Skip to content

python.md

Latest commit

 

History

History
109 lines (69 loc) · 5.11 KB

python.md

File metadata and controls

109 lines (69 loc) · 5.11 KB
Raw
title Python
description Connect one Python backend service to another with Graftcode - no REST clients, no DTOs, no handwritten integration code. Install a strongly-typed Graft and call remote methods directly from your service logic.

Goal

Connect a Python backend service to another remote service using Graftcode - so the remote integration stays strongly typed and reads like a normal method call.

What You'll See

  • Install a remote service as a strongly-typed Graft via pip.
  • Configure the generated client to point at the remote host.
  • Call a remote method from your own service logic as if it were a local class.
  • Use IDE autocompletion on the remote service's methods and classes.

Prerequisites

Step 1. Create a Python service

Create a new project folder for your backend service:

mkdir py-energy-consumer
cd py-energy-consumer

Step 2. Find the remote method in Graftcode Vision

We're hosting this sample service for you so you can see exactly what connecting to another team's service looks like in practice - open it in Graftcode Vision to explore.

Graftcode Vision shows all public classes and methods exposed by the remote service - their names, parameter types, and return types. It also gives you the exact package manager command needed to install that service as a Graft.

Step 3. Install the Graft

Open Graftcode Vision, pick PyPI, and copy the generated install command.

python -m pip install --extra-index-url https://grft.dev/simple/4b4e411f-60a0-4868-b8a6-46f5dee07448__free graft-nuget-energypriceservice==1.2.0

This adds the generated strongly-typed client for the remote service to your project.

Step 4. Call the remote method and run it

The exact configuration snippet for your language is available in Graftcode Vision under the Configuration installation tab. Create main.py:

The generated PyPI Graft uses a flat package layout (graft_config.py, meter_logic.py). Import those modules directly:

import os
from graft_nuget_energypriceservice.graft_config import GraftConfig
from graft_nuget_energypriceservice.meter_logic import MeterLogic

GraftConfig.host = "wss://g-d-ca-polc-demo-ecbe-01.nicedesert-fa74799d.polandcentral.azurecontainerapps.io/ws"

consumption = MeterLogic.net_consumption_k_wh(1000, 1150)
print(f"Net consumption: {consumption}", flush=True)
os._exit(0)

Run it:

python main.py

The Python Graft exposes net_consumption_k_wh (snake_case; the generator also splits KWh as k_wh). Use flush=True before os._exit(0) so the printed result is not lost when the process exits immediately.

You should see the net consumption value printed in your terminal. MeterLogic.net_consumption_k_wh(...) is a remote call, but your code reads like a normal method invocation - no HTTP request, no response parsing, no serialization.

Your IDE can autocomplete available methods on MeterLogic, BillingLogic, and any other class from that service because the Graft is a real installed package.

Step 5. Run with a Project Key (recommended for real-world usage)

Everything above works without any account - perfect for learning and local development. When you're ready for real-world usage, create a free account at portal.graftcode.com, set up a project, and copy its Project Key.

With a Project Key, point GraftConfig.host at your project's stable registry URL instead of a raw WebSocket address. A Project Key gives you:

  • Stable registry URL - the address for your Grafts stays permanent, so install commands don't change when you redeploy.
  • Portal visibility - see all your gateways and services in one place at gateways.graftcode.com.
  • Access control - decide who can download your Grafts using package manager authentication and permissions.

Without Graftcode

Connecting one backend service to another typically requires:

  • Designing and implementing REST or gRPC endpoints in the remote service
  • Defining request/response DTOs for every operation
  • Writing or generating an OpenAPI or Protobuf spec
  • Building or generating a client SDK in the consuming service's language
  • Manually keeping both sides in sync when signatures change
  • Adding error handling, retry logic, and serialization code

With Graftcode

  • Install the remote service as a strongly-typed Graft via pip install
  • Import its classes and call methods directly - no REST client code
  • When the remote service changes, update with one command - like any other package

Connecting two backend services with Graftcode is as simple as installing a pip package. No REST routes, no DTOs, no client generation - just import and call.

Old Way vs Graftcode