uniword 1.2.5 → 1.4.0

data/lib/uniword/wordprocessingml/settings.rb

@@ -2,34 +2,6 @@
 
 require "lutaml/model"
 require "nokogiri"
-require_relative "../properties/relationship_id"
-require_relative "zoom"
-require_relative "compat"
-require_relative "proof_state"
-require_relative "style_pane_format_filter"
-require_relative "default_tab_stop"
-require_relative "character_spacing_control"
-require_relative "do_not_display_page_boundaries"
-require_relative "rsids"
-require_relative "math_pr"
-require_relative "theme_font_lang"
-require_relative "clr_scheme_mapping"
-require_relative "shape_defaults"
-require_relative "decimal_symbol"
-require_relative "list_separator"
-require_relative "attached_template"
-require_relative "footnote_pr"
-require_relative "endnote_pr"
-require_relative "hdr_shape_defaults"
-require_relative "even_and_odd_headers"
-require_relative "mirror_margins"
-require_relative "do_not_include_subdocs_in_stats"
-require_relative "hyphenation_zone"
-require_relative "style_pane_sort_method"
-require_relative "doc_vars"
-require_relative "w14_doc_id"
-require_relative "w15_chart_tracking_ref_based"
-require_relative "w15_doc_id"
 
 module Uniword
   module Wordprocessingml
@@ -57,6 +29,7 @@ module Uniword
       attribute :footnote_pr, FootnotePr
       attribute :endnote_pr, EndnotePr
       attribute :hdr_shape_defaults, HdrShapeDefaults
+      attribute :update_fields, UpdateFields
       attribute :even_and_odd_headers, EvenAndOddHeaders
       attribute :mirror_margins, MirrorMargins
       attribute :do_not_include_subdocs_in_stats,
@@ -112,9 +85,9 @@ module Uniword
         map_element "doNotDisplayPageBoundaries",
                     to: :do_not_display_page_boundaries, render_nil: false
         map_element "mirrorMargins", to: :mirror_margins, render_nil: false
+        map_element "proofState", to: :proof_state, render_nil: false
         map_element "attachedTemplate", to: :attached_template,
                                         render_nil: false
-        map_element "proofState", to: :proof_state, render_nil: false
         map_element "stylePaneFormatFilter", to: :style_pane_format_filter,
                                              render_nil: false
         map_element "stylePaneSortMethod", to: :style_pane_sort_method,
@@ -126,6 +99,9 @@ module Uniword
                                          render_nil: false
         map_element "characterSpacingControl", to: :character_spacing_control,
                                                render_nil: false
+        map_element "updateFields", to: :update_fields, render_nil: false
+        map_element "hdrShapeDefaults", to: :hdr_shape_defaults,
+                                        render_nil: false
         map_element "footnotePr", to: :footnote_pr, render_nil: false
         map_element "endnotePr", to: :endnote_pr, render_nil: false
         map_element "compat", to: :compat, render_nil: false
@@ -137,11 +113,6 @@ module Uniword
                                         render_nil: false
         map_element "doNotIncludeSubdocsInStats",
                     to: :do_not_include_subdocs_in_stats, render_nil: false
-        map_element "hdrShapeDefaults", to: :hdr_shape_defaults,
-                                        render_nil: false
-        map_element "shapeDefaults", to: :shape_defaults, render_nil: false
-        map_element "decimalSymbol", to: :decimal_symbol, render_nil: false
-        map_element "listSeparator", to: :list_separator, render_nil: false
         # Both w14:docId and w15:docId use the same element name 'docId'
         # Separate map_element entries needed for namespace-aware matching
         # The namespace_scope ensures w14 and w15 namespaces are declared on root
@@ -150,6 +121,9 @@ module Uniword
         map_element "docId", to: :w14_doc_id, render_nil: false
         map_element "docId", to: :w15_doc_id, render_nil: false
         map_element "schemaLibrary", to: :schema_library, render_nil: false
+        map_element "shapeDefaults", to: :shape_defaults, render_nil: false
+        map_element "decimalSymbol", to: :decimal_symbol, render_nil: false
+        map_element "listSeparator", to: :list_separator, render_nil: false
       end
 
       # Override from_xml to manually deserialize w15:docId which has the same

data/lib/uniword/wordprocessingml/shape_defaults.rb

@@ -1,7 +1,6 @@
 # frozen_string_literal: true
 
 require "lutaml/model"
-require_relative "../vml_office"
 
 module Uniword
   module Wordprocessingml

data/lib/uniword/wordprocessingml/style.rb

@@ -49,7 +49,8 @@ module Uniword
         map "run_properties", to: :rPr
       end
 
-      # YAML transform methods (instance methods - called via send on an instance)
+      # YAML transform methods (instance methods called by lutaml-model's
+      # `with:` transform mechanism)
       def yaml_name_from(instance, value)
         instance.name = StyleName.new(val: value) if value
       end

data/lib/uniword/wordprocessingml/style_cleanup.rb

@@ -0,0 +1,198 @@
+# frozen_string_literal: true
+
+module Uniword
+  module Wordprocessingml
+    # Removes styles from a document's style definitions — Word's
+    # Styles-pane management as an API.
+    #
+    # `#remove_unused` deletes styles that no content references,
+    # directly (pStyle/rStyle/tblStyle in body, headers, footers, notes,
+    # comments) or transitively (basedOn/link/next chains from a used
+    # style, numbering level pStyle). Default styles (w:default="1")
+    # are never removed.
+    #
+    # @example Declutter a document
+    #   cleanup = StyleCleanup.new(document)
+    #   removed = cleanup.remove_unused # => ["ObsoleteStyle", ...]
+    class StyleCleanup
+      # @param document [DocumentRoot] Document to clean (mutated)
+      def initialize(document)
+        @document = document
+      end
+
+      # Remove one style by id (unless protected).
+      #
+      # @param style_id [String] Style id (w:styleId)
+      # @return [Boolean] true when the style was removed
+      def remove?(style_id)
+        return false if protected_id?(style_id)
+
+        before = styles.size
+        styles.delete_if { |style| style.styleId == style_id }
+        styles.size < before
+      end
+
+      # Ids of styles no content references.
+      #
+      # @return [Array<String>] Unreferenced, unprotected style ids
+      def unused_ids
+        used = used_ids
+        styles.filter_map(&:styleId).uniq - used - protected_ids
+      end
+
+      # Remove every unreferenced style.
+      #
+      # @return [Array<String>] Ids of removed styles
+      def remove_unused
+        ids = unused_ids
+        ids.each { |id| remove?(id) }
+        ids
+      end
+
+      private
+
+      def styles
+        @document.styles_configuration.styles
+      end
+
+      def protected_ids
+        styles.select(&:default).map(&:styleId)
+      end
+
+      def protected_id?(style_id)
+        protected_ids.include?(style_id)
+      end
+
+      # Ids referenced anywhere: content, notes, headers/footers,
+      # comments, numbering levels, and style-to-style chains.
+      def used_ids
+        used = content_style_ids + chain_seed_ids
+        expand_chains(used)
+      end
+
+      def content_style_ids
+        ids = []
+        walk_all_paragraphs do |paragraph|
+          ids.concat(paragraph_style_ids(paragraph))
+          ids.concat(run_style_ids(paragraph))
+        end
+        ids.concat(table_style_ids)
+        ids.concat(numbering_style_ids)
+        ids
+      end
+
+      # Styles referenced by other styles (basedOn/link/next) — chains
+      # are expanded from these too.
+      def chain_seed_ids
+        styles.flat_map { |style| chain_links(style) }
+      end
+
+      def chain_links(style)
+        [style.basedOn&.val, style.link&.val, style.nextStyle&.val].compact
+      end
+
+      def expand_chains(ids)
+        expanded = ids.to_set
+        loop do
+          additions = chain_additions(expanded)
+          break if additions.empty?
+
+          additions.each { |id| expanded << id }
+        end
+        expanded.to_a
+      end
+
+      def chain_additions(expanded)
+        styles.select { |style| expanded.include?(style.styleId) }
+          .flat_map { |style| chain_links(style) }
+          .reject { |id| expanded.include?(id) }
+      end
+
+      def paragraph_style_ids(paragraph)
+        Array(paragraph.properties&.style).filter_map(&:value)
+      end
+
+      def run_style_ids(paragraph)
+        paragraph_runs(paragraph)
+          .filter_map { |run| run.properties&.style&.value }
+      end
+
+      def paragraph_runs(paragraph)
+        (paragraph.runs || []) +
+          (paragraph.hyperlinks || []).flat_map(&:runs)
+      end
+
+      def table_style_ids
+        ids = []
+        walk_tables(@document.body) do |table|
+          ids << table.properties&.style&.val
+        end
+        ids.compact
+      end
+
+      def numbering_style_ids
+        numbering_definitions.flat_map do |definition|
+          (definition.levels || []).filter_map { |level| level.pStyle&.val }
+        end
+      end
+
+      def numbering_definitions
+        @document.numbering_configuration&.definitions || []
+      end
+
+      # Paragraphs in body (incl. table cells), headers/footers,
+      # notes, comments.
+      def walk_all_paragraphs(&block)
+        each_container { |container| walk_paragraphs(container, &block) }
+      end
+
+      def walk_paragraphs(container, &block)
+        return unless container
+
+        (container.paragraphs || []).each(&block)
+        walk_tables(container) { |table| walk_table_rows(table, &block) }
+      end
+
+      def walk_table_rows(table, &block)
+        (table.rows || []).each do |row|
+          (row.cells || []).each { |cell| walk_paragraphs(cell, &block) }
+        end
+      end
+
+      def walk_tables(container, &block)
+        return unless container
+
+        (container.tables || []).each do |table|
+          yield table
+          walk_table_cells(table, &block)
+        end
+      end
+
+      def walk_table_cells(table, &block)
+        (table.rows || []).each do |row|
+          (row.cells || []).each { |cell| walk_tables(cell, &block) }
+        end
+      end
+
+      def each_container(&block)
+        yield @document.body if @document.body
+        header_footer_containers.each(&block)
+        note_containers.each(&block)
+      end
+
+      def header_footer_containers
+        (@document.header_footer_parts || []).filter_map do |part|
+          part.content if part.content.is_a?(Lutaml::Model::Serializable)
+        end
+      end
+
+      def note_containers
+        [
+          @document.footnotes&.footnote_entries,
+          @document.endnotes&.endnote_entries,
+          @document.comments&.comments,
+        ].flatten.compact
+      end
+    end
+  end
+end

data/lib/uniword/wordprocessingml/table_cell_properties.rb

@@ -28,18 +28,18 @@ module Uniword
         namespace Uniword::Ooxml::Namespaces::WordProcessingML
         mixed_content
 
+        map_element "cnfStyle", to: :cnf_style, render_nil: false
         map_element "tcW", to: :cell_width, render_nil: false
-        map_element "tcBorders", to: :borders, render_nil: false
-        map_element "shd", to: :shading, render_nil: false
-        map_element "vAlign", to: :vertical_align, render_nil: false
         map_element "gridSpan", to: :grid_span, render_nil: false
         map_element "vMerge", to: :v_merge, render_nil: false
-        map_element "tcMar", to: :tc_mar, render_nil: false
-        map_element "cnfStyle", to: :cnf_style, render_nil: false
+        map_element "tcBorders", to: :borders, render_nil: false
+        map_element "shd", to: :shading, render_nil: false
         map_element "noWrap", to: :no_wrap, render_nil: false
-        map_element "hideMark", to: :hide_mark, render_nil: false
+        map_element "tcMar", to: :tc_mar, render_nil: false
         map_element "textDirection", to: :text_direction,
                                      render_nil: false
+        map_element "vAlign", to: :vertical_align, render_nil: false
+        map_element "hideMark", to: :hide_mark, render_nil: false
       end
     end
   end

data/lib/uniword/wordprocessingml/update_fields.rb

@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+
+require "lutaml/model"
+
+module Uniword
+  module Wordprocessingml
+    # Update fields when the document is opened
+    #
+    # Element: <w:updateFields>
+    #
+    # When true, Word updates all fields (TOC, page numbers, cross
+    # references) the first time the document is opened — generated
+    # documents show correct field values without a manual F9.
+    class UpdateFields < Lutaml::Model::Serializable
+      attribute :value, Ooxml::Types::OoxmlBoolean,
+                default: -> { true }
+
+      xml do
+        element "updateFields"
+        namespace Uniword::Ooxml::Namespaces::WordProcessingML
+
+        map_attribute "val", to: :value, render_nil: false,
+                                        render_default: false
+      end
+    end
+  end
+end

data/lib/uniword/wordprocessingml/w14_attributes.rb

@@ -16,15 +16,43 @@ module Uniword
     # W16CID namespace for Word 2016+ citation identifiers
     W16CID_NAMESPACE = Uniword::Ooxml::Namespaces::Word2016Cid
 
-    # Typed attribute for w14:paraId
+    # ST_LongHexNumber (ECMA-376): exactly 8 hexadecimal digits
+    LONG_HEX_NUMBER_PATTERN = /\A[0-9A-Fa-f]{8}\z/
+
+    # Validate an ST_LongHexNumber value at cast time
+    #
+    # @param value [Object] the raw value
+    # @return [String, nil, Object] the validated value
+    # @raise [Lutaml::Model::Type::InvalidValueError] when the value is
+    #   not 8 hexadecimal digits
+    def self.cast_long_hex_number(value)
+      casted = Lutaml::Model::Type::String.cast(value)
+      return casted if casted.nil? ||
+        Lutaml::Model::Utils.uninitialized?(casted)
+      return casted if LONG_HEX_NUMBER_PATTERN.match?(casted)
+
+      raise Lutaml::Model::Type::InvalidValueError.new(
+        value, ["ST_LongHexNumber: 8 hexadecimal digits"]
+      )
+    end
+
+    # Typed attribute for w14:paraId (ST_LongHexNumber)
     class W14ParaId < Lutaml::Model::Type::String
+      def self.cast(value, _options = {})
+        Wordprocessingml.cast_long_hex_number(value)
+      end
+
       xml do
         namespace W14_NAMESPACE
       end
     end
 
-    # Typed attribute for w14:textId
+    # Typed attribute for w14:textId (ST_LongHexNumber)
     class W14TextId < Lutaml::Model::Type::String
+      def self.cast(value, _options = {})
+        Wordprocessingml.cast_long_hex_number(value)
+      end
+
       xml do
         namespace W14_NAMESPACE
       end

data/lib/uniword/wordprocessingml.rb

@@ -4,6 +4,7 @@ module Uniword
   module Wordprocessingml
     # Core document structure
     autoload :DocumentRoot, "uniword/wordprocessingml/document_root"
+    autoload :DocumentStyling, "uniword/wordprocessingml/document_styling"
     autoload :Body, "uniword/wordprocessingml/body"
     autoload :Paragraph, "uniword/wordprocessingml/paragraph"
     autoload :Run, "uniword/wordprocessingml/run"
@@ -64,6 +65,7 @@ module Uniword
     # Structure and metadata
     autoload :Level, "uniword/wordprocessingml/level"
     autoload :Style, "uniword/wordprocessingml/style"
+    autoload :StyleCleanup, "uniword/wordprocessingml/style_cleanup"
     autoload :StructuredDocumentTag,
              "uniword/wordprocessingml/structured_document_tag"
     autoload :StructuredDocumentTagProperties,
@@ -72,6 +74,7 @@ module Uniword
     autoload :PageSize, "uniword/wordprocessingml/page_size"
     autoload :PageMargins, "uniword/wordprocessingml/page_margins"
     autoload :PageNumbering, "uniword/wordprocessingml/page_numbering"
+    autoload :PageSetup, "uniword/wordprocessingml/page_setup"
     autoload :Columns, "uniword/wordprocessingml/columns"
     autoload :HeaderReference, "uniword/wordprocessingml/header_reference"
     autoload :FooterReference, "uniword/wordprocessingml/footer_reference"
@@ -130,6 +133,7 @@ module Uniword
 
     # Fonts
     autoload :Font, "uniword/wordprocessingml/font"
+    autoload :FontReplacer, "uniword/wordprocessingml/font_replacer"
     autoload :Fonts, "uniword/wordprocessingml/fonts"
     autoload :FontTable, "uniword/wordprocessingml/font_table"
 
@@ -162,9 +166,7 @@ module Uniword
     autoload :NumFmt, "uniword/wordprocessingml/num_fmt"
     autoload :LvlText, "uniword/wordprocessingml/lvl_text"
     autoload :LvlJc, "uniword/wordprocessingml/lvl_jc"
-    autoload :Ind, "uniword/wordprocessingml/numbering_elements"
     autoload :RFonts, "uniword/wordprocessingml/numbering_elements"
-    autoload :Tabs, "uniword/wordprocessingml/level"
     autoload :PStyle, "uniword/wordprocessingml/level"
     autoload :NumberingConfiguration,
              "uniword/wordprocessingml/numbering_configuration"
@@ -191,6 +193,7 @@ module Uniword
     autoload :Next, "uniword/wordprocessingml/next"
     autoload :Link, "uniword/wordprocessingml/link"
     autoload :UiPriority, "uniword/wordprocessingml/ui_priority"
+    autoload :UpdateFields, "uniword/wordprocessingml/update_fields"
     autoload :LatentStyles, "uniword/wordprocessingml/latent_styles"
     autoload :ParagraphStyle, "uniword/wordprocessingml/paragraph_style"
     autoload :CharacterStyle, "uniword/wordprocessingml/character_style"

data/lib/uniword.rb

@@ -101,7 +101,6 @@ module Uniword
   autoload :Template, "uniword/template"
   autoload :TemplateManager, "uniword/template_manager"
   autoload :Visitor, "uniword/visitor"
-  autoload :Validators, "uniword/validators"
   autoload :Stylesets, "uniword/stylesets"
   autoload :Infrastructure, "uniword/infrastructure"
   autoload :Accessibility, "uniword/accessibility"
@@ -114,7 +113,6 @@ module Uniword
   autoload :Configuration, "uniword/configuration"
   autoload :Transformation, "uniword/transformation"
   autoload :Validation, "uniword/validation"
-  autoload :Warnings, "uniword/warnings"
   autoload :Mhtml, "uniword/mhtml"
   autoload :Themes, "uniword/themes"
   autoload :Resource, "uniword/resource"
@@ -135,6 +133,9 @@ module Uniword
   # CLI
   autoload :CLI, "uniword/cli/main"
   autoload :GenerateCLI, "uniword/cli/generate_cli"
+  autoload :FontsCLI, "uniword/cli/fonts_cli"
+  autoload :PageCLI, "uniword/cli/page_cli"
+  autoload :StylesCLI, "uniword/cli/styles_cli"
   autoload :DiffCLI, "uniword/cli/diff_cli"
   autoload :TemplateCLI, "uniword/cli/template_cli"
   autoload :CLIHelpers, "uniword/cli/helpers"
@@ -250,6 +251,30 @@ module Uniword
       Wordprocessingml::DocumentRoot.new
     end
 
+    # Get the global runtime configuration (memoized)
+    #
+    # @return [Configuration] The global configuration instance
+    #
+    # @example
+    #   Uniword.configuration.validate_on_save # => true
+    def configuration
+      @configuration ||= Configuration.new
+    end
+
+    # Configure global runtime policy
+    #
+    # @yieldparam config [Configuration] The global configuration instance
+    # @return [Configuration] The global configuration instance
+    #
+    # @example
+    #   Uniword.configure do |config|
+    #     config.xsd_validation = true
+    #   end
+    def configure
+      yield(configuration)
+      configuration
+    end
+
     # Load document from file
     #
     # @param path [String] File path