oracle

API which periodically fetches price data from Coingeckco, and Binance, then conditionally stores in Redis.

The program is separated into two services:
fetcher.py for fetching and storing data server.py for serving API /latest

Objectives

Notice

A snapshot has been made to branch snapshot.
I normally do not log debug this much, it is for your ease of debugging
server addr, password, should be loaded using

import os
from dotenv import load_dotenv

load_dotenv()
... = os.getenv('...')

but opted out for the sake of simplicity in testing. Sane options can be configured in fetcher.py constants.
Function pattern:

def func():
  guard
  body
  default

Rationale

Separating server.py, fetcher.py

Simply micro-service pattern. Each service should only do one thing.

DB: Redis

Constrin: easy to dev
RDBMS, too complex for this project.
NOSQL, yes, two choice, Mongo, or Redis.
Redis: Fast, easy, good for caching, fetching, storing JSON.
Mongo: all of about just not as fast.

API: FastAPI

Constrain: No need to write data
FastAPI: Fast dev, Fast response, modular, defining schemas optional.

Scheduler: Python APS

Constrain: Background job scheduler, simple to implement
Python APS: only one that fit the requirement
ZeroMQ: also works, but requries more complex/ elaborate setup

Usage

Local

Spin up redis

cd redis
docker compose up --build --force-recreate

Use env manager of your choice

pip install pipenv
pipenv install

Start fetcher

pipenv run python fetcher.py

Start apiserver

pipenv run uvicorn server:app --port 8888

Access at: http://127.0.0.1:8888/latest

Lodimup/oracle