Provides patches for widely used http libraries to make them work in Pyodide environments like JupyterLite.
# 1. Install this package
import micropip
await micropip.install('pyodide-http')
# 2. Patch requests
import pyodide_http
pyodide_http.patch_all() # Patch all libraries
# 3. Use requests
import requests
response = requests.get('https://raw.githubusercontent.com/statsbomb/open-data/master/data/lineups/15946.json')This package applies patches to common http libraries. How the patch works depends on the package.
All non-streaming requests are replaced with calls using XMLHttpRequest.
Streaming requests (i.e. calls with stream=True in requests) are replaced by calls to fetch in a separate web-worker if (and only if) you are in a state that can support web-threading correctly, which is that cross-origin isolation is enabled, and you are running pyodide in a web-worker. Otherwise it isn't possible until WebAssembly stack-switching becomes available, and it falls back to an implementation that fetches everything then returns a stream wrapper to a memory buffer.
The implementation of streaming requests makes use of Atomics.wait and SharedArrayBuffer to do the fetch in a separate web worker. For complicated web-security reasons, SharedArrayBuffers cannot be passed to a web-worker unless you have cross-origin isolation enabled. You enable this by serving the page using the following two headers:
Cross-Origin-Opener-Policy: same-origin
Cross-Origin-Embedder-Policy: require-corp
Be aware that this will have effects on what you are able to embed on the page - check out https://web.dev/cross-origin-isolation-guide/ for more details.
Currently the following packages can be patched:
| Package | Patched |
|---|---|
| requests | Session |
request |
|
head, get, post, put, patch, delete |
|
| urllib | urlopen |
OpenerDirector |
pyodide-http is used by a some awesome projects:
