.pre-commit-config.yaml

# Book Pre-commit Configuration
# =============================================================================
# Quarto content validation hooks for the ML Systems textbook
#
# This config is called by the root .pre-commit-config.yaml router
# Run manually: cd book && pre-commit run --config .pre-commit-config.yaml --all-files
# =============================================================================

repos:
  # ===========================================================================
  # AUTO-FORMATTERS
  # ===========================================================================

  - repo: https://github.com/executablebooks/mdformat
    rev: 0.7.9
    hooks:
      - id: mdformat
        name: "Format quarto markdown"
        additional_dependencies: [mdformat-frontmatter]
        files: ^quarto/contents/.*\.qmd$
        pass_filenames: true

  - repo: https://github.com/FlamingTempura/bibtex-tidy
    rev: v1.14.0
    hooks:
      - id: bibtex-tidy
        name: "Tidy bibtex files"
        args: [
          '--align=space',
          '--curly',
          '--sort=key',
          '--sort-fields',
          '--duplicates=key',
          '--remove-empty-fields',
          '--space=2',
          '--trailing-commas',
          '--escape',
          '--wrap=100',
          '--blank-lines'
        ]
        files: ^quarto/contents/.*\.bib$

  # ===========================================================================
  # CONTENT VALIDATORS
  # ===========================================================================

  - repo: local
    hooks:
      - id: collapse-blank-lines
        name: "Collapse extra blank lines"
        entry: python tools/scripts/content/format_blank_lines.py
        language: python
        pass_filenames: true
        files: ^quarto/contents/.*\.qmd$

      - id: format-python
        name: "Format Python code blocks (Black, 70 chars)"
        entry: python tools/scripts/content/format_python_in_qmd.py
        language: python
        additional_dependencies: [black>=23.0.0]
        pass_filenames: true
        files: ^quarto/contents/.*\.qmd$

      - id: check-list-formatting
        name: "Fix markdown list formatting"
        entry: python tools/scripts/utilities/check_list_formatting.py --fix
        language: python
        pass_filenames: true
        files: ^quarto/contents/.*\.qmd$

      - id: validate-json
        name: "Validate JSON files"
        entry: python tools/scripts/utilities/validate_json.py
        language: python
        files: \.json$
        pass_filenames: true

      - id: check-unreferenced-labels
        name: "Check for unreferenced labels"
        entry: python tools/scripts/content/check_unreferenced_labels.py ./quarto/contents/core
        language: python
        pass_filenames: false
        files: ^quarto/contents/.*\.qmd$

      - id: check-duplicate-labels
        name: "Check for duplicate labels"
        entry: python tools/scripts/content/check_duplicate_labels.py
        args: ['-d', 'quarto/contents/', '--figures', '--tables', '--listings', '--quiet', '--strict']
        language: python
        pass_filenames: false
        files: ^quarto/contents/.*\.qmd$

      - id: validate-citations
        name: "Validate citation references"
        entry: python tools/scripts/content/validate_citations.py --quiet
        language: python
        pass_filenames: true
        files: ^quarto/contents/.*\.qmd$

      - id: validate-footnotes
        name: "Validate footnote references"
        entry: python tools/scripts/content/footnote_cleanup.py -d quarto/contents/ --validate
        language: python
        pass_filenames: false
        files: ^quarto/contents/.*\.qmd$

      - id: check-forbidden-footnotes
        name: "Check for footnotes in tables/captions"
        entry: python tools/scripts/content/check_forbidden_footnotes.py -d quarto/contents/
        language: python
        pass_filenames: false
        files: ^quarto/contents/.*\.qmd$

      - id: check-purpose-unnumbered
        name: "Ensure Purpose sections are unnumbered"
        entry: bash -c 'grep -n "^## Purpose" "$@" | grep -v "\.unnumbered" && exit 1 || exit 0' --
        language: system
        pass_filenames: true
        files: ^quarto/contents/.*\.qmd$

      - id: check-table-formatting
        name: "Check table formatting"
        entry: python tools/scripts/content/format_tables.py --check -d quarto/contents/
        language: python
        pass_filenames: false
        files: ^quarto/contents/.*\.qmd$

      - id: validate-part-keys
        name: "Validate part keys"
        entry: python tools/scripts/utilities/validate_part_keys.py
        language: python
        additional_dependencies: [pyyaml]
        pass_filenames: false
        files: \.qmd$

  # ===========================================================================
  # IMAGE VALIDATORS
  # ===========================================================================

  - repo: local
    hooks:
      - id: validate-images
        name: "Validate image files"
        entry: python tools/scripts/images/manage_images.py
        language: python
        additional_dependencies: [pillow, rich]
        pass_filenames: true
        files: ^quarto/contents/.*\.(png|jpg|jpeg|gif)$

      - id: validate-external-images
        name: "Check for external images"
        entry: python tools/scripts/images/manage_external_images.py --validate quarto/contents/
        language: python
        additional_dependencies: [requests]
        pass_filenames: false
        files: ^quarto/contents/.*\.qmd$

      - id: validate-image-references
        name: "Check image references exist"
        entry: python tools/scripts/images/validate_image_references.py -d quarto/contents/ --quiet
        language: python
        pass_filenames: false
        files: ^quarto/contents/.*\.qmd$

      - id: prevent-svg
        name: "Prevent manual SVG files"
        entry: bash -c 'echo "❌ Manual SVG files are not allowed. Convert to PNG."; exit 1'
        language: system
        files: \.svg$
        exclude: ".*_files/mediabag/.*\\.svg$"

  # ===========================================================================
  # SYSTEM CHECKS
  # ===========================================================================

  - repo: local
    hooks:
      - id: cleanup-artifacts
        name: "Auto-cleanup build artifacts"
        entry: python tools/scripts/maintenance/cleanup_build_artifacts.py
        language: python
        additional_dependencies: [rich]
        pass_filenames: false
        stages: [pre-commit]