Stirling-PDF/.taskfiles/pre-commit.yml

version: '3'

# Repo-wide lint/format/secret checks - the single source of truth that the git
# pre-commit hook (.pre-commit-config.yaml) and CI (pre_commit.yml) both call.

vars:
  # File selections as git pathspecs: git does the include/exclude matching, so
  # there is no grep/xargs and it behaves identically on every platform.
  PY_FILES: >-
    'scripts/**/*.py'
    '.github/scripts/*.py'
    'app/core/src/main/resources/static/python/*.py'
    'testing/**/*.py'
    'engine/**/*.py'
    ':(exclude)engine/**/tool_io.py'
    ':(exclude)engine/**/tool_models.py'
  SPELL_FILES: >-
    '*.html'
    '*.css'
    '*.js'
    '*.py'
    '*.md'
    ':(exclude).vscode/*'
    ':(exclude).devcontainer/*'
    ':(exclude)app/core/src/main/resources/*'
    ':(exclude)app/proprietary/src/main/resources/*'
    ':(exclude)frontend/editor/public/vendor/*'
    ':(exclude)*Dockerfile*'
    ':(exclude)*pdfjs*'
    ':(exclude)*thirdParty*'
    ':(exclude)*bootstrap*'
    ':(exclude)*.min.*'
    ':(exclude)*diff.js'
  WS_FILES: >-
    '*.js'
    '*.java'
    '*.py'
    '*.yml'
    ':(exclude)*pdfjs*'
    ':(exclude)*thirdParty*'
    ':(exclude)*bootstrap*'
    ':(exclude)*.min.*'
    ':(exclude)*diff.js'
    ':(exclude).github/workflows/*'
  LOCALE_TOML: 'frontend/editor/public/locales/*/translation.toml'

  # gitleaks is pinned + checksum-verified by scripts/pre-commit/install_gitleaks.py,
  # which owns the version and caches the binary here.
  GITLEAKS_BIN: '.task/bin/gitleaks{{if eq OS "windows"}}.exe{{end}}'

env:
  # Keep repository-wide checks isolated from the engine runtime environment.
  UV_PROJECT_ENVIRONMENT: '.venv-pre-commit'

tasks:
  default:
    desc: "Check formatting, spelling, and secrets across the repo"
    cmds:
      - task: ruff
      - task: ruff-format
      - task: codespell
      - task: gitleaks
      - task: whitespace
      - task: toml-sort

  fix:
    desc: "Auto-fix formatting, spelling, and secrets issues across the repo"
    cmds:
      # Auto-fixers first, then the report-only tools (codespell, gitleaks) so a
      # finding there does not stop the fixers from running.
      - task: ruff
        vars: { FIX: '1' }
      - task: ruff-format
        vars: { FIX: '1' }
      - task: whitespace
        vars: { FIX: '1' }
      - task: toml-sort
        vars: { FIX: '1' }
      - task: codespell
      - task: gitleaks

  install:
    desc: "Install the pinned pre-commit Python tools"
    run: once
    cmds:
      - uv sync --locked --project engine --group pre-commit
    sources:
      - engine/uv.lock
      - engine/pyproject.toml
    status:
      - test -d engine/.venv-pre-commit

  clean:
    desc: "Remove the cached gitleaks binary and the pre-commit virtualenv"
    cmds:
      - cmd: rm -rf engine/.venv-pre-commit .task/bin/gitleaks
        platforms: [linux, darwin]
      - cmd: cmd /c "rmdir /s /q engine\.venv-pre-commit & del /q .task\bin\gitleaks.exe"
        platforms: [windows]
        ignore_error: true

  # Individual checks (hidden from `task --list`, but callable, e.g.
  # `task pre-commit:toml-sort FIX=1`). Pass FIX=1 to auto-fix where supported.
  ruff:
    deps: [install]
    cmds:
      - uv run --locked --project engine --group pre-commit ruff check $(git ls-files {{.PY_FILES}}) {{if .FIX}}--fix{{end}} --config engine/pyproject.toml

  ruff-format:
    deps: [install]
    cmds:
      - uv run --locked --project engine --group pre-commit ruff format $(git ls-files {{.PY_FILES}}) {{if .FIX}}{{else}}--check{{end}} --config engine/pyproject.toml

  ruff-format-diff:
    deps: [install]
    cmds:
      - uv run --locked --project engine --group pre-commit ruff format $(git ls-files {{.PY_FILES}}) --diff --config engine/pyproject.toml

  codespell:
    deps: [install]
    cmds:
      - uv run --locked --project engine --group pre-commit codespell --ignore-words-list=thirdParty,tabEl,tabEls,Sie,ist,fulfilment --quiet-level=2 $(git ls-files {{.SPELL_FILES}})

  toml-sort:
    deps: [install]
    cmds:
      - uv run --locked --project engine --group pre-commit python scripts/pre-commit/sort_locale_toml.py {{if .FIX}}--fix {{end}}{{.LOCALE_TOML}}

  whitespace:
    cmds:
      - uv run --locked --project engine --group pre-commit python scripts/pre-commit/whitespace.py {{if .FIX}}--fix {{end}}{{.WS_FILES}}

  gitleaks:
    deps: [gitleaks-bin]
    # Scan staged changes only, matching the old hook: the git-mode fingerprints
    # in .gitleaksignore (file:rule:line) still apply, and with nothing staged
    # this is a no-op. Secrets are never auto-fixed, so FIX has no effect.
    cmds:
      - "{{.GITLEAKS_BIN}} git --pre-commit --redact --staged --verbose"

  gitleaks-bin:
    internal: true
    desc: "Ensure the pinned, checksum-verified gitleaks binary is cached in .task/bin"
    cmds:
      - uv run --locked --project engine --group pre-commit python scripts/pre-commit/install_gitleaks.py