relaton-bipm 1.18.0 → 1.18.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: d51b6507b240f142eca33e8276587dc32d0e5e7a3a794d1ab7eaa2b6e2af09ac
-  data.tar.gz: 491c23ff3ef47f435ec18d605b0bc35806eb872755e338c1d06ce609284bab34
+  metadata.gz: a0ceb817fce307673ac91a29c26c09684bb8309aad1b522c33e4d76859db4c90
+  data.tar.gz: 075b945ebc0792e814e38d811579a9a91209339cf91874fa2af572402ee05f5e
 SHA512:
-  metadata.gz: b362e8a613b038c7890987beb9ebd84f42162a94f8f42149bb763426c2f8bf8536a815006f9a9587b12f1a89e714aaa832f619d22214a88e0549fb9eee11ca06
-  data.tar.gz: a71ec348afeeb1383bc1f3ea2c5ae11f9c1a34014f0f81462d5bc1cf248566ced81c2043991ff67bd0c40e42b9480a5472abf9d46ed2da1d0b6a3ec9a08a433e
+  metadata.gz: db4ad0e7175acc4f78f788f443aa4affddee29e9352cb27d5d0290f8dd27512506724a2c6f0e6a417456a8cfba24b041840e43473700d4501b107ec8b93f99a8
+  data.tar.gz: 23638af28bddb83784b67c89e7b5d523e7c09147bdc95cc920d17cbeef713f2c7f7d8d7f7790a9593b1e4716b7f1386c655a8cae5bb89fa889682789180b57ac

data/lib/relaton_bipm/bipm_bibliography.rb

@@ -46,7 +46,7 @@ module RelatonBipm
       # @return [RelatonBipm::BipmBibliographicItem]
       #
       def get_bipm(reference) # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
-        ref_id = Id.new reference
+        ref_id = Id.new.parse reference
         rows = index.search { |r| ref_id == r[:id] }
         return unless rows.any?
 
@@ -63,7 +63,7 @@ module RelatonBipm
 
       def index
         Relaton::Index.find_or_create(
-          :bipm, url: "#{GH_ENDPOINT}index2.zip", file: INDEX_FILE, id_keys: %i[group type number year corr]
+          :bipm, url: "#{GH_ENDPOINT}index2.zip", file: INDEX_FILE, id_keys: %i[group type number year corr part append]
         )
       end
 

data/lib/relaton_bipm/bipm_si_brochure_parser.rb

@@ -36,7 +36,7 @@ module RelatonBipm
         basename = File.join @data_fetcher.output, File.basename(f).sub(/(?:-(?:en|fr))?\.rxl$/, "")
         outfile = "#{basename}.#{@data_fetcher.ext}"
         key = hash1["docnumber"] || basename
-        @data_fetcher.index2.add_or_update Id.new(key).to_hash, outfile
+        @data_fetcher.index2.add_or_update Id.new.parse(key).to_hash, outfile
         hash = if File.exist? outfile
                  warn_duplicate = false
                  hash2 = YAML.load_file outfile
@@ -60,19 +60,19 @@ module RelatonBipm
     # @return [void]
     #
     def fix_si_brochure_id(hash) # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
-      did = hash["docid"].detect { |id| id["type"] == "BIPM" }
-      did["primary"] = true
-      return unless did["id"] == "BIPM Brochure"
+      # isbn = hash["docid"].detect { |id| id["type"] == "ISBN" }
+      # num = isbn && isbn["id"] == "978-92-822-2272-0" ?  "SI Brochure" : "SI Brochure, Appendix 4"
 
-      isbn = hash["docid"].detect { |id| id["type"] == "ISBN" }
-      num = if isbn && isbn["id"] == "978-92-822-2272-0"
-              "SI Brochure"
-            else
-              "SI Brochure, Appendix 4"
-            end
-      hash["id"] = hash["id"].sub(/(?<=^BIPM)Brochure$/i, num.gsub(/[,\s]/, ""))
-      hash["docnumber"] = hash["docnumber"].sub(/^Brochure$/i, num)
-      did["id"] = did["id"].sub(/(?<=^BIPM\s)Brochure$/i, num)
+      hash["docid"].each do |id|
+        next unless id["type"] == "BIPM" && id["id"].match?(/BIPM Brochure/i)
+
+        id["primary"] = true
+        id["id"].sub!(/(?<=^BIPM\s)(Brochure)/i, "SI \\1")
+      end
+
+      num = hash["docid"].detect { |id| id["primary"] && id["language"] == "en" }["id"]
+      hash["docnumber"].sub!(/^Brochure$/i, num.sub(/^BIPM\s/, ""))
+      hash["id"] = num.gsub(/[,\s]/, "")
     end
 
     #

data/lib/relaton_bipm/data_outcomes_parser.rb

@@ -230,7 +230,7 @@ module RelatonBipm
     # @param [String] path path to YAML file
     #
     def add_to_index(item, path) # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
-      key = Id.new(item.docnumber).to_hash
+      key = Id.new.parse(item.docnumber).to_hash
       @data_fetcher.index2.add_or_update key, path
     end
 

data/lib/relaton_bipm/id_parser.rb

@@ -1,60 +1,63 @@
 module RelatonBipm
   class Id
-    class Parser < Parslet::Parser
-      rule(:space) { match("\s").repeat(1) }
-      rule(:space?) { space.maybe }
-      rule(:comma) { str(",") >> space? }
-      rule(:lparen) { str("(") }
-      rule(:rparen) { str(")") }
-      rule(:slash) { str("/") }
-
-      rule(:delimeter) { str("--") >> space }
-      rule(:delimeter?) { delimeter.maybe }
-
-      rule(:lang) { comma >> space? >> match["A-Z"].repeat(1, 2).as(:lang) }
-      rule(:lang?) { lang.maybe }
-
-      rule(:numdash) { match["A-Z0-9-"].repeat(1).as(:number) }
-      rule(:number) { numdash >> space? }
-      rule(:number?) { number.maybe }
-      rule(:num_suff) { numdash >> match["a-z"].repeat(1, 2) >> space }
-
-      rule(:year) { match["0-9"].repeat(4, 4).as(:year) }
-      rule(:year_paren) { lparen >> year >> lang? >> rparen }
-      rule(:num_year) { number? >> year_paren }
-      rule(:year_num) { year >> str("-") >> number }
-      rule(:num_and_year) { num_year | year_num | number }
-
-      rule(:sect) { lparen >> match["IVX"].repeat >> rparen }
-      rule(:suff) { match["a-zA-Z-"].repeat(1) }
-      rule(:cgmp) { str("CGPM") }
-      rule(:cipm) { str("CIPM") >> (str(" MRA") | match["A-Z-"]).maybe }
-      rule(:cc) { str("CC") >> suff >> sect.maybe }
-      rule(:jc) { str("JC") >> suff }
-      rule(:cec) { str("CEC") }
-      rule(:wgms) { str("WG-MS") }
-      rule(:group) { (cgmp | cipm | cc | jc | cec | wgms).as(:group) }
-
-      rule(:type) { match["[:alpha:]"].repeat(1).as(:type) >> space }
-
-      rule(:type_group) { type >> group >> slash >> num_and_year }
-      rule(:group_type) { group >> space >> delimeter? >> type >> num_and_year }
-      rule(:group_num) { group >> space >> num_suff >> type >> year_paren }
-      rule(:outcome) { group_num | group_type | type_group }
-
-      rule(:append) { comma >> str("Appendix") >> space >> number }
-      rule(:brochure) { str("SI").as(:group) >> space >> str("Brochure").as(:type) >> append.maybe }
-
-      rule(:metrologia) { str("Metrologia").as(:group) >> (space >> match["a-zA-Z0-9\s"].repeat(1).as(:number)).maybe }
-
-      rule(:corr) { space >> str("Corrigendum").as(:corr) }
-      rule(:corr?) { corr.maybe }
-      rule(:jcgm) { group >> space >> numdash >> (str(":") >> year).maybe >> corr? }
-
-      rule(:result) { outcome | brochure | metrologia | jcgm }
-
-      root :result
-    end
+    # class Parser < Parslet::Parser
+    #   rule(:space) { match("\s").repeat(1) }
+    #   rule(:space?) { space.maybe }
+    #   rule(:comma) { str(",") >> space? }
+    #   rule(:lparen) { str("(") }
+    #   rule(:rparen) { str(")") }
+    #   rule(:slash) { str("/") }
+    #   rule(:num) { match["0-9"].repeat(1) }
+
+    #   rule(:delimeter) { str("--") >> space }
+    #   rule(:delimeter?) { delimeter.maybe }
+
+    #   rule(:lang) { comma >> space? >> match["A-Z"].repeat(1, 2).as(:lang) }
+    #   rule(:lang?) { lang.maybe }
+
+    #   rule(:numdash) { match["A-Z0-9-"].repeat(1).as(:number) }
+    #   rule(:number) { numdash >> space? }
+    #   rule(:number?) { number.maybe }
+    #   rule(:num_suff) { numdash >> match["a-z"].repeat(1, 2) >> space }
+
+    #   rule(:year) { match["0-9"].repeat(4, 4).as(:year) }
+    #   rule(:year_paren) { lparen >> year >> lang? >> rparen }
+    #   rule(:num_year) { number? >> year_paren }
+    #   rule(:year_num) { year >> str("-") >> number }
+    #   rule(:num_and_year) { num_year | year_num | number }
+
+    #   rule(:sect) { lparen >> match["IVX"].repeat >> rparen }
+    #   rule(:suff) { match["a-zA-Z-"].repeat(1) }
+    #   rule(:cgmp) { str("CGPM") }
+    #   rule(:cipm) { str("CIPM") >> (str(" MRA") | match["A-Z-"]).maybe }
+    #   rule(:cc) { str("CC") >> suff >> sect.maybe }
+    #   rule(:jc) { str("JC") >> suff }
+    #   rule(:cec) { str("CEC") }
+    #   rule(:wgms) { str("WG-MS") }
+    #   rule(:group) { (cgmp | cipm | cc | jc | cec | wgms).as(:group) }
+
+    #   rule(:type) { match["[:alpha:]"].repeat(1).as(:type) >> space }
+
+    #   rule(:type_group) { type >> group >> slash >> num_and_year }
+    #   rule(:group_type) { group >> space >> delimeter? >> type >> num_and_year }
+    #   rule(:group_num) { group >> space >> num_suff >> type >> year_paren }
+    #   rule(:outcome) { group_num | group_type | type_group }
+
+    #   rule(:part_partie) { str("Part") | str("Partie") }
+    #   rule(:part) { comma >> part_partie >> space >> num.as(:part) }
+    #   rule(:append) { (comma | space) >> (str("Appendix") | str("Annexe")) >> space >> num.as(:append) }
+    #   rule(:brochure) { str("SI").as(:group) >> space >> str("Brochure").as(:type) >> (part | append).maybe }
+
+    #   rule(:metrologia) { str("Metrologia").as(:group) >> (space >> match["a-zA-Z0-9\s"].repeat(1).as(:number)).maybe }
+
+    #   rule(:corr) { space >> str("Corrigendum").as(:corr) }
+    #   rule(:corr?) { corr.maybe }
+    #   rule(:jcgm) { group >> space >> numdash >> (str(":") >> year).maybe >> corr? }
+
+    #   rule(:result) { outcome | brochure | metrologia | jcgm }
+
+    #   root :result
+    # end
 
     TYPES = {
       "Resolution" => "RES",
@@ -75,16 +78,98 @@ module RelatonBipm
     #
     # Create a new Id object
     #
+    def initialize
+      # @id = Parser.new.parse(id)
+      # @id = parse(id)
+    # rescue Parslet::ParseFailed => e
+      # Util.warn "WARNING: Incorrect reference: `#{id}`"
+      # warn e.parse_failure_cause.ascii_tree
+      # raise RelatonBib::RequestError, e
+    end
+
     # @param [String] id id string
     #
-    def initialize(id)
-      @id = Parser.new.parse(id)
-    rescue Parslet::ParseFailed => e
+    def parse(id)
+      # str = StringScanner.new id
+      match = parse_outcome(id) || parse_brochure(id) || parse_metrologia(id) || parse_jcgm(id)
+      @id = match.named_captures.compact.transform_keys(&:to_sym)
+      self
+    rescue StandardError => e
       Util.warn "WARNING: Incorrect reference: `#{id}`"
-      # warn "[relaton-bipm] #{e.parse_failure_cause.ascii_tree}"
       raise RelatonBib::RequestError, e
     end
 
+    def parse_outcome(id)
+      parse_group_num(id) || parse_group_type(id) || parse_type_group(id)
+    end
+
+    def parse_group_num(id)
+      %r{^#{group}\s#{number}[a-z]{1,2}\s#{type}\s#{year_lang}$}.match(id)
+    end
+
+    def parse_group_type(id)
+      %r{^#{group}\s(?:--\s)?#{type}\s#{num_and_year}$}.match(id)
+    end
+
+    def parse_type_group(id)
+      %r{^#{type}\s#{group}\/#{num_and_year}$}.match(id)
+    end
+
+    def group
+      "(?<group>CGPM|CIPM(?:\\sMRA|[A-Z-])?|CC[a-zA-Z-]+[IVX]*|JC[a-zA-Z-]+|CEC|WG-MS)"
+    end
+
+    def type; "(?<type>[[:alpha:]]+)"; end
+    def number; "(?<number>[A-Z0-9-]+)"; end
+    def year; "(?<year>\\d{4})"; end
+    def lang; ",\\s?(?<lang>[A-Z]{1,2})"; end
+    def year_lang; "\\(#{year}(?:#{lang})?\\)"; end
+    def num_and_year; "(?:(?:#{number}\\s)?#{year_lang}|#{year}-#{number}|#{number})"; end
+
+    def parse_brochure(id)
+      %r{^
+        (?<group>SI)\s(?<type>Brochure)
+        (?:,?\s(?:(?:Part|Partie)\s(?<part>\d+)|(?:Appendix|Annexe)\s(?<append>\d+)))?
+      $}x.match(id)
+    end
+
+    def parse_metrologia(id)
+      %r{^(?<group>Metrologia)(?:\s(?<number>[a-zA-Z0-9\s]+))?$}.match(id)
+    end
+
+    def parse_jcgm(id)
+      %r{^#{group}\s#{number}(?::#{year})?(?:\s(?<corr>Corrigendum))?$}.match(id)
+    end
+
+    # def parse_gorup_num(str)
+    #   return unless group = parse_group(str)
+
+    #   return unless str.scan(" ") && num_suff = parse_num_suff(str)
+
+    #   return unless type = parse_type(str)
+
+    #   return unless year = parse_year_parent(str)
+
+    #   { group: group, number: num_suff, type: type, year: year }
+    # end
+
+    # def parse_group(str)
+    #   str.scan %r{CGPM|CIPM(?:\sMRA|[A-Z-])?|CC[a-zA-Z-]+[IVX]*|JC[a-zA-Z-]|CEC|WG-MS}
+    # end
+
+    # def parse_num_suff(str)
+    #   num = parse_numdash(str)
+    #   num if num && str.scan(/[a-z]{1,2}\s/)
+    # end
+
+    # def parse_number(str)
+    #   parse_numdash(str)
+    # end
+
+    # def parse_numdash(str)
+    #   str.scan(/[A-Z0-9-]+/)
+    # end
+
     #
     # Compare two Id objects
     #
@@ -129,6 +214,8 @@ module RelatonBipm
       hash[:number] = norm_num unless norm_num.nil? || norm_num.empty?
       hash[:year] = src[:year].to_s if src[:year]
       hash[:corr] = true if src[:corr]
+      hash[:part] = src[:part].to_s if src[:part]
+      hash[:append] = src[:append].to_s if src[:append]
       hash[:lang] = src[:lang].to_s if src[:lang]
       hash
     end

data/lib/relaton_bipm/rawdata_bipm_metrologia/article_parser.rb

@@ -2,7 +2,7 @@ module RelatonBipm
   module RawdataBipmMetrologia
     class ArticleParser
       ATTRS = %i[docid title contributor date copyright abstract relation series
-                 extent type doctype].freeze
+                 extent type doctype link].freeze
       #
       # Create new parser and parse document
       #
@@ -319,6 +319,14 @@ module RelatonBipm
       def parse_doctype
         DocumentType.new type: "article"
       end
+
+      def parse_link
+        @meta.xpath("./article-id[@pub-id-type='doi']").each_with_object([]) do |l, a|
+          url = "https://doi.org/#{l.text}"
+          a << RelatonBib::TypedUri.new(content: url, type: "src")
+          a << RelatonBib::TypedUri.new(content: url, type: "doi")
+        end
+      end
     end
   end
 end

data/lib/relaton_bipm/rawdata_bipm_metrologia/fetcher.rb

@@ -33,7 +33,7 @@ module RelatonBipm
           item = ArticleParser.parse path
           file = "#{item.docidentifier.first.id.downcase.gsub(' ', '-')}.#{@data_fetcher.ext}"
           out_path = File.join(@data_fetcher.output, file)
-          key = Id.new(item.docidentifier.first.id).to_hash
+          key = Id.new.parse(item.docidentifier.first.id).to_hash
           @data_fetcher.index2.add_or_update key, out_path
           @data_fetcher.write_file out_path, item
         end
@@ -76,7 +76,7 @@ module RelatonBipm
         )
         file = "#{id.downcase.gsub(' ', '-')}.#{@data_fetcher.ext}"
         path = File.join(@data_fetcher.output, file)
-        @data_fetcher.index2.add_or_update Id.new(id).to_hash, path
+        @data_fetcher.index2.add_or_update Id.new.parse(id).to_hash, path
         @data_fetcher.write_file path, item
       end
 
@@ -128,7 +128,7 @@ module RelatonBipm
       #
       # Fetch relations
       #
-      # @param (see #fetch_metrologia)
+      # @see #fetch_metrologia
       #
       # @return [Array<RelatonBib::DocumentRelation>] relations
       #

data/lib/relaton_bipm/version.rb

@@ -1,3 +1,3 @@
 module RelatonBipm
-  VERSION = "1.18.0".freeze
+  VERSION = "1.18.1".freeze
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: relaton-bipm
 version: !ruby/object:Gem::Version
-  version: 1.18.0
+  version: 1.18.1
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2024-01-06 00:00:00.000000000 Z
+date: 2024-03-23 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: faraday