relaton-bipm 1.19.2 → 1.19.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: d26cb426c0ebdcd6e4544117b76fabc440ce698ae86511d2fa13022fedc0fa33
-  data.tar.gz: b0dae1c393906430b924171c4bc9dc0ddc723108372fd7ac351092845153cb6b
+  metadata.gz: 3c9261622d7be459730444f639a2b7f91a6606600a8aa1d0e4806036a766235c
+  data.tar.gz: 0b285754154788c58a7cfb812c92c5cc54331ea6f52e41dd71c129e59ffa5660
 SHA512:
-  metadata.gz: bfcf6fc883da15803965d74ef0faee65f5aabc637b382e55586d63ace8647f988862748a88269fe3bc33c27308c41a2801c5d5bed7c408e736c451e1b7ba1f03
-  data.tar.gz: 6a1b674e7285565be0fd400f4b27fa9066483fb2500bea43bde9aec6f8c95aa483602d74946b1acb0a669203467a8739bd8e4e5b11eb2ef01958203058651780
+  metadata.gz: 573c4e55c1d0cc0ecfc2f96c3dccf31f0a85790a433c9332db3d0f8478bcc2d32330fb7727187102045378cbf9555ede82e05871c54024e04267a16b40629372
+  data.tar.gz: 61f92667f0154f11345ca3bec967a2fee19ac2d4795ef81ffc50561f5c554000501e5f0d26a337b903a3be2a2f76cd637026cbb7daaf4206e032bd609aef960e

data/lib/relaton_bipm/data_outcomes_parser.rb

@@ -88,8 +88,7 @@ module RelatonBipm
       if @data_fetcher.files.include?(path) && part
         add_part hash, part
         item = RelatonBipm::BipmBibliographicItem.new(**hash)
-        yaml = RelatonBib.parse_yaml(File.read(path, encoding: "UTF-8"), [Date])
-        has_part_item = RelatonBipm::BipmBibliographicItem.from_hash(yaml)
+        has_part_item = parse_file path
         has_part_item.relation << RelatonBib::DocumentRelation.new(type: "partOf", bibitem: item)
         @data_fetcher.write_file path, has_part_item, warn_duplicate: false
         path = File.join dir, "#{num}-#{part}.#{@data_fetcher.ext}"
@@ -111,6 +110,17 @@ module RelatonBipm
       fetch_resolution body: body, en: en, fr: fr, dir: dir, src: src, num: num
     end
 
+    def parse_file(path)
+        case @data_fetcher.format
+        when "yaml"
+          yaml = RelatonBib.parse_yaml(File.read(path, encoding: "UTF-8"), [Date])
+          RelatonBipm::BipmBibliographicItem.from_hash(yaml)
+        when "xml"
+          xml = File.read(path, encoding: "UTF-8")
+          RelatonBipm::XMLParser.from_xml xml
+        end
+    end
+
     #
     # Read English and French files
     #
@@ -348,8 +358,12 @@ module RelatonBipm
     #
     # @return [Hash] title
     #
-    def create_title(content, language)
-      { content: content, language: language, script: "Latn" }
+    def create_title(content, language, format = "text/plain")
+      if language == "fr"
+        content.sub!(/(\d+)(e)/, '\1<sup>\2</sup>')
+        format = "text/html" if content.match?(/<sup>/)
+      end
+      { content: content, language: language, script: "Latn", format: format }
     end
 
     #
@@ -541,11 +555,12 @@ module RelatonBipm
     end
 
     def create_meeting_docids(en_id)
-      fr_id = en_id.sub(/(\d+)(?:st|nd|rd|th)/, '\1e').sub("Meeting", "Réunion")
+      fr_id = en_id.sub(/(\d+)(?:st|nd|rd|th)/, '\1e').sub("Meeting", "réunion")
+      fr_id_sup = fr_id.sub(/(\d+)(e)/, '\1<sup>\2</sup>')
       [
         make_docid(id: en_id, type: "BIPM", primary: true, language: "en", script: "Latn"),
-        make_docid(id: fr_id, type: "BIPM", primary: true, language: "fr", script: "Latn"),
-        make_docid(id: "#{en_id} / #{fr_id}", type: "BIPM", primary: true),
+        make_docid(id: fr_id_sup, type: "BIPM", primary: true, language: "fr", script: "Latn"),
+        make_docid(id: "#{en_id} / #{fr_id_sup}", type: "BIPM", primary: true),
       ]
     end
 

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 = "1.19.2".freeze
+  VERSION = "1.19.3".freeze
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: relaton-bipm
 version: !ruby/object:Gem::Version
-  version: 1.19.2
+  version: 1.19.3
 platform: ruby
 authors:
 - Ribose Inc.
-autorequire: 
+autorequire:
 bindir: exe
 cert_chain: []
-date: 2024-08-08 00:00:00.000000000 Z
+date: 2024-12-13 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: faraday
@@ -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
@@ -164,8 +165,8 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.3.27
-signing_key: 
+rubygems_version: 3.5.19
+signing_key:
 specification_version: 4
 summary: 'RelatonBipm: retrieve BIPM Standards for bibliographic use using the BibliographicItem
   model'