webbrowser_open

prototyping opening things with the default webbrowser

Use

The simplest way to use this package is to replace:

import webbrowser

...
webbrowser.open(url)

with

import webbrowser_open

...
webbrowser_open.open(url)

You can also launch a URL with

python3 -m webbrowser_open URL

to see the difference (if any) in your environment.

The only difference in behavior is that webbrowser_open looks up the default browser application before opening the URL. This should only result in a change in behavior for opening file:// URLs where the default application associated with the file type is used instead of the default browser.

The $BROWSER environment variable still takes priority, if defined, in which case webbrowser_open.open just wraps a call to webbrowser.open with no changes.

This package uses the following APIs to explicitly launch the default browser:

Background

Most platforms have at least a semi-standard way to open URLs and discover the default browser.

webbrowser.open uses generic not-browser-specific APIs (e.g. open, xdg-open, os.startfile), which works fine with http[s] URLs. However, all of these systems associate file:// URLs with the default application for the file type, not necessarily a webbrowser, which webbrowser.open is meant to launch. The result is often webbrowser.open("file:///path/to/page.html") launching a file editor instead of a browser.

webbrowser.open does not work reliably with file:// URLs on any platform, though it may if the default application for HTML files is a browser, which it often is, except for developers.

Linux is in the best situation by default, because it does call xdg-settings get default-web-browser to actually lookup a browser, but it only uses this if a matching browser is already known ahead of time and found in register_X_browsers. This is still known to fail at least sometimes.

This is a prototype package for testing implementations to be submitted to the standard library.

ref: https://github.com/python/cpython/issues/128540