Double-click word selection: treat consecutive CJK characters as a single word

[l-qing]

Problem

When I double-click an English word in Ghostty, the whole word is selected — as expected. But when I double-click a Chinese character, only that single character is selected, even when surrounded by other Chinese characters with no whitespace or punctuation between them.

Example (the | marks where I double-clicked):

• hello wor|ld foo → world is selected ✅
• 配置文件已|经更新 → only 经 is selected ❌ (expected: 配置文件已经更新, or at least 已经更新)

The same applies to Japanese (Hiragana/Katakana/Han) and Korean (Hangul) text.

Comparison with other terminals

• iTerm2 treats a continuous run of CJK characters as one "word" by default, and the boundary set is configurable in Preferences → Selection.
• Terminal.app, Alacritty, WezTerm, and kitty also collapse consecutive CJK ideographs into a single word for double-click selection.

Ghostty is the only terminal in my daily rotation where I have to triple-click (whole line) or drag-select to grab a Chinese phrase, which is a noticeable papercut for CJK users.

Why selection-word-chars doesn't help

selection-word-chars only adds boundary characters; there's no way to express "treat this whole Unicode script as word-class". So the current behavior — each CJK codepoint ending up as its own word — cannot be worked around in config today.

Proposed behavior

Treat a continuous run of codepoints from any of these Unicode blocks/scripts as a single word for double-click selection (the same way ASCII letters/digits are):

• CJK Unified Ideographs and extensions (U+4E00–U+9FFF, U+3400–U+4DBF, U+20000–U+2A6DF, …)
• Hiragana (U+3040–U+309F)
• Katakana (U+30A0–U+30FF, U+31F0–U+31FF)
• Hangul Syllables and Jamo (U+AC00–U+D7AF, U+1100–U+11FF, U+3130–U+318F)

selection-word-chars would still apply on top, so users can punch holes (e.g. add 、，。 as boundaries) if they want finer breaks.

Optional: make it opt-in / configurable

If maintainers prefer not to change the default, a config like:

selection-word-script-mode = ascii   # current behavior (default)
selection-word-script-mode = unicode # consecutive same-script runs are one word

would also solve the problem and keep the door open for other scripts (Thai, Arabic, etc.) later.

Prior art / related discussions

Searched issues and discussions for CJK, chinese, ideograph, 双击, 汉字, selection-word-chars, smart selection, double-click word. Adjacent topics:

• Make the character selection boundary configurable #9639 — Make the character selection boundary configurable (discusses adding/removing boundary chars only)
• UX Improvements: Word Selection Boundaries + Trailing Space After Dragged Inserts #7575 — UX Improvements: Word Selection Boundaries
• Word selection boundaries are inconsistent in 1.3.0 #11404 / Unexpected behaviour with double-clicking to select a word #11415 / macOS: double-click selection doesn't stop on ":" and other characters even if they are in "selection-word-chars" #11525 — various ASCII-side word-boundary inconsistencies
• Issue Word selection is too greedy #904 — Word selection is too greedy (also ASCII-only)

None of them address the CJK / Unicode-script angle, hence this new discussion.

Environment

• Ghostty 1.3.x on macOS (Darwin 25.4.0)
• Theme: iTerm2 Solarized Dark (issue is independent of theme)

[00-kat]

Firefox and Chromium both select 已经.

配置文件已|经更新

In my Ghostty, double-clicking 经 selects 经, double-clicking 已 selects 已, and double-clicking exactly in between the two characters (which is where you placed the |) selects nothing.