relaton 2.2.0.pre.alpha.1 → 3.0.0.pre.alpha.2

data/lib/relaton/gb/yaml/prefixes.yaml

@@ -0,0 +1,200 @@
+# GbStandardNationalPrefix
+GB:
+  prefix: GB_national
+GBn:
+  prefix: GBn_confidential
+GJB:
+  prefix: GJB_military
+GSB:
+  prefix: GSB_physical
+
+# GbStandardSectorPrefix
+NY:
+  prefix: NY_agriculture
+  administration: 农业部
+SC:
+  prefix: SC_aquatic
+  administration: 农业部
+SL:
+  prefix: SL_water_resources
+  administration: 水利部
+QB:
+  prefix: QB_light_industry
+  administration: 国家发改委
+FZ:
+  prefix: FZ_textile
+  administration: 国家发改委
+YY:
+  prefix: YY_medicine
+  administration: 国家食品药品监督管理局
+MZ:
+  prefix: MZ_civil_affairs
+  administration: 民政部
+JY:
+  prefix: JY_education
+  administration: 教育部
+YC:
+  prefix: YC_tobacco
+  administration: 国家烟草专卖局
+YB:
+  prefix: YB_ferrous_metallurgy
+  administration: 国家发改委
+YS:
+  prefix: YS_nonferrous_metallurgy
+  administration: 国家发改委
+SY:
+  prefix: SY_natural_gas
+  administration: 国家发改委
+HG:
+  prefix: HG_chemical_industry
+  administration: 国家发改委
+SH:
+  prefix: SH_petrochemical
+  administration: 国家发改委
+JC:
+  prefix: JC_building_materials
+  administration: 国家发改委
+DZ:
+  prefix: DZ_geological
+  administration: 国土资源部
+TD:
+  prefix: TD_land_management
+  administration: 国土资源部
+CH:
+  prefix: CH_surveying
+  administration: 国家测绘局
+JB:
+  prefix: JB_mechanical
+  administration: 国家发改委
+QC:
+  prefix: QC_automotile
+  administration: 国家发改委
+MH:
+  prefix: MH_civil_aviation
+  administration: 中国民航管理总局
+WJ:
+  prefix: WJ_ordnance
+  administration: 国防科学工业委员会
+CB:
+  prefix: CB_ships
+  administration: 国防科学工业委员会
+HB:
+  prefix: HB_aviation
+  administration: 国防科学工业委员会
+QJ:
+  prefix: QJ_aerospace
+  administration: 国防科学工业委员会
+EJ:
+  prefix: EJ_nuclear_industry
+  administration: 国防科学工业委员会
+TB:
+  prefix: TB_rail_transport
+  administration: 铁道部
+JT:
+  prefix: JT_traffic
+  administration: 交通部
+LD:
+  prefix: LD_work_safety
+  administration: 劳动和社会保障部
+SJ:
+  prefix: SJ_electronics
+  administration: 信息产业部
+YD:
+  prefix: YD_communication
+  administration: 信息产业部
+GY:
+  prefix: GY_media
+  administration: 国家广播电视总局
+DY:
+  prefix: DY_film
+  administration: 国家电影局
+DL:
+  prefix: DL_electricity
+  administration: 国家发改委
+JR:
+  prefix: JR_financial
+  administration: 中国人民银行
+HY:
+  prefix: HY_oceanic
+  administration: 国家海洋局
+DA:
+  prefix: DA_archiving
+  administration: 国家档案局
+SN:
+  prefix: SN_inspection
+  administration: 国家质量监督检验检疫总局
+WH:
+  prefix: WH_culture
+  administration: 文化部
+TY:
+  prefix: TY_sports
+  administration: 国家体育总局
+SB:
+  prefix: SB_business
+  administration: 商务部
+WB:
+  prefix: WB_materials_management
+  administration: 国家发改委
+HJ:
+  prefix: HJ_environment_protection
+  administration: 国家环境保护总局
+XB:
+  prefix: XB_rare_earth
+  administration: 国家发改委稀土办公室
+CJ:
+  prefix: CJ_town_construction
+  administration: 建设部
+JG:
+  prefix: JG_construction_industry
+  administration: 建设部
+CY:
+  prefix: CY_news_publishing
+  administration: 国家新闻出版总署
+MT:
+  prefix: MT_coal
+  administration: 国家发改委
+WS:
+  prefix: WS_health
+  administration: 卫生部
+GA:
+  prefix: GA_public_safety
+  administration: 公安部
+BB:
+  prefix: BB_packaging
+  administration: 国家发改委
+DB:
+  prefix: DB_seismology
+  administration: 中国地震局
+LB:
+  prefix: LB_tourism
+  administration: 国家旅游局
+QX:
+  prefix: QX_meteorological
+  administration: 中国气象局
+WM:
+  prefix: WM_foreign_trade
+  administration: 外经贸部科技司
+HS:
+  prefix: HS_customs
+  administration: 海关总署
+YZ:
+  prefix: YZ_postal
+  administration: 国家邮政局
+GM:
+  prefix: GM_cryptography
+  administration: 国家密码管理局
+AQ:
+  prefix: AQ_production_safety
+  administration: 国家安全生产管理局
+GH:
+  prefix: GH_supply_marketing
+  administration: 中华全国供销合作总社
+LS:
+  prefix: LS_food
+  administration: 国家粮食局
+TJ:
+  prefix: TJ_railway_traffic
+  administration: 铁道部标准所
+ZY:
+  prefix: ZY_chinese_medicine
+  administration: 国家中医药管理局

data/lib/relaton/gb.rb

@@ -0,0 +1,33 @@
+require "digest/md5"
+require "mechanize"
+require "relaton/core"
+require "relaton/iso"
+require "cnccs"
+require_relative "version"
+require_relative "gb/util"
+require_relative "gb/item_data"
+require_relative "gb/item"
+require_relative "gb/bibitem"
+require_relative "gb/bibdata"
+require_relative "gb/bibliography"
+
+# if defined? Relaton
+#   require "relaton_gb/processor"
+#   # don't register the gem if it's required form relaton's registry
+#   return if caller.detect { |c| c.include? "register_gems" }
+
+#   Relaton::Registry.instance.register RelatonGb::Processor
+# end
+
+module Relaton
+  module Gb
+    # Returns hash of XML reammar
+    # @return [String]
+    def self.grammar_hash
+      # gem_path = File.expand_path "..", __dir__
+      # grammars_path = File.join gem_path, "grammars", "*"
+      # grammars = Dir[grammars_path].sort.map { |gp| File.read gp }.join
+      Digest::MD5.hexdigest Relaton::VERSION # grammars
+    end
+  end
+end

data/lib/relaton/gost/bibdata.rb

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

data/lib/relaton/gost/bibitem.rb

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

data/lib/relaton/gost/bibliography.rb

@@ -0,0 +1,107 @@
+require "net/http"
+
+module Relaton
+  module Gost
+    # Retrieval front-end for the GOST flavor. Parses the citation with
+    # Pubid::Gost, looks it up in the pubid-structured `relaton-data-gost`
+    # index-v2, and fetches the matching per-document YAML over HTTP.
+    #
+    # The index rows are keyed by Pubid::Gost identifiers (`_type: pubid:gost:
+    # {interstate,national}-standard`, number, year), so `Relaton::Index`
+    # narrows candidates by number via binary search before the block applies
+    # the precise pubid match. Both Latin "GOST"/"GOST R" and Cyrillic
+    # "ГОСТ"/"ГОСТ Р" surface forms parse (Pubid normalises Cyrillic to Latin).
+    module Bibliography
+      ENDPOINT = "https://raw.githubusercontent.com/relaton/relaton-data-gost/main/".freeze
+
+      class << self
+        # Search for a GOST publication by its identifier.
+        #
+        # @param text [String, Pubid::Gost::Identifier] the GOST reference to
+        #   look up (e.g. "GOST R 34.12-2015", "GOST 14946-82", "ГОСТ 1.0")
+        # @param year [String, nil] the edition year (optional; may also be
+        #   embedded in the reference)
+        # @param _opts [Hash] options (unused)
+        # @return [Relaton::Gost::Item, nil]
+        def search(text, year = nil, _opts = {})
+          pubid = text.is_a?(String) ? ::Pubid::Gost.parse(text) : text
+          Util.info "Fetching from Relaton repository ...", key: pubid.to_s
+          # Pass the pubid so Relaton::Index narrows candidates by number via
+          # binary search; pick the latest edition for an undated citation.
+          row = rows(pubid, year).max_by { |r| r[:id].year.to_i }
+          unless row
+            Util.info "Not found.", key: pubid.to_s
+            return
+          end
+
+          uri = URI("#{ENDPOINT}#{row[:file]}")
+          resp = Net::HTTP.get_response uri
+          unless resp.code == "200"
+            raise Relaton::RequestError, "Could not access #{uri}: HTTP #{resp.code}"
+          end
+
+          item = Relaton::Gost::Item.from_yaml resp.body
+          Util.info "Found: `#{item.docidentifier.first&.content}`", key: pubid.to_s
+          item.tap { |i| i.fetched = Date.today.to_s }
+        rescue SocketError, Errno::EINVAL, Errno::ECONNRESET, EOFError,
+               Net::HTTPBadResponse, Net::HTTPHeaderSyntaxError,
+               Net::ProtocolError, Net::ReadTimeout, OpenSSL::SSL::SSLError,
+               Errno::ETIMEDOUT => e
+          raise Relaton::RequestError, "Could not access #{uri}: #{e.message}"
+        end
+
+        # Fetch a GOST publication, suppressing the edition year for an undated
+        # citation so the designation renders undated (`GOST R 34.12`), the way
+        # the ISO/OIML fetchers do. A dated citation (`GOST R 34.12-2015`, or an
+        # explicit `year`) still pins that edition.
+        #
+        # @param opts [Hash] options
+        # @option opts [Boolean] :keep_year retain the edition year even for an
+        #   undated citation (or, when false, strip it even for a dated one)
+        # @see #search
+        def get(ref, year = nil, opts = {})
+          item = search(ref, year, opts)
+          return item unless item
+
+          pubid = ref.is_a?(String) ? ::Pubid::Gost.parse(ref) : ref
+          dated = pubid.year || year
+          return item if (dated && opts[:keep_year].nil?) || opts[:keep_year]
+
+          item.to_most_recent_reference
+        end
+
+        private
+
+        def index
+          Relaton::Index.find_or_create(
+            :gost,
+            url: "#{ENDPOINT}#{INDEXFILE}.zip",
+            file: "#{INDEXFILE}.yaml",
+            pubid_class: ::Pubid::Gost::Identifier,
+          )
+        end
+
+        # The index rows for a citation. A dated citation matches exactly
+        # (`exact: true`, pubid `==`). An undated one takes pubid's subset
+        # match `query === row`, so it matches every edition of the document:
+        # the omitted year matches any value. The class must be identical, so
+        # interstate and national standards stay distinct, and `1.1` does not
+        # match `1.10`. pubid declares `copublisher` strict for GOST, so an
+        # omitted copublisher means "none". The `year` argument pins an edition
+        # the reference string omitted.
+        #
+        # @param query [Pubid::Gost::Identifier]
+        # @param year [String, nil]
+        # @return [Array<Hash>]
+        def rows(query, year)
+          return index.search(query, exact: true) if query.year
+
+          rows = index.search(query)
+          return rows unless year
+
+          rows.select { |r| r[:id].year.to_s == year.to_s }
+        end
+      end
+    end
+  end
+end

data/lib/relaton/gost/docidentifier.rb

@@ -0,0 +1,80 @@
+module Relaton
+  module Gost
+    # Structured GOST document identifier. Wraps a Pubid::Gost identifier so the
+    # canonical citation form (e.g. "GOST R 34.12-2015", "GOST 14946-82") stays
+    # in sync with `content`, and undated-reference handling (return the latest
+    # edition) can strip the year via the pubid. Mirrors Relaton::Oiml::Docidentifier.
+    class Docidentifier < Bib::Docidentifier
+      attr_reader :pubid
+
+      # Capture the inherited content setter before overriding #content=, so
+      # #refresh_content! can write the re-rendered string back WITHOUT
+      # re-parsing. `remove_date!` used to re-sync through `self.content=`,
+      # which rebuilds @pubid from the string; harmless on its own, but a
+      # re-parse of an all-parts wrapper's own rendering (which carries
+      # pubid's "(all parts)" marker) would not round-trip at all.
+      alias_method :store_content, :content=
+
+      def initialize(attrs = {}, options = {})
+        pubid = attrs.is_a?(Hash) ? attrs.delete(:pubid) : nil
+        attrs[:content] ||= pubid.to_s if pubid
+        super
+        @pubid = pubid if pubid
+      end
+
+      def content=(value)
+        super
+        @pubid = ::Pubid::Gost.parse(value) if value
+      rescue StandardError
+        @pubid = nil
+      end
+
+      # `remove_date!` was implemented but `remove_part!` and `to_all_parts!`
+      # were not, so `Bib::ItemData#to_all_parts` still raised
+      # `NotImplementedError` on every GOST item — and it descends from
+      # `ScriptError`, so a caller's `rescue => e` did not catch it.
+      #
+      # GOST pubid carries the edition as `year` (not `date`), so `remove_date!`
+      # clears that: `GOST R 34.12-2015` -> `GOST R 34.12`. `part`/`subpart`
+      # come from the pubid base class and are unused by GOST today, so
+      # `remove_part!` is a no-op for the rendered string — implemented so it
+      # never raises. All three no-op safely when `@pubid` is nil.
+
+      def remove_part!
+        clear_attr!(:part)
+      end
+
+      def remove_date!
+        clear_attr!(:year)
+      end
+
+      def to_all_parts!
+        return if !@pubid || @pubid.all_parts?
+
+        # `remove_part!`/`remove_date!` mutate `@pubid` in place via direct
+        # attribute setters, so a bare reference (or a shallow `.dup`) would
+        # still see those mutations. `#exclude` with no args rebuilds the
+        # whole identifier as new objects — a real, independent snapshot —
+        # so the all-parts wrapper below keeps the original part and date.
+        original = @pubid.exclude
+        remove_part!
+        remove_date!
+        @pubid = original.to_all_parts
+      end
+
+      private
+
+      def clear_attr!(attr)
+        return unless @pubid && @pubid.respond_to?("#{attr}=")
+
+        @pubid.public_send("#{attr}=", nil)
+        @pubid.subpart = nil if attr == :part && @pubid.respond_to?(:subpart=)
+        refresh_content!
+      end
+
+      def refresh_content!
+        store_content(@pubid.to_s) if @pubid
+      end
+    end
+  end
+end

data/lib/relaton/gost/doctype.rb

@@ -0,0 +1,16 @@
+module Relaton
+  module Gost
+    # GOST document types. The value is carried in the inherited
+    # `content` attribute. The list mirrors the categories GOST
+    # publishes (interstate GOST, Russian national GOST R,
+    # preliminary standards, methodological documents).
+    class Doctype < Bib::Doctype
+      TYPES = %w[
+        interstate
+        national
+        preliminary
+        methodological
+      ].freeze
+    end
+  end
+end

data/lib/relaton/gost/ext.rb

@@ -0,0 +1,46 @@
+module Relaton
+  module Gost
+    # GOST bibliographic item extension. Adds GOST-specific structured
+    # metadata on top of the shared Relaton::Bib::Ext so the fields
+    # round-trip through both XML and YAML without per-repo merge hacks.
+    # Doctype is referenced lazily via autoload from lib/relaton/gost.rb.
+    class Ext < Bib::Ext
+      attribute :doctype,  Doctype
+      attribute :urn,      :string
+      attribute :webpage,  :string
+      # ICS classification code (МКС / Межгосударственный классификатор
+      # стандартов) — e.g. "13.080.20".
+      attribute :ics_code, :string
+      # Original-language developer of the standard (РАЗРАБОТЧИК).
+      attribute :developer, :string
+      # Keyword list (Ключевые слова).
+      attribute :keywords, :string, collection: true, initialize_empty: true
+      # Page count (Страниц).
+      attribute :pages,   :string
+      # Original-language designation as printed on the cover
+      # (e.g. "ГОСТ Р 71039— 2023") — preserves em-dash and spacing
+      # quirks that the parsed Pubid normalises away.
+      attribute :designation_original, :string
+
+      xml do
+        map_element "urn",                  to: :urn
+        map_element "webpage",              to: :webpage
+        map_element "ics_code",             to: :ics_code
+        map_element "developer",            to: :developer
+        map_element "keywords",             to: :keywords
+        map_element "pages",                to: :pages
+        map_element "designation_original", to: :designation_original
+      end
+
+      key_value do
+        map_element "urn",                  to: :urn
+        map_element "webpage",              to: :webpage
+        map_element "ics_code",             to: :ics_code
+        map_element "developer",            to: :developer
+        map_element "keywords",             to: :keywords
+        map_element "pages",                to: :pages
+        map_element "designation_original", to: :designation_original
+      end
+    end
+  end
+end

data/lib/relaton/gost/item.rb

@@ -0,0 +1,15 @@
+module Relaton
+  module Gost
+    # GOST bibliographic item. Re-declares +ext+ to use the typed Ext
+    # subclass so GOST-specific fields round-trip natively. Sub-files
+    # (ItemData, Ext, Docidentifier) are loaded lazily via the autoload
+    # entries declared in lib/relaton/gost.rb.
+    class Item < Bib::Item
+      model ItemData
+
+      attribute :ext, Ext
+      attribute :docidentifier, Docidentifier, collection: true,
+                                               initialize_empty: true
+    end
+  end
+end

data/lib/relaton/gost/item_base.rb

@@ -0,0 +1,18 @@
+module Relaton
+  module Gost
+    class ItemBase < Lutaml::Model::Serializable
+      include Bib::NamespaceHelper
+
+      attr_accessor :type
+
+      model ItemData
+
+      instance_exec(&Bib::ItemShared::ATTRIBUTES)
+
+      xml do
+        map_attribute "type", to: :type
+        instance_exec(&Bib::ItemShared::XML_BODY)
+      end
+    end
+  end
+end

data/lib/relaton/gost/item_data.rb

@@ -0,0 +1,6 @@
+module Relaton
+  module Gost
+    class ItemData < Bib::ItemData
+    end
+  end
+end

data/lib/relaton/gost/processor.rb

@@ -0,0 +1,49 @@
+require "relaton/core/processor"
+
+module Relaton
+  module Gost
+    # Relaton processor for the GOST flavor. Registered by
+    # Relaton::Db::Registry when the unified gem loads (see
+    # lib/relaton/db/registry.rb). Supports `relaton fetch gost ...`
+    # once the relaton-data-gost dataset is installed.
+    class Processor < Core::Processor
+      attr_reader :idtype
+
+      def initialize
+        @short = :relaton_gost
+        @prefix = "GOST"
+        # Both Latin "GOST" and Cyrillic "ГОСТ" route here. The trailing
+        # \b keeps the prefix from swallowing longer tokens ("GOSTA …").
+        @defaultprefix = %r{^(?:GOST|ГОСТ)\b}
+        @idtype = "GOST"
+      end
+
+      def get(code, date, opts)
+        require_relative "../gost"
+        Bibliography.get(code, date, opts)
+      end
+
+      def from_xml(xml)
+        require_relative "../gost"
+        Item.from_xml xml
+      end
+
+      def from_yaml(yaml)
+        require_relative "../gost"
+        Item.from_yaml(yaml)
+      end
+
+      def grammar_hash
+        require_relative "../gost"
+        @grammar_hash ||= ::Relaton::Gost.grammar_hash
+      end
+
+      def remove_index_file
+        require_relative "../gost"
+        Relaton::Index.find_or_create(
+          :gost, url: true, file: "#{INDEXFILE}.yaml"
+        ).remove_file
+      end
+    end
+  end
+end

data/lib/relaton/gost/util.rb

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

data/lib/relaton/gost.rb

@@ -0,0 +1,36 @@
+# frozen_string_literal: true
+
+require "relaton/bib"
+require "relaton/index"
+require "pubid"
+require "relaton/version"
+
+# Relaton::Gost is the GOST (Russian federal standards — Гостстандарт)
+# flavor of the unified relaton gem. Adds typed Ext fields (urn,
+# webpage) so they round-trip natively through YAML and XML without
+# per-repo merge hacks.
+module Relaton
+  module Gost
+    INDEXFILE = "index-v2".freeze
+
+    class Error < StandardError; end
+
+    autoload :Doctype,       "relaton/gost/doctype"
+    autoload :Ext,           "relaton/gost/ext"
+    autoload :ItemData,      "relaton/gost/item_data"
+    autoload :ItemBase,      "relaton/gost/item_base"
+    autoload :Item,          "relaton/gost/item"
+    autoload :Bibitem,       "relaton/gost/bibitem"
+    autoload :Bibdata,       "relaton/gost/bibdata"
+    autoload :Docidentifier, "relaton/gost/docidentifier"
+    autoload :Bibliography,  "relaton/gost/bibliography"
+    autoload :Util,          "relaton/gost/util"
+    autoload :Processor,     "relaton/gost/processor"
+
+    # Returns hash of XML grammar
+    # @return [String]
+    def self.grammar_hash
+      Digest::MD5.hexdigest Relaton::VERSION
+    end
+  end
+end

data/lib/relaton/iala/bibdata.rb

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

data/lib/relaton/iala/bibitem.rb

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