jsonschema_suite check fails with UnicodeDecodeError on Windows (non-UTF-8 default locale)

While running the sanity check script on Windows, I ran into an issue where python bin/jsonschema_suite check fails to parse test files that contain non-ASCII Unicode characters, such as idn-hostname.json and non-bmp-regex.json.

What happened?

It looks like path.read_text() is being called without an explicit encoding.

On Windows, Python uses the system's default locale encoding (in my case, cp1252) instead of UTF-8. Since these test files are UTF-8 encoded, this causes a UnicodeDecodeError when Python encounters certain characters.

Environment

• OS: Windows 10
• Python Version: Python 3.12
• Command: python bin/jsonschema_suite check

Proposed Fix

We can pass encoding="utf-8" to all read_text() calls in bin/jsonschema_suite so that files are always read as UTF-8, regardless of the host OS locale:

json.loads(path.read_text(encoding="utf-8"))

Screenshots

Start in bin/jsonschema_suite by locating every read_text() call and check how the suite loads files such as idn-hostname.json and non-bmp-regex.json. Run python bin/jsonschema_suite check on Windows or a non-UTF-8 locale; done means the check parses these UTF-8 files without UnicodeDecodeError.