https://github.com/nepalibhasha/varnavinyas
वर्णविन्यास — Open-source Nepali orthography toolkit based on Nepal Academy standards. Spell checking, sandhi analysis, punctuation diagnostics, and more.
https://github.com/nepalibhasha/varnavinyas
devanagari nepali nlp rust spell-checker wasm
Last synced: about 2 months ago
JSON representation
वर्णविन्यास — Open-source Nepali orthography toolkit based on Nepal Academy standards. Spell checking, sandhi analysis, punctuation diagnostics, and more.
- Host: GitHub
- URL: https://github.com/nepalibhasha/varnavinyas
- Owner: nepalibhasha
- License: apache-2.0
- Created: 2026-02-10T20:25:11.000Z (4 months ago)
- Default Branch: main
- Last Pushed: 2026-03-26T17:17:04.000Z (2 months ago)
- Last Synced: 2026-03-26T18:13:38.131Z (2 months ago)
- Topics: devanagari, nepali, nlp, rust, spell-checker, wasm
- Language: Rust
- Homepage: https://nepalibhasha.github.io/varnavinyas/
- Size: 3.14 MB
- Stars: 13
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- Contributing: .github/CONTRIBUTING.md
- License: LICENSE-APACHE
- Code of conduct: .github/CODE_OF_CONDUCT.md
- Security: .github/SECURITY.md
- Support: .github/SUPPORT.md
Awesome Lists containing this project
README
# Varnavinyas (वर्णविन्यास)
Open-source Nepali orthography tooling based on Nepal Academy standards.
*शुद्ध नेपाली, सबैका लागि।*
*(Correct Nepali, for everyone.)*
## What This Project Is
Varnavinyas is a Rust workspace for checking and normalizing Nepali text.
It is built for three kinds of users:
- writers, editors, teachers, students, and institutions that need standard Nepali spelling and punctuation
- developers who want a reusable Nepali orthography engine
- contributors who want Academy-aligned rules implemented transparently and auditable in code
The project focuses on:
- word-level orthography correction
- punctuation diagnostics
- rule tracing with Academy citations
- web, CLI, editor, and binding surfaces on top of the same core engine
Diagnostics use stable `category_code` values across the web app, CLI, LSP, and bindings so filtering and highlighting stay consistent.
## Use Varnavinyas
### Browser
Use the hosted web app linked from the repository homepage.
The web app includes:
- text checker
- word inspector
- rules reference
### CLI
Build the workspace:
```bash
cargo build --workspace
```
Run the checker:
```bash
cargo run -p varnavinyas -- check
```
Run with JSON output:
```bash
cargo run -p varnavinyas -- check --format json
```
### Editor / LSP
The workspace includes an LSP server in `crates/lsp` for editor integrations and diagnostic surfacing.
### Bindings
Public bindings exist for:
- WebAssembly: `crates/bindings-wasm`
- Python: `crates/bindings-python`
- C: `crates/bindings-c`
- UniFFI: `crates/bindings-uniffi`
## Architecture At A Glance
```mermaid
flowchart LR
A[kosha
lexicon + metadata]
B[prakriya
token-level correction]
C[parikshak
text-level checking]
D[CLI / LSP / Web / Bindings]
A --> B
B --> C
C --> D
```
The core flow is:
1. `kosha` provides lexicon lookup and metadata
2. `prakriya` decides token-level standard form and rule trace
3. `parikshak` runs text-level checking, span handling, padayog/padabiyog passes, punctuation, and heuristics
4. CLI, web, LSP, and bindings present those diagnostics to users
The main crates are:
- `crates/prakriya`: token-level orthography engine
- `crates/parikshak`: end-to-end text checker
- `crates/kosha`: lexicon and headword metadata
- `crates/lekhya`: punctuation diagnostics
- `web/`: browser UI backed by WASM
Inside `crates/prakriya`:
- `src/varna_vinyasa/` owns Academy orthography families
- `src/usage_fixes/` owns later cleanup-style rules
- `src/runtime.rs` assembles and caches runtime rule dispatch
- `src/model/` owns core derivation types
Inside `crates/parikshak`:
- `src/checker/word_level.rs` owns token-level integration
- `src/checker/padayog.rs` owns join/split text passes
- `src/checker/punctuation.rs` owns punctuation diagnostics
- `src/checker/style_variants.rs` and `src/checker/grammar.rs` own higher-level heuristics
## Workspace Layout
```text
crates/
core: akshar lipi shabda sandhi types
checking: kosha prakriya lekhya parikshak
analysis: vyakaran samasa eval
surfaces: cli lsp bindings-*
web/
docs/
```
For new contributors, the crate split is easiest to read as four groups:
- Foundation crates
- `crates/akshar`: Devanagari normalization, classification, and akshara utilities
- `crates/lipi`: transliteration and legacy-font conversion helpers
- `crates/types`: shared domain enums and data types
- Language knowledge crates
- `crates/kosha`: lexicon lookup and headword metadata
- `crates/shabda`: origin classification and lightweight decomposition
- `crates/sandhi`: sandhi rules and helpers
- Checking engine crates
- `crates/prakriya`: token-level orthography derivation and rule tracing
- `crates/lekhya`: punctuation diagnostics
- `crates/parikshak`: full-text checker pipeline that composes the lower layers
- Analysis and surface crates
- `crates/vyakaran`, `crates/samasa`: grammar and compound-analysis support
- `crates/eval`: evaluation harnesses over curated fixtures
- `crates/cli`, `crates/lsp`, `crates/bindings-*`: user-facing delivery surfaces
- `web/`: browser UI backed by the WASM bindings
## Build And Test
### Prerequisites
- Rust 1.85.0+
- Cargo
- optional for web builds: `wasm-pack` and `wasm-bindgen-cli`
### Main Commands
```bash
cargo build --workspace
cargo fmt --all --check
cargo clippy --workspace --all-targets -- -D warnings
cargo test --workspace -q
```
Build the web app:
```bash
bash web/build.sh
```
Smoke-test the web app:
```bash
bash web/smoke-test.sh
```
Serve the built web app locally:
```bash
python3 -m http.server 8080 --directory web/
```
## Documentation
Start here:
- [docs/README.md](docs/README.md)
Key docs:
- [docs/VISION.md](docs/VISION.md) — why the project exists
- [docs/ARCHITECTURE.md](docs/ARCHITECTURE.md) — system design and crate boundaries
- [docs/DEVELOPMENT.md](docs/DEVELOPMENT.md) — build and test workflow
- [docs/DATASETS.md](docs/DATASETS.md) — datasets and provenance
- [docs/RULES.md](docs/RULES.md) — rule implementation notes
- [docs/STATUS.md](docs/STATUS.md) — current feature matrix
- [docs/EXTENSION_WASM_CONTRACT.md](docs/EXTENSION_WASM_CONTRACT.md) — downstream browser artifact release and WASM contract
- [docs/Notices-pages-77-99.md](docs/Notices-pages-77-99.md) — Academy notice excerpt used for rule alignment
- [docs/PS-Saisanik-Vyakaran-Varnavinyas-Page-327-349.md](docs/PS-Saisanik-Vyakaran-Varnavinyas-Page-327-349.md) — Academy school-grammar reference also used for rule alignment
Crate-specific architecture docs:
- [crates/prakriya/ARCHITECTURE.md](crates/prakriya/ARCHITECTURE.md)
- [crates/parikshak/ARCHITECTURE.md](crates/parikshak/ARCHITECTURE.md)
## Contributing
Technical and non-technical contributions are welcome.
Start with:
- [.github/CONTRIBUTING.md](.github/CONTRIBUTING.md)
- [docs/RUST_GUIDE.md](docs/RUST_GUIDE.md)
- [docs/BACKLOG.md](docs/BACKLOG.md)
Community and process files:
- [.github/CODE_OF_CONDUCT.md](.github/CODE_OF_CONDUCT.md)
- [.github/SECURITY.md](.github/SECURITY.md)
- [.github/SUPPORT.md](.github/SUPPORT.md)
## License
Dual-licensed under MIT or Apache-2.0.
- [LICENSE-MIT](LICENSE-MIT)
- [LICENSE-APACHE](LICENSE-APACHE)