relaton-bipm 1.14.3 → 1.14.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 34d720b316dbd942e2c5d630d2ae0f07b74331e4ef07f68715e84304bad0fb13
-  data.tar.gz: 38d36e34b998db6e4fa9e9f1a6e5306fadacea45b9a878cd14629eaaca2ef50d
+  metadata.gz: e0d5eb5717b605c2d3eec083e4ca529ccafb140246b1b8a042137d1025b41abb
+  data.tar.gz: bdef48071e8e5ab07d643e54fed9f21c38d29119927620216a23d2d2d9c84a7b
 SHA512:
-  metadata.gz: a22261617d5c3de8aad7ed410091331698630f958332c5feb0b215b9fafe9167015530e9e6ecb71046307a6775aa7a2a0a71dd4c12e4f982cb0bd259e021a267
-  data.tar.gz: 376bb090dd4d273b8039357d78280c9bc4f1555918920a55b22ec72cb8be87c4ce0a8469254ffc000256fa95069a271b65b978dc767d735acd4b3e141c5dea24
+  metadata.gz: '09e75649f7ca561179506cf24c0eb9c06b2c2e6241aac712e3aaa01e85985542d7086d907068acc365b0b93386e83160646e41d67585b1c9a8e98c8b5c281f55'
+  data.tar.gz: ca8540f40846fe5d92b0b2da284c72cb3a44a4dfd580ad4a51eb6c625cfd400aa59b7a435c46af45244540482d02aff1bd9aa70b3cc5359a41b873602b5d5ecd

data/.github/workflows/rake.yml

@@ -5,6 +5,7 @@ name: rake
 on:
   push:
     branches: [ master, main ]
+    tags: [ v* ]
   pull_request:
 
 jobs:

data/.github/workflows/release.yml

@@ -10,8 +10,9 @@ on:
           Next release version. Possible values: x.y.z, major, minor, patch or pre|rc|etc
         required: true
         default: 'skip'
-  push:
-    tags: [ v* ]
+  repository_dispatch:
+    types: [ do-release ]
+
 
 jobs:
   release:

data/README.adoc

@@ -91,13 +91,13 @@ Short:
 {group name} {type-abbrev} {number} ({year}, {lang})
 ----
 
-`group name` - a name of the group, required. A full list of group names is available https://github.com/metanorma/bipm-editor-guides/blob/main/sources/bipm-outcomes-en.adoc#appendix-a-bipm-groups-and-codes[here].
-`type` - a type of document, required. A list of types is: Resolution (Résolution), Recommendation (Recommandation), Decision (Décision), Meeting (Réunion), Declaration (Déclaration).
-`type-abbrev` - an abbreviation of the type, required. A list of abbreviations: RES (Resolution), REC (Recommendation), DECN (Decision).
-`number` - a number of the document, optional. Can be with part, e.g. `1-2`.
-`zero_leading_number` - a number of the document with a leading zero, required. Can be used when a document has a 1 or 2 digits number. It's `00` for documents without a number.
-`year` - a year of the document, optional.
-`lang` - a language of the document, optional. Can be `EN` or `FR`.
+- `group name` - a name of the group, required. A full list of group names is available https://github.com/metanorma/bipm-editor-guides/blob/main/sources/bipm-outcomes-en.adoc#appendix-a-bipm-groups-and-codes[here].
+- `type` - a type of document, required. A list of types is: Resolution (Résolution), Recommendation (Recommandation), Decision (Décision), Meeting (Réunion), Declaration (Déclaration).
+- `type-abbrev` - an abbreviation of the type, required. A list of abbreviations: RES (Resolution), REC (Recommendation), DECN (Decision).
+- `number` - a number of the document, optional. Can be with part, e.g. `1-2`.
+- `zero_leading_number` - a number of the document with a leading zero, required. Can be used when a document has a 1 or 2 digits number. It's `00` for documents without a number.
+- `year` - a year of the document, optional.
+- `lang` - a language of the document, optional. Can be `EN` or `FR`.
 
 ===== Special case pattern
 
@@ -209,54 +209,54 @@ bib = RelatonBipm::BipmBibliography.get "BIPM Metrologia 29 6 001"
 ...
 
 # get CGPM meetings
-RelatonBipm::BipmBibliography.get "CGPM -- Meeting 1 (1889)"
-[relaton-bipm] ("CGPM -- Meeting 1 (1889)") fetching...
-[relaton-bipm] ("CGPM -- Meeting 1 (1889)") found CGPM -- Meeting 1 (1889)
+RelatonBipm::BipmBibliography.get "CGPM 1st Meeting (1889)"
+[relaton-bipm] ("CGPM 1st Meeting (1889)") fetching...
+[relaton-bipm] ("CGPM 1st Meeting (1889)") found CGPM 1th meeting (1889)
 => #<RelatonBipm::BipmBibliographicItem:0x00007f7fd02aba28
 ...
 
 # get CGPM resolutions
- RelatonBipm::BipmBibliography.get "CGPM -- RES (1889, EN)"
-[relaton-bipm] ("CGPM -- RES (1889, EN)") fetching...
-[relaton-bipm] ("CGPM -- RES (1889, EN)") found CGPM -- Resolution (1889)
+RelatonBipm::BipmBibliography.get "CGPM RES (1889, E)"
+[relaton-bipm] ("CGPM RES (1889, E)") fetching...
+[relaton-bipm] ("CGPM RES (1889, E)") found CGPM RES (1889)
 => #<RelatonBipm::BipmBibliographicItem:0x00007f80421f93d8
 ...
 
-RelatonBipm::BipmBibliography.get "CGPM -- Resolution (1889)"
-[relaton-bipm] ("CGPM -- Resolution (1889)") fetching...
-[relaton-bipm] ("CGPM -- Resolution (1889)") found CGPM -- Resolution (1889)
+RelatonBipm::BipmBibliography.get "CGPM Resolution (1889)"
+[relaton-bipm] ("CGPM Resolution (1889)") fetching...
+[relaton-bipm] ("CGPM Resolution (1889)") found CGPM RES (1889)
 => #<RelatonBipm::BipmBibliographicItem:0x00007f8017f60c18
 ...
 
-RelatonBipm::BipmBibliography.get "CGPM -- Résolution (1889)"
-[relaton-bipm] ("CGPM -- Résolution (1889)") fetching...
-[relaton-bipm] ("CGPM -- Résolution (1889)") found CGPM -- Resolution (1889)
+RelatonBipm::BipmBibliography.get "CGPM Résolution (1889)"
+[relaton-bipm] ("CGPM Résolution (1889)") fetching...
+[relaton-bipm] ("CGPM Résolution (1889)") found CGPM RES (1889)
 => #<RelatonBipm::BipmBibliographicItem:0x00007f8017f492e8
 ...
 
 # get CIPM decision by year and decision number
 RelatonBipm::BipmBibliography.get "BIPM Decision CIPM/101-1 (2012)"
 [relaton-bipm] ("BIPM Decision CIPM/101-1 (2012)") fetching...
-[relaton-bipm] ("BIPM Decision CIPM/101-1 (2012)") found Decision CIPM/101-1 (2012)
+[relaton-bipm] ("BIPM Decision CIPM/101-1 (2012)") found CIPM DECN 101-1 (2012)
 => #<RelatonBipm::BipmBibliographicItem:0x00007f8017f2bd88
 ...
 
-RelatonBipm::BipmBibliography.get "BIPM DECN CIPM/101-1 (2012, EN)"
-[relaton-bipm] ("BIPM DECN CIPM/101-1 (2012, EN)") fetching...
-[relaton-bipm] ("BIPM DECN CIPM/101-1 (2012, EN)") found Decision CIPM/101-1 (2012)
+RelatonBipm::BipmBibliography.get "BIPM DECN CIPM/101-1 (2012, E)"
+[relaton-bipm] ("BIPM DECN CIPM/101-1 (2012, E)") fetching...
+[relaton-bipm] ("BIPM DECN CIPM/101-1 (2012, E)") found CIPM DECN 101-1 (2012)
 => #<RelatonBipm::BipmBibliographicItem:0x00007f8017f39438
 ...
 
 # get CIPM recommendation
-RelatonBipm::BipmBibliography.get "CIPM -- Recommendation 1 (2005)"
-[relaton-bipm] ("CIPM -- Recommendation 1 (2005)") fetching...
-[relaton-bipm] ("CIPM -- Recommendation 1 (2005)") found CIPM -- Recommendation 1 (2005)
+RelatonBipm::BipmBibliography.get "CIPM Recommendation 1 (2005)"
+[relaton-bipm] ("CIPM Recommendation 1 (2005)") fetching...
+[relaton-bipm] ("CIPM Recommendation 1 (2005)") found CIPM REC 1 (2005)
 => #<RelatonBipm::BipmBibliographicItem:0x00007f8017f31da0
 ...
 
-RelatonBipm::BipmBibliography.get "CIPM -- REC 1 (2005, FR)"
-[relaton-bipm] ("CIPM -- REC 1 (2005, FR)") fetching...
-[relaton-bipm] ("CIPM -- REC 1 (2005, FR)") found CIPM -- Recommendation 1 (2005)
+RelatonBipm::BipmBibliography.get "CIPM REC 1 (2005, FR)"
+[relaton-bipm] ("CIPM REC 1 (2005, FR)") fetching...
+[relaton-bipm] ("CIPM REC 1 (2005, FR)") found CIPM REC 1 (2005)
 => #<RelatonBipm::BipmBibliographicItem:0x00007f80422100d8
 ...
 ----

data/lib/relaton_bipm/bipm_bibliographic_item.rb

@@ -2,9 +2,9 @@ module RelatonBipm
   class BipmBibliographicItem < RelatonBib::BibliographicItem
     include RelatonBib
 
-    TYPES = %w[brochure mise-en-pratique rapport monographie guide
-               meeting-report technical-report working-party-note strategy
-               cipm-mra resolutions].freeze
+    DOCTYPES = %w[brochure mise-en-pratique rapport monographie guide
+                  meeting-report technical-report working-party-note strategy
+                  cipm-mra resolutions].freeze
 
     STATUSES = %w[draft-proposal draft-development in-force retired].freeze
 

data/lib/relaton_bipm/bipm_bibliography.rb

@@ -44,7 +44,7 @@ module RelatonBipm
       # @return [RelatonBipm::BipmBibliographicItem]
       def get_bipm(reference, agent) # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
         ref_id = Id.new reference
-        index = Relaton::Index.find_or_create :BIPM, url: "#{GH_ENDPOINT}index2.zip"
+        index = Relaton::Index.find_or_create :BIPM, url: "#{GH_ENDPOINT}index2.zip", file: "index2.yaml"
         rows = index.search { |r| ref_id == r[:id] }
         return unless rows.any?
 

data/lib/relaton_bipm/bipm_si_brochure_parser.rb

@@ -38,7 +38,7 @@ module RelatonBipm
         key = hash1["docnumber"] || basename
         @data_fetcher.index[[key]] = outfile
         @data_fetcher.index_new.add_or_update [key], outfile
-        @data_fetcher.index2.add_or_update Id.new(key).normalized_hash, outfile
+        @data_fetcher.index2.add_or_update Id.new(key).to_hash, outfile
         hash = if File.exist? outfile
                  warn_duplicate = false
                  hash2 = YAML.load_file outfile

data/lib/relaton_bipm/data_outcomes_parser.rb

@@ -1,18 +1,20 @@
 module RelatonBipm
   class DataOutcomesParser
-    TYPEABBREV = {
+    SHORTTYPE = {
       "Resolution" => "RES",
       "Recommendation" => "REC",
       "Decision" => "DECN",
       "Statement" => "DECL",
+      "Declaration" => "DECL",
+      "Action" => "ACT",
     }.freeze
 
     TRANSLATIONS = {
-      "Déclaration" => "Declaration",
-      "Réunion" => "Meeting",
-      "Recommandation" => "Recommendation",
-      "Résolution" => "Resolution",
-      "Décision" => "Decision",
+      "Declaration" => "Déclaration",
+      "Meeting" => "Réunion",
+      "Recommendation" => "Recommandation",
+      "Resolution" => "Résolution",
+      "Decision" => "Décision",
     }.freeze
 
     #
@@ -228,19 +230,19 @@ module RelatonBipm
     # @param [String] path path to YAML file
     #
     def add_to_index(item, path) # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
-      key = [item.docnumber]
-      TYPEABBREV.each do |k, v|
-        if item.docnumber.include? k
-          key << item.docnumber.sub(k, v)
-          key << item.docnumber.sub(k, v).sub(/(\(\d{4})(\))/, "\\1, EN\\2")
-          key << item.docnumber.sub(k, v).sub(/(\(\d{4})(\))/, "\\1, FR\\2")
-          break
-        end
-      end
-      key << item.docidentifier.detect { |i| i.language == "fr" }.id
+      # key = [item.docnumber]
+      # SHORTTYPE.each do |k, v|
+      #   if item.docnumber.include? k
+      #     key << item.docnumber.sub(k, v)
+      #     key << item.docnumber.sub(k, v).sub(/(\(\d{4})(\))/, "\\1, EN\\2")
+      #     key << item.docnumber.sub(k, v).sub(/(\(\d{4})(\))/, "\\1, FR\\2")
+      #     break
+      #   end
+      # end
+      key = item.docidentifier.select { |i| i.type == "BIPM" }.map &:id
       @data_fetcher.index[key] = path
       @data_fetcher.index_new.add_or_update key, path
-      key2 = Id.new(item.docnumber).normalized_hash
+      key2 = Id.new(item.docnumber).to_hash
       @data_fetcher.index2.add_or_update key2, path
     end
 
@@ -349,12 +351,12 @@ module RelatonBipm
     # @param [String] session number of meeting
     #
     def add_part(hash, part)
-      regex = /(\p{L}+\s(?:--\s\p{L}+\s|\w+\/)\d+)/
+      regex = /(\p{L}+\s(?:\w+\/)?\d+)(?![\d-])/
       hash[:id] += "-#{part}"
       hash[:docnumber].sub!(regex) { |m| "#{m}-#{part}" }
       hash[:docid].select { |id| id.type == "BIPM" }.each do |did|
-        id = did.instance_variable_get(:@id).sub!(regex) { "#{$1}-#{part}" }
-        did.instance_variable_set(:@id, id)
+        did.instance_variable_get(:@id).sub!(regex) { "#{$1}-#{part}" }
+        # did.instance_variable_set(:@id, id)
       end
       hash[:structuredidentifier].instance_variable_set :@part, part
     end
@@ -374,12 +376,12 @@ module RelatonBipm
     # @return [Hash] Hash of BIPM meeting/resolution
     #
     def bibitem(**args) # rubocop:disable Metrics/MethodLength, Metrics/AbcSize, Metrics/CyclomaticComplexity
-      docnum = create_docnum args[:body], args[:type], args[:num], args[:en]["date"]
+      docnum = create_meeting_docnum args[:body], args[:type], args[:num], args[:en]["date"]
       hash = { title: [], type: "proceedings", doctype: args[:type],
                place: [RelatonBib::Place.new(city: "Paris")] }
       hash[:title] = create_titles args.slice(:en, :fr)
       hash[:date] = [{ type: "published", on: args[:en]["date"] }]
-      hash[:docid] = create_docids docnum
+      hash[:docid] = create_meeting_docids docnum
       hash[:docnumber] = docnum # .sub(" --", "").sub(/\s\(\d{4}\)/, "")
       hash[:id] = create_id(args[:body], args[:type], args[:num], args[:en]["date"])
       hash[:link] = create_links(**args)
@@ -426,16 +428,21 @@ module RelatonBipm
     #
     def create_docnum(body, type, num, date)
       year = Date.parse(date).year
-      if special_id_case? body, type, year
-        id = "#{type.capitalize} #{body}"
-        id += "/#{num}" if num.to_i.positive?
-      else
-        id = "#{body} -- #{type.capitalize}"
-        id += " #{num}" if num.to_i.positive?
-      end
+      # if special_id_case? body, type, year
+      #   id = "#{type.capitalize} #{body}"
+      #   id += "/#{num}" if num.to_i.positive?
+      # else
+      id = "#{body} #{SHORTTYPE[type.capitalize]}"
+      id += " #{num}" if num.to_i.positive?
+      # end
       "#{id} (#{year})"
     end
 
+    def create_meeting_docnum(body, type, num, date)
+      year = Date.parse(date).year
+      "#{body} #{num}th #{type} (#{year})"
+    end
+
     #
     # Create ID
     #
@@ -448,13 +455,11 @@ module RelatonBipm
     #
     def create_id(body, type, num, date)
       year = Date.parse(date).year
-      id = if special_id_case?(body, type, year)
-             "#{type.capitalize}-#{body}-#{year}"
-           else
-             "#{body}-#{type.capitalize}-#{year}"
-           end
-      id += "-#{num}" if num.to_i.positive?
-      id
+      # if special_id_case?(body, type, year)
+      #   [type.capitalize, body, year]
+      # else
+      [body, SHORTTYPE[type.capitalize], year, num].compact.join("-")
+      # end
     end
 
     #
@@ -466,10 +471,10 @@ module RelatonBipm
     #
     # @return [Boolean] is special case
     #
-    def special_id_case?(body, type, year)
-      (body == "CIPM" && type == "Decision" && year.to_i > 2011) ||
-        (body == "JCRB" && %w[recomendation resolution descision].include?(type))
-    end
+    # def special_id_case?(body, type, year)
+    #   (body == "CIPM" && type == "Decision" && year.to_i > 2011) ||
+    #     (body == "JCRB" && %w[recomendation resolution descision].include?(type))
+    # end
 
     #
     # Create documetn IDs
@@ -478,12 +483,22 @@ module RelatonBipm
     #
     # @return [Array<RelatonBib::DocumentIdentifier>] document IDs
     #
-    def create_docids(en_id)
-      id = en_id.clone
+    def create_docids(id)
+      en_id = id.sub(/(\s\(\d{4})(\))$/, '\1, E\2')
+      fr_id = id.sub(/(\s\(\d{4})(\))$/, '\1, F\2')
       [
         make_docid(id: id, type: "BIPM", primary: true),
-        make_docid(id: id.clone, type: "BIPM", primary: true, language: "en", script: "Latn"),
-        create_docid_fr(en_id),
+        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"),
+        # create_docid_fr(en_id),
+      ]
+    end
+
+    def create_meeting_docids(en_id)
+      fr_id = en_id.sub(/(\d+)th/, '\1e').sub("meeting", "réunion")
+      [
+        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"),
       ]
     end
 
@@ -494,11 +509,11 @@ module RelatonBipm
     #
     # @return [RelatonBib::DocumentIdentifier] french document ID
     #
-    def create_docid_fr(en_id)
-      tr = TRANSLATIONS.detect { |_, v| en_id.include? v }
-      id = tr ? en_id.sub(tr[1], tr[0]) : en_id
-      make_docid(id: id, type: "BIPM", primary: true, language: "fr", script: "Latn")
-    end
+    # def create_docid_fr(en_id)
+    #   tr = TRANSLATIONS.detect { |_, v| en_id.include? v }
+    #   id = tr ? en_id.sub(tr[1], tr[0]) : en_id
+    #   make_docid(id: id, type: "BIPM", primary: true, language: "fr", script: "Latn")
+    # end
 
     #
     # Create doucment ID

data/lib/relaton_bipm/id_parser.rb

@@ -11,11 +11,13 @@ module RelatonBipm
       rule(:delimeter) { str("--") >> space }
       rule(:delimeter?) { delimeter.maybe }
 
-      rule(:lang) { comma >> match["A-Z"].repeat(2, 2).as(:lang) }
+      rule(:lang) { comma >> space? >> match["A-Z"].repeat(1, 2).as(:lang) }
       rule(:lang?) { lang.maybe }
 
-      rule(:number) { match["0-9-"].repeat(1).as(:number) >> space? }
+      rule(:numdash) { match["0-9-"].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 }
@@ -37,7 +39,8 @@ module RelatonBipm
 
       rule(:type_group) { type >> group >> slash >> num_and_year }
       rule(:group_type) { group >> space >> delimeter? >> type >> num_and_year }
-      rule(:outcome) { group_type | type_group }
+      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 }
@@ -58,6 +61,7 @@ module RelatonBipm
       "Décision" => "DECN",
       "Declaration" => "Déclaration",
       "Réunion" => "Meeting",
+      "Action" => "ACT",
     }.freeze
 
     # @return [Hash] the parsed id components
@@ -84,8 +88,8 @@ module RelatonBipm
     # @return [Boolean] true if the two Id objects are equal
     #
     def ==(other)
-      other_hash = other.is_a?(Id) ? other.normalized_hash : other
-      hash = normalized_hash
+      other_hash = other.is_a?(Id) ? other.to_hash : normalize_hash(other)
+      hash = to_hash
       hash.delete(:year) unless other_hash[:year]
       other_hash.delete(:year) unless hash[:year]
       hash.delete(:lang) unless other_hash[:lang]
@@ -99,16 +103,26 @@ module RelatonBipm
     #
     # @return [Hash] the normalized ID parts
     #
-    def normalized_hash # rubocop:disable Metrics/AbcSize, Metrics/CyclomaticComplexity
-      @normalized_hash ||= begin
-        hash = { group: id[:group].to_s.sub("CCDS", "CCTF") }
-        hash[:type] = normalized_type if id[:type]
-        norm_num = normalized_number
-        hash[:number] = norm_num unless norm_num.nil? || norm_num.empty?
-        hash[:year] = id[:year].to_s if id[:year]
-        hash[:lang] = id[:lang].to_s if id[:lang]
-        hash
-      end
+    def to_hash
+      @to_hash ||= normalize_hash id
+    end
+
+    #
+    # Normalize ID parts
+    # Traslate type into abbreviation, remove leading zeros from number
+    #
+    # @param [RelatonBipm::Id, Hash] src the ID parts
+    #
+    # @return [Hash] the normalized ID parts
+    #
+    def normalize_hash(src) # rubocop:disable Metrics/AbcSize
+      hash = { group: src[:group].to_s.sub("CCDS", "CCTF") }
+      hash[:type] = normalized_type(src) if src[:type]
+      norm_num = normalized_number(src)
+      hash[:number] = norm_num unless norm_num.nil? || norm_num.empty?
+      hash[:year] = src[:year].to_s if src[:year]
+      hash[:lang] = src[:lang].to_s if src[:lang]
+      hash
     end
 
     #
@@ -116,8 +130,9 @@ module RelatonBipm
     #
     # @return [String] the normalized type
     #
-    def normalized_type
-      TYPES[id[:type].to_s] || id[:type].to_s
+    def normalized_type(src)
+      type = TYPES[src[:type].to_s.capitalize] || src[:type].to_s
+      type == type.upcase ? type : type.capitalize
     end
 
     #
@@ -125,10 +140,10 @@ module RelatonBipm
     #
     # @return [String, nil] the normalized number
     #
-    def normalized_number
-      return unless id[:number]
+    def normalized_number(src)
+      return unless src[:number]
 
-      id[:number].to_s.sub(/^0+/, "")
+      src[:number].to_s.sub(/^0+/, "")
     end
   end
 end

data/lib/relaton_bipm/processor.rb

@@ -7,7 +7,7 @@ module RelatonBipm
     def initialize
       @short = :relaton_bipm
       @prefix = "BIPM"
-      @defaultprefix = %r{^(?:BIPM|CCTF|CCDS|CGPM|CIPM)(?!\w)}
+      @defaultprefix = %r{^(?:BIPM|CCTF|CCDS|CGPM|CIPM|JCRB)(?!\w)}
       @idtype = "BIPM"
       @datasets = %w[bipm-data-outcomes bipm-si-brochure rawdata-bipm-metrologia]
     end

data/lib/relaton_bipm/rawdata_bipm_metrologia/article_parser.rb

@@ -6,22 +6,30 @@ module RelatonBipm
       #
       # Create new parser and parse document
       #
-      # @param [Nokogiri::XML::Element] doc document XML element
+      # @param [String] path path to XML file
       #
       # @return [RelatonBipm::BipmBibliographicItem] document
       #
-      def self.parse(doc)
-        new(doc).parse
+      def self.parse(path)
+        doc = Nokogiri::XML(File.read(path, encoding: "UTF-8"))
+        journal, volume, article = path.split("/")[-2].split("_")[1..]
+        new(doc, journal, volume, article).parse
       end
 
       #
       # Initialize parser
       #
-      # @param [Nokogiri::XML::Element] doc XML document
-      #
-      def initialize(doc)
-        @doc = doc
-        @meta = @doc.at("./front/article-meta")
+      # @param [Nokogiri::XML::Document] doc XML document
+      # @param [String] journal journal
+      # @param [String] volume volume
+      # @param [String] article article
+      #
+      def initialize(doc, journal, volume, article)
+        @doc = doc.at "/article"
+        @journal = journal
+        @volume = volume
+        @article = article
+        @meta = doc.at("/article/front/article-meta")
       end
 
       #
@@ -54,15 +62,15 @@ module RelatonBipm
       # @return [Array<String>] array of volume, issue and page
       #
       def volume_issue_article
-        volume = @meta.at("./volume").text
-        issue = @meta.at("./issue").text
+        # volume = @meta.at("./volume").text
+        # issue = @meta.at("./issue").text
         # page = @doc.at("./front/article-meta/fpage")&.text || manuscript
-        [volume, issue, article].join(" ")
+        [@journal, @volume, @article].compact.join(" ")
       end
 
-      def article
-        @meta.at("./article-id[@pub-id-type='manuscript']").text.match(/[^_]+$/).to_s
-      end
+      # def article
+      #   @meta.at("./article-id[@pub-id-type='manuscript']").text.match(/[^_]+$/).to_s
+      # end
 
       #
       # Parse journal title
@@ -95,7 +103,10 @@ module RelatonBipm
         @meta.xpath("./title-group/article-title").map do |t|
           next if t.text.empty?
 
-          RelatonBib::TypedTitleString.new content: t.text, language: t[:"xml:lang"], script: "Latn"
+          format = CGI.escapeHTML(t.inner_html) == t.inner_html ? "text/plain" : "text/html"
+          RelatonBib::TypedTitleString.new(
+            content: t.inner_html, language: t[:"xml:lang"], script: "Latn", format: format,
+          )
         end.compact
       end
 

data/lib/relaton_bipm/rawdata_bipm_metrologia/fetcher.rb

@@ -30,13 +30,12 @@ module RelatonBipm
       #
       def fetch_articles # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
         Dir["#{DIR}/**/*.xml"].each do |path|
-          doc = Nokogiri::XML File.read(path, encoding: "UTF-8")
-          item = ArticleParser.parse doc.at("/article")
+          item = ArticleParser.parse path
           file = "#{item.docidentifier.first.id.downcase.gsub(' ', '-')}.#{@data_fetcher.ext}"
           out_path = File.join(@data_fetcher.output, file)
           @data_fetcher.index[[item.docidentifier.first.id]] = out_path
           @data_fetcher.index_new.add_or_update [item.docidentifier.first.id], out_path
-          key = Id.new(item.docidentifier.first.id).normalized_hash
+          key = Id.new(item.docidentifier.first.id).to_hash
           @data_fetcher.index2.add_or_update key, out_path
           @data_fetcher.write_file out_path, item
         end
@@ -81,7 +80,7 @@ module RelatonBipm
         path = File.join(@data_fetcher.output, file)
         @data_fetcher.index[[id]] = path
         @data_fetcher.index_new.add_or_update [id], path
-        @data_fetcher.index2.add_or_update Id.new(id).normalized_hash, path
+        @data_fetcher.index2.add_or_update Id.new(id).to_hash, path
         @data_fetcher.write_file path, item
       end
 
@@ -139,11 +138,12 @@ module RelatonBipm
       #
       def relation(*args)
         dir = [DIR, *args].join("/")
-        Dir["#{dir}/*"].map do |path|
+        ids = Set.new
+        Dir["#{dir}/*"].each do |path|
           part = path.split("/").last
-          id = identifier(*args, part)
-          RelatonBib::DocumentRelation.new(type: "partOf", bibitem: rel_bibitem(id))
+          ids << identifier(*args, part)
         end
+        ids.map { |id| RelatonBib::DocumentRelation.new(type: "partOf", bibitem: rel_bibitem(id)) }
       end
 
       #

data/lib/relaton_bipm/version.rb

@@ -1,3 +1,3 @@
 module RelatonBipm
-  VERSION = "1.14.3".freeze
+  VERSION = "1.14.5".freeze
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: relaton-bipm
 version: !ruby/object:Gem::Version
-  version: 1.14.3
+  version: 1.14.5
 platform: ruby
 authors:
 - Ribose Inc.
-autorequire: 
+autorequire:
 bindir: exe
 cert_chain: []
-date: 2023-03-26 00:00:00.000000000 Z
+date: 2023-05-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: equivalent-xml
@@ -175,7 +175,7 @@ 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
@@ -190,8 +190,8 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.1.6
-signing_key: 
+rubygems_version: 3.4.9
+signing_key:
 specification_version: 4
 summary: 'RelatonBipm: retrieve BIPM Standards for bibliographic use using the BibliographicItem
   model'