pubid 2.0.0.pre.alpha.8 → 2.0.0.pre.alpha.9

data/lib/pubid/un.rb

@@ -0,0 +1,46 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Un
+    extend Pubid::PrefixesSupport
+
+    PREFIXES = ["UN"].freeze
+
+    autoload :Builder, "#{__dir__}/un/builder"
+    autoload :Identifier, "#{__dir__}/un/identifier"
+    autoload :Identifiers, "#{__dir__}/un/identifiers"
+    autoload :Parser, "#{__dir__}/un/parser"
+    autoload :Renderer, "#{__dir__}/un/renderer"
+
+    def self.parse(identifier)
+      Identifier.parse(identifier)
+    end
+
+    Identifier.format_registry = FormatRegistry.new(parent: Identifier.format_registry)
+    Identifier.format_registry.register(:human, renderer: Un::Renderer)
+
+    def self.identifier_types
+      @identifier_types ||= Identifiers.constants
+        .filter_map { |c| begin; Identifiers.const_get(c); rescue NameError; nil; end }
+        .select { |c| c.is_a?(Class) && c.singleton_methods(false).include?(:type) }
+        .select { |c| c.type.is_a?(Hash) }
+    end
+
+    def self.all_typed_stages
+      @all_typed_stages ||= identifier_types.flat_map do |klass|
+        klass.const_defined?(:TYPED_STAGES) ? klass.const_get(:TYPED_STAGES) : []
+      end
+    end
+
+    def self.locate_type(code)
+      identifier_types.find { |t| t.type[:key].to_s == code.to_s }
+    end
+
+    def self.locate_stage(abbr)
+      abbr_str = abbr.to_s.upcase
+      all_typed_stages.find { |s| s.abbr.any? { |a| a.to_s.upcase == abbr_str } }
+    end
+  end
+end
+
+Pubid::Registry.register(:un, Pubid::Un)

data/lib/pubid/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Pubid
-  VERSION = "2.0.0.pre.alpha.8"
+  VERSION = "2.0.0.pre.alpha.9"
 end

data/lib/pubid/w3c/builder.rb

@@ -0,0 +1,55 @@
+# frozen_string_literal: true
+
+module Pubid
+  module W3c
+    # Turns the parse tree into a concrete identifier: picks the class from the
+    # leading maturity token and splits the captured remainder into code + date.
+    class Builder
+      # Maturity token -> concrete identifier class name (resolved lazily so the
+      # Identifiers namespace need not be loaded at class-definition time).
+      TYPE_CLASSES = {
+        "NOTE" => "Note",
+        "DNOTE" => "DraftNote",
+        "WD" => "WorkingDraft",
+        "CR" => "CandidateRecommendation",
+        "CRD" => "CandidateRecommendationDraft",
+        "REC" => "Recommendation",
+        "PR" => "ProposedRecommendation",
+        "PER" => "ProposedEditedRecommendation",
+        "SPSD" => "SupersededRecommendation",
+        "OBSL" => "ObsoleteRecommendation",
+      }.freeze
+
+      # A trailing "-<digits>" group is the date only when the digit run is
+      # exactly 8 (YYYYMMDD), 6 (legacy YYMMDD) or 4 (legacy MMDD) wide. No real
+      # W3C code ends in such a run, so this never mis-splits a code that merely
+      # ends in a short digit run (e.g. "url-1", "ATAG10").
+      DATE_RE = /\A(.+)-(\d{8}|\d{6}|\d{4})\z/
+
+      def self.build(parsed_data)
+        new.build(parsed_data)
+      end
+
+      def build(data)
+        code, date = split_date(data[:rest].to_s)
+        klass_for(data[:type]&.to_s).new(code: code, date: date)
+      end
+
+      private
+
+      def klass_for(type)
+        name = TYPE_CLASSES[type] || "Standard"
+        Identifiers.const_get(name)
+      end
+
+      # @return [Array(String, String|nil)] the code and the optional date
+      def split_date(rest)
+        if (m = DATE_RE.match(rest))
+          [m[1], m[2]]
+        else
+          [rest, nil]
+        end
+      end
+    end
+  end
+end

data/lib/pubid/w3c/identifier.rb

@@ -0,0 +1,101 @@
+# frozen_string_literal: true
+
+module Pubid
+  module W3c
+    # Base class for every W3C identifier AND the flavor's parse/create entry
+    # point. Concrete identifiers under Pubid::W3c::Identifiers descend from
+    # this class (one per W3C maturity level), so a parsed W3C id is an
+    # instance of Pubid::W3c::Identifier.
+    #
+    # W3C identifiers are flat: a slug `code`, an optional publication `date`
+    # (kept verbatim as a string — width varies: 8-digit YYYYMMDD, or legacy
+    # 6-digit YYMMDD / 4-digit MMDD), and a document-type carried by the
+    # concrete class (its `type_prefix` returns the printed token, e.g. "WD").
+    class Identifier < ::Pubid::Identifier
+      # Slug, e.g. "charmod", "css-backgrounds-3", "07-WebData".
+      attribute :code, :string
+      # Publication date verbatim, or nil. Overrides the base Components::Date
+      # attribute with a plain string (same mechanism JIS uses for `number`):
+      # W3C dates are opaque digit runs that must round-trip exactly, and are
+      # never re-parsed as calendar dates.
+      attribute :date, :string
+
+      # Polymorphic type map for lutaml::Model key_value (de)serialization: maps
+      # each subclass's polymorphic_name to its class name so a stored hash
+      # rebuilds the correct identifier type via from_hash.
+      W3C_TYPE_MAP = {
+        "pubid:w3c:standard" => "Pubid::W3c::Identifiers::Standard",
+        "pubid:w3c:note" => "Pubid::W3c::Identifiers::Note",
+        "pubid:w3c:draft-note" => "Pubid::W3c::Identifiers::DraftNote",
+        "pubid:w3c:working-draft" => "Pubid::W3c::Identifiers::WorkingDraft",
+        "pubid:w3c:candidate-recommendation" =>
+          "Pubid::W3c::Identifiers::CandidateRecommendation",
+        "pubid:w3c:candidate-recommendation-draft" =>
+          "Pubid::W3c::Identifiers::CandidateRecommendationDraft",
+        "pubid:w3c:recommendation" =>
+          "Pubid::W3c::Identifiers::Recommendation",
+        "pubid:w3c:proposed-recommendation" =>
+          "Pubid::W3c::Identifiers::ProposedRecommendation",
+        "pubid:w3c:proposed-edited-recommendation" =>
+          "Pubid::W3c::Identifiers::ProposedEditedRecommendation",
+        "pubid:w3c:superseded-recommendation" =>
+          "Pubid::W3c::Identifiers::SupersededRecommendation",
+        "pubid:w3c:obsolete-recommendation" =>
+          "Pubid::W3c::Identifiers::ObsoleteRecommendation",
+      }.freeze
+
+      key_value do
+        map "_type", to: :_type, polymorphic_map: W3C_TYPE_MAP
+        map "code", to: :code
+        map "date", to: :date
+      end
+
+      # Publisher is always "W3C". A plain constant (not a `publisher` method)
+      # so it doesn't shadow the inherited lutaml `publisher` attribute, which
+      # would otherwise fail serialization type validation.
+      PUBLISHER = "W3C"
+
+      # The printed document-type token (e.g. "WD"). nil for the bare-code
+      # Standard form; concrete typed subclasses override it.
+      def type_prefix
+        nil
+      end
+
+      attr_reader :with_publisher
+
+      # Publication year derived from an 8-digit YYYYMMDD date, else nil. `date`
+      # is a plain string here (not a Components::Date), so the inherited
+      # `year` — which calls `date&.year` — would break; override it.
+      def year
+        return nil unless date.is_a?(::String) && date.length == 8
+
+        date[0, 4]
+      end
+
+      def to_s(with_publisher: true, **opts)
+        @with_publisher = with_publisher
+        render(format: :human, **opts)
+      end
+
+      # from_hash is the shared polymorphic dispatch on Pubid::Identifier;
+      # W3C_TYPE_MAP remains as the key_value polymorphic_map.
+
+      # Parse a W3C identifier string into an identifier object
+      # @param identifier [String] The W3C identifier string to parse
+      # @return [Identifier] The appropriate identifier object
+      # @raise [RuntimeError] If parsing fails
+      def self.parse(identifier)
+        # Reject pathological inputs before they reach the parser
+        # (CodeQL rb/polynomial-redos barrier — inline .length by design).
+        if identifier.length > Pubid::MAX_INPUT_LENGTH
+          raise ArgumentError, Pubid::INPUT_TOO_LONG_MESSAGE
+        end
+
+        parsed = Parser.parse(identifier)
+        Builder.build(parsed)
+      rescue Parslet::ParseFailed => e
+        raise "Failed to parse W3C identifier '#{identifier}': #{e.message}"
+      end
+    end
+  end
+end

data/lib/pubid/w3c/identifiers/candidate_recommendation.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module Pubid
+  module W3c
+    module Identifiers
+      # W3C Candidate Recommendation. Printed token: "CR".
+      # Example: "W3C CR-exi-20091208".
+      class CandidateRecommendation < Identifier
+        TYPED_STAGES = [
+          Pubid::Components::TypedStage.new(
+            code: :cr,
+            stage_code: :candidate_recommendation,
+            type_code: :cr,
+            abbr: ["CR"],
+            name: "Candidate Recommendation",
+            harmonized_stages: [],
+          ),
+        ].freeze
+
+        def self.type
+          { key: :cr, web: :candidate_recommendation,
+            title: "Candidate Recommendation", short: "CR" }
+        end
+
+        def type_prefix
+          "CR"
+        end
+      end
+    end
+  end
+end

data/lib/pubid/w3c/identifiers/candidate_recommendation_draft.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module Pubid
+  module W3c
+    module Identifiers
+      # W3C Candidate Recommendation Draft. Printed token: "CRD".
+      # Example: "W3C CRD-accelerometer-20250212".
+      class CandidateRecommendationDraft < Identifier
+        TYPED_STAGES = [
+          Pubid::Components::TypedStage.new(
+            code: :crd,
+            stage_code: :candidate_recommendation_draft,
+            type_code: :crd,
+            abbr: ["CRD"],
+            name: "Candidate Recommendation Draft",
+            harmonized_stages: [],
+          ),
+        ].freeze
+
+        def self.type
+          { key: :crd, web: :candidate_recommendation_draft,
+            title: "Candidate Recommendation Draft", short: "CRD" }
+        end
+
+        def type_prefix
+          "CRD"
+        end
+      end
+    end
+  end
+end

data/lib/pubid/w3c/identifiers/draft_note.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module Pubid
+  module W3c
+    module Identifiers
+      # Draft W3C Note. Printed token: "DNOTE".
+      # Example: "W3C DNOTE-webcodecs-flac-codec-registration-20240419".
+      class DraftNote < Identifier
+        TYPED_STAGES = [
+          Pubid::Components::TypedStage.new(
+            code: :dnote,
+            stage_code: :draft_note,
+            type_code: :dnote,
+            abbr: ["DNOTE"],
+            name: "Draft W3C Note",
+            harmonized_stages: [],
+          ),
+        ].freeze
+
+        def self.type
+          { key: :dnote, web: :draft_note, title: "Draft W3C Note",
+            short: "DNOTE" }
+        end
+
+        def type_prefix
+          "DNOTE"
+        end
+      end
+    end
+  end
+end

data/lib/pubid/w3c/identifiers/note.rb

@@ -0,0 +1,30 @@
+# frozen_string_literal: true
+
+module Pubid
+  module W3c
+    module Identifiers
+      # W3C Note (formerly Working Group / Interest Group Note).
+      # Printed token: "NOTE". Example: "W3C NOTE-xml-names".
+      class Note < Identifier
+        TYPED_STAGES = [
+          Pubid::Components::TypedStage.new(
+            code: :note,
+            stage_code: :note,
+            type_code: :note,
+            abbr: ["NOTE"],
+            name: "W3C Note",
+            harmonized_stages: [],
+          ),
+        ].freeze
+
+        def self.type
+          { key: :note, web: :note, title: "W3C Note", short: "NOTE" }
+        end
+
+        def type_prefix
+          "NOTE"
+        end
+      end
+    end
+  end
+end

data/lib/pubid/w3c/identifiers/obsolete_recommendation.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module Pubid
+  module W3c
+    module Identifiers
+      # Obsolete W3C Recommendation. Printed token: "OBSL".
+      # Example: "W3C OBSL-widgets-apis-20181011".
+      class ObsoleteRecommendation < Identifier
+        TYPED_STAGES = [
+          Pubid::Components::TypedStage.new(
+            code: :obsl,
+            stage_code: :obsolete_recommendation,
+            type_code: :obsl,
+            abbr: ["OBSL"],
+            name: "Obsolete Recommendation",
+            harmonized_stages: [],
+          ),
+        ].freeze
+
+        def self.type
+          { key: :obsl, web: :obsolete_recommendation,
+            title: "Obsolete Recommendation", short: "OBSL" }
+        end
+
+        def type_prefix
+          "OBSL"
+        end
+      end
+    end
+  end
+end

data/lib/pubid/w3c/identifiers/proposed_edited_recommendation.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module Pubid
+  module W3c
+    module Identifiers
+      # W3C Proposed Edited Recommendation. Printed token: "PER".
+      # Example: "W3C PER-rif-dtb-20121211".
+      class ProposedEditedRecommendation < Identifier
+        TYPED_STAGES = [
+          Pubid::Components::TypedStage.new(
+            code: :per,
+            stage_code: :proposed_edited_recommendation,
+            type_code: :per,
+            abbr: ["PER"],
+            name: "Proposed Edited Recommendation",
+            harmonized_stages: [],
+          ),
+        ].freeze
+
+        def self.type
+          { key: :per, web: :proposed_edited_recommendation,
+            title: "Proposed Edited Recommendation", short: "PER" }
+        end
+
+        def type_prefix
+          "PER"
+        end
+      end
+    end
+  end
+end

data/lib/pubid/w3c/identifiers/proposed_recommendation.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module Pubid
+  module W3c
+    module Identifiers
+      # W3C Proposed Recommendation. Printed token: "PR".
+      # Example: "W3C PR-InkML-20110510".
+      class ProposedRecommendation < Identifier
+        TYPED_STAGES = [
+          Pubid::Components::TypedStage.new(
+            code: :pr,
+            stage_code: :proposed_recommendation,
+            type_code: :pr,
+            abbr: ["PR"],
+            name: "Proposed Recommendation",
+            harmonized_stages: [],
+          ),
+        ].freeze
+
+        def self.type
+          { key: :pr, web: :proposed_recommendation,
+            title: "Proposed Recommendation", short: "PR" }
+        end
+
+        def type_prefix
+          "PR"
+        end
+      end
+    end
+  end
+end

data/lib/pubid/w3c/identifiers/recommendation.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module Pubid
+  module W3c
+    module Identifiers
+      # W3C Recommendation (a published standard). Printed token: "REC".
+      # Example: "W3C REC-ATAG10-20000203".
+      class Recommendation < Identifier
+        TYPED_STAGES = [
+          Pubid::Components::TypedStage.new(
+            code: :rec,
+            stage_code: :recommendation,
+            type_code: :rec,
+            abbr: ["REC"],
+            name: "Recommendation",
+            harmonized_stages: [],
+          ),
+        ].freeze
+
+        def self.type
+          { key: :rec, web: :recommendation, title: "Recommendation",
+            short: "REC" }
+        end
+
+        def type_prefix
+          "REC"
+        end
+      end
+    end
+  end
+end

data/lib/pubid/w3c/identifiers/standard.rb

@@ -0,0 +1,12 @@
+# frozen_string_literal: true
+
+module Pubid
+  module W3c
+    module Identifiers
+      # Bare-code W3C reference with no maturity token, e.g. "W3C 2dcontext",
+      # "W3C 07-WebData". The default type; carries no `type_prefix`.
+      class Standard < Identifier
+      end
+    end
+  end
+end

data/lib/pubid/w3c/identifiers/superseded_recommendation.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module Pubid
+  module W3c
+    module Identifiers
+      # Superseded W3C Recommendation. Printed token: "SPSD".
+      # Example: "W3C SPSD-2dcontext-20210128".
+      class SupersededRecommendation < Identifier
+        TYPED_STAGES = [
+          Pubid::Components::TypedStage.new(
+            code: :spsd,
+            stage_code: :superseded_recommendation,
+            type_code: :spsd,
+            abbr: ["SPSD"],
+            name: "Superseded Recommendation",
+            harmonized_stages: [],
+          ),
+        ].freeze
+
+        def self.type
+          { key: :spsd, web: :superseded_recommendation,
+            title: "Superseded Recommendation", short: "SPSD" }
+        end
+
+        def type_prefix
+          "SPSD"
+        end
+      end
+    end
+  end
+end

data/lib/pubid/w3c/identifiers/working_draft.rb

@@ -0,0 +1,30 @@
+# frozen_string_literal: true
+
+module Pubid
+  module W3c
+    module Identifiers
+      # W3C Working Draft. Printed token: "WD".
+      # Example: "W3C WD-charmod-19991129".
+      class WorkingDraft < Identifier
+        TYPED_STAGES = [
+          Pubid::Components::TypedStage.new(
+            code: :wd,
+            stage_code: :working_draft,
+            type_code: :wd,
+            abbr: ["WD"],
+            name: "Working Draft",
+            harmonized_stages: [],
+          ),
+        ].freeze
+
+        def self.type
+          { key: :wd, web: :working_draft, title: "Working Draft", short: "WD" }
+        end
+
+        def type_prefix
+          "WD"
+        end
+      end
+    end
+  end
+end

data/lib/pubid/w3c/identifiers.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+module Pubid
+  module W3c
+    module Identifiers
+      autoload :Standard, "#{__dir__}/identifiers/standard"
+      autoload :Note, "#{__dir__}/identifiers/note"
+      autoload :DraftNote, "#{__dir__}/identifiers/draft_note"
+      autoload :WorkingDraft, "#{__dir__}/identifiers/working_draft"
+      autoload :CandidateRecommendation,
+               "#{__dir__}/identifiers/candidate_recommendation"
+      autoload :CandidateRecommendationDraft,
+               "#{__dir__}/identifiers/candidate_recommendation_draft"
+      autoload :Recommendation, "#{__dir__}/identifiers/recommendation"
+      autoload :ProposedRecommendation,
+               "#{__dir__}/identifiers/proposed_recommendation"
+      autoload :ProposedEditedRecommendation,
+               "#{__dir__}/identifiers/proposed_edited_recommendation"
+      autoload :SupersededRecommendation,
+               "#{__dir__}/identifiers/superseded_recommendation"
+      autoload :ObsoleteRecommendation,
+               "#{__dir__}/identifiers/obsolete_recommendation"
+    end
+  end
+end

data/lib/pubid/w3c/parser.rb

@@ -0,0 +1,43 @@
+# frozen_string_literal: true
+
+require "parslet"
+
+module Pubid
+  module W3c
+    # Parslet grammar for printed W3C identifiers, e.g.
+    #   "W3C WD-charmod-19991129"   (type + code + date)
+    #   "W3C NOTE-xml-names"        (type + code)
+    #   "W3C 2dcontext"             (bare code)
+    #
+    # The grammar only splits off the leading maturity token; the code/date
+    # boundary is resolved in the Builder, because a PEG `code` rule would
+    # greedily consume the trailing date and make a `.maybe` date rule
+    # unreliable. This is safe: no real W3C code ends in a same-width (4/6/8)
+    # digit run, so the Builder's fixed-width trailing-date test never collides.
+    class Parser < Parslet::Parser
+      rule(:prefix) { str("W3C") >> str(" ") }
+
+      # Known W3C maturity tokens. Order matters for shared prefixes: CRD before
+      # CR, PER before PR, DNOTE before NOTE.
+      rule(:type_token) do
+        str("DNOTE") | str("NOTE") | str("WD") | str("CRD") | str("CR") |
+          str("REC") | str("PER") | str("PR") | str("SPSD") | str("OBSL")
+      end
+
+      # A leading token is only a type when immediately followed by "-".
+      rule(:type_part) { type_token.as(:type) >> str("-") }
+
+      # The remainder: code plus an optional trailing date. Captured whole and
+      # split in the Builder.
+      rule(:rest) { any.repeat(1).as(:rest) }
+
+      rule(:identifier) { prefix >> type_part.maybe >> rest }
+
+      rule(:root) { identifier }
+
+      def self.parse(input)
+        new.parse(input)
+      end
+    end
+  end
+end

data/lib/pubid/w3c/renderer.rb

@@ -0,0 +1,35 @@
+# frozen_string_literal: true
+
+module Pubid
+  module W3c
+    # Human-readable renderer for W3C identifiers.
+    #
+    # Produces strings like:
+    #   "W3C WD-charmod-19991129"
+    #   "W3C NOTE-xml-names"
+    #   "W3C 2dcontext"
+    #
+    # Registered as the :human format in the W3C format registry and invoked via
+    # `render(format: :human)`. When `id.with_publisher == false` the leading
+    # "W3C " publisher token is dropped.
+    class Renderer < ::Pubid::Renderers::Base
+      PUBLISHER = "W3C"
+
+      def render(context: nil, **opts)
+        id = @id
+        body = +""
+        body << "#{id.type_prefix}-" if id.type_prefix
+        body << id.code.to_s
+        body << "-#{id.date}" if id.date
+
+        with_publisher?(id) ? "#{PUBLISHER} #{body}" : body
+      end
+
+      private
+
+      def with_publisher?(id)
+        id.with_publisher != false
+      end
+    end
+  end
+end

data/lib/pubid/w3c/urn_generator.rb

@@ -0,0 +1,18 @@
+# frozen_string_literal: true
+
+module Pubid
+  module W3c
+    # Emits `urn:w3c:<type-down>:<code>[:<date>]`, e.g.
+    #   "W3C WD-charmod-19991129" -> "urn:w3c:wd:charmod:19991129"
+    #   "W3C 2dcontext"           -> "urn:w3c:2dcontext"
+    class UrnGenerator < Pubid::UrnGenerator::Base
+      def generate
+        parts = ["urn", "w3c"]
+        parts << identifier.type_prefix.downcase if identifier.type_prefix
+        parts << identifier.code.to_s
+        parts << identifier.date.to_s if identifier.date
+        parts.join(":")
+      end
+    end
+  end
+end