input_source_id and/or input_mode_id for Japanese?

[peterjc]

Reading https://karabiner-elements.pqrs.org/docs/json/complex-modifications-manipulator-definition/conditions/keyboard-type/ I can do this using something like this (working):

            "conditions": [
                {
                    "input_sources": [{ "language": "^ja$" }],
                    "type": "input_source_if"
                }
            ],

How can I find values for language, input_source_id and input_mode_id are for use in an input_sources conditions statement?

I am experimenting with remapping commands to enter Japanese text (unicode) and emoji. For this to work I need to tailor the rules according to the current input mode:

However, checking the language string is "ja" is not enough to tell the different Japanese input modes apart - but there are additional fields available:

        {
            "language": "language regex",
            "input_source_id": "input source id regex",
            "input_mode_id": "input mode id regex"
        },

Specifically there are three modes I want to be able to detect and handle differently:

• Unicode Hex Input (listed under languages "Others" in the settings). Here the special sequence to send a unicode character is press and hold left-alt aka option, then send the hex characters of the character (four for Japanese kana, typically eight for emoji), and release the option modifier.
• Japanese - Romanji (listed under languages "Japanese" in the settings). Here I need to send English letters, for example ha will be interpreted as は (hiragana). Depending on the language settings you can have shift type Katakana, so sending HA gives ハ instead.
• Japanese - Kana (listed under languages "Japanese" in the settings). Here to send the hiragana は, I consulted the keyboard layout, and sent the keycode for "f".

By experiment, Unicode Hex Input seems to have a blank language string, thus "language": "^$" works as the regex.

By experiment, both Japanese - Romanji and Japanese - Kana use "ja", and thus match "language": "^ja$", so I need the source or mode too to tell them apart.

Note that in the macOS input selection menu (top right), these three are listed as shown if all are installed. If only one of the Japanese modes is setup, it is instead shown as just "Japanese".

[peterjc]

I happened to be reading https://karabiner-elements.pqrs.org/docs/json/complex-modifications-manipulator-definition/to/select-input-source/ for the opposite issue, trying to change the input mode. That suggested using Karabiner-EventViewer's Variable screen.

In "ABC" mode (i.e. English for me):

    "input_source": {
        "input_source_id": "com.apple.keylayout.ABC",
        "language": "en"
    },

In romaji mode:

    "input_source": {
        "input_mode_id": "com.apple.inputmethod.Japanese",
        "input_source_id": "com.apple.inputmethod.Kotoeri.RomajiTyping.Japanese",
        "language": "ja"
    },

And in kana mode:

    "input_source": {
        "input_mode_id": "com.apple.inputmethod.Japanese",
        "input_source_id": "com.apple.inputmethod.Kotoeri.KanaTyping.Japanese",
        "language": "ja"
    },

Finally in Unicode Hex Input mode:

    "input_source": {
        "input_source_id": "com.apple.keylayout.UnicodeHexInput",
        "language": ""
    },

These values worked!