Taskfile.yaml

version: '3'

silent: true

includes:
  nvidia: Taskfile.nvidia.yaml
  privacy-filter: Taskfile.privacy-filter.yaml
  gliner2: Taskfile.gliner2.yaml
  cli: Taskfile.cli.yaml

dotenv:
  - .env

vars:
  PYTHON:
    sh: "command -v python3.13 || command -v python3.12 || command -v python3"
  VENV_DIR: .venv
  REQUIREMENTS: requirements.txt
  RUFF:
    sh: "if command -v ruff >/dev/null 2>&1; then command -v ruff; elif test -x .venv/bin/ruff; then printf '%s\n' .venv/bin/ruff; else printf '%s\n' ruff; fi"

tasks:
  # Environment ----------------------------------------------------------------

  setup:
    desc: Create venv and install runtime dependencies
    cmds:
      - task: venv
      - task: install
      - |
        printf '%s\n' "Setup complete. Next steps:"
        printf '%s\n' "  task check     - Run lint and tests"
        printf '%s\n' "  task run       - Run benchmark evaluation"
        printf '%s\n' "  task analyze   - Screen an unlabelled file (see README)"
        printf '%s\n' "  task privacy-filter:install / gliner2:install - Install optional engines"

  venv:
    desc: Create the Python virtual environment
    cmds:
      - "uv venv {{.VENV_DIR}} --python {{.PYTHON}}"
    status:
      - test -d {{.VENV_DIR}}
      - test -x {{.VENV_DIR}}/bin/python3 || test -x {{.VENV_DIR}}/bin/python

  install:
    desc: Install runtime dependencies into the venv
    deps: [venv]
    cmds:
      - "uv pip install --upgrade -r {{.REQUIREMENTS}} --python {{.VENV_DIR}}/bin/python"
      - "uv pip install -e . --python {{.VENV_DIR}}/bin/python"
      - "touch {{.VENV_DIR}}/.requirements.stamp"
    status:
      - test -f {{.VENV_DIR}}/.requirements.stamp
      - test {{.VENV_DIR}}/.requirements.stamp -nt {{.REQUIREMENTS}}
      - test {{.VENV_DIR}}/.requirements.stamp -nt pyproject.toml

  dev-install:
    desc: Install development dependencies (pytest)
    deps: [install]
    cmds:
      - "uv pip install -e '.[dev]' --python {{.VENV_DIR}}/bin/python"

  update:
    desc: Upgrade runtime dependencies to latest versions
    deps: [venv]
    cmds:
      - "uv pip install --upgrade -r {{.REQUIREMENTS}} --python {{.VENV_DIR}}/bin/python"
      - "uv pip install -e . --python {{.VENV_DIR}}/bin/python"
      - "touch {{.VENV_DIR}}/.requirements.stamp"

  clean:
    desc: Remove the virtual environment and build artefacts
    cmds:
      - |
        rm -rf {{.VENV_DIR}}
        find . -type d -name "__pycache__" -exec rm -rf {} + 2>/dev/null || true
        find . -type f -name "*.pyc" -delete 2>/dev/null || true
        rm -rf *.egg-info build dist .cache .pytest_cache .ruff_cache

  cache-clear:
    desc: Clear the PII analysis disk cache
    cmds:
      - "rm -rf .cache/pii-eval"

  # Quality --------------------------------------------------------------------

  check:
    desc: Run lint and tests
    deps: [install]
    cmds:
      - task: lint
      - task: test

  lint:
    desc: Run ruff lint and format check
    deps: [install]
    cmds:
      - |
        printf '%s\n' "Running ruff linter..."
        {{.RUFF}} check evaluation/ examples/ tests/
        printf '%s\n' "Running ruff format check..."
        {{.RUFF}} format --check evaluation/ examples/ tests/

  format:
    desc: Format code with ruff
    deps: [install]
    cmds:
      - |
        printf '%s\n' "Formatting with ruff..."
        {{.RUFF}} format evaluation/ examples/ tests/
        {{.RUFF}} check --fix evaluation/ examples/ tests/

  test:
    desc: Run the pytest suite
    deps: [dev-install]
    cmds:
      - "uv run --python {{.VENV_DIR}}/bin/python python -m pytest tests/ -v"

  # Evaluation CLI -------------------------------------------------------------

  run:
    desc: Run benchmark evaluation on labelled JSONL samples
    deps: [install]
    cmds:
      - "uv run --python {{.VENV_DIR}}/bin/python pii-eval run"

  compare:
    desc: Compare engines on labelled JSONL samples
    deps: [install]
    cmds:
      - "uv run --python {{.VENV_DIR}}/bin/python pii-eval compare {{.CLI_ARGS}}"

  sweep:
    desc: Sweep an engine across score thresholds
    deps: [install]
    cmds:
      - "uv run --python {{.VENV_DIR}}/bin/python pii-eval sweep"

  analyze:
    desc: Screen one unlabelled file for PII findings
    deps: [install]
    cmds:
      - "uv run --python {{.VENV_DIR}}/bin/python pii-eval analyze {{.CLI_ARGS}}"

  analyze-compare:
    desc: Compare engines on one unlabelled file
    deps: [install]
    cmds:
      - "uv run --python {{.VENV_DIR}}/bin/python pii-eval analyze-compare {{.CLI_ARGS}}"

  analyze-batch:
    desc: Screen all supported files in a directory
    deps: [install]
    cmds:
      - "uv run --python {{.VENV_DIR}}/bin/python pii-eval analyze-batch {{.CLI_ARGS}}"

  annotations-to-jsonl:
    desc: Convert reviewed annotations into benchmark JSONL
    deps: [install]
    cmds:
      - "uv run --python {{.VENV_DIR}}/bin/python pii-eval annotations-to-jsonl {{.CLI_ARGS}}"

  # Developer tooling ----------------------------------------------------------

  codex-review:
    desc: Run Codex code review on the codebase (if codex is installed)
    deps: [install]
    cmds:
      - |
        if command -v codex >/dev/null 2>&1; then
          printf '%s\n' "Running Codex code review..."
          codex --sandbox read-only exec "Review this Python project for potential issues and improvements."
        else
          printf '%s\n' "Codex CLI not found; skipping codex-review task."
        fi