Onym

Onym is a thesaurus and dictionary for GNOME, built on WordNet. You type a word and Onym shows its meanings grouped by part of speech, with example sentences, then its synonyms, its antonyms, and the lexical relations that connect it to other words. Those related terms are clickable, so one lookup leads to the next with a click. Search is live, with completion as you type and a suggestion when a word is misspelt. On wide windows a sidebar lists more words: the headwords that extend the current one and the other completions of its letters, each a click away. A dice button (or Ctrl+R) jumps to a random word, and a right click (or the Menu key) copies any word title or chip as plain text.

The word data comes from Open English WordNet, the actively maintained successor to Princeton's WordNet, with word origins from Wiktionary and translations from the Open Multilingual Wordnet. The lookup engine is onym-engine, a shared Rust core whose behaviour derives from Artha, an earlier WordNet thesaurus by Sundaram Ramaswamy. Onym is a fresh GTK4 and libadwaita application around that engine, built to feel at home on GNOME 50 and later.

Status

Onym is functional: the GTK4 and libadwaita application, the libonym library, and a headless command line tool. The application shows meanings grouped by part of speech with examples, then synonyms, antonyms, and the lexical relation trees, with live search, history, and an accessibility pass. It ships an icon, a desktop entry, AppStream metainfo, and a Flatpak that bundles WordNet.

• libonym is an installable library that turns a word into a structured result. It is the reusable core, with a stable model, a pkg-config file, and GObject Introspection so it binds from C, Python, JavaScript, and Vala.
• onym-cli is a small tool that looks words up from the terminal. It proves the library and drives the tests.

Building

Onym uses Meson. The library needs GLib, a Rust toolchain for the engine, zstd for the data preparation, and a core engine checkout beside this repository (or name one with -Donym_engine_dir). The WordNet data comes from the onym-data submodule, which meson prepares into the build tree, so no system WordNet is needed. On Debian or Ubuntu:

sudo apt install meson cargo rustc zstd libglib2.0-dev gobject-introspection libgirepository1.0-dev
git clone https://forge.ursa.nz/ursa-nz/onym-engine.git ../core
git submodule update --init
meson setup _build
meson compile -C _build
meson test -C _build

Try it:

./_build/tools/onym-cli serendipity
./_build/tools/onym-cli --complete sere

Flatpak

A Flatpak manifest is at build-aux/nz.ursa.Onym.yaml, building against org.gnome.Platform. It bundles the WordNet data from the onym-data submodule, so the Flatpak needs nothing else installed and the builder fetches no database:

flatpak-builder --user --install --force-clean _flatpak build-aux/nz.ursa.Onym.yaml
flatpak run nz.ursa.Onym

How the code is organised

Onym is three layers, so the lexical machinery stays sealed off and the rest stays easy to follow.

• The engine is the shared onym-engine Rust core, linked in as a static archive through its C ABI. It owns the WordNet file parsing, the morphology, the lookup rules, and the lemma index, and it is conformance-tested in its own repository.
• The bridge is the only code that calls the engine. It copies what the engine returns into a clean model of plain objects, then frees the engine's data.
• The library and application see only that model. They never know WordNet exists.

OnymEngine is the object you ask for a lookup. It hands back an OnymResult, which is a list of sections, each holding definitions, words, or antonyms. The application renders that result and never parses anything itself. ARCHITECTURE.md describes the layers and the files in more detail.

Licence and credits

Onym is free software under the GPL, version 3 or later. See COPYING.

• The word data is Open English WordNet 2025, under CC-BY-4.0, with the Princeton WordNet grant preserved. It is bundled from the onym-data submodule, which carries the full provenance.
• Etymology comes from Wiktionary, under CC-BY-SA-3.0, via the wiktextract dataset.
• Translations come from the Open Multilingual Wordnet: MultiWordNet (Italian, CC-BY-3.0), Wordnet Bahasa (Indonesian, MIT), and OpenWN-PT (Portuguese, CC-BY-SA-3.0), aligned through the Collaborative Interlingual Index (CC-BY-4.0).
• The lookup engine is onym-engine, GPL-3.0-or-later, whose behaviour derives from Artha by Sundaram Ramaswamy; the derivation is recorded in that repository's PROVENANCE.md.
• Onym is built with GTK, libadwaita, and GLib from the GNOME project.