relaton-bipm 2.0.0.pre.alpha.2 → 2.0.0.pre.alpha.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 28df241c309a5932e534f727cba44fed2414a8ca86c166b6d969b40edfea0154
-  data.tar.gz: 962edc994ca6435fbb6346e858702f58e7ed30b47d0acfdb2f1acee40e085da0
+  metadata.gz: 28e812b274f936f9499e5e3499dfd4aa1dbdf6eefb5e9faee499bc77b76e4b32
+  data.tar.gz: cb3d30c7d029c7ab1934867f164baaab2bc13363f07921bdd969225bfce4a63c
 SHA512:
-  metadata.gz: ece155d44403c0a38f4023b00fdc9f6aece11d33e6cf66a8c36223f744734b7154ded701a5c3a11a140bc59f2cb5b136d1cf61f78e1dc54ab75396e20d1a50ee
-  data.tar.gz: 04fbc52a458552adb636b1355615e3d8e7dbb512a788b429bb7576e3d0a8ae17d13512429a86699e667087844f8cfeecf7b1de8353d6cff6b14062a6d7ece43e
+  metadata.gz: 747287435b174e1102b6a21e99db6703b432b022bd04c4a5dc66d2a7e437902493e01510cb084e068773ecb0c28cc54b530a2efdd0c9187cbbbfeb973e686d40
+  data.tar.gz: 98eb1a4b7fd46adc7cefdc65238ff57fd90b30454cdd4a46f7a3a2cc814bda6522416a984c7bf8c453ad32500bf8a4ce069e058b52e66c48921c9430f93f5070

data/lib/relaton/bipm/bibliography.rb

@@ -4,7 +4,6 @@ require_relative "id_parser"
 module Relaton::Bipm
   class Bibliography
     GH_ENDPOINT = "https://raw.githubusercontent.com/relaton/relaton-data-bipm/refs/heads/data-v2/".freeze
-    INDEX_FILE = "index-v1.yaml".freeze
 
     class << self
       # @param text [String]
@@ -63,7 +62,7 @@ module Relaton::Bipm
 
       def index
         Relaton::Index.find_or_create(
-          :bipm, url: "#{GH_ENDPOINT}index-v1.zip", file: INDEX_FILE, id_keys: %i[group type number year corr part append]
+          :bipm, url: "#{GH_ENDPOINT}index-v1.zip", file: INDEXFILE, id_keys: %i[group type number year corr part append]
         )
       end
 

data/lib/relaton/bipm/data_fetcher.rb

@@ -9,7 +9,7 @@ module Relaton::Bipm
     attr_reader :output, :format, :ext, :files, :index
 
     def index
-      @index ||= Relaton::Index.find_or_create :bipm, file: Bibliography::INDEX_FILE
+      @index ||= Relaton::Index.find_or_create :bipm, file: INDEXFILE
     end
 
     #

data/lib/relaton/bipm/data_outcomes_parser.rb

@@ -268,26 +268,29 @@ module Relaton::Bipm
     #
     def contributors(date, body)
       contribs = [bipm_contrib]
-      author = author_org date, body
-      return contribs unless author
+      subdiv = committee_subdivision date, body
+      return contribs unless subdiv
 
-      role = Relaton::Bib::Contributor::Role.new(type: "author")
-      contribs << Relaton::Bib::Contributor.new(organization: author, role: [role])
+      bipm_name = [Relaton::Bib::TypedLocalizedString.new(content: "BIPM", script: "Latn")]
+      org = Relaton::Bib::Organization.new(name: bipm_name, subdivision: [subdiv])
+      desc = Relaton::Bib::LocalizedMarkedUpString.new(content: "committee")
+      role = Relaton::Bib::Contributor::Role.new(type: "author", description: [desc])
+      contribs << Relaton::Bib::Contributor.new(organization: org, role: [role])
     end
 
     #
-    # Create author organization
+    # Create committee subdivision
     #
     # @param [String] date date of publication
     # @param [String] body organization abbreviation (CCTF, CIPM, CGPM)
     #
-    # @return [Hash, nil] author organization
+    # @return [Relaton::Bib::Subdivision, nil] committee subdivision
     #
-    def author_org(date, body)
+    def committee_subdivision(date, body)
       case body
-      when "CCTF" then cctf_org date
-      when "CGPM" then cgpm_org
-      when "CIPM" then cipm_org
+      when "CCTF" then cctf_subdivision date
+      when "CGPM" then cgpm_subdivision
+      when "CIPM" then cipm_subdivision
       end
     end
 
@@ -307,25 +310,25 @@ module Relaton::Bipm
     end
 
     #
-    # Create CCTF organization
+    # Create CCTF subdivision
     #
     # @param [String] date date of meeting
     #
-    # @return [Hash] CCTF organization
+    # @return [Relaton::Bib::Subdivision] CCTF subdivision
     #
-    def cctf_org(date) # rubocop:disable Metrics/MethodLength
+    def cctf_subdivision(date) # rubocop:disable Metrics/MethodLength
       if ::Date.parse(date).year < 1999
         nms = [
           { content: "Consultative Committee for the Definition of the Second", language: "en" },
           { content: "Comité Consultatif pour la Définition de la Seconde", language: "fr" },
         ]
-        organization nms, "CCDS"
+        subdivision nms, "CCDS"
       else
         nms = [
           { content: "Consultative Committee for Time and Frequency", language: "en" },
           { content: "Comité consultatif du temps et des fréquences", language: "fr" },
         ]
-        organization nms, "CCTF"
+        subdivision nms, "CCTF"
       end
     end
 
@@ -347,29 +350,44 @@ module Relaton::Bipm
     end
 
     #
-    # Create CGPM organization
+    # Create subdivision
     #
-    # @return [Hash] CGPM organization
+    # @param [Array<Hash>] names subdivision names in different languages
+    # @param [String] abbr abbreviation
+    # @param [String] abbr_lang language of abbreviation
+    #
+    # @return [Relaton::Bib::Subdivision] subdivision
+    #
+    def subdivision(names, abbr, abbr_lang: "en")
+      name = names.map { |n| Relaton::Bib::TypedLocalizedString.new(**n, script: "Latn") }
+      abbreviation = Relaton::Bib::LocalizedString.new(content: abbr, language: abbr_lang, script: "Latn")
+      Relaton::Bib::Subdivision.new(type: "committee", name: name, abbreviation: abbreviation)
+    end
+
+    #
+    # Create CGPM subdivision
+    #
+    # @return [Relaton::Bib::Subdivision] CGPM subdivision
     #
-    def cgpm_org
+    def cgpm_subdivision
       nms = [
         { content: "General Conference on Weights and Measures", language: "en" },
         { content: "Conférence Générale des Poids et Mesures", language: "fr" },
       ]
-      organization nms, "CGPM", abbr_lang: "fr"
+      subdivision nms, "CGPM", abbr_lang: "fr"
     end
 
     #
-    # Create CIPM organization
+    # Create CIPM subdivision
     #
-    # @return [Array<Hash>] CIPM organization
+    # @return [Relaton::Bib::Subdivision] CIPM subdivision
     #
-    def cipm_org
+    def cipm_subdivision
       names = [
         { content: "International Committee for Weights and Measures", language: "en" },
         { content: "Comité international des poids et mesures", language: "fr" },
       ]
-      organization names, "CIPM", abbr_lang: "fr"
+      subdivision names, "CIPM", abbr_lang: "fr"
     end
 
     #

data/lib/relaton/bipm/processor.rb

@@ -62,7 +62,7 @@ module Relaton
       #
       def remove_index_file
         require_relative "../bipm"
-        Relaton::Index.find_or_create(:bipm, url: true, file: Bibliography::INDEX_FILE).remove_file
+        Relaton::Index.find_or_create(:bipm, url: true, file: INDEXFILE).remove_file
       end
     end
   end

data/lib/relaton/bipm/rawdata_bipm_metrologia/article_parser.rb

@@ -295,9 +295,49 @@ module Relaton::Bipm
       # @return [Array<Relaton::Bib::Relation>] array of document relations
       #
       def parse_relation
-        dates do |d, t|
+        rels = dates do |d, t|
           Relaton::Bib::Relation.new(type: "hasManifestation", bibitem: bibitem(d, t))
         end
+        rels + parse_references
+      end
+
+      #
+      # Parse back/ref-list references as "cites" relations
+      #
+      # @return [Array<Relaton::Bib::Relation>] array of "cites" relations
+      #
+      def parse_references
+        @doc.xpath("./back/ref-list/ref").filter_map do |ref|
+          citation = ref.at("./element-citation")
+          next unless citation
+
+          Relaton::Bib::Relation.new(type: "cites", bibitem: citation_bibitem(citation))
+        end
+      end
+
+      #
+      # Build bibitem from an element-citation
+      #
+      # @param [Nokogiri::XML::Element] citation element-citation node
+      #
+      # @return [Relaton::Bipm::ItemData] bibitem
+      #
+      def citation_bibitem(citation)
+        attrs = {}
+        doi = citation.at("./pub-id[@pub-id-type='doi']")
+        if doi
+          attrs[:docidentifier] = [Relaton::Bib::Docidentifier.new(content: doi.text, type: "doi")]
+          attrs[:source] = [Relaton::Bib::Uri.new(content: "https://doi.org/#{doi.text}", type: "doi")]
+        end
+        source = citation.at("./source")
+        if source
+          attrs[:title] = [Relaton::Bib::Title.new(content: source.text)]
+        end
+        year = citation.at("./year")
+        if year
+          attrs[:date] = [Relaton::Bib::Date.new(type: "published", at: year.text)]
+        end
+        ItemData.new(**attrs)
       end
 
       #

data/lib/relaton/bipm/rawdata_bipm_metrologia/fetcher.rb

@@ -33,7 +33,7 @@ module Relaton::Bipm
       #
       def fetch_articles # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
         # aff = Affiliations.parse DIR
-        Dir["#{DIR}/**/*.xml"].each do |path|
+        Dir["#{DIR}/**/*.xml"].sort_by { |p| archive_date(p) }.each do |path|
           item = ArticleParser.parse path # , aff
           file = "#{item.docidentifier.first.content.downcase.tr(' ', '-')}.#{@data_fetcher.ext}"
           out_path = File.join(@data_fetcher.output, file)
@@ -150,6 +150,17 @@ module Relaton::Bipm
         [Relaton::Bib::Uri.new(type: "src", content: link(*args))]
       end
 
+      #
+      # Extract archive date from path for sorting
+      #
+      # @param [String] path file path
+      #
+      # @return [String] date string for sorting
+      #
+      def archive_date(path)
+        path[%r{/data/(\d{4}-\d{2}-\d{2}T[\d_]+)_content/}, 1].to_s
+      end
+
       def link(*args)
         params = id_parts(*args).join("/")
         case args.size

data/lib/relaton/bipm/si_brochure_parser.rb

@@ -1,3 +1,4 @@
+require "nokogiri"
 require_relative "id_parser"
 
 module Relaton::Bipm
@@ -34,6 +35,9 @@ module Relaton::Bipm
         xml = File.read(f, encoding: "UTF-8")
         xml = xml.force_encoding("UTF-8") if xml.encoding != Encoding::UTF_8
         item1 = Bibdata.from_xml(xml)
+        unless has_committee_contributor?(item1)
+          extract_editorialgroup(xml).each { |c| item1.contributor << c }
+        end
         fix_si_brochure_id item1
         basename = File.join @data_fetcher.output, File.basename(f).sub(/(?:-(?:en|fr))?\.rxl$/, "")
         outfile = "#{basename}.#{@data_fetcher.ext}"
@@ -75,7 +79,7 @@ module Relaton::Bipm
       else
         item.docnumber = prid.sub(/^BIPM\s/, "")
       end
-      item.id = prid.gsub(/[,\s]/, "")
+      item.id = prid.gsub(/[,\s-]/, "")
     end
 
     def update_id(item)
@@ -93,6 +97,33 @@ module Relaton::Bipm
       end.content
     end
 
+    def has_committee_contributor?(item)
+      item.contributor.any? do |c|
+        c.role.any? { |r| r.description.any? { |d| d.content == "committee" } }
+      end
+    end
+
+    def extract_editorialgroup(xml)
+      doc = Nokogiri::XML(xml)
+      doc.xpath("//editorialgroup/committee").map do |committee|
+        acronym = committee["acronym"]
+        names = committee.xpath("variant").map do |v|
+          Relaton::Bib::TypedLocalizedString.new(
+            content: v.text, language: v["language"], script: v["script"],
+          )
+        end
+        subdiv = Relaton::Bib::Subdivision.new(
+          type: "committee", name: names,
+          abbreviation: Relaton::Bib::LocalizedString.new(content: acronym),
+        )
+        bipm_name = [Relaton::Bib::TypedLocalizedString.new(content: acronym)]
+        org = Relaton::Bib::Organization.new(name: bipm_name, subdivision: [subdiv])
+        desc = Relaton::Bib::LocalizedMarkedUpString.new(content: "committee")
+        role = Relaton::Bib::Contributor::Role.new(type: "author", description: [desc])
+        Relaton::Bib::Contributor.new(organization: org, role: [role])
+      end
+    end
+
     #
     # Deep merge two hashes
     #

data/lib/relaton/bipm/version.rb

@@ -1,5 +1,5 @@
 module Relaton
   module Bipm
-    VERSION = "2.0.0-alpha.2".freeze
+    VERSION = "2.0.0-alpha.3".freeze
   end
 end

data/lib/relaton/bipm.rb

@@ -16,6 +16,8 @@ module Relaton
   module Bipm
     class Error < StandardError; end
 
+    INDEXFILE = "index-v1.yaml".freeze
+
     # Returns hash of gems versions used to generate the data model.
     # @return [String]
     def grammar_hash

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: relaton-bipm
 version: !ruby/object:Gem::Version
-  version: 2.0.0.pre.alpha.2
+  version: 2.0.0.pre.alpha.3
 platform: ruby
 authors:
 - Ribose Inc.