A combining mark in a `Textbox`'s last cell loses the base character

Bug description:

Typing a combining mark into the lower-right cell of a curses.textpad.Textbox puts the mark in a cell of its own and pushes the character it should attach to off the window edge, so that character is lost. The same keystrokes one column earlier attach the mark correctly. Doc/library/curses.rst says of Textbox that "Entering and reading back the full Unicode range, including combining characters, is now supported when curses is built with wide-character support".

import curses, sys
from curses.textpad import Textbox

def typed(ncols):
    win = curses.newwin(1, ncols, 0, 0)
    box = Textbox(win)
    for ch in 'abe\u0301':      # COMBINING ACUTE ACCENT after the 'e'
        box.do_command(ch)
    return win.instr(0, 0, 40).decode()

out = curses.wrapper(lambda stdscr: (typed(3), typed(10)))
print('3 columns:  %a\n10 columns: %a' % out, file=sys.stderr)

Output:

3 columns:  'ab\u0301'
10 columns: 'abe\u0301       '

Expected 'abe\u0301' from both windows: the mark should join the e instead of taking its cell.

CPython versions tested on:

CPython main branch

Operating systems tested on:

Linux

Linked PRs

• gh-157211

從 curses.textpad.Textbox.do_command 開始，並在 Linux 上使用提供的三欄和十欄範例重現問題。檢查 Doc/library/curses.rst 中所述的 Unicode 行為；當兩個視窗都將基底字元和組合字元保留為 'abe\u0301' 時完成。