Making requests

requests is the de facto standard Python library for HTTP. It takes the raw mechanics of HTTP — connection management, encoding, redirects — and hides them behind a straightforward interface. Installing it (pip install requests) is almost always the right first step for any API work.

Sending a GET request

The minimal call is one line:

import requests

response = requests.get("https://jsonplaceholder.typicode.com/todos/1")

requests.get() opens a connection, sends the request, waits for the full response, and returns a Response object. Everything you need is on that object.

Reading the response

Three attributes cover the common cases:

• response.status_code — the integer status code (200, 404, etc.).
• response.json() — parses the body as JSON and returns a Python dict or list. Raises json.JSONDecodeError if the body is not valid JSON.
• response.text — the raw body as a string, regardless of content type.

Adding query parameters

Many endpoints accept optional parameters in the URL: ?page=2&per_page=50. Pass them as a dict to the params argument instead of building the URL by hand — requests handles encoding:

response = requests.get(
    "https://jsonplaceholder.typicode.com/todos",
    params={"userId": 1, "_limit": 5},
)

This produces the URL https://jsonplaceholder.typicode.com/todos?userId=1&_limit=5. Building URLs manually with string formatting is fragile and easy to get wrong.

Handling failure with raise_for_status()

requests.get() does not raise an exception just because the server returned a 4xx or 5xx code. The response object arrives normally; you have to check it. The cleanest way is raise_for_status():

response = requests.get("https://jsonplaceholder.typicode.com/todos/9999")
response.raise_for_status()  # raises requests.HTTPError if status >= 400

Call this immediately after every request unless you are intentionally handling specific non-200 codes yourself. It converts silent failures — where your script happily processes an error page — into loud exceptions.

Try it

The block below hits a real public API. Run it to see a live response:

The browser runner uses urllib (which is available in Pyodide) instead of requests, but the shape is the same: status code, parsed JSON body, explicit failure check. In a real script you would use requests and the cleaner .raise_for_status() call.

Where to go next

Next: authentication — most real APIs require credentials. There are three common patterns, and one of them (hardcoding keys) will cause you serious problems the moment you push to version control.