relaton-bipm 1.19.1 → 1.20.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 6169158d13f03ec83666123ca52d0e87b7960ba13fe5b6d5051228c54158f93c
-  data.tar.gz: 46236ffb5e1f875fd042dd89fe866251962a553497dd9a8a474edcd6aa18e2fa
+  metadata.gz: 98f4b8a04eb9738b1302ff84f91c11f5e604adfdefc2fdfac999279f4d62bf0d
+  data.tar.gz: 4e6dad8f65ca0472b24d178474b69a78fe20b3d784adbc1512d4c79aad576de0
 SHA512:
-  metadata.gz: d1b01efea043f825d7e08dcecc6396e21b56c75ab0131dffd6b5e936ea10e185e9f8412ddf9c30fa21f64eaf9e97408da92e06fedc2a7cbb1d9da581151ab0ea
-  data.tar.gz: 8cc33bc1d0b431ecd6d55c640ddcf5d715e640dfd2a926a5e2db5d72f442fefe93f527275f17236e2a815302cdb65ec9c47f71dd970b9d0eb708133765654d92
+  metadata.gz: d07764ecb8ffba9c394eb01fb17ec617c1209038390cd37e6035a897616e1f717b69ced0830da7d6b721733eeeb8d3ae503bb65a7efc9228142dd90852f1f665
+  data.tar.gz: a97d282c516026bd083a7ec534a0d1357e8eb4ab09f01495aff28e6be8fbabf78b071c422aa572690f12d5700d287ebb5cc2e34a23c7a1c88f041813834b7571

data/lib/relaton_bipm/bipm_bibliographic_item.rb

@@ -86,12 +86,16 @@ module RelatonBipm
     #
     def to_hash(embedded: false)
       hash = super
-      hash["comment_period"] = comment_period.to_hash if comment_period
-      hash["si_aspect"] = si_aspect if si_aspect
-      hash["meeting_note"] = meeting_note if meeting_note
+      hash["ext"]["comment_period"] = comment_period.to_hash if comment_period
+      hash["ext"]["si_aspect"] = si_aspect if si_aspect
+      hash["ext"]["meeting_note"] = meeting_note if meeting_note
       hash
     end
 
+    def has_ext?
+      super || comment_period || si_aspect || meeting_note
+    end
+
     # @param prefix [String]
     # @return [String]
     def to_asciibib(prefix = "")

data/lib/relaton_bipm/committee.rb

@@ -1,27 +1,31 @@
 module RelatonBipm
-  class Committee
+  class Committee < RelatonBib::LocalizedString
+    ACRONYMS = YAML.load_file File.join(__dir__, "acronyms.yaml")
+
     # @return [String]
     attr_reader :acronym
 
     # @return [RelatonBib::LocalizedString]
     attr_reader :content
 
-    # @param acronym [String]
-    # @param content [RelatonBib::LocalisedString, String, nil]
-    def initialize(acronym:, content: nil)
-      acronyms = YAML.load_file File.join(__dir__, "acronyms.yaml")
-      unless acronyms[acronym]
+    # @param [String] acronym
+    # @param [Hash] args
+    # @option args [RelatonBib::LocalisedString, String, nil] :content
+    # @option args [String, nil] :language
+    # @option args [String, nil] :script
+    def initialize(acronym:, **args)
+      unless ACRONYMS[acronym]
         Util.warn "Invalid acronym: `#{acronym}`. Allowed " \
-                  "values: `#{acronyms.map { |k, _v| k }.join '`, `'}`"
+                  "values: `#{ACRONYMS.map { |k, _v| k }.join '`, `'}`"
       end
 
       @acronym = acronym
-      @content = localized_content content, acronyms[acronym]
+      super(*localized_args(acronym, **args))
     end
 
     # @param builder [Nokogiri::XML::Builder]
     def to_xml(builder)
-      builder.committee(acronym: acronym) { |b| content.to_xml b }
+      builder.committee(acronym: acronym) { |b| super b }
     end
 
     # @param prefix [String]
@@ -32,13 +36,13 @@ module RelatonBipm
       pref += "committee"
       out = count > 1 ? "#{pref}::\n" : ""
       out += "#{pref}.acronym:: #{acronym}\n"
-      out + content.to_asciibib(pref)
+      out + super(pref)
     end
 
     # @return [Hash]
     def to_hash
       hash = { "acronym" => acronym }
-      cnt = content.to_hash
+      cnt = super
       case cnt
       when Array then hash["variants"] = cnt
       when Hash then hash.merge! cnt
@@ -49,12 +53,14 @@ module RelatonBipm
 
     private
 
-    def localized_content(cnt, acr)
-      if cnt.is_a? String
-        RelatonBib::LocalizedString.new cnt
-      elsif (cnt.nil? || cnt.empty?) && acr && acr["en"]
-        RelatonBib::LocalizedString.new(acr["en"], "en", "Latn")
-      else cnt
+    def localized_args(accronym, **args)
+      if args[:content].is_a? String
+        [args[:content], args[:language], args[:script]]
+      elsif args[:content].nil?
+        lang = args[:language] || ACRONYMS.dig(acronym, "en") ? "en" : ACRONYMS[acronym]&.keys&.first
+        script = args[:script] || lang == "en" ? "Latn" : nil
+        [ACRONYMS.dig(accronym, lang), lang, script]
+      else [args[:content]]
       end
     end
   end

data/lib/relaton_bipm/hash_converter.rb

@@ -17,6 +17,7 @@ module RelatonBipm
 
       # project_group_hash_to_bib ret
       commentperiod_hash_to_bib ret
+      ret[:si_aspect] = args["ext"]["si_aspect"] if args.dig("ext", "si_aspect")
       ret
     end
 
@@ -42,7 +43,10 @@ module RelatonBipm
 
     # @param ret [Hash]
     def commentperiod_hash_to_bib(ret)
-      ret[:comment_period] &&= CommentPeriond.new(**ret[:comment_period])
+      compr = ret.dig(:ext, :comment_period) || ret[:comment_period] # @TODO: remove ret[:comment_period] after all data is updated
+      return unless compr
+
+      ret[:comment_period] &&= CommentPeriond.new(**compr)
     end
 
     # @param ret [Hash]
@@ -69,17 +73,17 @@ module RelatonBipm
 
     # @param ret [Hash]
     def editorialgroup_hash_to_bib(ret) # rubocop:disable Metrics/AbcSize,Metrics/MethodLength
-      return unless ret[:editorialgroup]
+      ed = ret.dig(:ext, :editorialgroup) || ret[:editorialgroup] # @TODO: remove ret[:editorialgroup] after all data is updated
+      return unless ed
 
-      cmt = ret[:editorialgroup][:committee].map do |c|
+      cmt = ed[:committee].map do |c|
         if (vars = committee_variants c).any?
-          content = RelatonBib::LocalizedString.new vars
-          Committee.new acronym: c[:acronym], content: content
+          Committee.new acronym: c[:acronym], content: vars
         else
           Committee.new(**c)
         end
       end
-      wg = RelatonBib.array(ret[:editorialgroup][:workgroup]).map do |w|
+      wg = RelatonBib.array(ed[:workgroup]).map do |w|
         w.is_a?(Hash) ? WorkGroup.new(**w) : WorkGroup.new(content: w)
       end
       ret[:editorialgroup] = EditorialGroup.new committee: cmt, workgroup: wg
@@ -98,9 +102,10 @@ module RelatonBipm
 
     # @param ret [Hash]
     def structuredidentifier_hash_to_bib(ret)
-      ret[:structuredidentifier] &&= StructuredIdentifier.new(
-        **ret[:structuredidentifier],
-      )
+      struct_id = ret.dig(:ext, :structuredidentifier) || ret[:structuredidentifier] # @TODO: remove ret[:structuredidentifier] after all data is updated
+      return unless struct_id
+
+      ret[:structuredidentifier] = StructuredIdentifier.new(**struct_id)
     end
 
     def create_doctype(**args)

data/lib/relaton_bipm/rawdata_bipm_metrologia/affiliations.rb

@@ -0,0 +1,111 @@
+module RelatonBipm
+  module RawdataBipmMetrologia
+    class Affiliations
+      attr_reader :affiliations
+
+      #
+      # Initialize parser
+      #
+      # @param [Array<RelatonBib::Affiliation>] affiliations directory with affiliations
+      #
+      def initialize(affiliations)
+        @affiliations = affiliations
+      end
+
+      #
+      # Parse affiliations
+      #
+      # @return [RelatonBipm::RawdataBipmMetrologia::Affiliations] affiliations
+      #
+      def self.parse(dir)
+        affiliations = Dir["#{dir}/*.xml"].each_with_object([]) do |path, m|
+          doc = Nokogiri::XML(File.read(path, encoding: "UTF-8"))
+          doc.xpath("//aff").each do |aff|
+            m << parse_affiliation(aff) if aff.at("institution")
+          end
+        end.uniq { |a| a.organization.name.first.content }
+        new affiliations
+      end
+
+      #
+      # Parse affiliation organization
+      # https://github.com/relaton/relaton-data-bipm/issues/17#issuecomment-1367035444
+      #
+      # @param [Nokogiri::XML::Element] aff
+      #
+      # @return [RelatonBib::Affiliation] Organization name, country, division, street address
+      #
+      def self.parse_affiliation(aff)
+        text = aff.at("text()").text
+        return if text.include? "Permanent address:" || text.include?("1005 Southover Lane") ||
+          text == "Germany" || text.starts_with?("Guest") || text.starts_with?("Deceased") ||
+          text.include?("Author to whom any correspondence should be addressed")
+
+        args = {}
+        institution = aff.at('institution')
+        if institution
+          name = institution.text
+          return if name == "1005 Southover Lane"
+
+          args[:subdivision] = parse_division(aff)
+          args[:contact] = parse_address(aff)
+        else
+        #   div, name, city, country = aff.xpath("text()").text.strip.split(", ")
+        #   div, name = name, div if name.nil?
+        #   args[:subdivision] = [RelatonBib::LocalizedString.new(div)] if div
+        #   args[:contact] = [RelatonBib::Address.new(city: city, country: country)] if city && country
+          name = aff.text
+        end
+        args[:name] = [RelatonBib::LocalizedString.new(name)]
+        org = RelatonBib::Organization.new(**args)
+        RelatonBib::Affiliation.new(organization: org)
+      end
+
+      def self.parse_division(aff)
+        div = aff.xpath("text()[following-sibling::institution]").text.gsub(/^\W*|\W*$/, "")
+        return [] if div.empty?
+
+        [RelatonBib::LocalizedString.new(div)]
+      end
+
+      def self.parse_address(aff)
+        address = []
+        addr = aff.xpath("text()[preceding-sibling::institution]").text.gsub(/^\W*|\W*$/, "")
+        address << addr unless addr.empty?
+        country = aff.at('country')
+        address << country.text if country && !country.text.empty?
+        address = address.join(", ")
+        return [] if address.empty?
+
+        [RelatonBib::Address.new(formatted_address: address)]
+      end
+
+      def self.parse_elements(aff)
+        elements = aff.xpath("text()").text.strip.split(", ")
+        case elements.size
+        when 1 then { name: RelatonBib::LocalizedString.new(elements[0]) }
+        when 2
+          # name, country
+          { name: RelatonBib::LocalizedString.new(elements[0]),
+            contact: [RelatonBib::Address.new(formatted_address: elements[1])] }
+        when 3
+          # it can be name, country, city or name, city, country
+          # so use formatted_address instead of city and country
+          { name: RelatonBib::LocalizedString.new(elements[0]),
+            contact: RelatonBib::Address.new(formatted_address: elements[1, 2].join(", ")) }
+        end
+      end
+
+      #
+      # Find affiliation by organization name
+      #
+      # @param [Strign] text string with organization name in it
+      #
+      # @return [RelatonBib::Affiliation]
+      #
+      def find(text)
+        @affiliations.select { |a| text.include?(a.organization.name[0].content) }.sort.last
+      end
+    end
+  end
+end

data/lib/relaton_bipm/rawdata_bipm_metrologia/article_parser.rb

@@ -59,12 +59,9 @@ module RelatonBipm
       #
       # Parse volume, issue and page
       #
-      # @return [Array<String>] array of volume, issue and page
+      # @return [String] volume issue page
       #
       def volume_issue_article
-        # volume = @meta.at("./volume").text
-        # issue = @meta.at("./issue").text
-        # page = @doc.at("./front/article-meta/fpage")&.text || manuscript
         [@journal, @volume, @article].compact.join(" ")
       end
 
@@ -140,17 +137,53 @@ module RelatonBipm
       #
       # @return [Array<RelatonBib::Affiliation>] array of affiliations
       #
-      def affiliation(contrib) # rubocop:disable Metrics/AbcSize
+      def affiliation(contrib)
         contrib.xpath("./xref[@ref-type='aff']").map do |x|
-          a = @meta.at("./contrib-group/aff[@id='#{x[:rid]}']/label/following-sibling::node()")
-          parts = a.text.split(", ")
-          orgname = parts[0..-3].join(", ")
-          city, country = parts[-2..]
-          address = []
-          address << RelatonBib::Address.new(city: city, country: country) if city && country
-          org = RelatonBib::Organization.new name: orgname, contact: address
-          RelatonBib::Affiliation.new organization: org
+          a = @meta.at("./contrib-group/aff[@id='#{x[:rid]}']") # /label/following-sibling::node()")
+            parse_affiliation a
+        end.compact
+      end
+
+      def parse_affiliation(aff)
+        text = aff.xpath("text()|sup|sub").to_xml.split(",").map(&:strip).reject(&:empty?).join(", ")
+        text = CGI::unescapeHTML(text)
+        return if text.include?("Permanent address:") || text == "Germany" ||
+          text.start_with?("Guest") || text.start_with?("Deceased") ||
+          text.include?("Author to whom any correspondence should be addressed")
+
+        args = {}
+        institution = aff.at('institution')
+        if institution
+          name = institution.text
+          return if name == "1005 Southover Lane"
+
+          args[:subdivision] = parse_division(aff)
+          args[:contact] = parse_address(aff)
+        else
+          name = text
         end
+        args[:name] = [RelatonBib::LocalizedString.new(name)]
+        org = RelatonBib::Organization.new(**args)
+        RelatonBib::Affiliation.new(organization: org)
+      end
+
+      def parse_division(aff)
+        div = aff.xpath("text()[following-sibling::institution]").text.gsub(/^\W*|\W*$/, "")
+        return [] if div.empty?
+
+        [RelatonBib::LocalizedString.new(div)]
+      end
+
+      def parse_address(aff)
+        address = []
+        addr = aff.xpath("text()[preceding-sibling::institution]").text.gsub(/^\W*|\W*$/, "")
+        address << addr unless addr.empty?
+        country = aff.at('country')
+        address << country.text if country && !country.text.empty?
+        address = address.join(", ")
+        return [] if address.empty?
+
+        [RelatonBib::Address.new(formatted_address: address)]
       end
 
       #
@@ -173,20 +206,20 @@ module RelatonBipm
       #
       # @return [Array<RelatonBib::Forename>] array of forenames
       #
-      def forename(given_name) # rubocop:disable Metrics/MethodLength
-        return [] unless given_name
+      # def forename(given_name) # rubocop:disable Metrics/MethodLength
+      #   return [] unless given_name
 
-        given_name.text.scan(/(\w+)(?:\s(\w)(?:\s|$))?/).map do |nm, int|
-          if nm.size == 1
-            name = nil
-            init = nm
-          else
-            name = nm
-            init = int
-          end
-          RelatonBib::Forename.new(content: name, language: ["en"], script: ["Latn"], initial: init)
-        end
-      end
+      #   given_name.text.scan(/(\w+)(?:\s(\w)(?:\s|$))?/).map do |nm, int|
+      #     if nm.size == 1
+      #       name = nil
+      #       init = nm
+      #     else
+      #       name = nm
+      #       init = int
+      #     end
+      #     RelatonBib::Forename.new(content: name, language: ["en"], script: ["Latn"], initial: init)
+      #   end
+      # end
 
       #
       # Parse date

data/lib/relaton_bipm/rawdata_bipm_metrologia/fetcher.rb

@@ -29,8 +29,9 @@ module RelatonBipm
       # Fetch articles from rawdata-bipm-metrologia and save to files
       #
       def fetch_articles # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
+        # aff = Affiliations.parse DIR
         Dir["#{DIR}/**/*.xml"].each do |path|
-          item = ArticleParser.parse path
+          item = ArticleParser.parse path # , aff
           file = "#{item.docidentifier.first.id.downcase.tr(' ', '-')}.#{@data_fetcher.ext}"
           out_path = File.join(@data_fetcher.output, file)
           key = Id.new.parse(item.docidentifier.first.id).to_hash

data/lib/relaton_bipm/version.rb

@@ -1,3 +1,3 @@
 module RelatonBipm
-  VERSION = "1.19.1".freeze
+  VERSION = "1.20.0".freeze
 end

data/lib/relaton_bipm/xml_parser.rb

@@ -58,12 +58,9 @@ module RelatonBipm
 
         cm = eg.xpath("committee").map do |c|
           vars = variants c
-          cnt = if vars.any?
-                  RelatonBib::LocalizedString.new vars
-                else
-                  RelatonBib::LocalizedString.new c.text, c[:language], c[:script]
-                end
-          Committee.new acronym: c[:acronym], content: cnt
+          cnt = vars.any? ? vars : c.text
+          args = c.to_h.transform_keys(&:to_sym).select { |k, _| %i[language script locale].include?(k) }
+          Committee.new acronym: c[:acronym], content: cnt, **args
         end
         wg = eg.xpath("workgroup").map do |w|
           WorkGroup.new content: w.text, acronym: w[:acronym]

data/relaton_bipm.gemspec

@@ -35,7 +35,7 @@ Gem::Specification.new do |spec| # rubocop:disable Metrics/BlockLength
   spec.add_dependency "faraday", "~> 2.7.0"
   spec.add_dependency "mechanize", "~> 2.10"
   spec.add_dependency "parslet", "~> 2.0.0"
-  spec.add_dependency "relaton-bib", "~> 1.19.0"
+  spec.add_dependency "relaton-bib", "~> 1.20.0"
   spec.add_dependency "relaton-index", "~> 0.2.2"
   spec.add_dependency "rubyzip", "~> 2.3.0"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: relaton-bipm
 version: !ruby/object:Gem::Version
-  version: 1.19.1
+  version: 1.20.0
 platform: ruby
 authors:
 - Ribose Inc.
-autorequire: 
+autorequire:
 bindir: exe
 cert_chain: []
-date: 2024-07-31 00:00:00.000000000 Z
+date: 2024-12-12 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: faraday
@@ -58,14 +58,14 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 1.19.0
+        version: 1.20.0
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 1.19.0
+        version: 1.20.0
 - !ruby/object:Gem::Dependency
   name: relaton-index
   requirement: !ruby/object:Gem::Requirement
@@ -135,6 +135,7 @@ files:
 - lib/relaton_bipm/hash_converter.rb
 - lib/relaton_bipm/id_parser.rb
 - lib/relaton_bipm/processor.rb
+- lib/relaton_bipm/rawdata_bipm_metrologia/affiliations.rb
 - lib/relaton_bipm/rawdata_bipm_metrologia/article_parser.rb
 - lib/relaton_bipm/rawdata_bipm_metrologia/fetcher.rb
 - lib/relaton_bipm/structured_identifier.rb
@@ -149,7 +150,7 @@ licenses:
 metadata:
   homepage_uri: https://github.com/relaton/relaton-bipm
   source_code_uri: https://github.com/relaton/relaton-bipm
-post_install_message: 
+post_install_message:
 rdoc_options: []
 require_paths:
 - lib
@@ -165,7 +166,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubygems_version: 3.3.27
-signing_key: 
+signing_key:
 specification_version: 4
 summary: 'RelatonBipm: retrieve BIPM Standards for bibliographic use using the BibliographicItem
   model'