coradoc 2.0.27 → 2.0.29

data/lib/coradoc/query.rb

@@ -234,8 +234,8 @@ module Coradoc
         @elements = Array(elements).compact
       end
 
-      def each(&block)
-        @elements.each(&block)
+      def each(&)
+        @elements.each(&)
       end
 
       def [](index)
@@ -272,16 +272,16 @@ module Coradoc
         ResultSet.new(results.uniq)
       end
 
-      def map(&block)
-        ResultSet.new(@elements.map(&block))
+      def map(&)
+        ResultSet.new(@elements.map(&))
       end
 
-      def select(&block)
-        ResultSet.new(@elements.select(&block))
+      def select(&)
+        ResultSet.new(@elements.select(&))
       end
 
-      def reject(&block)
-        ResultSet.new(@elements.reject(&block))
+      def reject(&)
+        ResultSet.new(@elements.reject(&))
       end
 
       def to_a

data/lib/coradoc/reference/address/anchor.rb

@@ -0,0 +1,48 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module Reference
+    class Address < Lutaml::Model::Serializable
+      # In-document anchor — the most common navigation target.
+      # Matches raw strings starting with "#" plus bareword names that
+      # don't look like document IDs or external URLs.
+      #
+      #   "#intro"     => anchor target "intro"
+      #   "intro"      => anchor target "intro"
+      #   "footnote-1" => anchor target "footnote-1"
+      module Anchor
+        module_function
+
+        def scheme_name
+          :anchor
+        end
+
+        def matches?(raw)
+          return false if raw.nil? || raw.empty?
+
+          stripped = raw.to_s
+          return false if stripped == '#'
+          return true if stripped.start_with?('#')
+
+          stripped.match?(/\A[\w.\-]+\z/)
+        end
+
+        def parse(raw)
+          target = raw.to_s.delete_prefix('#')
+          new_address(target: target)
+        end
+
+        def serialize(address)
+          target = address.target
+          return '' if target.nil? || target.empty?
+
+          "##{target}"
+        end
+
+        def new_address(target:)
+          Address.new(scheme: 'anchor', target: target)
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/reference/address/doi.rb

@@ -0,0 +1,40 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module Reference
+    class Address < Lutaml::Model::Serializable
+      # DOI — Digital Object Identifier. Regex per the DOI handbook.
+      module Doi
+        module_function
+
+        DOI_PATTERN = %r{\A(10\.\d{4,}/[^\s#]+)(?:#(.*))?\z}
+
+        def scheme_name
+          :doi
+        end
+
+        def matches?(raw)
+          return false if raw.nil? || raw.empty?
+
+          DOI_PATTERN.match?(raw.to_s)
+        end
+
+        def parse(raw)
+          match = DOI_PATTERN.match(raw.to_s)
+          raise Address::ParseError, "Invalid DOI: #{raw.inspect}" unless match
+
+          target = match[1]
+          fragment = match[2]
+          Address.new(scheme: 'doi', target: target, fragment: fragment)
+        end
+
+        def serialize(address)
+          base = address.target.to_s
+          return base unless address.fragment && !address.fragment.empty?
+
+          "#{base}##{address.fragment}"
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/reference/address/isbn.rb

@@ -0,0 +1,39 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module Reference
+    class Address < Lutaml::Model::Serializable
+      # ISBN — book identifier. Recognizes an optional "ISBN" prefix
+      # (case insensitive) and either ISBN-10 or ISBN-13 digits. Bare
+      # numbers must be long enough to be ISBN-shaped so short numeric
+      # anchors are not claimed.
+      module Isbn
+        module_function
+
+        ISBN_PATTERN = /\A(?:ISBN\s+)?(\d[\d\-Xx]{8,16})\z/i
+
+        def scheme_name
+          :isbn
+        end
+
+        def matches?(raw)
+          return false if raw.nil? || raw.empty?
+
+          ISBN_PATTERN.match?(raw.to_s)
+        end
+
+        def parse(raw)
+          match = ISBN_PATTERN.match(raw.to_s)
+          raise Address::ParseError, "Invalid ISBN: #{raw.inspect}" unless match
+
+          target = match[1]
+          Address.new(scheme: 'isbn', target: target)
+        end
+
+        def serialize(address)
+          "ISBN #{address.target}"
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/reference/address/path.rb

@@ -0,0 +1,66 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module Reference
+    class Address < Lutaml::Model::Serializable
+      # Inter-document path — a document identifier plus optional fragment.
+      # Matches document-ID-shaped barewords (uppercase + dash + digit,
+      # e.g. "ELF-5005-1") and relative/absolute file paths
+      # ("images/foo.png", "../shared/common.adoc"). Locality is carried
+      # in fragment.
+      #
+      #   "ELF-5005-1"          => path target "ELF-5005-1"
+      #   "ELF-5005-1#sec-3"    => path target "ELF-5005-1", fragment "sec-3"
+      #   "images/foo.png"      => path target "images/foo.png"
+      module Path
+        module_function
+
+        # Doc-ID pattern: uppercase-led bareword ("ELF-5005-1"). Used
+        # together with a separate digit scan — two linear passes, no
+        # ambiguous adjacent quantifiers (polynomial-ReDoS-safe).
+        DOC_ID_PATTERN = /\A[A-Z][A-Z0-9_\-]*\z/
+        # Loose pattern used by +parse+ when the scheme is already chosen
+        # (via hint). Accepts any non-empty target plus optional fragment.
+        LOOSE_PATTERN = /\A([^#]+)(?:#(.*))?\z/
+
+        def scheme_name
+          :path
+        end
+
+        def matches?(raw)
+          return false if raw.nil? || raw.empty?
+
+          value = raw.to_s
+          return true if value.include?('/')
+
+          # Document IDs are uppercase-led and contain at least one digit
+          # (what distinguishes them from anchors like "SECTION").
+          id_part = value.split('#', 2).first
+          DOC_ID_PATTERN.match?(id_part) && id_part.match?(/\d/)
+        end
+
+        def parse(raw)
+          match = LOOSE_PATTERN.match(raw.to_s)
+          raise Address::ParseError, "Invalid path: #{raw.inspect}" unless match
+
+          Address.new(
+            scheme: 'path',
+            target: match[1],
+            fragment: normalize_fragment(match[2])
+          )
+        end
+
+        def serialize(address)
+          base = address.target.to_s
+          return base unless address.fragment && !address.fragment.empty?
+
+          "#{base}##{address.fragment}"
+        end
+
+        def normalize_fragment(fragment)
+          fragment&.empty? ? nil : fragment
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/reference/address/scoped_path.rb

@@ -0,0 +1,47 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module Reference
+    class Address < Lutaml::Model::Serializable
+      # Scoped path — a path within a collection's namespace. The scope
+      # identifies the collection (e.g. "ELF"); the target is the path
+      # inside it ("5005:1"); the fragment is locality within that doc.
+      #
+      #   "ELF:5005:1#sec-3"  => scope "ELF", target "5005:1", fragment "sec-3"
+      module ScopedPath
+        module_function
+
+        SCOPED_PATTERN = /\A([A-Z][A-Z0-9_\-]*):([\d:]+)(?:#(.*))?\z/
+
+        def scheme_name
+          :scoped_path
+        end
+
+        def matches?(raw)
+          return false if raw.nil? || raw.empty?
+
+          SCOPED_PATTERN.match?(raw.to_s)
+        end
+
+        def parse(raw)
+          match = SCOPED_PATTERN.match(raw.to_s)
+          raise Address::ParseError, "Invalid scoped path: #{raw.inspect}" unless match
+
+          Address.new(
+            scheme: 'scoped_path',
+            scope: match[1],
+            target: match[2],
+            fragment: match[3]
+          )
+        end
+
+        def serialize(address)
+          base = "#{address.scope}:#{address.target}"
+          return base unless address.fragment && !address.fragment.empty?
+
+          "#{base}##{address.fragment}"
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/reference/address/url.rb

@@ -0,0 +1,41 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module Reference
+    class Address < Lutaml::Model::Serializable
+      # Absolute URL — anything with a recognized URL scheme prefix.
+      # Fragment is the optional "#..." suffix.
+      module Url
+        module_function
+
+        URL_PATTERN = %r{\A([a-z][a-z0-9+\-.]*)://([^#]+)(?:#(.*))?\z}i
+
+        def scheme_name
+          :url
+        end
+
+        def matches?(raw)
+          return false if raw.nil? || raw.empty?
+
+          URL_PATTERN.match?(raw.to_s)
+        end
+
+        def parse(raw)
+          match = URL_PATTERN.match(raw.to_s)
+          raise Address::ParseError, "Invalid URL: #{raw.inspect}" unless match
+
+          target = "#{match[1]}://#{match[2]}"
+          fragment = match[3]
+          Address.new(scheme: 'url', target: target, fragment: fragment)
+        end
+
+        def serialize(address)
+          base = address.target.to_s
+          return base unless address.fragment && !address.fragment.empty?
+
+          "#{base}##{address.fragment}"
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/reference/address.rb

@@ -0,0 +1,153 @@
+# frozen_string_literal: true
+
+require 'lutaml/model'
+
+module Coradoc
+  module Reference
+    # Scheme-aware locator targeting a Content node.
+    #
+    # An Address is one of six built-in schemes — +:anchor+, +:path+,
+    # +:scoped_path+, +:url+, +:doi+, +:isbn+ — selected by heuristic on
+    # parse or by an explicit +hint:+. External gems register additional
+    # schemes via +Address.register_scheme+ (OCP).
+    #
+    #   Address.parse("foo")               # => anchor "foo"
+    #   Address.parse("ELF-5005-1#sec-3")  # => path "ELF-5005-1", fragment "sec-3"
+    #   Address.parse("ELF:5005:1#sec-3")  # => scoped_path scope "ELF"
+    #   Address.parse("https://x.y/z")     # => url
+    #   Address.parse("10.1234/abc")       # => doi
+    #   Address.parse("ISBN 978-1-2-3")    # => isbn
+    #
+    # Addresses are value types: two addresses are equal iff every
+    # attribute matches. Round-trip via +to_s+.
+    class Address < Lutaml::Model::Serializable
+      class ParseError < Coradoc::Reference::Error; end
+      class UnknownSchemeError < Coradoc::Reference::Error; end
+
+      attribute :scheme, :string
+      attribute :target, :string
+      attribute :fragment, :string
+      attribute :scope, :string
+
+      autoload :Url, "#{__dir__}/address/url"
+      autoload :Doi, "#{__dir__}/address/doi"
+      autoload :Isbn, "#{__dir__}/address/isbn"
+      autoload :ScopedPath, "#{__dir__}/address/scoped_path"
+      autoload :Path, "#{__dir__}/address/path"
+      autoload :Anchor, "#{__dir__}/address/anchor"
+
+      BUILTIN_SCHEME_ORDER = %i[url doi isbn scoped_path path anchor].freeze
+
+      class << self
+        def parse(raw, hint: nil)
+          Scheme.ensure_builtins_registered!
+          mod = hint ? scheme_for_hint!(hint) : Scheme.match(raw)
+          unless mod
+            raise ParseError,
+                  "Cannot determine address scheme for #{raw.inspect}"
+          end
+          mod.parse(raw)
+        end
+
+        def register_scheme(mod)
+          Scheme.register(mod)
+        end
+
+        def scheme_names
+          Scheme.ensure_builtins_registered!
+          Scheme.names
+        end
+
+        private
+
+        def scheme_for_hint!(hint)
+          Scheme.for(hint) ||
+            raise(UnknownSchemeError,
+                  "Unknown address scheme #{hint.inspect}. " \
+                  "Registered: #{Scheme.names.inspect}")
+        end
+      end
+
+      # Value equality (==/eql?/hash) comes from Lutaml::Model — all
+      # attributes compared, class-aware, safe as Hash keys.
+
+      def to_s
+        Scheme.ensure_builtins_registered!
+        mod = Scheme.for(scheme)
+        unless mod
+          raise UnknownSchemeError,
+                "No registered scheme for #{scheme.inspect}. " \
+                "Registered: #{Scheme.names.inspect}"
+        end
+        mod.serialize(self)
+      end
+
+      # Registry of scheme modules. Each module provides:
+      #   scheme_name      -> Symbol
+      #   matches?(raw)    -> Boolean
+      #   parse(raw)       -> Address
+      #   serialize(addr)  -> String
+      #
+      # +register+ always forces builtin registration first, so an
+      # external scheme registered before first use is never clobbered
+      # by the lazy builtin pass. The catch-all anchor scheme is always
+      # tried LAST on match, so external schemes are never shadowed by
+      # it no matter when they register.
+      module Scheme
+        @registered = []
+        @builtins_registered = false
+        @registering_builtins = false
+
+        CATCH_ALL_SCHEME = :anchor
+        private_constant :CATCH_ALL_SCHEME
+
+        MUTEX = Mutex.new
+        private_constant :MUTEX
+
+        class << self
+          def register(mod)
+            ensure_builtins_registered! unless @registering_builtins
+            @registered.delete_if { |m| m.scheme_name == mod.scheme_name }
+            @registered << mod
+          end
+
+          def names
+            @registered.map(&:scheme_name)
+          end
+
+          def for(name)
+            @registered.find { |m| m.scheme_name == name.to_sym }
+          end
+
+          def match(raw)
+            catch_all, specific =
+              @registered.partition { |m| m.scheme_name == CATCH_ALL_SCHEME }
+            (specific + catch_all).find { |m| m.matches?(raw) }
+          end
+
+          def ensure_builtins_registered!
+            MUTEX.synchronize do
+              return if @builtins_registered
+
+              @registering_builtins = true
+              begin
+                BUILTIN_SCHEME_ORDER.each do |name|
+                  const_name = name.to_s.split('_').map(&:capitalize).join
+                  register(Address.const_get(const_name))
+                end
+                @builtins_registered = true
+              ensure
+                @registering_builtins = false
+              end
+            end
+          end
+
+          def reset!
+            @registered.clear
+            @builtins_registered = false
+          end
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/reference/catalog/composite.rb

@@ -0,0 +1,59 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module Reference
+    module Catalog
+      # Composes multiple catalogs. Lookup asks EVERY child and merges
+      # the hits: exactly one hit returns the Content; several hits
+      # return an Array (which the Resolver surfaces per the
+      # +ambiguous:+ policy). Precedence between catalogs is a
+      # resolver/policy concern, not a lookup concern:
+      #
+      #   Composite.new(
+      #     Local.from_doc(doc),
+      #     Collection.from_manifest(...),
+      #     Remote.new(client: http_client)
+      #   )
+      class Composite
+        include Catalog::Protocol
+
+        attr_reader :children
+
+        def initialize(*children)
+          @children = children.flatten
+        end
+
+        def lookup(address)
+          results = []
+          children.each do |catalog|
+            result = catalog.lookup(address)
+            next if result.nil?
+
+            results.concat(Array(result))
+          end
+          return nil if results.empty?
+          return results.first if results.size == 1
+
+          results
+        end
+
+        def ambiguous?(address)
+          result = lookup(address)
+          result.is_a?(Array) && result.size > 1
+        end
+
+        def each_pair(&block)
+          return to_enum(:each_pair) unless block_given?
+
+          children.each do |catalog|
+            catalog.each_pair(&block)
+          end
+        end
+
+        def recognizes_scheme?(scheme)
+          children.any? { |catalog| catalog.recognizes_scheme?(scheme) }
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/reference/catalog/local.rb

@@ -0,0 +1,97 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module Reference
+    module Catalog
+      # Indexes one CoreModel document. Walks the tree once at
+      # construction; indexes every node carrying an +id+ by its anchor
+      # Address, plus the document itself by an optional path Address.
+      #
+      #   catalog = Catalog::Local.from_doc(doc, path: "ELF-5005-1")
+      #   catalog.lookup(Address.parse("ELF-5005-1"))        # => doc
+      #   catalog.lookup(Address.parse("sec-3"))             # => section
+      class Local
+        include Catalog::Protocol
+
+        attr_reader :document, :document_path
+
+        def initialize(document:, document_path: nil, index: nil)
+          @document = document
+          @document_path = document_path
+          @index = index || MemoryIndex.new
+          populate! unless index
+        end
+
+        class << self
+          def from_doc(document, path: nil)
+            new(document: document, document_path: path)
+          end
+        end
+
+        def lookup(address)
+          @index.lookup(address)
+        end
+
+        def ambiguous?(address)
+          @index.ambiguous?(address)
+        end
+
+        def each_pair(&)
+          @index.each_pair(&)
+        end
+
+        def recognizes_scheme?(scheme)
+          return true if scheme.to_sym == :anchor
+          return true if scheme.to_sym == :path && document_path
+          return true if scheme.to_sym == :scoped_path && document_path&.include?(':')
+
+          @index.recognizes_scheme?(scheme)
+        end
+
+        private
+
+        def populate!
+          index_document_root!
+          walk(document)
+        end
+
+        def index_document_root!
+          return unless document_path
+
+          address = Coradoc::Reference::Address.parse(document_path, hint: path_hint)
+          @index.add(address, document)
+        end
+
+        def path_hint
+          document_path.include?(':') ? :scoped_path : :path
+        end
+
+        def walk(node)
+          return unless node.is_a?(Coradoc::CoreModel::Base)
+
+          index_node!(node)
+          walk_children(node)
+        end
+
+        def index_node!(node)
+          return unless node.id
+
+          address = Coradoc::Reference::Address.new(
+            scheme: 'anchor',
+            target: node.id
+          )
+          @index.add(address, node)
+        end
+
+        def walk_children(node)
+          return unless node.is_a?(Coradoc::CoreModel::HasChildren)
+
+          children = node.children
+          return if children.nil? || children.empty?
+
+          children.each { |child| walk(child) }
+        end
+      end
+    end
+  end
+end

data/lib/coradoc/reference/catalog/memory_index.rb

@@ -0,0 +1,59 @@
+# frozen_string_literal: true
+
+module Coradoc
+  module Reference
+    module Catalog
+      # Shared in-memory index. Built-in catalogs compose this rather
+      # than reimplementing the storage layer (DRY). The index is
+      # immutable from the outside — entries are added by the catalog
+      # during construction, then read-only.
+      class MemoryIndex
+        include Enumerable
+
+        def initialize
+          @by_address = {}
+          @schemes = Set.new
+        end
+
+        def add(address, content)
+          raise ArgumentError, 'address required' unless address.is_a?(Coradoc::Reference::Address)
+          raise ArgumentError, 'content required' unless content
+
+          @by_address[address] = Array(@by_address[address]) << content
+          @schemes << address.scheme.to_sym
+          self
+        end
+
+        def lookup(address)
+          entries = @by_address[address]
+          return nil unless entries
+          return entries.first if entries.size == 1
+
+          # Copy: callers must not be able to corrupt the index.
+          entries.dup
+        end
+
+        def ambiguous?(address)
+          entries = @by_address[address]
+          entries && entries.size > 1
+        end
+
+        def each_pair
+          return to_enum(:each_pair) unless block_given?
+
+          @by_address.each do |address, contents|
+            contents.each { |c| yield address, c }
+          end
+        end
+
+        def recognizes_scheme?(scheme)
+          @schemes.include?(scheme.to_sym)
+        end
+
+        def size
+          @by_address.size
+        end
+      end
+    end
+  end
+end