Chinese characters become mojibake when using custom text on Windows build and the font cannot be specified by `--text-font` flag

[GitHubonline1396529]

Brief Description

I'm using Activate Linux on Windows. But When I specifying Chinese text via -t, --text-title/-m, --text-message flags in the Windows version (activate-windows64.exe), the rendered text shows mojibake (tofu characters). However, the default Chinese localization works correctly without any encoding issues.

Steps to Reproduce

In MSYS2 shell (UTF-8 environment):

./activate-windows64.exe -t "激活 MSYS 2" -m "转到“设置”以激活 MSYS 2"

Then observe corrupted text rendering:

���� MSYS 2
Л�������á��L��� MSYS 2

Expected Behavior

Correct display matching the default Chinese localization text.

激活 MSYS 2
转到“设置”以激活 MSYS 2

Environment

• OS: Windows 10 22H2 (Simplified Chinese locale)
• Shell: MSYS2 Zsh 5.9 (UTF-8 mode confirmed by $LANG=zh_CN.UTF-8)
• Tested binaries: Both prebuilt v1.1.0 release and self-compiled version
• Font issue: Custom fonts via --text-font flag have no effect

Critical Observations

Encoding Paradox

• Default localization works: Program can render Chinese correctly. If I just run activate-windows64.exe Chinese text would be shown correctly.
• English text works correctly: ./activate-windows64.exe -t "Activate MSYS 2" -m "Go to Settings to activate MSYS 2".
• Custom parameters fail: Possible UTF-8 (chcp 65001) / GBK (chcp 936) conversion mismatch in argument handling.

Font Rendering Anomaly

I've noticed that the text appears in a low-resolution serif font (see screenshot below), even when specifying --text-font Arial. I guess this suggests Cairo may not be loading system fonts properly.

Analysis

Based on mojibake patterns, 激活 → ���� and 转到设置 → ������� might suggests GBK bytes misdecoded as UTF-8.

[Lyn7Eiki]

you can try run "intl.cpl",

and reboot pc.

But it has some problems for beta things, so maybe you can try npp to convert chinese text.
ps1: with using Unicode UTF-8 for worldwide language support, ansi can't display
ps2: i find it that only even numbers of kanzi can be complete convert

[steve02081504]

But it has some problems for beta things, so maybe you can try npp to convert chinese text. ps1: with using Unicode UTF-8 for worldwide language support, ansi can't display ps2: i find it that only even numbers of kanzi can be complete convert

btw this will breaks JP galgames with lc convter.

[Lyn7Eiki]

@GitHubonline1396529
我尝试用vide coded了一个修复版（我fork了原版），你可以试试https://github.com/Lyn7Eiki/activate-linux

[Lyn7Eiki]

use msys2 and your sample-command, it is perfect