ci: add Release Please and auto-publish the compiled .dict on release

.github/workflows/assemble.yml

@@ -2,7 +2,7 @@ name: assemble dictionary
 
 # Stage B (lightweight): merge cached + live sources, filter, score, emit a
 # `.combined` wordlist; open a PR for human review. The compiled `.dict` is
-# produced from a merged PR by the release-dict workflow (not yet authored).
+# produced and published on release by release.yml (Release Please).
 
 on:
   schedule:

.github/workflows/release.yml

@@ -0,0 +1,70 @@
+name: release
+
+# Release Please maintains a release PR from Conventional Commits on main.
+# When that PR merges, it tags the commit and creates a GitHub Release; the
+# publish-dict job then compiles the puhekieli `.dict` and attaches it to the
+# release as a stable, versioned download (consumed by Workstream B).
+
+on:
+  push:
+    branches: [main]
+
+permissions:
+  contents: write
+  pull-requests: write
+
+concurrency:
+  group: release-${{ github.ref }}
+  cancel-in-progress: false
+
+jobs:
+  release-please:
+    runs-on: ubuntu-latest
+    outputs:
+      release_created: ${{ steps.release.outputs.release_created }}
+      tag_name: ${{ steps.release.outputs.tag_name }}
+    steps:
+      - uses: googleapis/release-please-action@v4
+        id: release
+        with:
+          config-file: release-please-config.json
+          manifest-file: .release-please-manifest.json
+
+  publish-dict:
+    needs: release-please
+    if: ${{ needs.release-please.outputs.release_created == 'true' }}
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          ref: ${{ needs.release-please.outputs.tag_name }}
+      - name: install libvoikko + Finnish dictionary
+        run: |
+          sudo apt-get update
+          sudo apt-get install -y libvoikko-dev voikko-fi
+      - uses: dtolnay/rust-toolchain@stable
+      - uses: Swatinem/rust-cache@v2
+      - uses: extractions/setup-just@v2
+      - name: set up JDK (for dicttool)
+        uses: actions/setup-java@v4
+        with:
+          distribution: temurin
+          java-version: "17"
+      - name: build .combined + compile .dict
+        # Downloads the OpenSubtitles corpus + dicttool jar, assembles the
+        # wordlist (always includes the committed curated seed; Mastodon is
+        # skipped — no cache in a fresh checkout), and compiles the .dict.
+        run: just generate
+      - name: stage versioned artifacts
+        env:
+          TAG: ${{ needs.release-please.outputs.tag_name }}
+        run: |
+          mkdir -p dist
+          cp data/out/puhekieli_fi.dict "dist/puhekieli_fi-${TAG}.dict"
+          cp data/out/niinku.combined "dist/niinku-${TAG}.combined"
+          ls -l dist
+      - name: attach artifacts to the release
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          TAG: ${{ needs.release-please.outputs.tag_name }}
+        run: gh release upload "$TAG" dist/* --clobber

.release-please-manifest.json

@@ -0,0 +1,3 @@
+{
+  ".": "0.0.1"
+}

Cargo.toml

@@ -7,7 +7,7 @@ edition = "2021"
 license = "MIT"
 repository = "https://github.com/laurigates/niinku"
 authors = ["Lauri Gates"]
-version = "0.0.1"
+version = "0.0.1" # x-release-please-version
 
 [workspace.dependencies]
 anyhow = "1"

README.md

@@ -42,7 +42,11 @@ just ingest-mastodon # pull Finnish hashtag posts → data/cached/mastodon-fi.tx
 just ingest        # pass through to the CLI (e.g. `just ingest mastodon --tags suomi`)
 ```
 
-The `.dict` file ships at `data/out/puhekieli_fi.dict`. To use it:
+Pre-built `.dict` files are attached to each
+[GitHub Release](https://github.com/laurigates/niinku/releases) as
+`puhekieli_fi-<version>.dict` (built automatically — see CI/CD below).
+You can also build one locally with `just generate`, which writes
+`data/out/puhekieli_fi.dict`. To use it:
 
 1. Transfer to an Android device that has HeliBoard installed.
 2. Settings → Languages → Finnish → "Add dictionary from file" → select the `.dict`.
@@ -164,8 +168,13 @@ it.
 - Scheduled `cron` (monthly suggested) runs Stage B.
 - The job does **not** auto-publish. It opens a PR containing the regenerated
   wordlist plus a diff of new-vs-removed entries. A human merges.
-- On merge, the compiled `.dict` is published as a GitHub Release artifact at
-  a stable URL (consumed by Workstream B).
+- Releases are cut by [Release Please](https://github.com/googleapis/release-please)
+  (`release.yml`): it maintains a release PR from Conventional Commits on
+  `main`; merging it tags the commit and creates a GitHub Release.
+- On release, `release.yml` compiles the `.dict` and attaches it (plus the
+  `.combined` source) to the release as a stable, versioned download
+  (consumed by Workstream B). Version bumps are driven by commit type
+  (`feat:` → minor, `fix:` → patch) — see [Conventional Commits](https://www.conventionalcommits.org).
 
 ### Repo layout
 
@@ -181,6 +190,7 @@ it.
 /.github/workflows
   ingest.yml     # Stage A, manual / version-triggered
   assemble.yml   # Stage B, cron + PR
+  release.yml    # Release Please + compile/publish .dict on release
   ci.yml         # fmt, clippy, test
 /tools
   dicttool_aosp.jar   # not yet vendored

release-please-config.json

@@ -0,0 +1,14 @@
+{
+  "$schema": "https://raw.githubusercontent.com/googleapis/release-please/main/schemas/config.json",
+  "packages": {
+    ".": {
+      "release-type": "simple",
+      "package-name": "niinku",
+      "changelog-path": "CHANGELOG.md",
+      "include-component-in-tag": false,
+      "extra-files": [
+        "Cargo.toml"
+      ]
+    }
+  }
+}