[AbdullahAljelamneh]

Pull Request

Discussion:

Description

AI Disclaimer

AI transcript

Checklist

• This PR links to a GitHub Discussion for the proposed code change.
• I added tests for the change.
• The new or updated tests fail on the main branch and pass on this PR.
• Coverage stays at 100%.
• The documentation explains the change if needed.

[zyzkon]

Confirming this fix works — reproduced independently on a different machine, and it resolves the issue.

Environment

• Windows, system default code page 936 (GBK), PowerShell
• Python 3.14.10, FastAPI 0.141.1
• Fresh clone of master, no unrelated changes

Before the fix

POST /api/v1/password-recovery/admin@example.com → 500 Internal Server Error

UnicodeDecodeError: 'gbk' codec can't decode byte 0x80 in position 658: illegal multibyte sequence
  File "backend/app/utils.py", line 28, in render_email_template
    ).read_text()

After applying encoding="utf-8"

Same request → 200, and the recovery email is delivered to Mailpit correctly.

Two things that may help this get merged:

1. TRAINING.md looks accidentally committed. It's a 125-line personal study log unrelated to the fix, and it's what makes this PR hard to review. Dropping it leaves a clean one-line diff that maintainers can merge immediately.
2. The repo asks for a linked GitHub Discussion before opening a PR (see the PR template). Adding that link should unblock review.

One more note for the fix itself: render_email_template() is the only Path.read_text() call in the backend without an explicit encoding, so this single change covers all three templates (test_email.html, reset_password.html, new_account.html).

Regarding the missing regression test: this bug does not reproduce on UTF-8 CI runners, so a portable test is difficult. I'd suggest that shouldn't block the merge.

For release notes, this would fall under bug (the latest-changes/label check is still pending).