relaton 2.2.0.pre.alpha.1 → 3.0.0.pre.alpha.1

data/lib/relaton/iec/hit_collection.rb

@@ -0,0 +1,138 @@
+# frozen_string_literal: true
+
+require "addressable/uri"
+require_relative "hit"
+
+module Relaton
+  module Iec
+    # Page of hit collection.
+    class HitCollection < Core::HitCollection
+      def_delegators :@array, :detect, :last, :max_by, :sort_by
+
+      # @param exclude [Array<Symbol>] keys to exclude from comparison (e.g. :year, :part, :type)
+      def search(exclude: [:year])
+        @array = fetch_from_index exclude
+        self
+      end
+
+      # @return [Relaton::Iec::ItemData, nil]
+      def to_all_parts(r_year, opts = {}) # rubocop:disable Metrics/AbcSize,Metrics/MethodLength,Metrics/CyclomaticComplexity,Metrics/PerceivedComplexity
+        parts = @array.select { |h| h.part && (!r_year || h.hit[:id]&.date&.year&.to_s == r_year) }
+        if opts[:publication_date_before] || opts[:publication_date_after]
+          parts = parts.select { |h| Bibliography.send(:year_in_range?, h.hit[:id].date&.year.to_i, opts) }
+        end
+        hit = parts.min_by { |h| h.part.to_i }
+        return @array.first&.item unless hit
+
+        bibitem = hit.item
+        all_parts_item = bibitem.to_all_parts
+        others = parts.reject { |h| h.hit[:id] == hit.hit[:id] }
+        others.sort_by { |h| part_sort_key(h.hit[:id]) }.each do |hi|
+          code = hi.hit[:id].to_s
+          bib = ItemData.new(
+            formattedref: Bib::Formattedref.new(content: code),
+            docidentifier: [Docidentifier.new(content: code, type: "IEC", primary: true)],
+          )
+          all_parts_item.relation << Relation.new(type: "partOf", bibitem: bib)
+        end
+        all_parts_item
+      end
+
+      private
+
+      # Ascending sort key for part ordering, e.g. 61326-2-1 < 61326-2-6.
+      # part/subpart are pubid Code components; coerce via to_s, missing → 0.
+      def part_sort_key(pubid)
+        sub = pubid.respond_to?(:subpart) ? pubid.subpart : nil
+        [pubid.part.to_s.to_i, sub.to_s.to_i]
+      end
+
+      def index
+        @index ||= Relaton::Index.find_or_create(
+          :IEC,
+          url: "#{Hit::GHURL}#{INDEXFILE}.zip",
+          file: "#{INDEXFILE}.yaml",
+          pubid_class: ::Pubid::Iec::Identifier
+        )
+      end
+
+      # Returns array of integers for sorting compound parts like "2-1", "2-6"
+      # @param part [String, nil] part string e.g. "1", "2-1", "2-6"
+      # @return [Array<Integer>] e.g. [2, 1] for "2-1", [6] for "6"
+      def part_sort_key(part)
+        return [] unless part
+
+        part.to_s.split("-").map(&:to_i)
+      end
+
+      # Keys pubid re-emits at a default value after `from_hash` but omits on a
+      # fresh `parse` (a lutaml render_default asymmetry: a deserialized
+      # attribute loses its using_default flag, so render_default: false no
+      # longer suppresses it). Equality must ignore them so a deserialized index
+      # row compares equal to the freshly-parsed query. Values are the
+      # stringified defaults (see #stringify).
+      LEAKING_DEFAULTS = { "publisher" => "IEC", "all_parts" => "false", "database" => "false" }.freeze
+
+      # Map a flat exclude symbol to the lean to_hash key it removes (at every
+      # nesting level). :type removes the polymorphic `_type` discriminator.
+      EXCLUDE_KEYS = { year: "year", part: "part", subpart: "subpart", type: "_type" }.freeze
+
+      # Compare pubids for matching, excluding specified fields.
+      # @param row_pubid [Pubid::Iec::Identifier] pubid from index row
+      # @return [Boolean]
+      def pubid_matches?(row_pubid, exclude)
+        return false unless row_pubid
+
+        canonical_id(@ref, exclude) == canonical_id(row_pubid, exclude)
+      end
+
+      # Build-path-independent comparison key: the lean `to_hash` with keys and
+      # scalars stringified, leaking defaults dropped, and excluded fields
+      # removed at every nesting level. More robust than comparing attribute
+      # objects, whose derived `type`/`stage` differ between a parsed identifier
+      # (component object) and a deserialized one (nil/symbol) — at the top level
+      # and inside `base_identifier`.
+      def canonical_id(pubid, exclude)
+        drop = exclude.filter_map { |e| EXCLUDE_KEYS[e] }
+        prune(stringify(pubid.to_hash), drop)
+      end
+
+      # Stringify hash keys and scalar values so comparison ignores YAML scalar
+      # typing (1 vs "1") and string/symbol key differences.
+      def stringify(value)
+        case value
+        when Hash  then value.to_h { |k, v| [k.to_s, stringify(v)] }
+        when Array then value.map { |v| stringify(v) }
+        when nil   then nil
+        else value.to_s
+        end
+      end
+
+      # Recursively drop nil values, leaking defaults, and excluded keys.
+      def prune(value, drop)
+        case value
+        when Hash
+          value.each_with_object({}) do |(k, v), h|
+            next if v.nil? || drop.include?(k) || LEAKING_DEFAULTS[k] == v
+
+            h[k] = prune(v, drop)
+          end
+        when Array then value.map { |v| prune(v, drop) }
+        else value
+        end
+      end
+
+      def fetch_from_index(exclude) # rubocop:disable Metrics/MethodLength
+        return [] unless @ref
+
+        index.search(@ref) do |row|
+          pubid_matches?(row[:id], exclude)
+        end.sort_by do |row|
+          [row[:id].date&.year.to_i, *part_sort_key(row[:id].part)]
+        end.map do |row|
+          Hit.new(row, self)
+        end
+      end
+    end
+  end
+end

data/lib/relaton/iec/item_data.rb

@@ -0,0 +1,7 @@
+module Relaton
+  module Iec
+    class ItemData < Bib::ItemData
+
+    end
+  end
+end

data/lib/relaton/iec/model/bibdata.rb

@@ -0,0 +1,8 @@
+module Relaton
+  module Iec
+    class Bibdata < Item
+      model ItemData
+      include Bib::BibdataShared
+    end
+  end
+end

data/lib/relaton/iec/model/bibitem.rb

@@ -0,0 +1,8 @@
+module Relaton
+  module Iec
+    class Bibitem < Item
+      model ItemData
+      include Bib::BibitemShared
+    end
+  end
+end

data/lib/relaton/iec/model/docidentifier.rb

@@ -0,0 +1,131 @@
+module Relaton
+  module Iec
+    class Docidentifier < Bib::Docidentifier
+      attribute :content, :string
+
+      attr_reader :pubid
+
+      def initialize(arg = nil, **kwargs)
+        arg.is_a?(Hash) ? super(arg) : super(**kwargs)
+        raw = arg.is_a?(Hash) ? (arg["content"] || arg[:content]) : kwargs[:content]
+        self.content = raw if raw
+      end
+
+      alias_method :original_content=, :content=
+      alias_method :original_content, :content
+
+      def content=(value)
+        @pubid = nil
+        @raw_content = nil
+
+        parsed =
+          case value
+          when ::Pubid::Iec::Identifier then value
+          when String
+            begin
+              ::Pubid::Iec::Identifier.parse(value)
+            rescue StandardError
+              Util.warn "Failed to parse Pubid: #{value}"
+              nil
+            end
+          end
+
+        if parsed
+          @pubid = parsed
+        elsif value.is_a?(String)
+          @raw_content = value
+        end
+
+        send(:original_content=, to_s)
+      end
+
+      def content
+        return @raw_content if @raw_content
+        return render_pubid(@pubid) if @pubid
+
+        original_content
+      end
+
+      def to_s
+        content.to_s
+      end
+
+      def to_all_parts!
+        return unless @pubid
+
+        remove_part!
+        remove_date!
+        remove_stage!
+        @pubid.all_parts = true if @pubid.respond_to?(:all_parts=)
+        refresh_content!
+      end
+
+      def remove_stage!
+        remove_attr!(:stage)
+      end
+
+      def remove_part!
+        remove_attr!(:part)
+      end
+
+      def remove_date!
+        remove_attr!(:date)
+      end
+
+      private
+
+      def render_pubid(pubid)
+        case type
+        # pubid owns the legacy positional IEC URN format (and the all-parts
+        # ":::ser" series form); render through it.
+        when "URN" then pubid.to_urn.to_s
+        else pubid.to_s
+        end
+      end
+
+      def remove_attr!(attr)
+        return unless @pubid
+
+        # For supplements (Amendment/Corrigendum, which carry their own
+        # date/version as an addition to the base), preserve the
+        # supplement's own attrs and only clear the base chain — e.g.
+        # "IEC 60027-1:1992/AMD1:1997" should drop the base year but keep
+        # the amendment year. For wrappers (VapIdentifier, ConsolidatedIdentifier),
+        # clear the outer attr too — they re-state the base year and need it
+        # gone everywhere.
+        unless @pubid.is_a?(::Pubid::Iec::SupplementIdentifier)
+          clear_attr_on(@pubid, attr)
+        end
+
+        node = @pubid.base_identifier
+        while node
+          clear_attr_on(node, attr)
+          # ConsolidatedIdentifier carries a sibling collection of bundled
+          # identifiers; clear the attr on each non-supplement entry too.
+          if node.respond_to?(:identifiers) && node.identifiers
+            node.identifiers.each do |id|
+              next if id.is_a?(::Pubid::Iec::SupplementIdentifier)
+
+              clear_attr_on(id, attr)
+            end
+          end
+          node = node.base_identifier
+        end
+        refresh_content!
+      end
+
+      # Clear one attribute on a single pubid; for :part also clear the
+      # paired :subpart (pubid 2.x splits "16-1-1" into part="1" + subpart="1").
+      def clear_attr_on(pubid, attr)
+        pubid.send("#{attr}=", nil) if pubid.respond_to?("#{attr}=")
+        return unless attr == :part && pubid.respond_to?(:subpart=)
+
+        pubid.subpart = nil
+      end
+
+      def refresh_content!
+        send(:original_content=, to_s)
+      end
+    end
+  end
+end

data/lib/relaton/iec/model/doctype.rb

@@ -0,0 +1,12 @@
+module Relaton
+  module Iec
+    class Doctype < Bib::Doctype
+      TYPES = %w[
+        international-standard technical-specification technical-report publicly-available-specification
+        international-workshop-agreement guide industry-technical-agreement system-reference-deliverable
+      ].freeze
+
+      attribute :content, :string, values: TYPES
+    end
+  end
+end

data/lib/relaton/iec/model/ext.rb

@@ -0,0 +1,53 @@
+require_relative "doctype"
+require_relative "stage_name"
+
+module Relaton
+  module Iec
+    class Ext < Bib::Ext
+      attribute :doctype, Doctype
+      attribute :subdoctype, :string, values: %w[specification method-of-test vocabulary code-of-practice]
+      attribute :structuredidentifier, Iso::StructuredIdentifier
+      attribute :horizontal, :boolean
+      attribute :stagename, StageName
+      attribute :updates_document_type, :string, values: Doctype::TYPES
+      attribute :fast_track, :boolean
+      attribute :price_code, :string
+      attribute :function, :string, values: %w[emc safety environment quality-assurance]
+      attribute :accessibility_color_inside, :boolean
+      attribute :cen_processing, :boolean
+      attribute :secretary, :string
+      attribute :interest_to_committees, :string
+      attribute :tc_sc_officers_note, :string, raw: true
+
+      xml do
+        map_element "horizontal", to: :horizontal
+        map_element "stagename", to: :stagename
+        map_element "updates-document-type", to: :updates_document_type
+        map_element "fast-track", to: :fast_track
+        map_element "price-code", to: :price_code
+        map_element "function", to: :function
+        map_element "accessibility-color-inside", to: :accessibility_color_inside
+        map_element "cen-processing", to: :cen_processing
+        map_element "secretary", to: :secretary
+        map_element "interest-to-committees", to: :interest_to_committees
+        map_element "tc-sc-officers-note", to: :tc_sc_officers_note
+      end
+
+      key_value do
+        map_element "horizontal", to: :horizontal
+        map_element "stagename", to: :stagename
+        map_element "updates_document_type", to: :updates_document_type
+        map_element "fast_track", to: :fast_track
+        map_element "price_code", to: :price_code
+        map_element "function", to: :function
+        map_element "accessibility_color_inside", to: :accessibility_color_inside
+        map_element "cen_processing", to: :cen_processing
+        map_element "secretary", to: :secretary
+        map_element "interest_to_committees", to: :interest_to_committees
+        map_element "tc_sc_officers_note", to: :tc_sc_officers_note
+      end
+
+      def get_schema_version = Relaton.schema_versions["relaton-model-iec"]
+    end
+  end
+end

data/lib/relaton/iec/model/item.rb

@@ -0,0 +1,20 @@
+require_relative "docidentifier"
+require_relative "ext"
+
+module Relaton
+  module Iec
+    class Relation < Bib::Relation
+    end
+
+    class Item < Iso::Item
+      model ItemData
+
+      attribute :docidentifier, Docidentifier, collection: true, initialize_empty: true
+      attribute :relation, Relation, collection: true, initialize_empty: true
+      attribute :ext, Ext
+    end
+  end
+end
+
+require_relative "item_base"
+require_relative "relation"

data/lib/relaton/iec/model/item_base.rb

@@ -0,0 +1,12 @@
+require_relative "docidentifier"
+
+module Relaton
+  module Iec
+    class ItemBase < Iso::ItemBase
+      model ItemData
+
+      attribute :docidentifier, Docidentifier, collection: true, initialize_empty: true
+      attribute :relation, Relation, collection: true, initialize_empty: true
+    end
+  end
+end

data/lib/relaton/iec/model/relation.rb

@@ -0,0 +1,7 @@
+module Relaton
+  module Iec
+    class Relation
+      attribute :bibitem, ItemBase
+    end
+  end
+end

data/lib/relaton/iec/model/stage_name.rb

@@ -0,0 +1,13 @@
+module Relaton
+  module Iec
+    class StageName < Lutaml::Model::Serializable
+      attribute :abbreviation, :string
+      attribute :content, :string
+
+      xml do
+        map_attribute "abbreviation", to: :abbreviation
+        map_content to: :content
+      end
+    end
+  end
+end

data/lib/relaton/iec/processor.rb

@@ -0,0 +1,73 @@
+require "relaton/core/processor"
+
+module Relaton
+  module Iec
+    class Processor < Core::Processor
+      def initialize
+        @short = :relaton_iec
+        @prefix = "IEC"
+        @defaultprefix = %r{^(IEC\s|CISPR\s|IEV($|\s))}
+        @idtype = "IEC"
+        @datasets = %w[iec-harmonized-all iec-harmonized-latest]
+      end
+
+      # @param code [String]
+      # @param date [String, nil] year
+      # @param opts [Hash]
+      # @return [Relaton::Iec::ItemData, nil]
+      def get(code, date, opts)
+        require_relative "../iec"
+        Bibliography.get(code, date, opts)
+      end
+
+      #
+      # Fetch all the documents from a source
+      #
+      # @param [String] source source name (iec-harmonized-all, iec-harmonized-latest)
+      # @param [Hash] opts
+      # @option opts [String] :output directory to output documents
+      # @option opts [String] :format output format (xml, yaml, bibxml)
+      #
+      def fetch_data(source, opts)
+        require_relative "data_fetcher"
+        DataFetcher.fetch(source, **opts)
+      end
+
+      # @param xml [String]
+      # @return [Relaton::Iec::ItemData]
+      def from_xml(xml)
+        require_relative "../iec"
+        Item.from_xml xml
+      end
+
+      # @param yaml [String]
+      # @return [Relaton::Iec::ItemData]
+      def from_yaml(yaml)
+        require_relative "../iec"
+        Item.from_yaml yaml
+      end
+
+      # Returns hash of XML grammar
+      # @return [String]
+      def grammar_hash
+        require_relative "../iec"
+        @grammar_hash ||= ::Relaton::Iec.grammar_hash
+      end
+
+      # @param code [String]
+      # @return [String, nil]
+      def urn_to_code(code)
+        require_relative "../iec"
+        Relaton::Iec.urn_to_code code
+      end
+
+      #
+      # Remove index file
+      #
+      def remove_index_file
+        require_relative "../iec"
+        Relaton::Index.find_or_create(:iec, url: true, file: "#{INDEXFILE}.yaml").remove_file
+      end
+    end
+  end
+end

data/lib/relaton/iec/statuses.yml

@@ -0,0 +1,199 @@
+ACD:
+  status: Approved for CD
+  stage: '20.99'
+ACDV:
+  status: Approved for CDV
+  stage: '30.99'
+ADISSB:
+  status: Preparation of text subcontracted to CO
+  stage: '40.95'
+ADTR:
+  status: Approved for DTR
+  stage: '40.99'
+ADTS:
+  status: Approved for DTS
+  stage: '40.99'
+AFDIS:
+  status: Approved for FDIS
+  stage: '40.99'
+AMW:
+  status: Document under revision
+  stage: '92.20'
+ANW:
+  status: Registration of new project
+  stage: '20.00'
+APUB:
+  status: Approved for publication
+  stage: '50.99'
+APUBSB:
+  status: Preparation of text subcontracted to CO
+  stage: '50.95'
+BPUB:
+  status: Being published
+  stage: '60.00'
+BWG:
+  status: Return to drafting phase or redefine project
+  stage: '30.92'
+CAN:
+  status: Draft cancelled
+  stage: '20.98'
+CCDV:
+  status: Draft circulated as CDV
+  stage: '40.00'
+CD:
+  status: Draft circulated as CD
+  stage: '30.00'
+CDISH:
+  status: Draft circulated as DISH
+  stage: '50.20'
+CDM:
+  status: CD to be discussed at meeting
+  stage: '30.20'
+CDPAS:
+  status: Draft circulated as DPAS
+  stage: '50.20'
+CDTR:
+  status: Draft circulated as DTR
+  stage: '50.20'
+CDTS:
+  status: Draft circulated as DTS
+  stage: '50.20'
+CDVM:
+  status: Rejected CDV to be discussed at a meeting
+  stage: '40.93'
+CFDIS:
+  status: Draft circulated as FDIS
+  stage: '50.20'
+DECDISH:
+  status: DISH at editing check
+  stage: '40.99'
+DECFDIS:
+  status: FDIS at editing check
+  stage: '50.60'
+DECPUB:
+  status: Publication at editing check
+  stage: '60.00'
+DEL:
+  status: Deleted/abandoned
+  stage: '00.99'
+DELPUB:
+  status: Deleted publication
+  stage: '90.99'
+DREJ:
+  status: Abandon
+  stage: '30.98'
+DTRM:
+  status: Rejected DTR to be discussed at meeting
+  stage: '50.92'
+DTSM:
+  status: Rejected DTS to be discussed at meeting
+  stage: '50.92'
+MERGED:
+  status: Fragment merged
+  stage: '30.97'
+NADIS:
+  status: Repeat enquiry
+  stage: '40.93'
+NCDV:
+  status: CDV rejected
+  stage: '40.98'
+NDTR:
+  status: DTR rejected
+  stage: '50.98'
+NDTS:
+  status: DTS rejected
+  stage: '50.98'
+NFDIS:
+  status: FDIS rejected
+  stage: '50.98'
+PCC:
+  status: Preparation of CC
+  stage: '30.92'
+PNW:
+  status: New work item proposal
+  stage: '10.00'
+PPUB:
+  status: Publication issued
+  stage: '60.60'
+PRVC:
+  status: Preparation of RVC
+  stage: '40.92'
+PRVD:
+  status: Preparation of RVD
+  stage: '40.92'
+PRVDISH:
+  status: Preparation of RVDISH
+  stage: '40.92' # ?
+PRVDPAS:
+  status: Preparation of RVDPAS
+  stage: '40.92'
+PRVDTR:
+  status: Preparation of RVDTR
+  stage: '40.92'
+PRVDTS:
+  status: Preparation of RVDTS
+  stage: '40.92'
+PRVN:
+  status: Preparation of RVN
+  stage: '40.92'
+PWI:
+  status: Preliminary work item
+  stage: '00.00'
+RDIS:
+  status: Registration for formal approval
+  stage: '50.00'
+RDISH:
+  status: DISH received and registered
+  stage: '50.00'
+RFDIS:
+  status: FDIS received and registered
+  stage: '50.00'
+RPUB:
+  status: Publication received and registered
+  stage: '60.60'
+SPE:
+  stage: SPE # ?
+SPLIT:
+  status: Project Fragmented
+  stage: SPLIT # ?
+SRP:
+  stage: SRP
+SUSPENDED:
+  status: Project Suspended
+  stage: SUSPENDED # ?
+TCDV:
+  status: Translation of CDV
+  stage: '50.00'
+TDISH:
+  status: Translation of DISH
+  stage: '50.00' # ?
+TDTR:
+  status: Translation of DTR
+  stage: '50.00'
+TDTS:
+  status: Translation of DTS
+  stage: '50.00'
+TFDIS:
+  status: Translation of FDIS
+  stage: '50.00'
+TPUB:
+  status: Translation of publication
+  stage: '60.00'
+WPUB:
+  status: Publication withdrawn
+  stage: '95.99'
+preCD:
+  status: Preparation of CD document
+  stage: preCD
+preCDPAS:
+  status: Preparation of DPAS
+  stage: preCDPAS
+preDISH:
+  status: Preparation of DISH
+  stage: preDISH
+preDTR:
+  status: Preparation of DTR document
+  stage: preDTR
+prePNW:
+  status: Preparation of NP document
+  stage: prePNW

data/lib/relaton/iec/util.rb

@@ -0,0 +1,8 @@
+module Relaton
+  module Iec
+    module Util
+      extend Relaton::Bib::Util
+      PROGNAME = "relaton-iec".freeze
+    end
+  end
+end