Rules for the Vietnamese input methods VNI (and additionally Telex for comparison purpose).

Producing rules was done by parsing words from the Hunspell-vi dictionary and written to generated-im-vni.json, generated-im-telex.json.

For Android Florisboard users, import floris-im-vni-extension.flex to type with VNI.

Install:

(package-vc-install '(viet-im :url "https://gitlab.com/daanturo/viet-im"))

Then call set-input-method/toggle-input-method (toggle-input-method need prefix argument if it was already called in this Emacs session) and select vietnamese-vni-x or vietnamese-telex-x.

The package currently supports VNI, Telex (secondary), VIQR and other input methods aren't planned at the moment. VIQR is conceptually similar to VNI and Telex, the only (IIRC) difference among the 3 IMs is at triggering characters so adding support for it shouldn't be hard if needed.

Related issue: https://lists.gnu.org/archive/html/help-gnu-emacs/2025-09/msg00205.html. Tan Ky mode is supported for both VNI and Telex, allows typing diacritics anywhere after the base letters.

With uv installed:

uv venv
source .venv/bin/activate
python3 main.py

Processing steps on a high-level:

• From words in dictionary, extract rhymes
• Generate generated-preliminary-table.json: core information - rhymes with their tone positions
• Generate generated-rhyme-table.json: describes how each rhyme response to each diacritic change
• Generate generated-im-telex.json & generated-im-vni.json: rules for those input methods

• https://en.wikipedia.org/wiki/Vietnamese_language_and_computers#Unicode_code_points
• https://github.com/1ec5/hunspell-vi/blob/main/dictionaries/vi-DauMoi.dic