relaton-ieee 1.19.1 → 1.19.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: d868eabc41398aaf2703d4f318117057d4936e3a77f06c9cbdced96fa04445dc
-  data.tar.gz: 3aceeac56b681d0b327340ebdde53479a58b6f0d03404223ae33514202f28843
+  metadata.gz: 295351404fc4f7272682fa1dff9dbfdaa43c9b0846e52f4891d5340136b92dda
+  data.tar.gz: b3d70bf86c583d09d000852c0669c74fb72c3b892553ab5d127f123ec37d8d5c
 SHA512:
-  metadata.gz: 91a548d6535f9267f818869ee32fbad3faa25831f658031f978ccbea2ef844046fd79a07501560cf0b193832557ee36c7d300a220b54b220453745afddefaafb
-  data.tar.gz: ed854341d691cff8777b91a82a2588ceb25fbac6b643adc5d14a27ebe0441253b14de1e76a615b426f27bd97e4578e2d424d75e9aa6cd685184e9cc4cae4b44b
+  metadata.gz: ec9f90d5dcd6c4817222fd655de5fa6f05af253ef648cd25611458d8759dfb9a915fb84eb98d91bb43bbdde96944dfc23cd5221cf8af21940eb8c37527ad0f8c
+  data.tar.gz: d0f32bb45d68aa4410859c74468abf22e2604da838fc213beae327911e792c16b996f7f32ee7050b51a4655981e0bdde5fc80d317976d8121a1a59176cc518ca

data/lib/relaton_ieee/data_fetcher.rb

@@ -1,5 +1,5 @@
 require "zip"
-require "relaton_ieee/data_parser"
+require "relaton_ieee/idams_parser"
 require "relaton_ieee/rawbib_id_parser"
 
 module RelatonIeee
@@ -88,27 +88,25 @@ module RelatonIeee
     # @param [String] filename source file
     #
     def fetch_doc(xml, filename) # rubocop:disable Metrics/AbcSize,Metrics/MethodLength,Metrics/CyclomaticComplexity,Metrics/PerceivedComplexity
-      doc = Nokogiri::XML(xml).at("/publication")
-      unless doc
+      begin
+        doc = Ieee::Idams::Publication.from_xml(xml)
+      rescue StandardError
         Util.warn "Empty file: `#{filename}`"
         return
       end
-      stdid = doc.at("./publicationinfo/standard_id")&.text
-      return if stdid == "0"
+      return if doc.publicationinfo&.standard_id == "0"
 
-      fetcher = DataParser.new doc, self
-      bib = fetcher.parse
+      bib = IdamsParser.new(doc, self).parse
       if bib.docnumber.nil?
-        nt = doc&.at("./normtitle")&.text
-        Util.warn "PubID parse error. Normtitle: `#{nt}`, file: `#{filename}`"
+        Util.warn "PubID parse error. Normtitle: `#{doc.normtitle}`, file: `#{filename}`"
         return
       end
-      amsid = doc.at("./publicationinfo/amsid").text
+      amsid = doc.publicationinfo.amsid
       if backrefs.value?(bib.docidentifier[0].id) && /updates\.\d+/ !~ filename
         oamsid = backrefs.key bib.docidentifier[0].id
         Util.warn "Document exists ID: `#{bib.docidentifier[0].id}` AMSID: " \
              "`#{amsid}` source: `#{filename}`. Other AMSID: `#{oamsid}`"
-        if bib.docidentifier[0].id.include?(doc.at("./publicationinfo/stdnumber").text)
+        if bib.docidentifier.find(&:primary).id.include?(doc.publicationinfo.stdnumber)
           save_doc bib # rewrite file if the PubID matches to the stdnumber
           backrefs[amsid] = bib.docidentifier[0].id
         end
@@ -125,9 +123,9 @@ module RelatonIeee
     # @param [Nokogiri::XML::Element] amsid relation data
     #
     def add_crossref(docnumber, amsid)
-      return if RELATION_TYPES[amsid[:type]] == false
+      return if RELATION_TYPES[amsid.type] == false
 
-      ref = { amsid: amsid.text, type: amsid[:type] }
+      ref = { amsid: amsid.date_string, type: amsid.type }
       if @crossrefs[docnumber]
         @crossrefs[docnumber] << ref
       else @crossrefs[docnumber] = [ref]

data/lib/relaton_ieee/idams_parser.rb

@@ -0,0 +1,252 @@
+require "ieee-idams"
+
+module RelatonIeee
+  class IdamsParser
+    ATTRS = %i[
+      docnumber title date docid contributor abstract copyright docstatus
+      relation link keyword ics editorialgroup standard_status standard_modified
+      pubstatus holdstatus doctype
+    ].freeze
+
+    def initialize(doc, fetcher)
+      @doc = doc
+      @fetcher = fetcher
+    end
+
+    #
+    # Parse IEEE document
+    #
+    # @return [RelatonIeee::IeeeBibliographicItem]
+    #
+    def parse
+      args = { type: "standard", language: ["en"], script: ["Latn"] }
+      ATTRS.each { |attr| args[attr] = send("parse_#{attr}") }
+      IeeeBibliographicItem.new(**args)
+    end
+
+    def parse_docnumber
+      docnumber
+    end
+
+    #
+    # Parse docnumber
+    #
+    # @return [String] PubID
+    #
+    def docnumber
+      @docnumber ||= pubid&.to_id
+    end
+
+    #
+    # Create PubID
+    #
+    # @return [RelatonIeee::RawbibIdParser] PubID
+    #
+    def pubid
+      @pubid ||= begin
+        normtitle = @doc.normtitle
+        stdnumber = @doc.publicationinfo.stdnumber
+        RawbibIdParser.parse(normtitle, stdnumber)
+      end
+    end
+
+    #
+    # Parse title
+    #
+    # @return [Array<RelatonBib::TypedTitleString>]
+    #
+    def parse_title
+      @doc.btitle.map { |args| RelatonBib::TypedTitleString.new(**args) }
+    end
+
+    #
+    # Parse date
+    #
+    # @return [Array<RelatonBib::BibliographicDate>]
+    #
+    def parse_date
+      @doc.bdate.map { |args| RelatonBib::BibliographicDate.new(**args) }
+    end
+
+    #
+    # Parse identifiers
+    #
+    # @return [Array<RelatonBib::DocumentIdentifier>]
+    #
+    def parse_docid # rubocop:disable Metrics/MethodLength
+      ids = @doc.isbn_doi
+
+      ids.unshift(id: pubid.to_s(trademark: true), scope: "trademark", type: "IEEE", primary: true)
+      ids.unshift(id: pubid.to_s, type: "IEEE", primary: true)
+
+      ids.map { |dcid| RelatonBib::DocumentIdentifier.new(**dcid) }
+    end
+
+    #
+    # Parse contributors
+    #
+    # @return [Array<RelatonBib::ContributionInfo>]
+    #
+    def parse_contributor
+      name, addr = @doc.contrib_name_addr { |args| RelatonBib::Address.new(**args) }
+
+      entity = create_org name, addr
+      [RelatonBib::ContributionInfo.new(entity: entity, role: [type: "publisher"])]
+    end
+
+    #
+    # Parse abstract
+    #
+    # @return [Array<RelatonBib::FormattedString>]
+    #
+    def parse_abstract
+      @doc.volume.article.articleinfo.abstract.map do |abs|
+        RelatonBib::FormattedString.new(content: abs.value, language: "en", script: "Latn")
+      end
+    end
+
+    #
+    # Parse copyright
+    #
+    # @return [Array<RelatonBib::CopyrightAssociation>]
+    #
+    def parse_copyright
+      @doc.copyright.map do |owner, year|
+        contrib = owner.map { |own| RelatonBib::ContributionInfo.new entity: create_org(own) }
+        RelatonBib::CopyrightAssociation.new(owner: contrib, from: year)
+      end
+    end
+
+    #
+    # Parse status
+    #
+    # @return [RelatonIee::DocumentStatus, nil]
+    #
+    def parse_docstatus
+      @doc.docstatus { |args| DocumentStatus.new(**args) }
+    end
+
+    #
+    # Parse relation
+    #
+    # @return [RelatonBib::DocRelationCollection]
+    #
+    def parse_relation # rubocop:disable Metrics/AbcSize
+      rels = []
+      @doc.publicationinfo.standard_relationship&.each do |relation|
+        if (ref = @fetcher.backrefs[relation.date_string])
+          rel = @fetcher.create_relation(relation.type, ref)
+          rels << rel if rel
+        elsif !relation.date_string.include?("Inactive Date") && docnumber
+          @fetcher.add_crossref(docnumber, relation)
+        end
+      end
+      RelatonBib::DocRelationCollection.new rels
+    end
+
+    #
+    # Parce link
+    #
+    # @return [Array<RelatonBib::TypedUri>]
+    #
+    def parse_link
+      @doc.link { |url| RelatonBib::TypedUri.new(content: url, type: "src") }
+    end
+
+    #
+    # Parse keyword
+    #
+    # @return [Array<Strign>]
+    #
+    def parse_keyword
+      @doc.keyword
+    end
+
+    #
+    # Parse ICS
+    #
+    # @return [Array<RelatonBib::ICS>]
+    #
+    def parse_ics
+      @doc.ics { |ics| RelatonBib::ICS.new(**ics) }
+    end
+
+    #
+    # Parse editorialgroup
+    #
+    # @return [RelatonIeee::EditorialGroup, nil] editorialgroup or nil
+    #
+    def parse_editorialgroup
+      @doc.editorialgroup { |committee| EditorialGroup.new committee: committee }
+    end
+
+    #
+    # Parse standard status
+    #
+    # @return [String, nil] standard status or nil
+    #
+    def parse_standard_status
+      @doc.publicationinfo.standard_status
+    end
+
+    #
+    # Parse standard modifier
+    #
+    # @return [String, nil] standard modifier or nil
+    #
+    def parse_standard_modified
+      @doc.standard_modifier
+    end
+
+    #
+    # Parse pubstatus
+    #
+    # @return [String, nil] pubstatus or nil
+    #
+    def parse_pubstatus
+      @doc.publicationinfo.pubstatus
+    end
+
+    #
+    # Pasrse holdstatus
+    #
+    # @return [String, nil] holdstatus or nil
+    #
+    def parse_holdstatus
+      @doc.publicationinfo.holdstatus
+    end
+
+    #
+    # Parse doctype
+    #
+    # @return [String] doctype
+    #
+    def parse_doctype
+      DocumentType.new type: @doc.doctype
+    end
+
+    private
+
+    #
+    # Create organization
+    #
+    # @param [String] name organization's name
+    # @param [Array<Hash>] addr address
+    #
+    # @return [RelatonBib::Organization]
+    def create_org(name, addr = []) # rubocop:disable Metrics/MethodLength
+      case name
+      when "IEEE"
+        abbr = name
+        n = "Institute of Electrical and Electronics Engineers"
+        url = "http://www.ieee.org"
+      when "ANSI"
+        abbr = name
+        n = "American National Standards Institute"
+        url = "https://www.ansi.org"
+      else n = name
+      end
+      RelatonBib::Organization.new(name: n, abbreviation: abbr, url: url, contact: addr)
+    end
+  end
+end

data/lib/relaton_ieee/version.rb

@@ -1,3 +1,3 @@
 module RelatonIeee
-  VERSION = "1.19.1".freeze
+  VERSION = "1.19.2".freeze
 end

data/relaton_ieee.gemspec

@@ -32,6 +32,7 @@ Gem::Specification.new do |spec|
   spec.require_paths = ["lib"]
 
   spec.add_dependency "faraday", "~> 2.7.0"
+  spec.add_dependency "ieee-idams", "~> 0.2.10"
   spec.add_dependency "mini_portile2", "~> 2.8.0"
   spec.add_dependency "relaton-bib", "~> 1.19.0"
   spec.add_dependency "relaton-index", "~> 0.2.0"

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: relaton-ieee
 version: !ruby/object:Gem::Version
-  version: 1.19.1
+  version: 1.19.2
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2024-10-25 00:00:00.000000000 Z
+date: 2024-11-27 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: faraday
@@ -24,6 +24,20 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: 2.7.0
+- !ruby/object:Gem::Dependency
+  name: ieee-idams
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: 0.2.10
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: 0.2.10
 - !ruby/object:Gem::Dependency
   name: mini_portile2
   requirement: !ruby/object:Gem::Requirement
@@ -109,11 +123,11 @@ files:
 - lib/relaton_ieee/balloting_group.rb
 - lib/relaton_ieee/bibxml_parser.rb
 - lib/relaton_ieee/data_fetcher.rb
-- lib/relaton_ieee/data_parser.rb
 - lib/relaton_ieee/document_status.rb
 - lib/relaton_ieee/document_type.rb
 - lib/relaton_ieee/editorial_group.rb
 - lib/relaton_ieee/hash_converter.rb
+- lib/relaton_ieee/idams_parser.rb
 - lib/relaton_ieee/ieee_bibliographic_item.rb
 - lib/relaton_ieee/ieee_bibliography.rb
 - lib/relaton_ieee/processor.rb

data/lib/relaton_ieee/data_parser.rb

@@ -1,350 +0,0 @@
-module RelatonIeee
-  class DataParser
-    DATETYPES = { "OriginalPub" => "created", "ePub" => "published",
-                  "LastInspecUpd" => "updated" }.freeze
-    ATTRS = %i[
-      docnumber title date docid contributor abstract copyright docstatus
-      relation link keyword ics editorialgroup standard_status standard_modified
-      pubstatus holdstatus doctype
-    ].freeze
-
-    attr_reader :doc, :fetcher
-
-    #
-    # Create RelatonIeee::DataParser instance
-    #
-    # @param [Nokogiri::XML::Element] doc document
-    # @param [RelatonIeee::DataFetcher] fetcher
-    #
-    def initialize(doc, fetcher)
-      @doc = doc
-      @fetcher = fetcher
-    end
-
-    #
-    # Parse IEEE document
-    #
-    # @param [Nokogiri::XML::Element] doc document
-    # @param [RelatonIeee::DataFetcher] fetcher <description>
-    #
-    # @return [RelatonIeee::IeeeBibliographicItem]
-    #
-    def self.parse(doc, fetcher)
-      new(doc, fetcher).parse
-    end
-
-    #
-    # Parse IEEE document
-    #
-    # @return [RelatonIeee::IeeeBibliographicItem]
-    #
-    def parse
-      args = { type: "standard", language: ["en"], script: ["Latn"] }
-      ATTRS.each { |attr| args[attr] = send("parse_#{attr}") }
-      IeeeBibliographicItem.new(**args)
-    end
-
-    #
-    # Parse title
-    #
-    # @return [Array<RelatonBib::TypedTitleString>]
-    #
-    def parse_title
-      t = []
-      content = CGI.unescapeHTML doc.at("./volume/article/title").text
-      if content =~ /\A(.+)\s[-\u2014]\sredline\z/i
-        t << RelatonBib::TypedTitleString.new(content: $1, type: "title-main")
-      end
-      t << RelatonBib::TypedTitleString.new(content: content, type: "main")
-    end
-
-    #
-    # Parse date
-    #
-    # @return [Array<RelatonBib::BibliographicDate>]
-    #
-    def parse_date # rubocop:disable Metrics/AbcSize,Metrics/CyclomaticComplexity,Metrics/MethodLength,Metrics/PerceivedComplexity
-      dates = doc.xpath("./volume/article/articleinfo/date").map do |d|
-        da = [d.at("./year").text]
-        m = d.at("./month")&.text
-        if m
-          /^(?:(?<day>\d{1,2})\s)?(?<mon>\w+)/ =~ m
-          month = Date::ABBR_MONTHNAMES.index(mon) || Date::MONTHNAMES.index(mon) || m
-          da << month.to_s.rjust(2, "0")
-        end
-        day = d.at("./day")&.text || day
-        da << day.rjust(2, "0") if day
-        on = da.compact.join "-"
-        RelatonBib::BibliographicDate.new type: DATETYPES[d[:datetype]], on: on
-      end
-      pad = doc.at("./publicationinfo/PubApprovalDate")
-      if pad
-        issued = parse_date_string pad.text
-        dates << RelatonBib::BibliographicDate.new(type: "issued", on: issued)
-      end
-      dates
-    end
-
-    #
-    # Convert date string with month name to numeric date
-    #
-    # @param [String] date source date
-    #
-    # @return [String] numeric date
-    #
-    def parse_date_string(date)
-      case date
-      when /^\d{4}$/ then date
-      when /^\d{1,2}\s\w+\.?\s\d{4}/ then Date.parse(date).to_s
-      end
-    end
-
-    #
-    # Parse identifiers
-    #
-    # @return [Array<RelatonBib::DocumentIdentifier>]
-    #
-    def parse_docid # rubocop:disable Metrics/MethodLength
-      ids = [
-        { id: pubid.to_s, type: "IEEE", primary: true },
-        { id: pubid.to_s(trademark: true), scope: "trademark", type: "IEEE", primary: true },
-      ]
-      isbn = doc.at("./publicationinfo/isbn")
-      ids << { id: isbn.text, type: "ISBN" } if isbn
-      doi = doc.at("./volume/article/articleinfo/articledoi")
-      ids << { id: doi.text, type: "DOI" } if doi
-      ids.map do |dcid|
-        RelatonBib::DocumentIdentifier.new(**dcid)
-      end
-    end
-
-    #
-    # Create PubID
-    #
-    # @return [RelatonIeee::RawbibIdParser] PubID
-    #
-    def pubid
-      @pubid ||= begin
-        normtitle = doc.at("./normtitle").text
-        stdnumber = doc.at("./publicationinfo/stdnumber")&.text
-        RawbibIdParser.parse(normtitle, stdnumber)
-      end
-    end
-
-    def parse_docnumber
-      docnumber
-    end
-
-    #
-    # Parse docnumber
-    #
-    # @return [String] PubID
-    #
-    def docnumber
-      @docnumber ||= pubid&.to_id # doc.at("./publicationinfo/stdnumber").text
-    end
-
-    #
-    # Parse contributors
-    #
-    # @return [Array<RelatonBib::ContributionInfo>]
-    #
-    def parse_contributor # rubocop:disable Metrics/AbcSize,Metrics/MethodLength
-      doc.xpath("./publicationinfo/publisher").map do |contrib|
-        n = contrib.at("./publishername").text
-        addr = contrib.xpath("./address").each_with_object([]) do |adr, ob|
-          city, country, state = parse_country_city adr
-          next unless city && country
-
-          ob << RelatonBib::Address.new(street: [], city: city, state: state, country: country)
-        end
-        e = create_org n, addr
-        RelatonBib::ContributionInfo.new entity: e, role: [type: "publisher"]
-      end
-    end
-
-    def parse_country_city(address)
-      city = address.at("./city")
-      return unless city
-
-      city, state = city.text.split(", ")
-      country = address.at("./country")&.text || "USA"
-      [city, country, state]
-    end
-
-    #
-    # Create organization
-    #
-    # @param [String] name organization's name
-    # @param [Array<Hash>] addr address
-    #
-    # @return [RelatonBib::Organization]
-    def create_org(name, addr = []) # rubocop:disable Metrics/MethodLength
-      case name
-      when "IEEE"
-        abbr = name
-        n = "Institute of Electrical and Electronics Engineers"
-        url = "http://www.ieee.org"
-      when "ANSI"
-        abbr = name
-        n = "American National Standards Institute"
-        url = "https://www.ansi.org"
-      else n = name
-      end
-      RelatonBib::Organization.new(
-        name: n, abbreviation: abbr, url: url, contact: addr,
-      )
-    end
-
-    #
-    # Parse abstract
-    #
-    # @return [Array<RelatonBib::FormattedString>]
-    #
-    def parse_abstract
-      doc.xpath("./volume/article/articleinfo/abstract")[0...1].map do |a|
-        RelatonBib::FormattedString.new(
-          content: a.text, language: "en", script: "Latn",
-        )
-      end
-    end
-
-    #
-    # Parse copyright
-    #
-    # @return [Array<RelatonBib::CopyrightAssociation>]
-    #
-    def parse_copyright
-      doc.xpath("./publicationinfo/copyrightgroup/copyright").map do |c|
-        owner = c.at("./holder").text.split("/").map do |own|
-          RelatonBib::ContributionInfo.new entity: create_org(own)
-        end
-        RelatonBib::CopyrightAssociation.new(
-          owner: owner, from: c.at("./year").text,
-        )
-      end
-    end
-
-    #
-    # Parse status
-    #
-    # @return [RelatonIee::DocumentStatus, nil]
-    #
-    def parse_docstatus
-      st = parse_standard_modified
-      return unless %w[Draft Approved Superseded Withdrawn].include?(st)
-
-      DocumentStatus.new stage: st.downcase
-    end
-
-    #
-    # Parse relation
-    #
-    # @return [RelatonBib::DocRelationCollection]
-    #
-    def parse_relation # rubocop:disable Metrics/AbcSize
-      rels = []
-      doc.xpath("./publicationinfo/standard_relationship").each do |r|
-        if (ref = fetcher.backrefs[r.text])
-          rel = fetcher.create_relation(r[:type], ref)
-          rels << rel if rel
-        elsif !"Inactive Date".include?(r) && docnumber
-          fetcher.add_crossref(docnumber, r)
-        end
-      end
-      RelatonBib::DocRelationCollection.new rels
-    end
-
-    #
-    # Parce link
-    #
-    # @return [Array<RelatonBib::TypedUri>]
-    #
-    def parse_link
-      doc.xpath("./volume/article/articleinfo/amsid").map do |id|
-        l = "https://ieeexplore.ieee.org/document/#{id.text}"
-        RelatonBib::TypedUri.new content: l, type: "src"
-      end
-    end
-
-    #
-    # Parse keyword
-    #
-    # @return [Array<Strign>]
-    #
-    def parse_keyword
-      doc.xpath(
-        "./volume/article/articleinfo/keywordset/keyword/keywordterm",
-      ).map &:text
-    end
-
-    #
-    # Parse ICS
-    #
-    # @return [Array<RelatonBib::ICS>]
-    #
-    def parse_ics
-      doc.xpath("./publicationinfo/icscodes/code_term").map do |ics|
-        RelatonBib::ICS.new code: ics[:codenum], text: ics.text
-      end
-    end
-
-    #
-    # Parse editorialgroup
-    #
-    # @return [RelatonIeee::EditorialGroup, nil] editorialgroup or nil
-    #
-    def parse_editorialgroup
-      committee = doc.xpath(
-        "./publicationinfo/pubsponsoringcommitteeset/pubsponsoringcommittee",
-      ).map &:text
-      EditorialGroup.new committee: committee if committee.any?
-    end
-
-    #
-    # Parse standard status
-    #
-    # @return [String, nil] standard status or nil
-    #
-    def parse_standard_status
-      doc.at("./publicationinfo/standard_status")&.text
-    end
-
-    #
-    # Parse standard modifier
-    #
-    # @return [String, nil] standard modifier or nil
-    #
-    def parse_standard_modified
-      doc.at("./publicationinfo/standardmodifierset/standard_modifier")&.text
-    end
-
-    #
-    # Parse pubstatus
-    #
-    # @return [String, nil] pubstatus or nil
-    #
-    def parse_pubstatus
-      doc.at("./publicationinfo/pubstatus")&.text
-    end
-
-    #
-    # Pasrse holdstatus
-    #
-    # @return [String, nil] holdstatus or nil
-    #
-    def parse_holdstatus
-      doc.at("./publicationinfo/holdstatus")&.text
-    end
-
-    #
-    # Parse doctype
-    #
-    # @return [String] doctype
-    #
-    def parse_doctype
-      type = parse_standard_modified == "Redline" ? "redline" : "standard"
-      DocumentType.new type: type
-    end
-  end
-end