relaton-nist 1.9.4 → 1.9.9

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 847b161ffdf47b18befc71b090e2397e815594f4c20820b057c93608d171487f
-  data.tar.gz: 73748bbb1ff975876b6346f24a89ec235f4224d61aa4f714e13b8b981ece81cd
+  metadata.gz: e35a427023ba3ef9662b0e1c005717e82ea3086854c615d4a69302b62f20803f
+  data.tar.gz: 92ee933270aaabe628a8639e8d24cf550b031bb918c04a4aa72cc7e59ad5f314
 SHA512:
-  metadata.gz: 57640ab1c61feddcf2ac798128d46aff89b190670efa11b624afe9270c94c832c4f53b1cae07dc44f2d189d14b828c1ef1cd1b55d40ac5b94bd65954cd875efa
-  data.tar.gz: 8df687fcfae413e7efa74d6fbe1fbda4835d61b2c6e3751629365497e83e5f288ec5dcebce0b4472193b81a734d5f401babf1472c212370cfc817b8bc7d130bc
+  metadata.gz: 3d5a16eab60a743ee45012c494fb88af160498cb031fe66f176356dde77a29a80e87106d6c99d3cba9e70826ba47dbc987e2b0514d513c5470377492604325b4
+  data.tar.gz: 5ad90f3a527f7381e98510d71f2d0507569ce0b36bdf6f47b1b1e15510edbc54342ec53063e7e25ec36d97699ff586dbfc84b787efcaf546ba03e1d5fb3de2f9

data/grammars/biblio.rng

@@ -401,9 +401,9 @@
       <choice>
         <!-- iso191606 TODO -->
         <group>
-          <oneOrMore>
+          <zeroOrMore>
             <ref name="street"/>
-          </oneOrMore>
+          </zeroOrMore>
           <ref name="city"/>
           <optional>
             <ref name="state"/>

data/lib/relaton_nist/data_fetcher.rb

@@ -11,6 +11,8 @@ module RelatonNist
       "isTranslationOf" => "translatedFrom",
       "hasPreprint" => "hasReprint",
       "isSupplementTo" => "complements",
+      "isPartOf" => "partOf",
+      "hasPart" => "hasPart",
     }.freeze
     URL = "https://raw.githubusercontent.com/usnistgov/NIST-Tech-Pubs/nist-pages/xml/allrecords.xml"
 
@@ -20,24 +22,48 @@ module RelatonNist
       @ext = format.sub(/^bib/, "")
     end
 
-    def parse_docid(doc)
-      doi = doc.at("doi_data/doi").text
-      id = doc.at("publisher_item/item_number", "publisher_item/identifier").text.sub(%r{^/}, "")
-      case doi
-      when "10.6028/NBS.CIRC.12e2revjune" then id.sub!("13e", "12e")
-      when "10.6028/NBS.CIRC.36e2" then id.sub!("46e", "36e")
-      when "10.6028/NBS.HB.67suppJune1967" then id.sub!("1965", "1967")
-      when "10.6028/NBS.HB.105-1r1990" then id.sub!("105-1-1990", "105-1r1990")
-      when "10.6028/NIST.HB.150-10-1995" then id.sub!(/150-10$/, "150-10-1995")
+    def parse_docid(doc) # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
+      # case doi
+      # when "10.6028/NBS.CIRC.12e2revjune" then doi.sub!("13e", "12e")
+      # when "10.6028/NBS.CIRC.36e2" then doi.sub!("46e", "36e")
+      # when "10.6028/NBS.HB.67suppJune1967" then doi.sub!("1965", "1967")
+      # when "10.6028/NBS.HB.105-1r1990" then doi.sub!("105-1-1990", "105-1r1990")
+      # when "10.6028/NIST.HB.150-10-1995" then doi.sub!(/150-10$/, "150-10-1995")
+      # end
+      # anchor = doi.split("/")[1..-1].join "/"
+      [
+        { type: "NIST", id: pub_id(doc) },
+        { type: "DOI", id: doi(doc) },
+        { type: "NIST", id: anchor(doc), scope: "anchor" },
+      ]
+    end
+
+    def pub_id(doc)
+      anchor(doc).gsub(".", " ")
+    end
+
+    def doi(doc) # rubocop:disable Metrics/CyclomaticComplexity
+      id = doc.at("doi_data/doi").text
+      case id
+      when "10.6028/NBS.CIRC.e2e" then "10.6028/NBS.CIRC.2e2"
+      when "10.6028/NBS.CIRC.sup" then "10.6028/NBS.CIRC.24e7sup"
+      when "10.6028/NBS.CIRC.supJun1925-Jun1926" then "10.6028/NBS.CIRC.24e7sup2"
+      when "10.6028/NBS.CIRC.supJun1925-Jun1927" then "10.6028/NBS.CIRC.24e7sup3"
+      when "10.6028/NBS.CIRC.24supJuly1922" then "10.6028/NBS.CIRC.24e6sup"
+      when "10.6028/NBS.CIRC.24supJan1924" then "10.6028/NBS.CIRC.24e6sup2"
+      else id
       end
-      [{ type: "NIST", id: id }, { type: "DOI", id: doi }]
+    end
+
+    def anchor(doc)
+      doi(doc).split("/")[1..-1].join "/"
     end
 
     # @param doc [Nokogiri::XML::Element]
     # @return [Array<RelatonBib::DocumentIdentifier>]
     def fetch_docid(doc)
       parse_docid(doc).map do |id|
-        RelatonBib::DocumentIdentifier.new(type: id[:type], id: id[:id])
+        RelatonBib::DocumentIdentifier.new(**id)
       end
     end
 
@@ -47,7 +73,7 @@ module RelatonNist
       t = doc.xpath("titles/title|titles/subtitle")
       return [] unless t.any?
 
-      RelatonBib::TypedTitleString.from_string t.map(&:text).join(" "), "en", "Latn"
+      RelatonBib::TypedTitleString.from_string t.map(&:text).join, "en", "Latn"
     end
 
     # @param doc [Nokogiri::XML::Element]
@@ -76,12 +102,11 @@ module RelatonNist
     def fetch_relation(doc)
       ns = "http://www.crossref.org/relations.xsd"
       doc.xpath("./ns:program/ns:related_item", ns: ns).map do |rel|
-        doi = rel.at_xpath("ns:intra_work_relation|ns:inter_work_relation", ns: ns)
-        # ref = doi_to_id doi.text
-        # ref, = parse_docid doc
-        fref = RelatonBib::FormattedRef.new content: doi.text
+        rdoi = rel.at_xpath("ns:intra_work_relation|ns:inter_work_relation", ns: ns)
+        fref = RelatonBib::FormattedRef.new content: rdoi.text
         bibitem = RelatonBib::BibliographicItem.new formattedref: fref
-        type = RELATION_TYPES[doi["relationship-type"]]
+        type = RELATION_TYPES[rdoi["relationship-type"]]
+        warn "Relation type #{rdoi['relationship-type']} not found" unless type
         { type: type, bibitem: bibitem }
       end
     end
@@ -116,29 +141,48 @@ module RelatonNist
         end
         sname = p.at("surname").text
         surname = RelatonBib::LocalizedString.new sname, doc["language"], "Latn"
-        initial = []
         ident = p.xpath("ORCID").map do |id|
           RelatonBib::PersonIdentifier.new "orcid", id.text
         end
         fullname = RelatonBib::FullName.new(
           surname: surname, forename: forename, initial: initial, identifier: ident,
         )
-        person = RelatonBib::Person.new name: fullname
+        person = RelatonBib::Person.new name: fullname, affiliation: affiliation(doc)
         { entity: person, role: [{ type: p["contributor_role"] }] }
       end
       contribs + doc.xpath("publisher").map do |p|
         abbr = p.at("../institution/institution_acronym")&.text
-        org = RelatonBib::Organization.new(name: p.at("publisher_name").text, abbreviation: abbr)
+        place = p.at("./publisher_place")
+        cont = []
+        if place
+          city, state = place.text.split(", ")
+          cont << RelatonBib::Address.new(street: [], city: city, state: state, country: "US")
+        end
+        org = RelatonBib::Organization.new(
+          name: p.at("publisher_name").text, abbreviation: abbr, contact: cont,
+        )
         { entity: org, role: [{ type: "publisher" }] }
       end
     end
 
+    def affiliation(doc)
+      doc.xpath("./institution/institution_department").map do |id|
+        org = RelatonBib::Organization.new name: id.text
+        RelatonBib::Affiliation.new organization: org
+      end
+    end
+
     # @param doc [Nokogiri::XML::Element]
     # @return [Array<String>]
     def fetch_place(doc)
       doc.xpath("institution/institution_place").map(&:text)
     end
 
+    def fetch_series(doc)
+      title = RelatonBib::TypedTitleString.new(content: "NIST")
+      [RelatonBib::Series.new(title: title, number: pub_id(doc))]
+    end
+
     #
     # Save document
     #
@@ -174,14 +218,15 @@ module RelatonNist
         link: fetch_link(doc), abstract: fetch_abstract(doc),
         date: fetch_date(doc), edition: fetch_edition(doc),
         contributor: fetch_contributor(doc), relation: fetch_relation(doc),
-        place: fetch_place(doc),
+        place: fetch_place(doc), series: fetch_series(doc),
         language: [doc["language"]], script: ["Latn"], doctype: "standard"
       )
       write_file item
     rescue StandardError => e
       warn "Document: #{doc.at('doi').text}"
       warn e.message
-      raise e
+      warn e.backtrace[0..5].join("\n")
+      # raise e
     end
 
     #
@@ -202,6 +247,7 @@ module RelatonNist
       puts "Done in: #{(t2 - t1).round} sec."
     rescue StandardError => e
       warn e.message
+      warn e.backtrace[0..5].join("\n")
     end
 
     #

data/lib/relaton_nist/nist_bibliography.rb

@@ -28,17 +28,18 @@ module RelatonNist
       # @option opts [TrueClass, FalseClass] :bibdata
       #
       # @return [String] Relaton XML serialisation of reference
-      def get(code, year = nil, opts = {})
-        return fetch_ref_err(code, year, []) if code.match? /\sEP$/
+      def get(code, year = nil, opts = {}) # rubocop:disable Metrics/AbcSize, Metrics/MethodLength, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
+        return fetch_ref_err(code, year, []) if code.match?(/\sEP$/)
 
-        /^(?<code2>[^\(]+)(\((?<date2>\w+\s(\d{2},\s)?\d{4})\))?\s?\(?((?<=\()(?<stage>[^\)]+))?/ =~ code
+        /^(?<code2>[^(]+)(?:\((?<date2>\w+\s(?:\d{2},\s)?\d{4})\))?\s?\(?(?:(?<=\()(?<stage>[^\)]+))?/ =~ code
         stage ||= /(?<=\.)PD-\w+(?=\.)/.match(code)&.to_s
         if code2
           code = code2.strip
           if date2
-            if /\w+\s\d{4}/.match? date2
+            case date2
+            when /\w+\s\d{4}/
               opts[:issued_date] = Date.strptime date2, "%B %Y"
-            elsif /\w+\s\d{2},\s\d{4}/.match? date2
+            when /\w+\s\d{2},\s\d{4}/
               opts[:updated_date] = Date.strptime date2, "%B %d, %Y"
             end
           end

data/lib/relaton_nist/scrapper.rb

@@ -209,9 +209,15 @@ module RelatonNist
       # @param ref [String]
       # @param uri [String]
       # @return [RelatonNist::DocumentRelation]
-      def doc_relation(type, ref, uri, lang = "en", script = "Latn")
+      def doc_relation(type, ref, uri, lang = "en", script = "Latn") # rubocop:disable Metrics/MethodLength
+        if type == "supersedes"
+          descr = RelatonBib::FormattedString.new(content: "supersedes", language: lang, script: script)
+          t = "obsoletes"
+        else t = type
+        end
         DocumentRelation.new(
-          type: type,
+          type: t,
+          description: descr,
           bibitem: RelatonBib::BibliographicItem.new(
             formattedref: RelatonBib::FormattedRef.new(
               content: ref, language: lang, script: script, format: "text/plain",

data/lib/relaton_nist/version.rb

@@ -1,3 +1,3 @@
 module RelatonNist
-  VERSION = "1.9.4".freeze
+  VERSION = "1.9.9".freeze
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: relaton-nist
 version: !ruby/object:Gem::Version
-  version: 1.9.4
+  version: 1.9.9
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2022-01-12 00:00:00.000000000 Z
+date: 2022-01-21 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: equivalent-xml