IBus does not use the specified (underlying) layout for Mozc Japanese

Fytch · 2024-10-26 00:53:52

I followed https://wiki.archlinux.org/title/Localization/Japanese and installed IBus and added two keyboard layouts in the IBus Preferences:

German, Swiss - German (Switzerland)
Japanese - Mozc:あ

(When I click on "Japanese - Mozc:あ" and then on "About", it says "Keyboard layout: default".)
I also unchecked "Use system keyboard layout" in IBus Preferences.

When I now type Japanese with Mozc, under the hood it uses the Swiss German layout instead of a true Japanese layout (y/z are the same for my layouts when they're supposed to be flipped). I want to change this because üöä are useless and because some vital keys are inaccessible. Basically, I want it to work like it does on Windows.
I found https://github.com/google/mozc/blob/mas … rations.md and followed the instructions under "Specify the keyboard layout", i.e. I opened ~/.config/mozc/ibus_config.textproto and changed layout : "default" to layout : "ja" (or layout : "jp") for all three engines. (The .md says "ja" but I'm pretty sure it has to be "jp"; doesn't work either way.)

After ibus write-cache; ibus restart, it still didn't not work, i.e. when I switch to Mozc it still uses the Swiss German layout and I can still type üöä.

Apparently, Mozc forwarded everything correctly to ibus:

$ ibus read-config
SCHEMA: org.freedesktop.ibus
SCHEMA: org.freedesktop.ibus.general
  enable-by-default: false
  xkb-latin-layouts: ['af', 'af(fa-olpc)', 'af(ps-olpc)', 'af(ps)', 'af(uz)', 'af(uz-olpc)', 'am', 'am(eastern)', 'am(eastern-alt)', 'am(phonetic)', 'am(phonetic-alt)', 'am(western)', 'ara', 'ara(azerty)', 'ara(azerty_digits)', 'ara(buckwalter)', 'ara(digits)', 'ara(qwerty)', 'ara(qwerty_digits)', 'az(cyrillic)', 'bd', 'bd(probhat)', 'bg', 'bg(bas_phonetic)', 'bg(phonetic)', 'brai', 'brai(left_hand)', 'brai(right_hand)', 'bt', 'by', 'by(phonetic)', 'by(legacy)', 'ca(ike)', 'cn(tib)', 'cn(tib_asciinum)', 'cn(ug)', 'cz', 'cz(ucw)', 'de(ru)', 'dev', 'et', 'fr(geo)', 'ge', 'ge(os)', 'gr', 'gr(extended)', 'gr(nodeadkeys)', 'gr(polytonic)', 'gr(simple)', 'gur', 'il', 'il(biblical)', 'il(lyx)', 'il(phonetic)', 'id(melayu-phonetic)', 'id(melayu-phoneticx)', 'id(pegon-phonetic)', 'in', 'in(ben)', 'in(ben_baishakhi)', 'in(ben_bornona)', 'in(ben_gitanjali)', 'in(ben_inscript)', 'in(ben_probhat)', 'in(bolnagri)', 'in(deva)', 'in(guj)', 'in(guj-kagapa)', 'in(guru)', 'in(hin-kagapa)', 'in(hin-wx)', 'in(jhelum)', 'in(kan)', 'in(kan-kagapa)', 'in(mal)', 'in(mal_enhanced)', 'in(mal_lalitha)', 'in(mar-kagapa)', 'in(ori)', 'in(san-kagapa)', 'in(tam)', 'in(tamilnet)', 'in(tamilnet_TAB)', 'in(tamilnet_TSCII)', 'in(tamilnet_tamilnumbers)', 'in(tam_tamilnumbers)', 'in(tel)', 'in(tel-kagapa)', 'in(urd-phonetic)', 'in(urd-phonetic3)', 'in(urd-winkeys)', 'iq', 'ir', 'ir(azb)', 'ir(pes_keypad)', 'jp(kana)', 'jp(mac)', 'kg', 'kg(phonetic)', 'kh', 'kz', 'kz(kazrus)', 'kz(ruskaz)', 'la', 'la(stea)', 'lk', 'lk(tam_TAB)', 'lk(tam_unicode)', 'lv(modern-cyr)', 'ma', 'ma(tifinagh)', 'ma(tifinagh-alt)', 'ma(tifinagh-alt-phonetic)', 'ma(tifinagh-extended)', 'ma(tifinagh-extended-phonetic)', 'ma(tifinagh-phonetic)', 'me(cyrillic)', 'me(cyrillicalternatequotes)', 'me(cyrillicyz)', 'mk', 'mk(nodeadkeys)', 'mm', 'mn', 'mv', 'np', 'ph(capewell-dvorak-bay)', 'ph(capewell-qwerf2k6-bay)', 'ph(colemak-bay)', 'ph(dvorak-bay)', 'ph(qwerty-bay)', 'pk', 'pk(ara)', 'pk(snd)', 'pk(urd-crulp)', 'pk(urd-nla)', 'pl(ru_phonetic_dvorak)', 'rs', 'rs(alternatequotes)', 'rs(rue)', 'rs(yz)', 'ru', 'ru(bak)', 'ru(chm)', 'ru(cv)', 'ru(dos)', 'ru(kom)', 'ru(legacy)', 'ru(mac)', 'ru(os_legacy)', 'ru(os_winkeys)', 'ru(phonetic)', 'ru(phonetic_winkeys)', 'ru(sah)', 'ru(srp)', 'ru(tt)', 'ru(typewriter)', 'ru(typewriter-legacy)', 'ru(udm)', 'ru(xal)', 'se(rus)', 'se(swl)', 'sy', 'sy(syc)', 'sy(syc_phonetic)', 'th', 'th(pat)', 'th(tis)', 'tj', 'tj(legacy)', 'tz', 'ua', 'ua(homophonic)', 'ua(legacy)', 'ua(phonetic)', 'ua(typewriter)', 'ua(winkeys)', 'us(chr)', 'us(rus)', 'uz']
  use-xmodmap: true
  preload-engines: ['xkb:ch::gsw', 'mozc-on']
  dconf-preserve-name-prefixes: ['/desktop/ibus/engine/pinyin', '/desktop/ibus/engine/bopomofo', '/desktop/ibus/engine/hangul']
  use-global-engine: true
  engines-order: ['xkb:ch::gsw', 'mozc-on']
  embed-preedit-text: true
  switcher-delay-time: 400
  use-system-keyboard-layout: false
  version: '1.5.30'
SCHEMA: org.freedesktop.ibus.general.hotkey
  previous-engine: @as []
  enable-unconditional: @as []
  disable-unconditional: @as []
  next-engine: ['Alt+Shift_L']
  trigger: ['Control+space', 'Zenkaku_Hankaku', 'Alt+Kanji', 'Alt+grave', 'Hangul', 'Alt+Release+Alt_R']
  next-engine-in-menu: ['Alt+Shift_L']
  prev-engine: @as []
  triggers: ['<Control>space']
SCHEMA: org.freedesktop.ibus.panel
  show-im-name: false
  custom-font: 'Sans 10'
  custom-icon: 'Adwaita'
  show: 0
  auto-hide-timeout: 10000
  custom-theme: 'Adwaita'
  x: -1
  use-custom-theme: false
  property-icon-delay-time: 500
  lookup-table-orientation: 1
  use-custom-font: false
  follow-input-cursor-when-always-shown: false
  use-custom-icon: false
  y: -1
  show-icon-on-systray: true
  use-glyph-from-engine-lang: true
  xkb-icon-rgba: '#51a2da'
SCHEMA: org.freedesktop.ibus.panel.emoji
  has-partial-match: false
  favorite-annotations: @as []
  load-unicode-at-startup: false
  partial-match-length: 3
  favorites: ['?️']
  hotkey: ['<Super>period', '<Super>semicolon']
  lang: 'en'
  font: 'Monospace 16'
  load-emoji-at-startup: true
  unicode-hotkey: ['<Control><Shift>u']
  partial-match-condition: 0

and

$ ibus read-cache
//snip
        <component>
            <name>com.google.IBus.Mozc</name>
            <description>Mozc Component</description>
            <version>0.0.0.0</version>
            <license>New BSD</license>
            <author>Google Inc.</author>
            <homepage>https://github.com/google/mozc</homepage>
            <exec>/usr/lib/ibus-mozc/ibus-engine-mozc --ibus</exec>
            <textdomain>ibus-mozc</textdomain>
            <observed-paths>
                <path mtime="1729890940" >/usr/share/ibus/component/mozc.xml</path>
            </observed-paths>
            <engines>
                    <engine>
                        <name>mozc-jp</name>
                        <longname>Mozc</longname>
                        <description>Mozc (Japanese Input Method)</description>
                        <language>ja</language>
                        <license></license>
                        <author></author>
                        <icon>/usr/share/ibus-mozc/product_icon.png</icon>
                        <layout>jp</layout>
                        <layout_variant></layout_variant>
                        <layout_option></layout_option>
                        <hotkeys></hotkeys>
                        <symbol>あ</symbol>
                        <setup>/usr/lib/mozc/mozc_tool --mode=config_dialog</setup>
                        <version></version>
                        <textdomain></textdomain>
                        <icon_prop_key>InputMode</icon_prop_key>
                        <rank>80</rank>
                    </engine>
                    <engine>
                        <name>mozc-on</name>
                        <longname>Mozc:あ</longname>
                        <description>Mozc (Japanese Input Method)</description>
                        <language>ja</language>
                        <license></license>
                        <author></author>
                        <icon>/usr/share/ibus-mozc/product_icon.png</icon>
                        <layout>jp</layout>
                        <layout_variant></layout_variant>
                        <layout_option></layout_option>
                        <hotkeys></hotkeys>
                        <symbol>あ</symbol>
                        <setup>/usr/lib/mozc/mozc_tool --mode=config_dialog</setup>
                        <version></version>
                        <textdomain></textdomain>
                        <icon_prop_key>InputMode</icon_prop_key>
                        <rank>99</rank>
                    </engine>
                    <engine>
                        <name>mozc-off</name>
                        <longname>Mozc:A_</longname>
                        <description>Mozc (Japanese Input Method)</description>
                        <language>ja</language>
                        <license></license>
                        <author></author>
                        <icon>/usr/share/ibus-mozc/product_icon.png</icon>
                        <layout>jp</layout>
                        <layout_variant></layout_variant>
                        <layout_option></layout_option>
                        <hotkeys></hotkeys>
                        <symbol>A</symbol>
                        <setup>/usr/lib/mozc/mozc_tool --mode=config_dialog</setup>
                        <version></version>
                        <textdomain></textdomain>
                        <icon_prop_key>InputMode</icon_prop_key>
                        <rank>99</rank>
                    </engine>
            </engines>
        </component>
//snip

So the layout was correctly passed on to IBus.
BTW, ibus read-cache | grep jp produces multiple hits in the non-Mozc part for <layout>s which is why I think the .md is incorrect for using "ja".

There was also this small tidbit in the wiki https://wiki.archlinux.org/title/IBus#Ibus_IME_packages_only_input_keyboard's_native_charset: Setting GTK_IM_MODULE=xim.
Unfortunately, this didn't change anything for me either.

Using "us" instead of "jp"/"ja" for layout didn't change anything either.

My localectl is:

$ localectl
System Locale: LANG=en_US.UTF-8
    VC Keymap: de_CH-latin1
   X11 Layout: ch
  X11 Variant: de_nodeadkeys

Packages:

ibus 1.5.30-3
ibus-mozc 2.30.5618.102-1
ibus-mozc-debug 2.30.5618.102-1
libibus 1.5.30-3
mozc-ut 2.30.5618.102.20241010-1
mozc-ut-debug 2.30.5618.102.20241010-1

seth · 2024-10-26 13:24:39

What's your testcase and on what desktop environment?
If the answer doesn't involve gnome, have you considered fcitx5?

Do you use ibus for input at all?
Because https://wiki.archlinux.org/title/IBus#Ibus_IME_packages_only_input_keyboard's_native_charset deliberately does NOT use ibus for gtk clients (the reasoning in that paragraph kinda escapes me)

Have you tried adding a second layout "setxkbmap -layout ch,jp" (this will probably not work on wayland)

Fytch · 2024-10-26 14:36:47

My testcase is to just open a few programs like xfce4-terminal, mousepad and Firefox and see what pressing different keys (like üöä$) produces with each of the IBus profiles.

I use Xfce4 with Xfwm4 on X11:

xfdesktop 4.18.1-3
xfwm4 4.18.0-3
xorg-server 21.1.13-1

In Xfce's setting dialog "Keyboard", I've had "Use system defaults" toggled on.
If I switch it off, it allows me to define layouts in that dialog. If I define a Swiss German and a Japanese layout, and if I use both Xfce's and IBus's toggling hotkeys (one after another), I get the desired result (but having two toggles is not really practical for me).

I haven't tried fcitx5 but I'll consider switching to it if I can't get IBus to work as desired.

I strongly assume I use IBus for input given that Mozc is running and working whenever I use IBus's toggle. I've set the environment variables as per "2 Integration"; =xim was just a test.

setxkbmap does not seem to work. It also doesn't show up in localectl's output

fytch@abc:~$ setxkbmap -layout ch,jp -variant de_nodeadkeys,
fytch@abc:~$ localectl
System Locale: LANG=en_US.UTF-8
    VC Keymap: de_CH-latin1
   X11 Layout: ch
  X11 Variant: de_nodeadkeys
fytch@abc:~$ setxkbmap -layout ch,jp
fytch@abc:~$ localectl
System Locale: LANG=en_US.UTF-8
    VC Keymap: de_CH-latin1
   X11 Layout: ch
  X11 Variant: de_nodeadkeys

seth · 2024-10-26 15:23:06

localectl simply reads /etc/vconsole.conf (well, through dbus - it's systemd after all)

setxkbmap -print -query

will get you the current set of layouts.

If I define a Swiss German and a Japanese layout, and if I use both Xfce's and IBus's toggling hotkeys (one after another), I get the desired result (but having two toggles is not really practical for me).

If ibus cannot change the layout itself, you can "xdotool key ISO_First_Group" resp. "xdotool key ISO_Last_Group" to programatically switch them.

Fytch · 2024-10-26 19:25:28

Ok, now I set:

localectl set-x11-keymap "ch,jp" "de_nodeadkeys,"
localectl set-keymap --no-convert de_CH-latin1

I get:

fytch@abc:~$ localectl
System Locale: LANG=en_US.UTF-8
    VC Keymap: de_CH-latin1
   X11 Layout: ch,jp
    X11 Model: de_nodeadkeys,

IBus still fails to change layouts when I use its toggle functionality even after a reboot.

Now, following your suggestion, I removed the key binding from IBus and instead added these two application shortcuts to my Xfce keyboard tool:

bash -c "setxkbmap -layout ch -variant de_nodeadkeys && ibus engine xkb:ch::gsw"
bash -c "setxkbmap -layout jp && ibus engine mozc-on"

It only works with this order (first setxkbmap and then ibus)! It's an ugly hack but I guess if IBus refuses to work any other way then I have to go with this.

seth · 2024-10-26 19:56:13

localectl set-x11-keymap "ch,jp" "de_nodeadkeys,"

X11 Model: de_nodeadkeys,

https://wiki.archlinux.org/title/Xorg/K … _localectl isn't very smart, you need to provide the correct parameters in correct order or it'll try to turn your variant into a model
Also always use setxkbmap to query the actual layout, again: localectl just reads a file it has written itself. It doesn't tell you anything about the current constitution of the X11 server.

bash -c "setxkbmap -layout jp && ibus engine mozc-on"

Ewww… don't use setxkbmap to change layouts, use the layout switch ("xdotool key ISO_First_Group" resp. "xdotool key ISO_Last_Group")
setxkbmap is slow AF and will steamroll over/reset other keyboard settings.

Fytch · 2024-11-11 01:05:19

Thanks for pointing out that mistake. It should have been:

localectl set-x11-keymap "ch,jp" "" "de_nodeadkeys,"

Unfortunately, I haven't had any success with xdotool:

fytch@abc:~$ setxkbmap -print -query
xkb_keymap {
	xkb_keycodes  { include "evdev+aliases(qwertz)"	};
	xkb_types     { include "complete"	};
	xkb_compat    { include "complete"	};
	xkb_symbols   { include "pc+ch(de_nodeadkeys)+jp:2+inet(evdev)"	};
	xkb_geometry  { include "pc(pc105)"	};
};
rules:      evdev
model:      pc105
layout:     ch,jp
variant:    de_nodeadkeys,
fytch@abc:~$ xkblayout-state print "%s"; echo
ch
fytch@abc:~$ xdotool key ISO_Last_Group
fytch@abc:~$ xkblayout-state print "%s"; echo
ch
fytch@abc:~$ xdotool key ISO_First_Group
fytch@abc:~$ xkblayout-state print "%s"; echo
ch

It's not just xkblayout-state; when I press äöü on my keyboard I actually get these letters, which shouldn't happen with the Japanese layout.

I'm fine with using the hacky setxkbmap solution if the proper solution doesn't work.

seth · 2024-11-11 08:02:09

You don't have any groub switch configured,

grep grp /usr/share/X11/xkb/rules/base.lst

Arch Linux

#1 2024-10-26 00:53:52

IBus does not use the specified (underlying) layout for Mozc Japanese

#2 2024-10-26 13:24:39

Re: IBus does not use the specified (underlying) layout for Mozc Japanese

#3 2024-10-26 14:36:47

Re: IBus does not use the specified (underlying) layout for Mozc Japanese

#4 2024-10-26 15:23:06

Re: IBus does not use the specified (underlying) layout for Mozc Japanese

#5 2024-10-26 19:25:28

Re: IBus does not use the specified (underlying) layout for Mozc Japanese

#6 2024-10-26 19:56:13

Re: IBus does not use the specified (underlying) layout for Mozc Japanese

#7 2024-11-11 01:05:19

Re: IBus does not use the specified (underlying) layout for Mozc Japanese

#8 2024-11-11 08:02:09

Re: IBus does not use the specified (underlying) layout for Mozc Japanese

Board footer