coradoc 2.0.28 → 2.0.30

data/lib/coradoc/reference/edge_search.rb

@@ -0,0 +1,128 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module Reference
+    # Walks a CoreModel tree and yields every Edge found in it. For
+    # Phase 1, Edges are wrapped inside existing node types —
+    # +CrossReferenceElement+, +LinkElement+, +Image+, +Include+,
+    # +FootnoteElement+. The Phase 3 migration will move Edges to
+    # first-class attributes; this module abstracts the extraction so
+    # the rest of the resolver doesn't change.
+    #
+    # The walker is read-only: it never mutates the input tree.
+    module EdgeSearch
+      module_function
+
+      # Yields (parent_node, edge) for every Edge found in the tree.
+      # The parent_node is the node that owns the edge — the
+      # materializer replaces the edge in-place inside parent_node's
+      # children list.
+      def each_edge(root)
+        return to_enum(:each_edge, root) unless block_given?
+
+        walk(root) { |node| edges_for(node).each { |edge| yield node, edge } }
+      end
+
+      # Walk tree, yield each Base node. Stops at inline leaves.
+      def walk(node, &block)
+        return unless node.is_a?(Coradoc::CoreModel::Base)
+        return unless block
+
+        yield(node)
+        return unless node.is_a?(Coradoc::CoreModel::HasChildren)
+
+        children = node.children
+        return unless children
+
+        children.each { |child| walk(child, &block) }
+      end
+
+      # Extract every Edge the given node owns. Pure function of node.
+      def edges_for(node)
+        edge_extractor = EDGE_EXTRACTORS[node.class]
+        return [] unless edge_extractor
+
+        edge = edge_extractor.call(node)
+        edge ? [edge] : []
+      end
+
+      def edge_from_cross_reference(node)
+        Edge.build(
+          kind: :navigation,
+          address: parse_address(node.target, hint: xref_hint(node.target)),
+          source_id: node.id,
+          label: node.content
+        )
+      end
+
+      def edge_from_link(node)
+        Edge.build(
+          kind: :link,
+          address: parse_address(node.target),
+          source_id: node.id,
+          label: node.content
+        )
+      end
+
+      def edge_from_include(node)
+        Edge.build(
+          kind: :include,
+          address: parse_address(node.target, hint: path_or_url_hint(node.target)),
+          source_id: node.id,
+          options: { include_options: node.options }
+        )
+      end
+
+      def edge_from_image(node)
+        Edge.build(
+          kind: :image_ref,
+          address: parse_address(node.src, hint: path_or_url_hint(node.src)),
+          source_id: node.id,
+          label: node.alt,
+          options: { alt_text: node.alt }
+        )
+      end
+
+      def edge_from_footnote(node)
+        Edge.build(
+          kind: :footnote_ref,
+          address: parse_address(node.target || node.id, hint: :anchor),
+          source_id: node.id,
+          options: { footnote_id: node.id }
+        )
+      end
+
+      # Xrefs are document-internal by definition (AsciiDoc): bare
+      # targets are anchors, even uppercase document-ID-shaped ones
+      # ("SEC-2"). Only an explicit "document#fragment" shape or a URL
+      # points outside the current document.
+      def xref_hint(target)
+        raw = target.to_s
+        return :url if Coradoc::Reference::Address::Url.matches?(raw)
+        return :path if raw.include?('#') && !raw.start_with?('#')
+
+        :anchor
+      end
+
+      # File-like targets (include, image): URLs stay urls, everything
+      # else is a path — including bare filenames ("foo.png") that the
+      # anchor bareword heuristic would otherwise claim.
+      def path_or_url_hint(target)
+        Coradoc::Reference::Address::Url.matches?(target.to_s) ? :url : :path
+      end
+
+      def parse_address(target, hint: nil)
+        Coradoc::Reference::Address.parse(target.to_s, hint: hint)
+      end
+
+      EDGE_EXTRACTORS = {
+        Coradoc::CoreModel::CrossReferenceElement => method(:edge_from_cross_reference),
+        Coradoc::CoreModel::LinkElement => method(:edge_from_link),
+        Coradoc::CoreModel::Include => method(:edge_from_include),
+        Coradoc::CoreModel::Image => method(:edge_from_image),
+        Coradoc::CoreModel::FootnoteElement => method(:edge_from_footnote)
+      }.freeze
+      private_constant :EDGE_EXTRACTORS
+    end
+  end
+end

data/lib/coradoc/reference/materializer/base.rb

@@ -0,0 +1,44 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module Reference
+    module Materializer
+      # Protocol base. Subclasses declare their (kind, presentation,
+      # format) tuple via class-level +.kind+, +.presentation+, +.format+
+      # methods, then implement +#materialize+ to produce a
+      # CoreModel::InlineElement subtree.
+      #
+      # Materializers consume Results — they never touch the catalog
+      # directly. The Presentation tells them where the target lives.
+      class Base
+        class << self
+          def kind
+            nil
+          end
+
+          def presentation
+            :any
+          end
+
+          def format
+            :any
+          end
+        end
+
+        # Render one resolved edge.
+        #
+        # @param edge [Edge]
+        # @param result [Result::Base] the resolved outcome
+        # @param node [CoreModel::Base] the original edge-bearing node;
+        #   return it unchanged to keep the node as authored
+        # @param presentation [Presentation::Base]
+        # @param pages [Array<Presentation::Page>]
+        # @return [CoreModel::Base, nil] replacement node, the original
+        #   +node+ to keep it, or nil to drop it from the tree
+        def materialize(edge:, result:, node:, presentation:, pages:)
+          raise NotImplementedError
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/reference/materializer/passthrough.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module Reference
+    module Materializer
+      # Default fallback. Returns the original edge-bearing node
+      # unchanged — an unresolvable or unrenderable reference survives
+      # materialization exactly as authored, so round-tripping is
+      # always safe and no content is ever lost.
+      class Passthrough < Base
+        class << self
+          def kind
+            :any
+          end
+
+          def presentation
+            :any
+          end
+
+          def format
+            :any
+          end
+        end
+
+        def materialize(node:, **)
+          node
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/reference/materializer/registry.rb

@@ -0,0 +1,116 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module Reference
+    module Materializer
+      # Registry of materializers keyed by [kind, presentation, format].
+      # Lookup falls back on +:any+ — so a materializer registered for
+      # +[:link, :any, :html]+ handles every presentation in HTML.
+      # Most specific key wins (concrete > :any).
+      #
+      # Format gems register their materializers process-wide via
+      # +.register_global+ (OCP): every registry instance picks them up
+      # on first use, and dispatch code never changes.
+      class Registry
+        # Core ships only the node-preserving fallback. Format-specific
+        # materializers live in the format gems (coradoc-html,
+        # coradoc-adoc, ...) and register globally at load time.
+        BUILTINS = [Materializer::Passthrough].freeze
+
+        @global_registrations = []
+
+        GLOBAL_MUTEX = Mutex.new
+        private_constant :GLOBAL_MUTEX
+
+        class << self
+          def register_global(klass)
+            GLOBAL_MUTEX.synchronize do
+              @global_registrations << klass unless @global_registrations.include?(klass)
+            end
+          end
+
+          def global_registrations
+            GLOBAL_MUTEX.synchronize { @global_registrations.dup }
+          end
+
+          def reset_globals!
+            GLOBAL_MUTEX.synchronize { @global_registrations.clear }
+          end
+        end
+
+        def initialize
+          @by_key = {}
+          @builtins_registered = false
+        end
+
+        def register(klass)
+          ensure_builtins_registered!
+          @by_key[EntryKey.new(klass.kind, klass.presentation, klass.format)] = klass
+        end
+
+        def lookup(kind:, presentation:, format:)
+          ensure_builtins_registered!
+          find_most_specific(kind, presentation, format) ||
+            find_most_specific(kind, presentation, :any) ||
+            find_most_specific(kind, :any, format) ||
+            find_most_specific(kind, :any, :any) ||
+            find_most_specific(:any, :any, :any)
+        end
+
+        def registered
+          ensure_builtins_registered!
+          @by_key.dup
+        end
+
+        def reset!
+          @by_key.clear
+          @builtins_registered = false
+        end
+
+        private
+
+        def find_most_specific(kind, presentation, format)
+          @by_key[EntryKey.new(kind.to_sym, presentation.to_sym, format.to_sym)]
+        end
+
+        def ensure_builtins_registered!
+          return if @builtins_registered
+
+          register_builtins!
+          @builtins_registered = true
+        end
+
+        def register_builtins!
+          (BUILTINS + self.class.global_registrations).each do |k|
+            @by_key[EntryKey.new(k.kind, k.presentation, k.format)] = k
+          end
+        end
+
+        # Internal composite key for the registry. Equality based on
+        # the three symbol values so two equivalent tuples compare equal.
+        class EntryKey
+          attr_reader :kind, :presentation, :format
+
+          def initialize(kind, presentation, format)
+            @kind = kind.to_sym
+            @presentation = presentation.to_sym
+            @format = format.to_sym
+          end
+
+          def ==(other)
+            other.is_a?(EntryKey) &&
+              kind == other.kind &&
+              presentation == other.presentation &&
+              format == other.format
+          end
+          alias eql? ==
+
+          def hash
+            [kind, presentation, format].hash
+          end
+        end
+        private_constant :EntryKey
+      end
+    end
+  end
+end

data/lib/coradoc/reference/materializer.rb

@@ -0,0 +1,14 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module Reference
+    # Render an Edge per (kind, presentation, format). Registry-based:
+    # adding a new output format or citation style is registering a
+    # new materializer, not editing switch statements.
+    module Materializer
+      autoload :Base, "#{__dir__}/materializer/base"
+      autoload :Registry, "#{__dir__}/materializer/registry"
+      autoload :Passthrough, "#{__dir__}/materializer/passthrough"
+    end
+  end
+end

data/lib/coradoc/reference/presentation/base.rb

@@ -0,0 +1,47 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module Reference
+    module Presentation
+      # Protocol base. Subclasses implement +#layout+ (resolved graph →
+      # Pages) and +#locate_page+ (find which page a target lives on).
+      #
+      # A Presentation NEVER reads the original document — it consumes
+      # the resolved graph produced by the Resolver. It never renders.
+      class Base
+        # Registry key used in materializer lookup tuples
+        # [kind, presentation, format]. Subclasses override so
+        # materializers can target a specific layout.
+        def self.key
+          :any
+        end
+
+        def key
+          self.class.key
+        end
+
+        # Lay out the resolved graph as a tree of Pages. Subclasses
+        # decide slicing, ordering, hierarchy.
+        #
+        # @param resolved_graph [CoreModel::Base] the document the
+        #   caller wants to present. Already resolved (edges → targets).
+        # @return [Array<Page>]
+        def layout(resolved_graph)
+          raise NotImplementedError
+        end
+
+        # Given an Edge and its resolved target, find the Page where
+        # the target lives. This is what makes cross-references
+        # survive re-pagination.
+        #
+        # @param edge [Edge]
+        # @param target_content [CoreModel::Base]
+        # @param pages [Array<Page>]
+        # @return [Page, nil]
+        def locate_page(edge, target_content, pages:)
+          raise NotImplementedError
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/reference/presentation/custom_hierarchy.rb

@@ -0,0 +1,104 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module Reference
+    module Presentation
+      # Caller-supplied hierarchy. The user provides a tree of node ids
+      # mapping to pages; the presentation lays out content accordingly.
+      # This is the presentation website builds use.
+      #
+      #   CustomHierarchy.new(hierarchy: [
+      #     { id: "intro", title: "Introduction", children: [...] },
+      #     { id: "chap1", title: "Chapter 1", children: [
+      #       { id: "chap1-1", title: "1.1" }
+      #     ] }
+      #   ])
+      class CustomHierarchy < Base
+        attr_reader :hierarchy
+
+        def self.key
+          :custom_hierarchy
+        end
+
+        def initialize(hierarchy:)
+          super()
+          @hierarchy = normalize_hierarchy(hierarchy)
+        end
+
+        def layout(resolved_graph)
+          pages = []
+          walk_hierarchy(@hierarchy, parent_id: nil, pages: pages, root: resolved_graph)
+          pages
+        end
+
+        def locate_page(_edge, target_content, pages:)
+          pages.find { |page| page.content.equal?(target_content) }
+        end
+
+        private
+
+        def normalize_hierarchy(tree)
+          Array(tree).map do |entry|
+            entry.merge(children: normalize_hierarchy(entry[:children]))
+          end
+        end
+
+        def walk_hierarchy(nodes, parent_id:, pages:, root:)
+          nodes.each do |entry|
+            content = find_content_by_id(root, entry[:id]) || root
+            pages << Page.new(
+              id: entry[:id],
+              title: entry[:title] || content.title,
+              content: content,
+              parent_id: parent_id,
+              order: pages.size
+            )
+            walk_hierarchy(entry[:children], parent_id: entry[:id], pages: pages, root: root)
+          end
+        end
+
+        def find_content_by_id(root, id)
+          return nil unless id
+
+          visitor = VisitorById.new(id)
+          visitor.visit(root)
+          visitor.found
+        end
+
+        # Single-purpose visitor: walks a CoreModel tree looking for
+        # one node by id. Cleaner than a recursive method on the
+        # presentation class itself.
+        class VisitorById
+          attr_reader :found
+
+          def initialize(target_id)
+            @target_id = target_id
+            @found = nil
+          end
+
+          def visit(node)
+            return @found if @found
+            return unless node.is_a?(Coradoc::CoreModel::Base)
+
+            @found = node if node.id == @target_id
+            return if @found
+
+            visit_children(node)
+          end
+
+          private
+
+          def visit_children(node)
+            return unless node.is_a?(Coradoc::CoreModel::HasChildren)
+
+            children = node.children
+            return unless children
+
+            children.each { |c| visit(c) }
+            nil
+          end
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/reference/presentation/page.rb

@@ -0,0 +1,20 @@
+# frozen_string_literal: true
+
+require 'lutaml/model'
+
+module Coradoc
+  module Reference
+    module Presentation
+      # A page is the unit of output a Presentation produces. One HTML
+      # file, one PDF page, one EPUB chapter — same data, different
+      # materialization. The Materializer consumes Pages.
+      class Page < Lutaml::Model::Serializable
+        attribute :id, :string
+        attribute :title, :string
+        attribute :content, Coradoc::CoreModel::Base
+        attribute :parent_id, :string
+        attribute :order, :integer
+      end
+    end
+  end
+end

data/lib/coradoc/reference/presentation/single_document.rb

@@ -0,0 +1,35 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module Reference
+    module Presentation
+      # One page = the whole document. Use for fixed-structure output
+      # (single HTML, single PDF, EPUB chapter). Cross-references
+      # always resolve to the single page.
+      class SingleDocument < Base
+        def self.key
+          :single_document
+        end
+
+        def layout(resolved_graph)
+          [Page.new(
+            id: page_id_for(resolved_graph),
+            title: resolved_graph.title,
+            content: resolved_graph,
+            order: 0
+          )]
+        end
+
+        def locate_page(_edge, _target_content, pages:)
+          pages.first
+        end
+
+        private
+
+        def page_id_for(node)
+          node.id || 'root'
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/reference/presentation/split_pages.rb

@@ -0,0 +1,118 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module Reference
+    module Presentation
+      # Splits the document by top-level section (or a configurable
+      # boundary). Each section becomes a Page. Cross-references are
+      # rewritten to point at whichever page the target content lives on.
+      #
+      #   SplitPages.new(split_at: :section)  # default
+      #   SplitPages.new(split_at: :chapter) # HeaderElement only
+      class SplitPages < Base
+        attr_reader :split_at
+
+        def self.key
+          :split_pages
+        end
+
+        def initialize(split_at: :section)
+          super()
+          @split_at = split_at.to_sym
+          @page_index = nil
+        end
+
+        def layout(resolved_graph)
+          children = read_children(resolved_graph)
+          return [single_page_for(resolved_graph)] if children.nil? || children.empty?
+
+          sections = children.select { |child| matches_split?(child) }
+          return [single_page_for(resolved_graph)] if sections.empty?
+
+          pages = sections.map.with_index { |section, idx| page_for(section, resolved_graph, idx) }
+          index_pages!(pages)
+          pages
+        end
+
+        def locate_page(_edge, target_content, pages:)
+          indexed = @page_index&.[](target_content)
+          return indexed if indexed
+
+          pages.find { |page| owns_target?(page, target_content) }
+        end
+
+        private
+
+        # O(1) target→page lookup, built once per layout instead of
+        # rescanning every page subtree per edge.
+        def index_pages!(pages)
+          @page_index = {}.compare_by_identity
+          pages.each { |page| index_node!(page.content, page) }
+        end
+
+        def index_node!(node, page)
+          @page_index[node] = page
+          return unless node.is_a?(Coradoc::CoreModel::HasChildren)
+
+          children = node.children
+          return unless children
+
+          children.each { |child| index_node!(child, page) }
+        end
+
+        def page_for(section, parent, idx)
+          Page.new(
+            id: section.id || "page-#{idx}",
+            title: section.title || "Page #{idx}",
+            content: section,
+            parent_id: parent.id,
+            order: idx
+          )
+        end
+
+        def read_children(node)
+          return nil unless node.is_a?(Coradoc::CoreModel::HasChildren)
+
+          node.children
+        end
+
+        def matches_split?(node)
+          return false unless node.is_a?(Coradoc::CoreModel::StructuralElement)
+
+          klass = boundary_class
+          node.is_a?(klass)
+        end
+
+        def boundary_class
+          split_at == :chapter ? Coradoc::CoreModel::HeaderElement : Coradoc::CoreModel::SectionElement
+        end
+
+        def single_page_for(node)
+          Page.new(
+            id: node.id || 'root',
+            title: node.title,
+            content: node,
+            order: 0
+          )
+        end
+
+        def owns_target?(page, target)
+          return true if page.content.equal?(target)
+
+          descendant_of?(page.content, target)
+        end
+
+        def descendant_of?(ancestor, target)
+          return false unless ancestor.is_a?(Coradoc::CoreModel::HasChildren)
+
+          children = ancestor.children
+          return false unless children
+
+          children.any? do |child|
+            child.equal?(target) || descendant_of?(child, target)
+          end
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/reference/presentation.rb

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module Reference
+    # Model-View boundary. Same Content graph → N Presentations. A
+    # Presentation defines slicing, page boundaries, ordering, and
+    # hierarchy — never the rendering (that's the Materializer's job).
+    module Presentation
+      autoload :Base, "#{__dir__}/presentation/base"
+      autoload :Page, "#{__dir__}/presentation/page"
+      autoload :SingleDocument, "#{__dir__}/presentation/single_document"
+      autoload :SplitPages, "#{__dir__}/presentation/split_pages"
+      autoload :CustomHierarchy, "#{__dir__}/presentation/custom_hierarchy"
+    end
+  end
+end