uniword 1.3.0 → 1.4.1

data/lib/uniword/toc/toc_generator.rb

@@ -91,14 +91,13 @@ module Uniword
         # Add the SDT to element_order so it serializes correctly
         return unless body.element_order
 
-        # element_order from the parser may be frozen; never mutate it.
-        body.element_order = body.element_order.dup
         insert_element = Lutaml::Xml::Element.new("Element", "sdt")
-        if position.positive? && body.element_order.size >= position
-          body.element_order.insert(position, insert_element)
-        else
-          body.element_order.insert(0, insert_element)
-        end
+        index = if position.positive? && body.element_order.size >= position
+                  position
+                else
+                  0
+                end
+        Ooxml::ElementOrder.insert_at(body, index, insert_element)
       end
 
       # Update an existing TOC in the document.

data/lib/uniword/transformation/mhtml_element_renderer.rb

@@ -219,11 +219,7 @@ module Uniword
       def resolve_image_target(embed_id)
         return nil unless @image_parts
 
-        entry = @image_parts.find { |pair| pair[0] == embed_id }
-        return nil unless entry
-
-        image_data = entry[1]
-        image_data[:target]
+        @image_parts[embed_id]&.target
       end
 
       # Apply run formatting (text must already be HTML-escaped)

data/lib/uniword/transformation/mhtml_metadata_builder.rb

@@ -63,8 +63,8 @@ module Uniword
       # Build filelist.xml part
       def build_filelist_part
         image_entries = if @document.image_parts && !@document.image_parts.empty?
-                          @document.image_parts.map do |_r_id, image_data|
-                            %(<o:File HRef="#{image_data[:target]}"/>)
+                          @document.image_parts.map do |_r_id, part|
+                            %(<o:File HRef="#{part.target}"/>)
                           end.join("\n            ")
                         else
                           ""
@@ -91,12 +91,12 @@ module Uniword
       def build_image_parts
         return [] unless @document.image_parts && !@document.image_parts.empty?
 
-        @document.image_parts.map do |_r_id, image_data|
+        @document.image_parts.map do |_r_id, image_part|
           part = Uniword::Mhtml::ImagePart.new
-          part.content_type = image_data[:content_type] || "image/png"
+          part.content_type = image_part.content_type || "image/png"
           part.content_transfer_encoding = "base64"
-          part.raw_content = image_data[:data]
-          part.content_location = "file:///C:/D057922B/#{document_name}.fld/#{image_data[:target]}"
+          part.raw_content = image_part.data
+          part.content_location = "file:///C:/D057922B/#{document_name}.fld/#{image_part.target}"
           part
         end
       end

data/lib/uniword/validation/opc_validator.rb

@@ -221,7 +221,22 @@ module Uniword
       def resolve_target(base_dir, target)
         return target[1..] if target.start_with?("/")
 
-        base_dir.empty? ? target : File.join(base_dir, target)
+        normalize_package_path(
+          base_dir.empty? ? target : File.join(base_dir, target),
+        )
+      end
+
+      # Lexically normalize a package-relative path (resolve "." and
+      # ".." segments). Deliberately avoids File.expand_path, which
+      # prepends a drive letter on Windows.
+      def normalize_package_path(path)
+        segments = []
+        path.split("/").each do |segment|
+          next if segment.empty? || segment == "."
+
+          segment == ".." ? segments.pop : segments << segment
+        end
+        segments.join("/")
       end
     end
   end

data/lib/uniword/validation/validators/xml_schema_validator.rb

@@ -20,6 +20,11 @@ module Uniword
       #   validator = XmlSchemaValidator.new("xml_schema" => { "xsd_validation" => true })
       #   result = validator.validate('/path/to/document.docx')
       class XmlSchemaValidator < LayerValidator
+        # Markup Compatibility namespace — the mc:Ignorable attribute
+        # itself lives here and is stripped during MCE preprocessing.
+        MCE_NAMESPACE_URI =
+          "http://schemas.openxmlformats.org/markup-compatibility/2006"
+
         def layer_name
           "XSD Schema"
         end
@@ -83,6 +88,7 @@ module Uniword
               content = entry.get_input_stream.read
               schema = registry.load_schema(schema_path)
               doc = Nokogiri::XML(content)
+              preprocess_mce(doc)
 
               schema.validate(doc).each do |error|
                 issues << Report::ValidationIssue.new(
@@ -114,6 +120,46 @@ module Uniword
           end
         end
 
+        # Strip MCE-ignorable ATTRIBUTES per the part's own mc:Ignorable
+        # declaration (w14:paraId/textId and the mc:Ignorable attribute
+        # itself), so XSD validation is not drowned in false errors for
+        # content every real document carries.
+        #
+        # Extension ELEMENTS are deliberately kept: removing them changes
+        # libxml2's content-model evaluation and can expose pre-existing
+        # Word-vs-schema quirks (e.g. rsids placed late, tolerated via the
+        # sequence's trailing xsd:any) as false baseline errors.
+        # Operates in place on the parsed document; never touches the
+        # packaged bytes.
+        #
+        # @param doc [Nokogiri::XML::Document] parsed part to clean
+        # @return [void]
+        def preprocess_mce(doc)
+          root = doc.root
+          return unless root
+
+          ignorable = root["Ignorable"] || root["mc:Ignorable"]
+          return if ignorable.nil? || ignorable.empty?
+
+          uris = ignorable.split(/\s+/).filter_map do |prefix|
+            root.namespaces["xmlns:#{prefix}"]
+          end
+          # The mc:Ignorable attribute itself is MCE machinery and is
+          # undeclared on most root elements in the transitional XSDs.
+          strip_mce_attributes(root, uris + [MCE_NAMESPACE_URI])
+        end
+
+        # Remove attributes in ignorable namespaces from every element.
+        def strip_mce_attributes(doc, uris)
+          doc.traverse do |node|
+            next unless node.element?
+
+            node.attribute_nodes.each do |attr|
+              attr.remove if uris.include?(attr.namespace&.href)
+            end
+          end
+        end
+
         def check_unknown_namespaces(registry, content, part_name, issues)
           ns_uris = registry.detect_namespaces(content)
           unknown = registry.unknown_namespaces(ns_uris)

data/lib/uniword/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Uniword
-  VERSION = "1.3.0"
+  VERSION = "1.4.1"
 end

data/lib/uniword/wordprocessingml/document_root.rb

@@ -12,6 +12,7 @@ module Uniword
       autoload :FeatureFacade, "#{__dir__}/document_root/feature_facade"
 
       include FeatureFacade
+      include DocumentStyling
       include Uniword::DocumentInput
 
       attribute :mc_ignorable, Uniword::Ooxml::Types::McIgnorable
@@ -96,9 +97,6 @@ module Uniword
       attr_accessor :theme, :raw_html, :revisions, :comments
       # Footnotes and endnotes (separate XML parts in DOCX package)
       attr_accessor :footnotes, :endnotes
-      # Image parts to embed in the DOCX package
-      # Hash: r_id => { path: String, data: String, content_type: String, target: String }
-      attr_accessor :image_parts
       # Bibliography sources for sources.xml
       attr_accessor :bibliography_sources
       # Round-trip parts (copied from DocxPackage during load)
@@ -154,6 +152,19 @@ module Uniword
         footers.replace(value)
       end
 
+      # Image parts (word/media/* binaries), keyed by rId.
+      #
+      # @return [Docx::PartCollection] rId => Docx::ImagePart
+      def image_parts
+        @image_parts ||= Docx::PartCollection.new(:r_id, Docx::ImagePart)
+      end
+
+      # Bulk-assign image parts (Hash of rId => ImagePart/legacy part
+      # hash; nil clears).
+      def image_parts=(value)
+        image_parts.replace_all(value)
+      end
+
       # Chart parts to embed in the DOCX package, keyed by rId.
       #
       # @return [Docx::PartCollection] rId => ChartPart
@@ -178,6 +189,21 @@ module Uniword
         embeddings.replace_all(value)
       end
 
+      # Raw passthrough parts (unmodelled package parts carried
+      # byte-for-byte), keyed by package path. Mirrored from the
+      # package on load and back on save via the registry-driven
+      # document↔package copies (:raw_parts definition).
+      #
+      # @return [Docx::PartCollection] package path => Docx::RawPart
+      def raw_parts
+        @raw_parts ||= Docx::PartCollection.new(:path, Docx::RawPart)
+      end
+
+      # Bulk-assign raw parts (Hash of path => RawPart/hash; nil clears).
+      def raw_parts=(value)
+        raw_parts.replace_all(value)
+      end
+
       # Get app_properties (lazy initialization)
       #
       # @return [Uniword::Ooxml::AppProperties] The app properties object
@@ -327,211 +353,6 @@ module Uniword
 
       # Apply theme to document
       #
-      # Applies a Uniword theme by name, updating doc defaults and
-      # built-in heading/hyperlink styles to reference the theme.
-      #
-      # @param name [String, Symbol] Theme slug (e.g., 'meridian', 'corporate')
-      # @param options [Hash] optional overrides
-      # @option options [Hash] :colors override specific color keys
-      # @option options [String] :major_font override major font
-      # @option options [String] :minor_font override minor font
-      # @return [self] For method chaining
-      def apply_theme(name, **options)
-        friendly = Themes::Theme.load(name.to_s)
-
-        options[:colors]&.each do |key, value|
-          friendly.color_scheme[key.to_s] = value
-        end
-        friendly.font_scheme.major_font = options[:major_font] if options[:major_font]
-        friendly.font_scheme.minor_font = options[:minor_font] if options[:minor_font]
-
-        word_theme = Themes::ThemeTransformation.new.to_word(friendly)
-        Themes::ThemeApplicator.new.apply(word_theme, self)
-        self
-      end
-
-      # Apply theme from .thmx file
-      #
-      # @param path [String] Path to .thmx file
-      # @param variant [String, Integer, nil] Optional variant
-      # @return [self] For method chaining
-      def apply_theme_file(path, variant: nil)
-        loader = Themes::ThemeLoader.new
-        self.theme = if variant
-                       loader.load_with_variant(path, variant)
-                     else
-                       loader.load(path)
-                     end
-        self
-      end
-
-      # Apply StyleSet to document
-      #
-      # @param name [String, Symbol] StyleSet slug (e.g., 'signature', 'heritage')
-      # @param strategy [Symbol] Application strategy (:keep_existing, :replace, :rename)
-      # @return [self] For method chaining
-      def apply_styleset(name, strategy: :keep_existing)
-        styleset = Uniword::Stylesets::YamlStyleSetLoader.load_bundled(name.to_s)
-        styleset.apply_to(self, strategy: strategy)
-        self
-      end
-
-      # Apply a bundled font scheme to the document's theme
-      #
-      # Mirrors Word's Design → Fonts gallery: replaces only the theme's
-      # fontScheme — colors and formats are untouched. Creates the theme
-      # part from the bundled Office theme when the document has none.
-      #
-      # @param name [String, Symbol] Font scheme slug (data/font_schemes/)
-      # @return [self] For method chaining
-      # @raise [ArgumentError] if the scheme is unknown
-      def apply_font_scheme(name)
-        fonts = Resource::FontSchemeLoader.load(name.to_s)
-        transformation = Themes::ThemeTransformation.new
-        ensure_theme!(transformation)
-        theme.theme_elements.font_scheme =
-          transformation.build_font_scheme(fonts)
-        self
-      end
-
-      # Apply a bundled color scheme to the document's theme
-      #
-      # Mirrors Word's Design → Colors gallery: replaces only the theme's
-      # clrScheme — fonts and formats are untouched. Creates the theme
-      # part from the bundled Office theme when the document has none.
-      #
-      # @param name [String, Symbol] Color scheme slug (data/color_schemes/)
-      # @return [self] For method chaining
-      # @raise [ArgumentError] if the scheme is unknown
-      def apply_color_scheme(name)
-        colors = Resource::ColorSchemeLoader.load(name.to_s)
-        transformation = Themes::ThemeTransformation.new
-        ensure_theme!(transformation)
-        theme.theme_elements.clr_scheme =
-          transformation.build_color_scheme(colors)
-        self
-      end
-
-      # Replace one font family with another across the document
-      #
-      # Mirrors Word's Home → Replace → Replace Fonts: rewrites rFonts
-      # references in styles and defaults, body content, headers,
-      # footers, notes, comments, and numbering. Theme font references
-      # (asciiTheme etc.) are untouched — use #apply_font_scheme for
-      # those.
-      #
-      # @param from [String] Font family to replace (exact match)
-      # @param to [String] Replacement font family
-      # @return [Integer] Number of rFonts attribute values rewritten
-      def replace_font(from:, to:)
-        replacer = FontReplacer.new(from: from, to: to)
-        replacer.replace(self)
-      end
-
-      # Apply uniform page setup to every section of the document
-      #
-      # Mirrors Word's Layout dialog: named paper sizes, orientation
-      # (with Word-style dimension swap), and margins.
-      #
-      # @param size [String, nil] Paper size: letter, legal, a4, a5,
-      #   executive
-      # @param orientation [String, nil] "portrait" or "landscape"
-      # @param margins [Integer, String, nil] Uniform margin for all
-      #   sides (twips, or "1in" / "2.5cm" / "25mm")
-      # @param top, right, bottom, left [Integer, String, nil] Per-side
-      #   margin overrides
-      # @return [Integer] Number of sections updated
-      def apply_page_setup(size: nil, orientation: nil, margins: nil,
-                           top: nil, right: nil, bottom: nil, left: nil)
-        setup = PageSetup.new(size: size, orientation: orientation,
-                              margins: margins, top: top, right: right,
-                              bottom: bottom, left: left)
-        setup.apply(self)
-      end
-
-      # Remove one style from the document's style definitions
-      #
-      # Default styles (w:default="1") are never removed.
-      #
-      # @param style_id [String] Style id (w:styleId)
-      # @return [Boolean] true when the style was removed
-      def remove_style(style_id)
-        StyleCleanup.new(self).remove?(style_id)
-      end
-
-      # Rename a style's display name (w:name) — Word's style rename.
-      # References (pStyle/rStyle/tblStyle) target the styleId, so they
-      # keep working unchanged.
-      #
-      # @param identifier [String] Style id (w:styleId) or current
-      #   display name (w:name)
-      # @param new_name [String] New display name
-      # @return [Style, nil] The renamed style, or nil when not found
-      def rename_style(identifier, new_name)
-        style = styles_configuration.style(identifier)
-        return unless style
-
-        style.name = StyleName.new(val: new_name)
-        style
-      end
-
-      # Remove every style that no content references — Word's
-      # Styles-pane decluttering as one call. Default styles are kept.
-      #
-      # @return [Array<String>] Ids of removed styles
-      def remove_unused_styles
-        StyleCleanup.new(self).remove_unused
-      end
-
-      # Auto-transition from MS theme to Uniword equivalent
-      #
-      # Detects the MS theme in the document's embedded theme and replaces
-      # it with the corresponding Uniword theme (font-substituted, renamed).
-      #
-      # @return [Hash, nil] { uniword_slug:, ms_name: } or nil if no match
-      #
-      # @example
-      #   result = doc.auto_transition_theme
-      #   puts "Transitioned from #{result[:ms_name]} to #{result[:uniword_slug]}"
-      def auto_transition_theme
-        Resource::ThemeTransition.auto_transition!(self)
-      end
-
-      # Apply theme from another document
-      #
-      # @param source_path [String] Path to source .docx file
-      # @return [self] For method chaining
-      def apply_theme_from(source_path)
-        source_doc = Uniword.load(source_path)
-        self.theme = source_doc.theme.dup if source_doc.theme
-        self
-      end
-
-      # Apply styles from another document
-      #
-      # @param source_path [String] Path to source .docx file
-      # @param strategy [Symbol] Conflict resolution strategy
-      # @return [self] For method chaining
-      def apply_styles_from(source_path, strategy: :keep_existing)
-        source_doc = Uniword.load(source_path)
-        styles_configuration.merge(source_doc.styles_configuration,
-                                   conflict_resolution: strategy)
-        self
-      end
-
-      # Apply both theme and styles from a template document
-      #
-      # @param template_path [String] Path to template .docx file
-      # @param strategy [Symbol] Conflict resolution strategy for styles
-      # @return [self] For method chaining
-      def apply_template(template_path, strategy: :keep_existing)
-        template_doc = Uniword.load(template_path)
-        self.theme = template_doc.theme.dup if template_doc.theme
-        styles_configuration.merge(template_doc.styles_configuration,
-                                   conflict_resolution: strategy)
-        self
-      end
-
       # Custom inspect for readable output
       #
       # @return [String] Human-readable representation
@@ -548,18 +369,6 @@ module Uniword
 
       private
 
-      # Ensure the document carries a theme part, creating it from a
-      # fresh parse of the bundled Office theme when absent (every
-      # document gets its own copy — no shared state).
-      #
-      # @param transformation [Themes::ThemeTransformation] Converter
-      # @return [void]
-      def ensure_theme!(transformation)
-        return if theme&.theme_elements
-
-        self.theme = transformation.default_office_theme
-      end
-
       # Run model-level validation rules against this document.
       #
       # @return [Array<Validation::Report::ValidationIssue>] issues found