The Problem

For the following code in git HEAD:

Looks like its size is computed based on text size, after applying translation:

...which works great in the default scenario:

-> % python3
Python 3.13.12 (main, Feb 4 2026, 15:06:39) [GCC 15.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> len('Filter by tag, payee, ...')+1
26

...but fails to handle CJK characters with wide characters:

-> % python3
Python 3.13.12 (main, Feb 4 2026, 15:06:39) [GCC 15.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> len('筛选(标签,收款人等等)')+1
13

This is expected since the size attribute of <input> element does not consider wide characters and is an approximation anyway. In general, wide characters like CJK characters takes approximately the width of 2 narrow characters.

Possible solutions

I don't think native JS provides any elegant solution to this issue.

• We either identify wide characters in the string and manually assign 2x size value to them (ugly),
• or compute rendered result and explicitly set the width of <input> width.

I don't really like either of the solution and this bug is not that important, so I would like to hear from others first on how to handle it.