FastAPI Tutorial - UMD Classes -- Tutorials

This folder contains the setup for running FastAPI and uvicorn tutorials within a containerized environment

Quick Start¶

From the root of the repository, change your directory to the FastAPI tutorial folder:
```
> cd tutorials/fastapi
```
Once the location has been changed to the repo run the command to build the image to run dockers:
```
> ./docker_build.sh
```
Once the docker has been built you can then go ahead and run the container and launch jupyter notebook using the created image using the command:
```
> ./docker_jupyter.sh
```
For more information on the Docker build system refer to Project template README
Once the ./docker_jupyter.sh script is running, work through the following notebooks in order

Tutorial Notebooks¶

Work through the following notebooks in order:
- fastapi.API.ipynb: Core FastAPI fundamentals
  - Path and query parameters
  - Request body validation with pydantic models
  - Dependency injection with Depends()
  - Error handling with HTTPException
  - The automatic /docs, /redoc, and /openapi.json
- fastapi_utils.py: Utility functions supporting the tutorial notebooks
- fastapi.example.ipynb: End-to-end Book Catalog API
  - Runs a real uvicorn server on a background thread
  - Exercises the API with real HTTP requests via httpx
  - Covers create, read, update, delete, filtering, and error responses

Changelog¶

2026-08-10: Initial release

TODO¶

In-process, no server, no network (what the test suite does)

fastapi.testclient.TestClient wraps create_app() directly — same interface as requests, but no socket, no uvicorn process needed:

import fastapi.testclient import research.Noesis.batch_call_auction as rnbacaau import research.Noesis.passthrough_proxy as rnopapro import research.Noesis.platform_api as rnoplapi

order_book = rnbacaau.OrderBook() gateway = rnopapro.Gateway() app = rnoplapi.create_app(order_book, gateway, {“key1”: “acct_1”}) client = fastapi.testclient.TestClient(app)

r = client.post( “/bids”, headers={“X-API-Key”: “key1”}, json={“buyer_id”: “buyer_1”, “n_tasks”: 10000, “c_level_min”: “frontier”, “l_max”: 2.0, “r_min”: 0.999, “p_max”: 0.02}, ) print(r.status_code, r.json()) Good for scripts, notebooks, or driving the market without paying HTTP overhead.

Real HTTP client against a running server (uvicorn up from earlier)

import requests

BASE_URL = “http://127.0.0.1:8000” HEADERS = {“X-API-Key”: “key1”}

r = requests.post(f"{BASE_URL}/bids", headers=HEADERS, json={ “buyer_id”: “buyer_1”, “n_tasks”: 10000, “c_level_min”: “frontier”, “l_max”: 2.0, “r_min”: 0.999, “p_max”: 0.02, }) r.raise_for_status() print(r.json())

r = requests.post(f"{BASE_URL}/rounds/clear") print(r.json()) Swap requests for httpx (sync or async with httpx.AsyncClient()) if you need async.

Generated typed client from the OpenAPI spec (fits the openapi.json from last turn)

pip install openapi-python-client openapi-python-client generate --path openapi.json --meta none Produces a package with a typed function per endpoint (bids_post.sync(client=..., body=BidRequest(...)), response parsed into the ContractResponse/RoundClearResponse models) — best when a client needs IDE autocomplete/type checking against the exact schema, or the API surface will keep changing and you want it to fail loudly on drift.