faststan 0.1.4

Build data streaming pipelines using faststan

FastSTAN

Easily deploy NATS Streaming subscribers using Python.

Features

• Define subscribers using sync and async python functions
• Automatic data parsing and validation using type annotations and pydantic
• Support custom validation using any function
• Allow several subscribers on same channel
• Support all subscription configuration available in stan.py
• Healthcheck available using HTTP GET request to monitor the applications
• (TODO) Metrics available using HTTP GET requests to monitor subsriptions status
• All of FastAPI features

Quick start

• Install the package from pypi:

pip install faststan

• Create your first subscriber. Create a file named app.py and write the following lines:

from faststan import FastSTAN

app = FastSTAN()

@app.stan.subscribe("demo")
def on_event(message: str):
    print(f"INFO :: Received new message: {message}")

• Start your subscriber:

uvicorn app:app

• Or if you are in a jupyter notebook environment, start the subscriptions:

await app.stan.run()

Advanced features

Using error callbacks

from faststan import FastSTAN


app = FastSTAN()


def handle_error(error):
    print("ERROR: {error}")


@app.stan.subscribe("demo", error_cb=handle_error)
def on_event(message: str):
    print(f"INFO :: Received new message: {message}")

Using pydantic models

You can use pydantic models in order to automatically parse incoming messages:

from pydantic import BaseModel
from faststan import FastSTAN


class Event(BaseModel):
    timestamp: int
    temperature: float
    humidity: float


app = FastSTAN()


@app.stan.subscribe("event")
def on_event(event):
    msg = f"INFO :: {event.timestamp} :: Temperature: {event.temperature} | Humidity: {event.humidity}"
    print(msg)

Using pydantic models with numpy or pandas

import numpy as np
from pydantic import BaseModel
from faststan import FastSTAN


class NumpyEvent(BaseModel):
    values: np.ndarray
    timestamp: int

    @validator("temperature", pre=True)
    def validate_array(cls, value):
        return np.array(value, dtype=np.float32)

    @validator("humidity", pre=True)
    def validate_array(cls, value):
        return np.array(value, dtype=np.float32)

    class Config:
        arbitrary_types_allowed = True

@app.stan.subscribe("event")
def on_event(event: NumpyEvent):
    print(
        f"INFO :: {event.timestamp} :: Temperature values: {event.values[0]} | Humidity values: {event.values[1]}"
    )