pubid 2.0.0.alpha.10 → 2.0.0.pre.alpha.1

data/lib/pubid/iala/identifier.rb

@@ -1,90 +0,0 @@
-# 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

@@ -1,15 +0,0 @@
-# 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

@@ -1,57 +0,0 @@
-# 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

@@ -1,19 +0,0 @@
-# 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

@@ -1,17 +0,0 @@
-# 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

@@ -1,16 +0,0 @@
-# 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

@@ -1,18 +0,0 @@
-# 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

@@ -1,17 +0,0 @@
-# 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

@@ -1,17 +0,0 @@
-# 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

@@ -1,16 +0,0 @@
-# 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

@@ -1,16 +0,0 @@
-# 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

@@ -1,17 +0,0 @@
-# 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

@@ -1,19 +0,0 @@
-# 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

data/lib/pubid/iala/parser.rb

@@ -1,138 +0,0 @@
-# frozen_string_literal: true
-
-require "parslet"
-
-module Pubid
-  module Iala
-    # Parslet grammar for IALA identifiers.
-    #
-    # Accepts every shape that appears on the IALA website or cover page:
-    #
-    #   S1070
-    #   S1070 Ed 2.0
-    #   IALA S1070 Ed 2.0
-    #   S1070 Ed 2.0 (F)
-    #   R1016:ed2.0(F)
-    #   C0103-1
-    #   C0103-1 Ed 3.0
-    class Parser < ::Pubid::Parser::Grammar
-      include ::Pubid::Parser::CommonParseRules
-
-      root :identifier
-
-      rule(:space)  { str(" ") }
-      rule(:space?) { space.maybe }
-      rule(:dash)   { str("-") }
-      rule(:colon)  { str(":") }
-      rule(:dot)    { str(".") }
-      rule(:lparen) { str("(") }
-      rule(:rparen) { str(")") }
-
-      # Type prefix — IALA document classes. GA must be tried before G so it
-      # isn't left with a dangling "A". Captured to route to the right
-      # Identifiers::* subclass.
-      rule(:type_letter) do
-        (str("GA") | str("S") | str("R") | str("G") | str("M") |
-         str("C") | str("A") | str("L") | str("X") | str("P")).as(:type_letter)
-      end
-
-      # Document number — one or more decimal digits. The dataset historically
-      # zero-pads to 4 (S1070, R0126, M0001) but the canonical form going
-      # forward is unpadded (M1, GA1.1). Accept any length so both shapes
-      # round-trip; the number attribute preserves whatever was parsed.
-      rule(:doc_number) do
-        match("[0-9]").repeat(1).as(:doc_number)
-      end
-
-      # Dotted continuation that some prefixes use as part of the number:
-      # GA01.01 (series.index), L2.1.11 (series.sub.item). Captured into the
-      # same :doc_number atom so the number attribute stays a single string
-      # (e.g. "01.01", "2.1.11") and the existing builder needs no changes.
-      rule(:dotted_number) do
-        dot >> match("[0-9]").repeat(1)
-      end
-
-      rule(:number_with_dots) do
-        doc_number >> dotted_number.repeat.as(:doc_number_dots)
-      end
-
-      # Numeric sub-part suffix(es): "-1", "-9-10", "-11". The catalogue
-      # uses ranges like R0124-9-10 & 11 — that compound is captured
-      # verbatim and resolved by the caller, not here.
-      rule(:subpart) do
-        (dash >> digits.as(:subpart_number)).repeat(1).as(:subpart)
-      end
-
-      rule(:code) do
-        type_letter >> number_with_dots >> subpart.maybe
-      end
-
-      # Edition forms observed on covers and listing pages:
-      #   " Ed 2.0"      — cover-page human form
-      #   ":ed2.0"       — listing-page compact form
-      rule(:edition_human) do
-        (space >> str("Ed") >> space >> edition_value).as(:edition)
-      end
-
-      rule(:edition_compact) do
-        (colon >> str("ed") >> edition_value).as(:edition)
-      end
-
-      rule(:edition_value) do
-        (digits >> (dot >> digits).repeat).as(:edition_value)
-      end
-
-      rule(:edition) do
-        edition_human | edition_compact
-      end
-
-      # Language is a single uppercase letter inside parens: (E), (F), …
-      rule(:language) do
-        (space? >> lparen >>
-          (str("E") | str("F") | str("S") | str("C") | str("A") | str("R")).as(:language) >>
-          rparen).as(:language_group)
-      end
-
-      # Optional "IALA " prefix.
-      rule(:publisher) do
-        (str("IALA") >> space).maybe
-      end
-
-      rule(:identifier) do
-        annex_identifier | (publisher >> code >> edition.maybe >> language.maybe)
-      end
-
-      # Annex to a base publication. Wraps the base code and adds an
-      # "Annex" / "ANNEX" marker (case preserved verbatim), an optional
-      # single-letter suffix (A/B/C/D), then the usual edition + language.
-      # Tried before the flat `identifier` so the annex marker isn't left
-      # dangling after a successful base-code match.
-      #
-      # The letter suffix uses a negative lookahead for a following lowercase
-      # letter so the "E" in "Ed 1" isn't mis-captured as annex letter "E".
-      rule(:annex_identifier) do
-        (publisher >> type_letter >> number_with_dots >> subpart.maybe).as(:base) >>
-          space >> (str("Annex") | str("ANNEX")).as(:annex_marker) >>
-          (space >> match("[A-Z]").as(:annex_letter) >> match("[a-z]").absent?).maybe >>
-          edition.maybe >> language.maybe
-      end
-
-      # Parse a string and return the raw parslet tree.
-      # @param string [String]
-      # @return [Hash]
-      def self.parse(string)
-        unless string.is_a?(String)
-          raise ::Pubid::Errors::InvalidInputError,
-                ::Pubid::INPUT_NOT_A_STRING_MESSAGE
-        end
-
-        if string.length > ::Pubid::MAX_INPUT_LENGTH
-          raise ::Pubid::Errors::InvalidInputError,
-                ::Pubid::INPUT_TOO_LONG_MESSAGE
-        end
-
-        new.parse(string.strip)
-      end
-    end
-  end
-end

data/lib/pubid/iala/renderer.rb

@@ -1,36 +0,0 @@
-# frozen_string_literal: true
-
-module Pubid
-  module Iala
-    # Human-readable renderer for IALA identifiers.
-    #
-    # Produces strings like:
-    #   "IALA S1070"
-    #   "IALA S1070 Ed 2.0"
-    #   "IALA R1016 Ed 2.0 (F)"
-    class Renderer < ::Pubid::Renderers::Base
-      def render(context: nil, **_opts)
-        id = @id
-
-        return render_annex(id) if id.is_a?(Identifiers::Annex)
-
-        rendered = "#{id.publisher} #{id.type_letter}#{id.number}"
-        rendered << " Ed #{id.edition}" if id.edition
-        rendered << " (#{id.language})" if id.language
-        rendered
-      end
-
-      # Render an Annex wrapper. The base identifier is rendered first, then
-      # the marker form is preserved verbatim ("Annex" vs "ANNEX"), then the
-      # optional letter, edition, and language.
-      def render_annex(id)
-        base_str = id.base.to_s
-        rendered = "#{base_str} #{id.annex_form}"
-        rendered << " #{id.letter}" if id.letter
-        rendered << " Ed #{id.edition}" if id.edition
-        rendered << " (#{id.language})" if id.language
-        rendered
-      end
-    end
-  end
-end

data/lib/pubid/iala/urn_generator.rb

@@ -1,49 +0,0 @@
-# frozen_string_literal: true
-
-module Pubid
-  module Iala
-    # Generates URN strings for IALA identifiers using the Maritime Resource
-    # Name (MRN) namespace that IALA prints on its cover pages.
-    #
-    # Format: urn:mrn:iala:pub:<type-lower><number>[:ed<edition>][:<lang-lower>]
-    # Example: urn:mrn:iala:pub:s1070:ed2.0
-    #          urn:mrn:iala:pub:r1016:ed2.0:f
-    class UrnGenerator
-      attr_reader :identifier
-
-      def initialize(identifier)
-        @identifier = identifier
-      end
-
-      def generate
-        return generate_annex if identifier.is_a?(Identifiers::Annex)
-
-        parts = ["urn:mrn:iala:pub", code_segment]
-        parts << "ed#{identifier.edition}" if identifier.edition
-        parts << identifier.language.downcase if identifier.language
-        parts.join(":")
-      end
-
-      private
-
-      # URN for an Annex: embeds the base's code, then an "annex" segment
-      # (with optional letter suffix), then the annex's own edition/lang.
-      #   G1045 Annex Ed 1           → urn:mrn:iala:pub:g1045:annex:ed1
-      #   G1128 ANNEX A Ed 1.6 (E)    → urn:mrn:iala:pub:g1128:annex-a:ed1.6:e
-      def generate_annex
-        base = identifier.base
-        parts = ["urn:mrn:iala:pub",
-                 "#{base.type_letter.downcase}#{base.number}"]
-        annex_seg = identifier.letter ? "annex-#{identifier.letter.downcase}" : "annex"
-        parts << annex_seg
-        parts << "ed#{identifier.edition}" if identifier.edition
-        parts << identifier.language.downcase if identifier.language
-        parts.join(":")
-      end
-
-      def code_segment
-        "#{identifier.type_letter.downcase}#{identifier.number}"
-      end
-    end
-  end
-end

data/lib/pubid/iala/urn_parser.rb

@@ -1,67 +0,0 @@
-# frozen_string_literal: true
-
-module Pubid
-  module Iala
-    # Parses IALA MRN URNs back into identifiers.
-    #
-    # UrnGenerator emits:
-    #   urn:mrn:iala:pub:<type-lower><number>[:ed<edition>][:<lang-lower>]
-    # and for Annex identifiers:
-    #   urn:mrn:iala:pub:<base-code>:annex[-<letter>][:ed<edition>][:<lang>]
-    #
-    # Examples:
-    #   urn:mrn:iala:pub:s1070:ed2.0          → IALA S1070 Ed 2.0
-    #   urn:mrn:iala:pub:r1016:ed2.0:f        → IALA R1016 Ed 2.0 (F)
-    #   urn:mrn:iala:pub:c0103-1              → IALA C0103-1
-    #   urn:mrn:iala:pub:g1045:annex:ed1      → IALA G1045 Annex Ed 1
-    #   urn:mrn:iala:pub:g1128:annex-a:ed1.6  → IALA G1128 ANNEX A Ed 1.6
-    class UrnParser < Pubid::UrnParser::Base
-      PREFIX = "urn:mrn:iala:pub:".freeze
-
-      def parse_urn(urn)
-        body = strip_namespace(urn)
-        parts = split_parts(body)
-        code = parts.fetch(0)
-
-        edition = nil
-        language = nil
-        annex_form = nil
-        annex_letter = nil
-
-        parts.drop(1).each do |seg|
-          case seg
-          when /\Aannex(-([a-z]))?\z/i
-            # UrnGenerator emits "annex" (bare) for the Annex-without-letter
-            # form and "annex-X" (lettered) for the uppercase ANNEX form.
-            # The human parser accepts both "Annex" and "ANNEX"; mirror the
-            # generator's casing so URN round-trip is exact.
-            annex_letter = Regexp.last_match(2)&.upcase
-            annex_form = annex_letter ? "ANNEX" : "Annex"
-          when /\Aed/
-            edition = seg.sub(/\Aed/, "")
-          when /\A[a-z]\z/i
-            language = seg.upcase
-          end
-        end
-
-        text = "IALA #{code.upcase}"
-        text += " #{annex_form}" if annex_form
-        text += " #{annex_letter}" if annex_letter
-        text += " Ed #{edition}" if edition
-        text += " (#{language})" if language
-        flavor_parse(text)
-      end
-
-      private
-
-      def strip_namespace(urn)
-        unless urn.downcase.start_with?(PREFIX)
-          raise Pubid::UrnParser::Errors::ParseError,
-                "Invalid IALA URN: #{urn.inspect}"
-        end
-
-        urn[PREFIX.length..]
-      end
-    end
-  end
-end