pygettext.py crashes with UnicodeEncodeError when writing .pot files containing emoji

[mtfreitasf]

Bug report

Bug description:

Summary:
pygettext.py fails with a UnicodeEncodeError when generating .pot files that contain Unicode characters such as emoji (e.g. ✅). This happens on Windows systems where the default encoding is cp1252, which cannot represent many Unicode symbols.

Steps to Reproduce:

• Add a translatable string with an emoji to a Python file:
  _("Operation complete ✅")
• Run pygettext.py to generate the .pot file:
  python pygettext.py -d messages -o messages.pot your_script.py
• Observe the crash:
  UnicodeEncodeError: 'charmap' codec can't encode character '\u2705' in position 1: character maps to

Cause:
The .pot file is opened using:
fp = open(output_file, 'w')

This defaults to the system encoding (cp1252 on Windows), which cannot encode emoji or other extended Unicode characters.

Suggested Fix:
Update the file-writing line in pygettext.py to explicitly use UTF-8:
fp = open(output_file, 'w', encoding='utf-8')

This ensures compatibility with all Unicode characters and aligns with modern Python practices.

Environment:

• Python version: 3.14
• OS: Windows 10
• Locale: pt_BR
• Default encoding: cp1252

Related Pull Request:
This issue appears related to PR #132244, which discusses encoding behavior in pygettext.py. However, the PR focuses on testing and does not yet address the default encoding used when writing .pot files.

CPython versions tested on:

3.14

Operating systems tested on:

Windows

Linked PRs

• gh-139873: pygettext: Open files with encoding='utf-8' #139915

[StanFromIreland]

xgettext's newer versions default to UTF-8 now IIRC, so I propose we do the same.

[mtfreitasf]

xgettext's newer versions default to UTF-8 now IIRC, so I propose we do the same.

@StanFromIreland, thank you. I have never done this before but I am available to help if you want.

Best regards,

Marcos