Taskfile.yml

version: "3"

includes:
    rust:
        taskfile: .task/rust.yml
    python:
        taskfile: .task/python.yml
    node:
        taskfile: .task/node.yml
    go:
        taskfile: .task/go.yml
    java:
        taskfile: .task/java.yml
    elixir:
        taskfile: .task/elixir.yml
    c:
        taskfile: .task/c.yml
    ruby:
        taskfile: .task/ruby.yml
    wasm:
        taskfile: .task/wasm.yml
    php:
        taskfile: .task/php.yml
    csharp:
        taskfile: .task/csharp.yml
    dart:
        taskfile: .task/dart.yml
    kotlin-android:
        taskfile: .task/kotlin-android.yml
    swift:
        taskfile: .task/swift.yml
    zig:
        taskfile: .task/zig.yml
    docs:
        taskfile: ./.task/tools/docs.yml
    version:
        taskfile: .task/version.yml
    test-apps:
        taskfile: .task/test-apps.yml
    _e2e_wf:
        taskfile: .task/workflows/e2e.yml

vars:
    PROJECT_ROOT:
        sh: pwd
    VERSION:
        sh: python3 -c "import tomllib; print(tomllib.load(open('Cargo.toml','rb'))['workspace']['package']['version'])"
    FFI_TEST_LANGUAGES: "python,rust,javascript,typescript,go,html,css,json,mojo,nim,norg"

# Preconditions for task setup:
# - Rust toolchain (rustup) — install via https://rustup.rs
# - alef CLI — install via cargo install --git https://github.com/kreuzberg-dev/alef
# - prek — install via cargo install prek
# - uv (Python package manager) — for clone_vendors.py and other Python scripts
# - git, network access for grammar cloning
# - Per-language tools auto-skipped via alef preconditions

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

    setup:
        desc: "Install all dependencies and dev tools"
        cmds:
            - task: rust:install
            - alef setup
            - prek install
            - prek install --hook-type commit-msg
            - task: clone:if-missing

    ai-rulez:
        desc: Generate ai-rulez configuration files (CLAUDE.md, etc.)
        cmds:
            - pnpm dlx ai-rulez@latest generate

    clone:
        desc: "Clone all language grammar repositories (full re-clone)"
        cmds:
            - uv run --no-sync scripts/clone_vendors.py

    clone:if-missing:
        desc: "Clone only missing language grammar repositories"
        status:
            - test -d parsers && [ "$(ls -A parsers 2>/dev/null | wc -l)" -gt 0 ]
        cmds:
            - uv run --no-sync scripts/clone_vendors.py

    build:
        desc: Build Rust core crates only
        cmds:
            - task: build:core

    build:dev:
        desc: Build in dev mode with a few languages for fast iteration
        cmds:
            - task: rust:build:dev

    build:core:
        desc: Build Rust core crates only
        cmds:
            - task: rust:build

    build:all:
        desc: Build Rust core crates and all language bindings
        cmds:
            - task: build:core
            - task: build:bindings

    build:release:
        desc: Build Rust core crates in release mode
        cmds:
            - task: rust:build:release

    build:ci:
        desc: Build Rust core crates in CI mode
        cmds:
            - task: rust:build:release

    build:bindings:
        desc: Build all language bindings
        cmds:
            - task: alef:build

    test:
        desc: Run Rust core tests only
        cmds:
            - task: rust:test

    test:all:
        desc: Run all test suites across all languages
        cmds:
            - task: test
            - alef test

    test:bindings:
        desc: Run binding test suites only
        cmds:
            - alef test

    test:ci:
        desc: Run test suites in CI mode
        cmds:
            - task: test:all

    lint:
        desc: Run all linters and formatters via prek (pre-commit framework)
        cmds:
            - prek run --all-files

    lint:check:
        desc: Run check-only linting
        cmds:
            - task: rust:lint:check

    lint:licenses:
        desc: Verify all grammar repositories use permissive licenses (no GPL/AGPL/copyleft)
        cmds:
            - uv run --no-sync python scripts/lint_grammar_licenses.py --update-cache

    check:
        desc: Run check-only validation
        cmds:
            - task: format:check
            - task: lint:check

    format:
        desc: Format Rust core code
        cmds:
            - task: rust:format

    format:check:
        desc: Check Rust core formatting
        cmds:
            - task: rust:format:check

    clean:
        desc: Clean all build artifacts
        cmds:
            - task: rust:clean
            - alef clean

    update:
        desc: Update all dependencies within current major versions (non-breaking)
        cmds:
            - echo "Starting within-major update (non-breaking minor/patch bumps)"
            - echo "Updating Rust dependencies..."
            - task: rust:update
            - echo "Updating Python..."
            - task: python:update
            - echo "Updating Node/TypeScript..."
            - task: node:update
            - echo "Updating WASM..."
            - task: wasm:update
            - echo "Updating Ruby..."
            - task: ruby:update
            - echo "Updating Go..."
            - task: go:update
            - echo "Updating Java..."
            - task: java:update
            - echo "Updating C#..."
            - task: csharp:update
            - echo "Updating PHP..."
            - task: php:update
            - echo "Updating Elixir..."
            - task: elixir:update
            - echo "Updating Dart..."
            - task: dart:update
            - echo "Updating Swift..."
            - task: swift:update
            - echo "Updating Zig..."
            - task: zig:update
            - echo "Updating Kotlin-Android..."
            - task: kotlin-android:update
            - echo "Updating language grammar revisions..."
            - uv run --no-sync scripts/pin_vendors.py
            - echo "Refreshing pre-commit hooks..."
            - prek autoupdate
            - echo "Updating GitHub Actions pins..."
            - uvx --from gh-actions-updater gau --update --recursive --pin-style major --missing-ref fallback .github
            - echo "Within-major update complete (non-breaking changes only)"

    upgrade:
        desc: Upgrade all dependencies to latest versions (cross-major / breaking allowed)
        cmds:
            - echo "Starting cross-major upgrade (breaking changes allowed)"
            - echo "Upgrading Rust dependencies (--incompatible)..."
            - task: rust:upgrade
            - echo "Upgrading Python..."
            - task: python:upgrade
            - echo "Upgrading Node/TypeScript..."
            - task: node:upgrade
            - echo "Upgrading WASM..."
            - task: wasm:upgrade
            - echo "Upgrading Ruby..."
            - task: ruby:upgrade
            - echo "Upgrading Go..."
            - task: go:upgrade
            - echo "Upgrading Java..."
            - task: java:upgrade
            - echo "Upgrading C#..."
            - task: csharp:upgrade
            - echo "Upgrading PHP..."
            - task: php:upgrade
            - echo "Upgrading Elixir..."
            - task: elixir:upgrade
            - echo "Upgrading Dart..."
            - task: dart:upgrade
            - echo "Upgrading Swift..."
            - task: swift:upgrade
            - echo "Upgrading Zig..."
            - task: zig:upgrade
            - echo "Upgrading Kotlin-Android..."
            - task: kotlin-android:upgrade
            - echo "Updating language grammar revisions..."
            - uv run --no-sync scripts/pin_vendors.py
            - echo "Refreshing pre-commit hooks..."
            - prek autoupdate
            - echo "Updating GitHub Actions pins..."
            - uvx --from gh-actions-updater gau --update --recursive --pin-style major --missing-ref fallback .github
            - echo "Cross-major upgrade complete (breaking bumps applied)"

    update:grammars:
        desc: Update language grammar revisions
        cmds:
            - uv run --no-sync scripts/pin_vendors.py

    update:grammars:missing:
        desc: Pin only languages missing revisions
        cmds:
            - uv run --no-sync scripts/pin_vendors.py --only-missing

    update:grammars:check:
        desc: Check for grammar updates against upstream (reports stale revisions without writing)
        cmds:
            - uv run --no-sync scripts/check_grammar_updates.py

    # Publish tasks (via alef)
    publish:validate:
        desc: "Validate publish manifests are consistent"
        cmds:
            - alef publish validate

    publish:prepare:
        desc: "Prepare packages for publishing (vendor deps, stage FFI)"
        cmds:
            - alef publish prepare

    publish:build:
        desc: "Build release artifacts for all languages"
        cmds:
            - alef publish build

    publish:package:
        desc: "Package built artifacts into distributable archives"
        cmds:
            - alef publish package -o dist/

    # Alef lifecycle
    alef:generate:
        desc: Regenerate all Alef-managed outputs without formatting or building
        cmds:
            - alef all --clean --format=false

    alef:verify:
        desc: Check Alef-generated files are up to date
        cmds:
            - alef verify --exit-code

    alef:build:
        desc: Build all language bindings
        cmds:
            - alef build

    alef:format:
        desc: Format Alef-managed language bindings
        cmds:
            - alef fmt

    alef:sync:
        desc: Sync version from Cargo.toml to all manifests
        cmds:
            - alef sync-versions

    alef:docs:
        desc: Generate API reference documentation
        cmds:
            - alef docs

    # E2E aggregator tasks (canonical kreuzberg-dev surface).
    e2e:generate:
        desc: Generate E2E tests for all languages from fixtures
        cmds:
            - task: _e2e_wf:generate

    e2e:build:
        desc: Build all language bindings for E2E tests
        cmds:
            - task: alef:build

    e2e:test:
        desc: Run e2e tests across all bindings (FFI built first)
        cmds:
            - task: _e2e_wf:test

    e2e:all:
        desc: Build all bindings and run all e2e tests
        cmds:
            - task: e2e:build
            - task: e2e:test

    e2e:quick:
        desc: Quick e2e (Rust only — fastest feedback)
        cmds:
            - task: _e2e_wf:quick

    e2e:verify:
        desc: Full e2e pipeline — generate, lint, test all suites
        cmds:
            - task: _e2e_wf:verify

    e2e:lint:
        desc: Lint all generated E2E test code
        cmds:
            - task: _e2e_wf:lint

    # Documentation generation
    docs:generate:languages:
        desc: Generate supported languages documentation from language_definitions.json
        cmds:
            - uv run --no-sync scripts/generate_grammar_table.py
        sources:
            - sources/language_definitions.json
        generates:
            - docs/languages.md

    docs:generate:api:
        desc: Generate API reference docs (via alef)
        cmds:
            - alef docs

    # README generation (via alef)
    readme:generate:
        desc: Generate all package READMEs
        cmds:
            - alef readme