# Third-Party Notices

This repository contains project code, bundled vocabulary data, and development tooling with different provenance.

## 1. Project Code

- The `eitango` application code is licensed under Apache License 2.0.
- See [LICENSE](LICENSE) for the project license text.

## 2. Runtime Dependencies Distributed with the App

The Go application is built from modules that use permissive upstream licenses. The runtime dependency set currently referenced in `go.mod` includes:

| Component | Role | Upstream license | Reference |
| --- | --- | --- | --- |
| `charm.land/bubbletea/v2` | TUI runtime | MIT | `third_party/licenses/MIT.txt` |
| `charm.land/bubbles/v2` | TUI components | MIT | `third_party/licenses/MIT.txt` |
| `charm.land/lipgloss/v2` | terminal styling | MIT | `third_party/licenses/MIT.txt` |
| `github.com/BurntSushi/toml` | TOML parsing | MIT | `third_party/licenses/MIT.txt` |
| `github.com/spf13/cobra` | CLI framework | Apache-2.0 | `third_party/licenses/Apache-2.0.txt` |
| `github.com/google/uuid` | UUID generation | BSD-3-Clause style | `third_party/licenses/BSD-3-Clause.txt` |
| `github.com/mattn/go-runewidth` | rune/string display width handling | MIT | `third_party/licenses/MIT.txt` |
| `modernc.org/sqlite` | pure Go SQLite driver | BSD-style | `third_party/licenses/BSD-3-Clause.txt` |

These dependencies are part of the source distribution and linked binary builds. Retain the accompanying license texts when redistributing release artifacts.

## 3. Repository-Distributed Tooling for Vocabulary Generation

The repository also ships Python tooling under `scripts/vocab/`. The current pipeline uses only the Python standard library together with locally supplied corpus/database inputs and does not require third-party Python packages at runtime.

These tools are for repository maintenance and are not required for normal end-user execution of the compiled CLI.

## 4. Bundled Vocabulary Data and Upstream Data Lineage

The file `assets/words_core.jsonl` is bundled with the repository and seeded into the local database at runtime. It is a project-curated vocabulary file, and the current repository treats it conservatively as an edited dataset produced from upstream lexical resources.

Field-level provenance summary for `assets/words_core.jsonl`:

- `meaning_ja` contains Japanese meaning text curated with Japanese WordNet as an upstream lexical source.
- `frequency_rank` is derived from the Leipzig Corpora Collection English News 2024 1M word list.
- `level` is an `eitango`-specific `core-1` to `core-4` bucket, not an upstream source label.

Relevant upstream references:

| Upstream source | How it is used here | License reference |
| --- | --- | --- |
| Japanese WordNet (`wnjpn.db`) | queried by scripts under `scripts/vocab/` for meanings and examples; published results that use derived data should preserve the Japanese WordNet attribution guidance | `third_party/licenses/Japanese-WordNet.txt` |
| Leipzig Corpora Collection English News 2024 1M word list | consulted for frequency-ranked seed generation and bundled-core ranking | `third_party/licenses/CC-BY-3.0.txt` |
| Princeton WordNet / WordNet-family notices | relevant to WordNet-derived corpora and notices | `third_party/licenses/Princeton-WordNet.txt` |

Important distribution note:

- this repository does **not** ship raw `wnjpn.db`
- this repository does **not** ship the raw Leipzig corpus files used during local regeneration
- this repository **does** ship `assets/words_core.jsonl`, which should be redistributed together with this notice and the referenced upstream license materials
- redistributions that publish or embed Japanese-WordNet-influenced data should also keep the version-appropriate attribution link wording documented in `third_party/licenses/Japanese-WordNet.txt`

Recommended public attribution example for the current expected Japanese WordNet input:

```text
Japanese Wordnet (v1.1) © 2009-2011 NICT, 2012-2015 Francis Bond and 2016-2024 Francis Bond, Takayuki Kuribayashi
https://bond-lab.github.io/wnja/index.en.html
```

```text
日本語ワードネット（1.1版）© 2009-2011 NICT, 2012-2015 Francis Bond and 2016-2024 Francis Bond, Takayuki Kuribayashi
https://bond-lab.github.io/wnja/index.ja.html
```

If you redistribute `words_core.jsonl` or a derivative package that embeds it, keep this notice, the referenced upstream license materials, and the Japanese WordNet attribution guidance before making additional licensing claims.

## 5. Practical Guidance for Redistributors

- Treat the application code license and the bundled data provenance as separate concerns.
- Do not describe the entire repository as if every bundled artifact were covered only by Apache-2.0.
- Include `LICENSE`, this file, and `third_party/licenses/` in source or binary redistributions.
- For Japanese-WordNet-influenced data, preserve the version-appropriate attribution link text and companion license note from `third_party/licenses/Japanese-WordNet.txt`.
- When in doubt about data provenance, keep the attribution and upstream links intact rather than removing them.