Generating gRPC implementation of Services

Troubleshooting

  • I get TypeError: Plain typing.NoReturn is not valid as type argument

Clever bits to document...

  • Code Generation (e.g. foo.proto) + how to build
    • Dataclasses -> foo_dc.py + how to use (+ DataclassBase freebies)
      • Extending Dataclasses (a no-no for pb2 files apparently)
    • Service API -> foo_api.py + how to use and implement
      • Automatic parameter unpacking
      • Return value packing
      • Raising protoplasm.errors.api.* on errors or non-ok returns
      • Using protoplasm.decorators for param and type checking
      • The takes_context decorator and how to use it
    • GRPC DC Service Servicer -> foo_dc_grpc.py + how to use
  • Utilities
    • Proto <-> dict <-> Dataclass casters
    • The mkproto and mkdataclass helpers
    • The unpack_dataclass_request and pack_dataclass_response helpers

About

Utilities for working with Protobuf & gRPC in Python, e.g; compiling Python 3 Dataclasses from compiled proto-message-python code and casting between the two, quickly initializing large/nested Protobuf object, simplifying gRPC service interfaces etc.

Resources

License

Stars

Watchers

Forks

Packages