gapst

Gapst is a test repo for using gRPC to communicate between a go API and a python ML model, without building a python REST API.

Structure

The repo is structured into two main components:

• The go API (found in the api folder)
• The python ML model (found in the classifier folder)

Requirements

To run the code in this repo, you will need the following:

• Go 1.22
• Python 3.11
• Make
• Sed
• Protobuf
• Protoc-gen-go
• Protoc-gen-go-grpc

and the following python packages:

• grpcio
• grpcio-tools
• numpy
• pandas
• scikit-learn

Other versions of go and python might work, but those are the versions I used.

These are all provided in the corresponding flake.nix and flake.lock files in the two folders. To use them, use the following command in either the api or classifier (depending on if you're running the go API or python server):

$ nix develop

Note

If you are not using nix, you will have to set up the dependencies yourself, including setting up the python environment.

Usage

Python Server

To try it out, you'll want to first start the python ML model server. This can be done by using the following:

$ cd classifier

# If using nix, otherwise activate your python environment
$ nix develop

$ python server.py

Go API Client

Then start the go API with the following in a separate terminal:

$ cd api

# If using nix
$ nix develop

$ go run main.go

Using the Go API as the Client

You can then make a POST request to http://localhost:8080/predict to make predictions using a model trained on the iris dataset using curl, postman, or any other tool:

# Example using curl
$ curl -X POST http://localhost:8080/predict -d "{\"SepalLengthCm\": 2.4, \"SepalWidthCm\": 1.2, \"PetalLengthCm\": 1.1, \"PetalWidthCm\": 1.0}"

Protobuf Modifications

If you make any protobuf modifications, you use use make generate to re-generate the gRPC code.