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:
  GITLEAKS: '8.30.0'

  # 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'
    ':(exclude)*split_photos.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_BIN: '.task/bin/gitleaks-{{.GITLEAKS}}{{if eq OS "windows"}}.exe{{end}}'

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 (ruff, codespell, toml-sort)"
    run: once
    cmds:
      - uv sync --project scripts/pre-commit --locked
    sources:
      - scripts/pre-commit/uv.lock
      - scripts/pre-commit/pyproject.toml
    status:
      - test -d scripts/pre-commit/.venv

  clean:
    desc: "Remove the cache/build artifacts"
    cmds:
      - task: '{{if eq OS "windows"}}clean-windows{{else}}clean-unix{{end}}'

  clean-unix:
    internal: true
    cmds:
      - rm -rf scripts/pre-commit/.venv .task/bin/gitleaks-*

  # On Windows, use PowerShell so it matches the same paths and tolerates absent
  # files without erroring.
  clean-windows:
    internal: true
    ignore_error: true
    cmds:
      - powershell -NoProfile -Command "Remove-Item -Recurse -Force -ErrorAction SilentlyContinue scripts/pre-commit/.venv, .task/bin/gitleaks-*"

  # 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 --project scripts/pre-commit --no-sync ruff check --line-length=127 {{if .FIX}}--fix {{end}}$(git ls-files {{.PY_FILES}})

  ruff-format:
    deps: [install]
    cmds:
      - uv run --project scripts/pre-commit --no-sync ruff format {{if .FIX}}{{else}}--check {{end}}$(git ls-files {{.PY_FILES}})

  codespell:
    deps: [install]
    cmds:
      - uv run --project scripts/pre-commit --no-sync 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 --project scripts/pre-commit --no-sync toml-sort --all --ignore-case {{if .FIX}}--in-place{{else}}--check{{end}} {{.LOCALE_TOML}}

  whitespace:
    cmds:
      - uv run --no-project python scripts/pre-commit/whitespace.py {{if .FIX}}--fix {{end}}$(git ls-files {{.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 gitleaks binary is cached in .task/bin"
    status:
      - test -f {{.GITLEAKS_BIN}}
    vars:
      GL_ARCH: '{{if eq ARCH "amd64"}}x64{{else if eq ARCH "arm64"}}arm64{{else if eq ARCH "386"}}x32{{else}}{{ARCH}}{{end}}'
      GL_PLATFORM: '{{OS}}_{{.GL_ARCH}}'
      GL_URL: 'https://github.com/gitleaks/gitleaks/releases/download/v{{.GITLEAKS}}/gitleaks_{{.GITLEAKS}}_{{.GL_PLATFORM}}'
      # SHA-256 of each release asset, from gitleaks_{{.GITLEAKS}}_checksums.txt.
      GL_SHA: >-
        {{if eq .GL_PLATFORM "linux_x64"}}79a3ab579b53f71efd634f3aaf7e04a0fa0cf206b7ed434638d1547a2470a66e
        {{- else if eq .GL_PLATFORM "linux_arm64"}}b4cbbb6ddf7d1b2a603088cd03a4e3f7ce48ee7fd449b51f7de6ee2906f5fa2f
        {{- else if eq .GL_PLATFORM "darwin_x64"}}ca221d012d247080c2f6f61f4b7a83bffa2453806b0c195c795bbe9a8c775ed5
        {{- else if eq .GL_PLATFORM "darwin_arm64"}}b251ab2bcd4cd8ba9e56ff37698c033ebf38582b477d21ebd86586d927cf87e7
        {{- else if eq .GL_PLATFORM "windows_x64"}}54fe94f644b832dd08e8c3a5915efb3bfa862386d59fb27ca0792cb687a83573
        {{- end}}
    cmds:
      - cmd: bash scripts/pre-commit/install-gitleaks.sh "{{.GL_URL}}.tar.gz" "{{.GL_SHA}}" "{{.GITLEAKS_BIN}}"
        platforms: [linux, darwin]
      - cmd: powershell -NoProfile -File scripts/pre-commit/install-gitleaks.ps1 -Url "{{.GL_URL}}.zip" -Sha "{{.GL_SHA}}" -Dest "{{.GITLEAKS_BIN}}"
        platforms: [windows]