Taskfile.yml

version: "3"

vars:
  PYTHON: uv run --no-sync python
  MATURIN: uv run --no-sync maturin
  # On NixOS, maturin's hard-link install can silently leave the old .so in
  # site-packages. SO_CP copies the freshly-built .so to the correct location.
  PY_SO_DST: uv run --no-sync python -c "import sysconfig,os; print(os.path.join(sysconfig.get_path('platlib'),'object_storage_proxy','object_storage_proxy'+sysconfig.get_config_var('EXT_SUFFIX')))"

tasks:
  # ── setup ──────────────────────────────────────────────────────────────────

  setup:
    desc: "Install Python dev dependencies and install git hooks"
    cmds:
      - uv sync
      - uv run pre-commit install
      - echo "✅ pre-commit hooks installed"

  hooks:install:
    desc: "Install (or re-install) the pre-commit git hooks"
    cmds:
      - uv run pre-commit install

  hooks:run:
    desc: "Run pre-commit checks against all files"
    cmds:
      - uv run pre-commit run --all-files

  hooks:update:
    desc: "Bump pre-commit hook revisions to latest"
    cmds:
      - uv run pre-commit autoupdate

  # ── changelog ─────────────────────────────────────────────────────────────

  changelog:
    desc: "Regenerate CHANGELOG.md from git history (all tags)"
    cmds:
      - git cliff --config cliff.toml -o CHANGELOG.md

  changelog:unreleased:
    desc: "Preview unreleased changes since the last tag"
    cmds:
      - git cliff --config cliff.toml --unreleased

  changelog:tag:
    desc: "Generate changelog for a specific tag range, e.g. task changelog:tag -- v0.5.3..v0.5.4"
    cmds:
      - git cliff --config cliff.toml {{.CLI_ARGS}}

  # ── release ────────────────────────────────────────────────────────────────
  # Usage: bump version in Cargo.toml, then run `task release`

  release:
    desc: "Read version from Cargo.toml, update CHANGELOG, commit, and tag"
    cmds:
      - |
        VERSION=$(grep '^version' Cargo.toml | head -1 | sed 's/version = "//;s/"//')
        TAG="v${VERSION}"
        echo "Releasing ${TAG}"

        # 1. Regenerate CHANGELOG.md for the new tag
        git cliff --config cliff.toml --tag "${TAG}" -o CHANGELOG.md

        # 2. Stage and commit
        git add Cargo.toml CHANGELOG.md
        # Run pre-commit; if it auto-fixes files, stage again and retry
        git diff --cached --quiet && echo "Nothing to commit" || \
          git commit -m "chore: release ${TAG}" || \
          { git add Cargo.toml CHANGELOG.md && git commit -m "chore: release ${TAG}"; }

        # 3. Create the annotated tag
        git tag -a "${TAG}" -m "Release ${TAG}"

        # 4. Push commit and tag
        git push
        git push origin "${TAG}"
        echo "✅ Released ${TAG}"

  # ── build ──────────────────────────────────────────────────────────────────

  build:
    desc: "Compile the Rust extension and install it into .venv (dev build)"
    cmds:
      - cargo build
      - cp target/debug/libobject_storage_proxy.so $({{.PY_SO_DST}})

  build:release:
    desc: "Compile with release optimisations and install into .venv"
    cmds:
      - cargo build --release
      - cp target/release/libobject_storage_proxy.so $({{.PY_SO_DST}})

  wheel:
    desc: "Build a debug wheel into target/wheels/"
    cmds:
      - uv run maturin build

  wheel:release:
    desc: "Build a release wheel into target/wheels/"
    cmds:
      - "{{.MATURIN}} build --release"

  # ── run ────────────────────────────────────────────────────────────────────

  run:
    desc: "Build (dev) and start the proxy test server"
    deps: [build]
    cmds:
      - "{{.PYTHON}} test_server.py"

  run:release:
    desc: "Build (release) and start the proxy test server"
    deps: [build:release]
    cmds:
      - "{{.PYTHON}} test_server.py"

  # ── test ───────────────────────────────────────────────────────────────────

  test:rust:
    desc: "Run Rust unit tests with cargo nextest (falls back to cargo test)"
    cmds:
      - cargo nextest run
    ignore_error: false

  test:rust:cargo:
    desc: "Run Rust unit tests with plain cargo test"
    cmds:
      - cargo test

  test:integration:
    desc: "Run the integration test shell script"
    cmds:
      - bash test_integration.sh

  test:
    desc: "Run all tests (Rust + integration)"
    deps: [build]
    cmds:
      - task: test:rust
      - task: test:integration

  # ── integration environment ────────────────────────────────────────────────
  # Lifecycle is intentionally split so you can keep Garage running for
  # manual dev/testing while restarting just the proxy (or vice-versa).
  #
  # Quick reference:
  #   task integration:up             # full stack (garage + bootstrap + proxy)
  #   task integration:down           # tear everything down
  #   task integration:test           # run tests against whatever is running
  #   task integration:run            # automated: up -> test -> down
  #
  #   task integration:garage:up      # just Garage (stays running)
  #   task integration:garage:bootstrap   # create bucket + key, write .env
  #   task integration:server:start   # just the OSP proxy (background)
  #   task integration:server:stop    # kill the proxy
  #
  #   task integration:minio:up       # start MinIO container
  #   task integration:minio:bootstrap # create bucket, write .env.minio
  #   task integration:minio:down     # stop MinIO container
  #   task integration:minio:destroy  # stop MinIO AND remove volumes
  #   task integration:test:spark         # Spark tests against all backends
  #   task integration:test:spark:garage  # Spark tests (Garage only)
  #   task integration:test:spark:minio   # Spark tests (MinIO only)
  #   task integration:test:duckdb        # DuckDB tests against all backends
  #   task integration:test:duckdb:fast   # DuckDB tests, stop on first failure

  integration:setup:
    desc: "Install Python deps for the integration test project"
    dir: integration/test_server
    cmds:
      - uv sync

  integration:garage:up:
    desc: "Start the Garage S3 container (stays running)"
    dir: integration/test_server
    cmds:
      - docker compose up -d
      - echo "Garage starting — waiting for healthy status …"
      - |
        for i in $(seq 1 30); do
          status=$(docker inspect --format='{{"{{"}}{{".State.Health.Status"}}{{"}}"}}' osp-garage 2>/dev/null)
          if [ "$status" = "healthy" ]; then echo "Garage is healthy"; exit 0; fi
          echo "  waiting… ($i/30) status=$status"; sleep 2
        done
        echo "Garage did not become healthy in time" >&2; exit 1

  integration:garage:down:
    desc: "Stop and remove the Garage container"
    dir: integration/test_server
    cmds:
      - docker compose down

  integration:garage:destroy:
    desc: "Stop Garage AND remove its data volumes"
    dir: integration/test_server
    cmds:
      - docker compose down -v

  integration:garage:logs:
    desc: "Follow Garage container logs"
    dir: integration/test_server
    cmds:
      - docker compose logs -f

  integration:garage:status:
    desc: "Show Garage cluster status via admin API"
    cmds:
      - |
        curl -sf -H "Authorization: Bearer osp-integration-admin-token" \
          http://localhost:3903/v1/status | python3 -m json.tool

  integration:garage:bootstrap:
    desc: "Create bucket + HMAC key in Garage, write integration/test_server/.env"
    dir: integration/test_server
    cmds:
      - uv run python bootstrap.py

  integration:minio:up:
    desc: "Start the MinIO container (stays running)"
    dir: integration/test_server
    cmds:
      - docker compose up -d minio
      - echo "MinIO starting — waiting for healthy status …"
      - |
        for i in $(seq 1 30); do
          status=$(docker inspect --format='{{"{{"}}{{".State.Health.Status"}}{{"}}"}}' osp-minio 2>/dev/null)
          if [ "$status" = "healthy" ]; then echo "MinIO is healthy"; exit 0; fi
          echo "  waiting… ($i/30) status=$status"; sleep 2
        done
        echo "MinIO did not become healthy in time" >&2; exit 1

  integration:minio:bootstrap:
    desc: "Create MinIO bucket and write integration/test_server/.env.minio"
    dir: integration/test_server
    cmds:
      - uv run python minio_bootstrap.py

  integration:minio:down:
    desc: "Stop and remove the MinIO container"
    dir: integration/test_server
    cmds:
      - docker compose stop minio
      - docker compose rm -f minio

  integration:minio:destroy:
    desc: "Stop MinIO AND remove its data volumes"
    dir: integration/test_server
    cmds:
      - docker compose down -v minio

  integration:server:start:
    desc: "Start the OSP proxy in the background (logs -> integration/test_server/proxy.log)"
    cmds:
      - |
        nohup uv run --no-sync python integration/test_server/server.py \
          > integration/test_server/proxy.log 2>&1 &
        sleep 2
        pgrep -f "server\.py" | head -1 > integration/test_server/.proxy.pid
        if kill -0 $(cat integration/test_server/.proxy.pid) 2>/dev/null; then
          echo "✅ Proxy started (PID=$(cat integration/test_server/.proxy.pid))"
          echo "   logs: integration/test_server/proxy.log"
          echo "   http://localhost:6190   metrics: http://localhost:9091/metrics"
        else
          echo "❌ Proxy failed to start — check integration/test_server/proxy.log"
          exit 1
        fi

  integration:server:stop:
    desc: "Stop the background OSP proxy"
    cmds:
      - |
        PID_FILE=integration/test_server/.proxy.pid
        if [ -f "$PID_FILE" ]; then
          PID=$(cat "$PID_FILE")
          kill "$PID" 2>/dev/null && echo "Proxy (PID=$PID) stopped." || echo "Process not found."
          rm -f "$PID_FILE"
        else
          echo "No .proxy.pid file found — proxy may not be running."
        fi

  integration:server:logs:
    desc: "Tail the proxy log"
    cmds:
      - tail -f integration/test_server/proxy.log

  integration:bootstrap:
    desc: "Bootstrap both backends: create Garage bucket+key and MinIO bucket, write .env files"
    cmds:
      - task: integration:garage:bootstrap
      - task: integration:minio:bootstrap

  integration:up:
    desc: "Full environment: Garage + MinIO up -> bootstrap both -> proxy start"
    cmds:
      - task: integration:garage:up
      - task: integration:minio:up
      - task: integration:bootstrap
      - task: integration:server:start

  integration:down:
    desc: "Tear down: stop proxy -> stop Garage + MinIO"
    cmds:
      - task: integration:server:stop
      - task: integration:garage:down
      - task: integration:minio:down

  integration:test:
    desc: "Run integration tests against the running environment (excludes Spark and DuckDB)"
    dir: integration/test_server
    cmds:
      - uv run pytest tests/ -q -m 'not spark and not spark_minio and not duckdb'

  integration:test:fast:
    desc: "Run integration tests, stop on first failure (excludes Spark and DuckDB)"
    dir: integration/test_server
    cmds:
      - uv run pytest tests/ -q -x -m 'not spark and not spark_minio and not duckdb'

  integration:test:spark:
    desc: "Run Spark s3a tests against all backends (Garage + MinIO)"
    cmds:
      - task: integration:test:spark:garage
      - task: integration:test:spark:minio

  integration:test:spark:garage:
    desc: "Run only the Spark s3a tests (Garage backend)"
    dir: integration/test_server
    cmds:
      - uv run pytest tests/test_spark.py -q -m spark

  integration:test:spark:garage:fast:
    desc: "Run Spark tests (Garage), stop on first failure"
    dir: integration/test_server
    cmds:
      - uv run pytest tests/test_spark.py -q -m spark -x

  integration:test:spark:fast:
    desc: "Run Spark tests against all backends, stop on first failure"
    dir: integration/test_server
    cmds:
      - uv run pytest tests/test_spark.py tests/test_spark_minio.py -q -m 'spark or spark_minio' -x

  integration:test:spark:minio:
    desc: "Run only the Spark s3a tests (MinIO backend)"
    dir: integration/test_server
    cmds:
      - uv run pytest tests/test_spark_minio.py -q -m spark_minio

  integration:test:spark:minio:fast:
    desc: "Run Spark tests (MinIO), stop on first failure"
    dir: integration/test_server
    cmds:
      - uv run pytest tests/test_spark_minio.py -q -m spark_minio -x

  integration:test:duckdb:
    desc: "Run DuckDB httpfs tests against all backends (Garage + MinIO)"
    dir: integration/test_server
    cmds:
      - uv run pytest tests/test_duckdb.py -q -m duckdb

  integration:test:duckdb:fast:
    desc: "Run DuckDB httpfs tests, stop on first failure"
    dir: integration/test_server
    cmds:
      - uv run pytest tests/test_duckdb.py -q -m duckdb -x

  integration:test:all:
    desc: "Run full test suite including Spark and DuckDB tests"
    cmds:
      - task: integration:test
      - task: integration:test:spark
      - task: integration:test:duckdb

  integration:run:
    desc: "Automated: up -> test -> down (tears down even on failure)"
    cmds:
      - task: integration:up
      - defer: {task: integration:down}
      - task: integration:test

  # ── lint / fmt ─────────────────────────────────────────────────────────────

  fmt:
    desc: "Format Rust code"
    cmds:
      - cargo fmt

  lint:rust:
    desc: "Run clippy on the Rust code"
    cmds:
      - cargo clippy -- -D warnings

  lint:
    desc: "Lint everything (Rust clippy)"
    cmds:
      - task: lint:rust

  # ── clean ──────────────────────────────────────────────────────────────────

  clean:
    desc: "Remove Rust build artefacts and the Python virtual environment"
    cmds:
      - cargo clean
      - rm -rf .venv

  clean:wheels:
    desc: "Remove built wheels only"
    cmds:
      - rm -rf target/wheels

  # ── default ────────────────────────────────────────────────────────────────

  default:
    desc: "Show available tasks"
    cmds:
      - task --list