$ cd ../lessons
pyodide: loading…

[practice]Web & APIs

JSON from APIs

# theory

real network

Lesson 26 introduced pyfetch. This lesson uses it for actual work: hitting a public REST API, getting JSON, turning it into a DataFrame.

Every example below makes a real network request. The test API is https://jsonplaceholder.typicode.com, a free CORS-friendly stand-in for production APIs. Same shape (users, posts, comments, todos), no auth required.

the shape you'll keep writing

from pyodide.http import pyfetch
import pandas as pd

response = await pyfetch("https://jsonplaceholder.typicode.com/users")
if response.status != 200:
    raise RuntimeError(f"API returned {response.status}")
data = await response.json()
df = pd.DataFrame(data)

Four steps. pyfetch → check status → .json() → pd.DataFrame. That sequence covers 80% of real API work.

picking columns

A typical API returns way more fields than you need. Strip down right away:

df = pd.DataFrame(data)[["id", "name", "email"]]

If the response is nested (a list of dicts where one value is itself a dict), use pd.json_normalize to flatten:

from pandas import json_normalize

# {"users": [{"id": 1, "address": {"city": "Boston"}}, ...]}
df = json_normalize(data["users"])
# Yields columns: id, address.city, address.zipcode, ...

counting by foreign key

A common API has two endpoints that join on an id. JsonPlaceholder gives you /posts (with a userId field) and /users (with that user's name). Fetch both, merge, aggregate.

posts = pd.DataFrame(await (await pyfetch(".../posts")).json())
users = pd.DataFrame(await (await pyfetch(".../users")).json())
joined = posts.merge(users, left_on="userId", right_on="id")
posts_per_user = joined.groupby("name").size()

handling non-200

Always read response.status before .json(). A 404 or 500 page might still have a JSON body, but it won't have the shape you're expecting.

resp = await pyfetch(".../users/9999")  # doesn't exist
if resp.status == 404:
    print("user not found")
elif resp.status >= 500:
    print("server error, retry later")
else:
    user = await resp.json()

# examples [3]

# example 01 · fetching one record and reading the JSON

GET a single user by id. Check status first, then parse JSON.

1
2
3
4
5
6
7
8
9
🐍
Loading PythonSetting up pandas & numpy...
# example 02 · merging two endpoints and aggregating

Fetch posts and users, join on userId, count posts per user. Real two-table API work in 6 lines.

1
2
3
4
5
6
7
8
9
10
11
12
🐍
Loading PythonSetting up pandas & numpy...
# example 03 · handling a 404 gracefully

Always check response.status before .json(). Missing resources shouldn't crash the script.

1
2
3
4
5
6
7
8
9
10
11
12
🐍
Loading PythonSetting up pandas & numpy...

# challenges [2]

# challenge 01/02todo
Fetch user id 3 from https://jsonplaceholder.typicode.com/users/3 and print only their email address.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
🐍
Loading PythonSetting up pandas & numpy...
# challenge 02/02todo
Fetch all posts from https://jsonplaceholder.typicode.com/posts, group by userId, and print the user with the most posts in the format 'user X has Y posts' (X and Y are integers).
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
🐍
Loading PythonSetting up pandas & numpy...

# project

# project-challenge

thread: Sales Performance Dashboard · reward: 50 xp

# brief

The analytics API returns a nested JSON report with sales grouped by region. Parse this structure to create a flat DataFrame showing regional performance.

# task

Parse Nested Sales Report

# your code
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
🐍
Loading PythonSetting up pandas & numpy...