pubid 1.15.21 → 2.0.0.alpha.10

data/lib/pubid/gost/urn_parser.rb

@@ -0,0 +1,52 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Gost
+    # Parses GOST URNs back into identifiers.
+    #
+    # UrnGenerator emits:
+    #   urn:gost:std:<number>[:<year>]              (interstate)
+    #   urn:gost:std:r:<number>[:<year>]            (Russian national)
+    #
+    # Examples:
+    #   urn:gost:std:14946:82           → GOST 14946-82
+    #   urn:gost:std:r:34.12:2015       → GOST R 34.12-2015
+    class UrnParser < Pubid::UrnParser::Base
+      PREFIX = "urn:gost:std:".freeze
+
+      def parse_urn(urn)
+        body = strip_namespace(urn)
+        parts = split_parts(body)
+
+        if parts.first == "r"
+          number = parts[1]
+          year = parts[2]
+          Identifiers::NationalStandard.new(number: number, year: year)
+        else
+          number = parts[0]
+          year = parts[1]
+          Identifiers::InterstateStandard.new(number: number, year: year)
+        end
+      end
+
+      private
+
+      def strip_namespace(urn)
+        unless urn.downcase.start_with?(PREFIX)
+          raise Pubid::UrnParser::Errors::ParseError,
+                "Invalid GOST URN: #{urn.inspect}"
+        end
+
+        urn[PREFIX.length..]
+      end
+
+      def split_parts(body)
+        parts = body.split(":")
+        raise Pubid::UrnParser::Errors::ParseError,
+              "GOST URN missing number: #{body.inspect}" if parts.empty?
+
+        parts
+      end
+    end
+  end
+end

data/lib/pubid/gost.rb

@@ -0,0 +1,49 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Gost
+    extend Pubid::PrefixesSupport
+
+    # Leading identifier tokens that a GOST document id can start with.
+    # Latin "GOST" and Cyrillic "ГОСТ" are both official; "R" (Russian
+    # national, e.g. "GOST R 34.12-2015") routes to this flavor too.
+    PREFIXES = ["GOST", "ГОСТ"].freeze
+
+    autoload :Builder,      "#{__dir__}/gost/builder"
+    autoload :Identifier,   "#{__dir__}/gost/identifier"
+    autoload :Identifiers,  "#{__dir__}/gost/identifiers"
+    autoload :Parser,       "#{__dir__}/gost/parser"
+    autoload :Renderer,     "#{__dir__}/gost/renderer"
+    autoload :UrnGenerator, "#{__dir__}/gost/urn_generator"
+    autoload :UrnParser,    "#{__dir__}/gost/urn_parser"
+
+    # Parse a GOST identifier string into an identifier object.
+    # @param identifier [String]
+    # @return [Pubid::Gost::Identifier]
+    def self.parse(identifier)
+      Identifier.parse(identifier)
+    end
+
+    # Per-flavor format registry: inherits global formats, overrides :human.
+    Identifier.format_registry = FormatRegistry.new(parent: ::Pubid::Identifier.format_registry)
+    Identifier.format_registry.register(:human, renderer: Gost::Renderer)
+
+    # Auto-discover identifier types from the Identifiers namespace.
+    # @return [Array<Class>]
+    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 < Pubid::Identifier }
+        .reject { |c| c == Identifier }
+    end
+
+    # Lookup: type key → identifier class
+    # @param code [String, Symbol]
+    # @return [Class, nil]
+    def self.locate_type(code)
+      identifier_types.find { |t| t.type[:key].to_s == code.to_s }
+    end
+  end
+end
+
+Pubid::Registry.register(:gost, Pubid::Gost)

data/lib/pubid/iala/builder.rb

@@ -0,0 +1,102 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Iala
+    # Builds an IALA identifier object from the Parslet parse tree.
+    class Builder
+      def build(parsed)
+        hash = parsed
+
+        return build_annex(hash) if hash[:annex_marker]
+
+        type_letter = stringify(hash[:type_letter])&.upcase
+        klass = Iala.identifier_klass_for_type_letter(type_letter)
+
+        attrs = {
+          number:   build_number(hash, klass),
+          edition:  stringify(hash.dig(:edition, :edition_value)),
+          language: stringify(hash.dig(:language_group, :language))&.upcase,
+        }.compact
+
+        klass.new(**attrs)
+      end
+
+      def self.build(parsed)
+        new.build(parsed)
+      end
+
+      private
+
+      # Build the Annex wrapper from the parse tree. The base sub-tree
+      # (captured under :base) is recursively built via the same logic as
+      # a flat identifier, then wrapped.
+      def build_annex(hash)
+        base_hash = hash[:base]
+        base_type = stringify(base_hash[:type_letter])&.upcase
+        base_klass = Iala.identifier_klass_for_type_letter(base_type)
+        base = base_klass.new(number: build_number(base_hash, base_klass))
+
+        attrs = {
+          base: base,
+          annex_form:      stringify(hash[:annex_marker]),
+          letter:          stringify(hash[:annex_letter]),
+          edition:         stringify(hash.dig(:edition, :edition_value)),
+          language:        stringify(hash.dig(:language_group, :language))&.upcase,
+        }.compact
+
+        Identifiers::Annex.new(**attrs)
+      end
+
+      # Compose the full number string from the leading doc_number plus any
+      # dotted continuation (GA01.13 → "01.13", L2.1.11 → "2.1.11") plus
+      # the optional dash-separated subpart (C0103-1 → "0103-1", L2.7.1-2
+      # → "2.7.1-2"). Padding is class-driven: the resolved identifier
+      # class declares its canonical IALA cover-page widths via
+      # `canonical_number_width` / `canonical_dotted_segment_width`, so a
+      # new type declares its padding on its own class instead of editing
+      # this builder's case statement.
+      def build_number(hash, klass)
+        base = stringify(hash[:doc_number])
+        return base unless base
+
+        base = canonical_base(base, klass)
+
+        dots = hash[:doc_number_dots]
+        dots_str = dots.is_a?(Array) ? "" : canonical_dots(stringify(dots).to_s, klass)
+
+        subpart_str = hash[:subpart] ? "-#{subpart_to_s(hash[:subpart])}" : ""
+        subpart_str = "" if subpart_str == "-"
+        "#{base}#{dots_str}#{subpart_str}"
+      end
+
+      def canonical_base(base, klass)
+        width = klass&.canonical_number_width
+        width ? base.rjust(width, "0") : base
+      end
+
+      def canonical_dots(str, klass)
+        return "" if str.empty?
+
+        width = klass&.canonical_dotted_segment_width
+        return str unless width
+
+        str.split(".", -1).map { |seg| seg.empty? ? seg : seg.rjust(width, "0") }.join(".")
+      end
+
+      # The subpart capture may be an array of {subpart_number: "1"},
+      # {subpart_number: "9"}, … (repeated matches). Join with "-".
+      def subpart_to_s(subpart)
+        return stringify(subpart).to_s unless subpart.is_a?(Array)
+
+        subpart.filter_map { |h| h.is_a?(Hash) ? stringify(h[:subpart_number]) : nil }.join("-")
+      end
+
+      def stringify(value)
+        return nil if value.nil?
+
+        str = value.to_s
+        str.empty? ? nil : str
+      end
+    end
+  end
+end

data/lib/pubid/iala/identifier.rb

@@ -0,0 +1,90 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Iala
+    # Base class for all IALA identifiers. Canonical name
+    # Pubid::Iala::Identifier; every concrete IALA identifier
+    # (Identifiers::*) descends from it.
+    #
+    # IALA identifiers have the form:
+    #   [IALA ]{S|R|G|M|C|X|P}<4-digit number>[-<subpart>][ Ed <edition>][ (<LangLetter>)]
+    #
+    # Examples:
+    #   S1070
+    #   IALA S1070 Ed 2.0
+    #   R1016:ed2.0(F)
+    #   C0103-1 Ed 3.0
+    class Identifier < ::Pubid::Identifier
+      # Parse an IALA identifier string into an identifier object.
+      # @param identifier [String]
+      # @return [Pubid::Iala::Identifier]
+      # @raise [Pubid::Errors::ParseError] If parsing fails
+      def self.parse(identifier)
+        parsed = Parser.parse(identifier)
+        Builder.build(parsed)
+      end
+
+      attribute :publisher, :string, default: "IALA"
+      attribute :number,    :string    # "1070", "0126", "0103-1"
+      attribute :edition,   :string    # "2.0", "1.3", nil
+      attribute :language,  :string    # "E", "F", "S", "C", "A", "R", nil
+
+      IALA_TYPE_MAP = {
+        "pubid:iala:standard"          => "Pubid::Iala::Identifiers::Standard",
+        "pubid:iala:recommendation"    => "Pubid::Iala::Identifiers::Recommendation",
+        "pubid:iala:guideline"         => "Pubid::Iala::Identifiers::Guideline",
+        "pubid:iala:manual"            => "Pubid::Iala::Identifiers::Manual",
+        "pubid:iala:model-course"      => "Pubid::Iala::Identifiers::ModelCourse",
+        "pubid:iala:advice"            => "Pubid::Iala::Identifiers::Advice",
+        "pubid:iala:general-assembly"  => "Pubid::Iala::Identifiers::GeneralAssembly",
+        "pubid:iala:letter"            => "Pubid::Iala::Identifiers::Letter",
+        "pubid:iala:annex"             => "Pubid::Iala::Identifiers::Annex",
+        "pubid:iala:report"            => "Pubid::Iala::Identifiers::Report",
+        "pubid:iala:resolution"        => "Pubid::Iala::Identifiers::Resolution",
+      }.freeze
+
+      key_value do
+        map "_type",    to: :_type, polymorphic_map: IALA_TYPE_MAP
+        map "number",   to: :number
+        map "edition",  to: :edition
+        map "language", to: :language
+      end
+
+      def to_urn
+        UrnGenerator.new(self).generate
+      end
+
+      # Single-letter type code (S, R, G, …) — convenience accessor used
+      # by Renderer and UrnGenerator.
+      def type_letter
+        self.class.type[:short]
+      end
+
+      class << self
+        # Zero-pad the leading numeric run of the document number to this
+        # width. nil means preserve the input verbatim. Subclasses declare
+        # their canonical IALA cover-page form via `number_width` (e.g.
+        # Standard declares 4 → M0001; GeneralAssembly declares 2 → GA01).
+        # The builder consults this instead of branching on type_letter.
+        attr_reader :canonical_number_width
+
+        # Zero-pad each dotted continuation segment of the number to this
+        # width. nil means preserve verbatim. Only GeneralAssembly (GA)
+        # sets this today — its series.index form (01.01, 01.13) is the
+        # canonical cover-page shape.
+        attr_reader :canonical_dotted_segment_width
+
+        # DSL: declare the canonical zero-pad width for the leading number.
+        private
+
+        def number_width(width)
+          @canonical_number_width = width
+        end
+
+        def dotted_segment_width(width)
+          @canonical_dotted_segment_width = width
+        end
+      end
+    end
+  end
+end

data/lib/pubid/iala/identifiers/advice.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Iala
+    module Identifiers
+      # IALA Advices (A prefix). Numbering is dashed: A12-01, A13-04.
+      # Carries no edition in the corpus.
+      class Advice < Identifier
+        def self.type
+          { key: :advice, title: "Advice", short: "A" }
+        end
+      end
+    end
+  end
+end

data/lib/pubid/iala/identifiers/annex.rb

@@ -0,0 +1,57 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Iala
+    module Identifiers
+      # IALA Annex to a base publication. Two surface forms appear in the
+      # corpus, both wrapping a Guideline:
+      #
+      #   "G1045 Annex Ed 1"           — bare Annex, no letter
+      #   "G1128 ANNEX A Ed 1.6"        — uppercase + letter (A/B/C/D)
+      #
+      # The case of the marker is preserved (Annex vs ANNEX) for round-trip
+      # fidelity. Edition and language apply to the Annex itself, not the base.
+      class Annex < Identifier
+        # The publication this annex supplements. Always present; carries the
+        # type letter and number (e.g. G1045).
+        attribute :base, ::Pubid::Iala::Identifier, polymorphic: true
+        # "Annex" or "ANNEX" — preserved verbatim for exact round-trip.
+        attribute :annex_form, :string
+        # "A", "B", "C", "D", or nil for the bare form.
+        attribute :letter, :string
+
+        key_value do
+          map "base",
+              with: { to: :base_to_kv, 
+                      from: :base_from_kv }
+          map "annex_form", to: :annex_form
+          map "letter",     to: :letter
+        end
+
+        def self.type
+          { key: :annex, title: "Annex", short: nil }
+        end
+
+        # Public because lutaml-model invokes the `to:` / `from:` mapping
+        # methods via public_send during (de)serialization — a `private`
+        # section here raises NoMethodError on to_hash (matches the public
+        # visibility of Oiml::SupplementIdentifier's identical helpers).
+        def base_to_kv(model, doc)
+          base = model.base
+          return unless base
+
+          doc.add_child(
+            Lutaml::KeyValue::DataModel::Element.new("base",
+                                                     base.to_hash),
+          )
+        end
+
+        def base_from_kv(model, value)
+          return unless value
+
+          model.base = ::Pubid::Iala::Identifier.from_hash(value)
+        end
+      end
+    end
+  end
+end

data/lib/pubid/iala/identifiers/general_assembly.rb

@@ -0,0 +1,19 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Iala
+    module Identifiers
+      # IALA General Assembly resolutions (GA prefix). Numbering is dotted
+      # with a series and an index: GA01.01, GA01.13. No edition.
+      class GeneralAssembly < Identifier
+        number_width 2
+        dotted_segment_width 2
+
+        def self.type
+          { key: :"general-assembly", title: "General Assembly Resolution",
+            short: "GA" }
+        end
+      end
+    end
+  end
+end

data/lib/pubid/iala/identifiers/guideline.rb

@@ -0,0 +1,17 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Iala
+    module Identifiers
+      # IALA Guidelines (G series).
+      # Examples: G1015 Ed 2.2, G1078 Ed 2.1.
+      class Guideline < Identifier
+        number_width 4
+
+        def self.type
+          { key: :guideline, title: "Guideline", short: "G" }
+        end
+      end
+    end
+  end
+end

data/lib/pubid/iala/identifiers/letter.rb

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Iala
+    module Identifiers
+      # IALA Letters (L prefix). Numbering is dotted with a series, sub-series,
+      # and item: L2.1.11, L2.7.1-2 (the trailing -2 is a sub-part on top of
+      # the dotted number).
+      class Letter < Identifier
+        def self.type
+          { key: :letter, title: "Letter", short: "L" }
+        end
+      end
+    end
+  end
+end

data/lib/pubid/iala/identifiers/manual.rb

@@ -0,0 +1,18 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Iala
+    module Identifiers
+      # IALA Manuals (M prefix, planned).
+      # Today manuals carry no code (NAVGUIDE, VTS Manual, …); M is
+      # reserved for future numbered manuals.
+      class Manual < Identifier
+        number_width 4
+
+        def self.type
+          { key: :manual, title: "Manual", short: "M" }
+        end
+      end
+    end
+  end
+end

data/lib/pubid/iala/identifiers/model_course.rb

@@ -0,0 +1,17 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Iala
+    module Identifiers
+      # IALA Model Courses (C series).
+      # Examples: C0103-1 Ed 3.0, C2001-1 Ed 1.0.
+      class ModelCourse < Identifier
+        number_width 4
+
+        def self.type
+          { key: :"model-course", title: "Model Course", short: "C" }
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,17 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Iala
+    module Identifiers
+      # IALA Recommendations (R series).
+      # Examples: R0103 Ed 3.1, R0126 Ed 2.0, R1024 Ed 1.0.
+      class Recommendation < Identifier
+        number_width 4
+
+        def self.type
+          { key: :recommendation, title: "Recommendation", short: "R" }
+        end
+      end
+    end
+  end
+end

data/lib/pubid/iala/identifiers/report.rb

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Iala
+    module Identifiers
+      # IALA Reports & Proceedings (X prefix, planned).
+      # Today reports carry no code; X is reserved for future numbered
+      # reports.
+      class Report < Identifier
+        def self.type
+          { key: :report, title: "Report", short: "X" }
+        end
+      end
+    end
+  end
+end

data/lib/pubid/iala/identifiers/resolution.rb

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Iala
+    module Identifiers
+      # IALA Resolutions and other Council publications (P prefix,
+      # planned). Today resolutions carry no code; P is reserved for
+      # future numbered resolutions.
+      class Resolution < Identifier
+        def self.type
+          { key: :resolution, title: "Resolution", short: "P" }
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,17 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Iala
+    module Identifiers
+      # IALA Standards (S series).
+      # Examples: S1010, S1020 Ed 2.0, S1070 Ed 2.0.
+      class Standard < Identifier
+        number_width 4
+
+        def self.type
+          { key: :standard, title: "Standard", short: "S" }
+        end
+      end
+    end
+  end
+end

data/lib/pubid/iala/identifiers.rb

@@ -0,0 +1,19 @@
+# frozen_string_literal: true
+
+module Pubid
+  module Iala
+    module Identifiers
+      autoload :Advice,           "#{__dir__}/identifiers/advice"
+      autoload :Annex,            "#{__dir__}/identifiers/annex"
+      autoload :Standard,         "#{__dir__}/identifiers/standard"
+      autoload :Recommendation,   "#{__dir__}/identifiers/recommendation"
+      autoload :Guideline,        "#{__dir__}/identifiers/guideline"
+      autoload :Manual,           "#{__dir__}/identifiers/manual"
+      autoload :ModelCourse,      "#{__dir__}/identifiers/model_course"
+      autoload :GeneralAssembly,  "#{__dir__}/identifiers/general_assembly"
+      autoload :Letter,           "#{__dir__}/identifiers/letter"
+      autoload :Report,           "#{__dir__}/identifiers/report"
+      autoload :Resolution,       "#{__dir__}/identifiers/resolution"
+    end
+  end
+end