relaton 2.1.0 → 3.0.0.pre.alpha.1

data/lib/relaton/ieee/idams_parser.rb

@@ -0,0 +1,323 @@
+require "cgi"
+require "ieee-idams"
+
+module Relaton
+  module Ieee
+    class IdamsParser
+      include Core::ArrayWrapper
+
+      ATTRS = %i[
+        docnumber title date docidentifier contributor abstract copyright status
+        relation source keyword ext
+      ].freeze
+
+      # Upstream IDAMS abstracts sometimes carry escaped ASCII control
+      # characters as printable tokens like `<<ETX>>`. They are meaningless
+      # in output, and `<<…>>` blows up XML serialization downstream
+      # (libxml2 reads `<<` as the start of a tag). Strip the whole family.
+      CONTROL_PLACEHOLDER_RE = /<<[A-Z]{2,5}>>/.freeze
+
+      def initialize(doc, fetcher, errors = {})
+        @doc = doc
+        @fetcher = fetcher
+        @errors = errors
+      end
+
+      #
+      # Parse IEEE document
+      #
+      # @return [Relaton::Ieee::ItemData] bibliographic item data
+      #
+      def parse
+        args = { type: "standard", language: ["en"], script: ["Latn"] }
+        ATTRS.each { |attr| args[attr] = send("parse_#{attr}") }
+        ItemData.new(**args)
+      end
+
+      def parse_docnumber
+        result = docnumber
+        @errors[:docnumber] &&= result.nil?
+        result
+      end
+
+      #
+      # Parse docnumber
+      #
+      # @return [String] PubID
+      #
+      def docnumber
+        @docnumber ||= pubid&.to_id
+      end
+
+      #
+      # Create PubID
+      #
+      # @return [Relaton::Ieee::RawbibIdParser] PubID
+      #
+      def pubid
+        @pubid ||= begin
+          normtitle = @doc.normtitle
+          stdnumber = @doc.publicationinfo.stdnumber
+          RawbibIdParser.parse(normtitle, stdnumber)
+        end
+      end
+
+      #
+      # Parse title
+      #
+      # @return [Array<Relaton::Bib::Title>]
+      #
+      def parse_title
+        result = @doc.btitle.map { |args| Bib::Title.new(**args) }
+        @errors[:title] &&= result.empty?
+        result
+      end
+
+      #
+      # Parse date
+      #
+      # @return [Array<Relaton::Bib::Date>]
+      #
+      def parse_date
+        result = @doc.bdate.map { |args| Bib::Date.new(type: args[:type], at: args[:on]) }
+        @errors[:date] &&= result.empty?
+        result
+      end
+
+      #
+      # Parse identifiers
+      #
+      # @return [Array<Relaton::Bib::Docidentifier>]
+      #
+      def parse_docidentifier # rubocop:disable Metrics/MethodLength
+        ids = @doc.isbn_doi.map { |id| id[:content] = id.delete(:id); id }
+
+        ids.unshift(content: pubid.to_s(trademark: true), scope: "trademark", type: "IEEE", primary: true)
+        ids.unshift(content: pubid.to_s, type: "IEEE", primary: true)
+
+        result = ids.map { |dcid| Bib::Docidentifier.new(**dcid) }
+        @errors[:docidentifier] &&= result.empty?
+        result
+      end
+
+      #
+      # Parse contributors
+      #
+      # @return [Array<Relaton::Bib::Contributor>]
+      #
+      def parse_contributor
+        contributors = []
+
+        # Add publisher contributor
+        name, addr = @doc.contrib_name_addr { |args| Relaton::Bib::Address.new(**args) }
+        org = create_org name, addr
+        role = Bib::Contributor::Role.new type: "publisher"
+        contributors << Relaton::Bib::Contributor.new(organization: org, role: [role])
+
+        # Add committee contributors from editorial group
+        result = contributors + parse_committee_contributors
+        @errors[:contributor] &&= result.empty?
+        result
+      end
+
+      #
+      # Parse committee contributors from editorial group
+      #
+      # @return [Array<Relaton::Bib::Contributor>]
+      #
+      def parse_committee_contributors
+        committees = @doc.editorialgroup
+        return [] unless committees
+
+        result = committees.map do |committee|
+          create_committee_contributor(committee)
+        end
+        @errors[:committee] &&= result.empty?
+        result
+      end
+
+      #
+      # Create a committee contributor
+      #
+      # @param [String] committee committee name
+      #
+      # @return [Relaton::Bib::Contributor]
+      #
+      def create_committee_contributor(committee)
+        desc = Bib::LocalizedMarkedUpString.new(content: "committee", language: "en", script: "Latn")
+        role = Bib::Contributor::Role.new(type: "author", description: [desc])
+
+        # Create IEEE organization with committee as subdivision
+        orgname = Bib::TypedLocalizedString.new(
+          content: "Institute of Electrical and Electronics Engineers", language: "en", script: "Latn"
+        )
+        abbr = Bib::LocalizedString.new(content: "IEEE")
+
+        # Create subdivision for the committee
+        subdiv_name = Bib::TypedLocalizedString.new(content: CGI.unescapeHTML(committee), language: "en", script: "Latn")
+        subdivision = Bib::Subdivision.new(type: "committee", name: [subdiv_name])
+
+        org = Bib::Organization.new(name: [orgname], abbreviation: abbr, subdivision: [subdivision])
+        Relaton::Bib::Contributor.new(organization: org, role: [role])
+      end
+
+      #
+      # Parse abstract
+      #
+      # @return [Array<Relaton::Bib::LocalizedMarkedUpString>]
+      #
+      def parse_abstract
+        result = @doc.volume.article.articleinfo.abstract.each_with_object([]) do |abs, acc|
+          next unless abs.abstract_type == "Standard"
+
+          content = abs.value.gsub(CONTROL_PLACEHOLDER_RE, "").strip
+          next if content.empty?
+
+          acc << Bib::Abstract.new(content: content, language: "en", script: "Latn")
+        end
+        @errors[:abstract] &&= result.empty?
+        result
+      end
+
+      #
+      # Parse copyright
+      #
+      # @return [Array<Relaton::Bib::Copyright>]
+      #
+      def parse_copyright
+        result = @doc.copyright.map do |owner, year|
+          contrib = owner.map { |own| Bib::ContributionInfo.new organization: create_org(own) }
+          Bib::Copyright.new(owner: contrib, from: year)
+        end
+        @errors[:copyright] &&= result.empty?
+        result
+      end
+
+      #
+      # Parse status
+      #
+      # @return [Relaton::Bib::Status, nil]
+      #
+      def parse_status
+        return if @doc.docstatus.nil? || @doc.docstatus.empty?
+
+        @errors[:status] &&= @doc.docstatus[:stage].nil?
+        stage = Bib::Status::Stage.new content: @doc.docstatus[:stage]
+        Bib::Status.new stage: stage
+      end
+
+      #
+      # Parse relation
+      #
+      # @return [Array<Relaton::Bib::Relation>]
+      #
+      def parse_relation # rubocop:disable Metrics/AbcSize
+        result = array(@doc.publicationinfo.standard_relationship).each_with_object([]) do |relation, acc|
+          if (ref = @fetcher.backrefs[relation.date_string])
+            rel = @fetcher.create_relation(relation.type, ref)
+            acc << rel if rel
+          elsif !relation.date_string.include?("Inactive Date") && docnumber
+            @fetcher.add_crossref(docnumber, relation)
+          end
+        end
+        @errors[:relation] &&= result.empty?
+        result
+      end
+
+      #
+      # Parce source link
+      #
+      # @return [Array<Relaton::Bib::Uri>]
+      #
+      def parse_source
+        result = @doc.link { |url| Bib::Uri.new(content: url, type: "src") }
+        @errors[:source] &&= result.empty?
+        result
+      end
+
+      #
+      # Parse keyword
+      #
+      # @return [Array<Strign>]
+      #
+      def parse_keyword
+        result = @doc.keyword.map do |kw|
+          Bib::Keyword.new(vocab: Bib::LocalizedString.new(content: CGI.unescapeHTML(kw), language: "en", script: "Latn"))
+        end
+        @errors[:keyword] &&= result.empty?
+        result
+      end
+
+      private
+
+      #
+      # Create organization
+      #
+      # @param [String] name organization's name
+      # @param [Array<Hash>] addr address
+      #
+      # @return [Relaton::Bib::Organization]
+      def create_org(name, addr = []) # rubocop:disable Metrics/MethodLength
+        case name
+        when "IEEE"
+          abbr = Bib::LocalizedString.new content: name
+          n = "Institute of Electrical and Electronics Engineers"
+          uri = Bib::Uri.new content: "http://www.ieee.org", type: "org"
+        when "ANSI"
+          abbr = Bib::LocalizedString.new content: name
+          n = "American National Standards Institute"
+          uri = Bib::Uri.new content: "https://www.ansi.org", type: "org"
+        else n = name
+        end
+        orgname = Bib::TypedLocalizedString.new(content: n, language: "en", script: "Latn")
+        Bib::Organization.new(name: [orgname], abbreviation: abbr, uri: [uri], address: create_address(addr))
+      end
+
+      def create_address(addr)
+        addr.map { |ad| Bib::Address.new(**ad) }
+      end
+
+      def parse_ext
+        standard_status = @doc.publicationinfo.standard_status
+        standard_modified = @doc.standard_modifier
+        pubstatus = @doc.publicationinfo.pubstatus
+        holdstatus = @doc.publicationinfo.holdstatus
+        @errors[:standard_status] &&= standard_status.nil?
+        @errors[:standard_modified] &&= standard_modified.nil?
+        @errors[:pubstatus] &&= pubstatus.nil?
+        @errors[:holdstatus] &&= holdstatus.nil?
+        Ext.new(
+          doctype: parse_doctype,
+          flavor: "ieee",
+          ics: parse_ics,
+          standard_status: standard_status,
+          standard_modified: standard_modified,
+          pubstatus: pubstatus,
+          holdstatus: holdstatus,
+        )
+      end
+
+      #
+      # Parse doctype
+      #
+      # @return [Relaton::Ieee::Doctype] doctype
+      #
+      def parse_doctype
+        Doctype.new content: @doc.doctype
+      end
+
+      #
+      # Parse ICS
+      #
+      # @return [Array<Relaton::Bib::ICS>]
+      #
+      def parse_ics
+        result = @doc.ics.each_with_object([]) do |ics, acc|
+          acc << Bib::ICS.new(**ics) if ics[:code] && !ics[:code].empty?
+        end
+        @errors[:ics] &&= result.empty?
+        result
+      end
+    end
+  end
+end

data/lib/relaton/ieee/item.rb

@@ -0,0 +1,11 @@
+require_relative "ext"
+
+module Relaton
+  module Ieee
+    class Item < Bib::Item
+      model ItemData
+
+      attribute :ext, Ext
+    end
+  end
+end

data/lib/relaton/ieee/item_data.rb

@@ -0,0 +1,7 @@
+module Relaton
+  module Ieee
+    class ItemData < Bib::ItemData
+
+    end
+  end
+end

data/lib/relaton/ieee/processor.rb

@@ -0,0 +1,68 @@
+require "relaton/core/processor"
+
+module Relaton
+  module Ieee
+    class Processor < Core::Processor
+      attr_reader :idtype
+
+      def initialize # rubocop:disable Lint/MissingSuper
+        @short = :relaton_ieee
+        @prefix = "IEEE"
+        @defaultprefix = %r{^(?:(?:(?:ANSI|NACE)/)?IEEE|ANSI|AIEE|ASA|NACE|IRE)\s}
+        @idtype = "IEEE"
+        @datasets = %w[ieee-rawbib]
+      end
+
+      # @param code [String]
+      # @param date [String, NilClass] year
+      # @param opts [Hash]
+      # @return [Relaton::Ieee::ItemData, nil]
+      def get(code, date, opts)
+        require_relative "../ieee"
+        Bibliography.get(code, date, opts)
+      end
+
+      #
+      # Fetch all the documents from ./iee-rawbib directory
+      #
+      # @param [String] source source name
+      # @param [Hash] opts
+      # @option opts [String] :output directory to output documents
+      # @option opts [String] :format
+      #
+      def fetch_data(source, opts)
+        require_relative "data_fetcher"
+        DataFetcher.fetch(source, **opts)
+      end
+
+      # @param xml [String]
+      # @return [Relaton::Ieee::ItemData]
+      def from_xml(xml)
+        require_relative "../ieee"
+        Item.from_xml xml
+      end
+
+      # @param yaml [String]
+      # @return [Relaton::Ieee::ItemData]
+      def hash_to_bib(yaml)
+        require_relative "../ieee"
+        Item.from_yaml yaml
+      end
+
+      # Returns hash of XML grammar
+      # @return [String]
+      def grammar_hash
+        require_relative "../ieee"
+        @grammar_hash ||= Ieee.grammar_hash
+      end
+
+      #
+      # Remove index file
+      #
+      def remove_index_file
+        require_relative "../ieee"
+        Relaton::Index.find_or_create(:ieee, url: true, file: "#{INDEXFILE}.yaml").remove_file
+      end
+    end
+  end
+end

data/lib/relaton/ieee/pub_id.rb

@@ -0,0 +1,161 @@
+module Relaton
+  module Ieee
+    class PubId
+      class Id
+        # @return [String]
+        attr_reader :number
+
+        # @return [String, nil]
+        attr_reader :publisher, :std, :stage, :part, :status, :approval, :edition,
+                    :draft, :rev, :corr, :amd, :redline, :year, :month
+
+        #
+        # PubId constructor
+        #
+        # @param [String] number
+        # @param [<Hash>] **args
+        # @option args [String] :number
+        # @option args [String] :publisher
+        # @option args [Boolean] :std
+        # @option args [String] :stage
+        # @option args [String] :part
+        # @option args [String] :status
+        # @option args [String] :approval
+        # @option args [String] :edition
+        # @option args [String] :draft
+        # @option args [String] :rev
+        # @option args [String] :corr
+        # @option args [String] :amd
+        # @option args [Boolean] :redline
+        # @option args [String] :year
+        # @option args [String] :month
+        #
+        def initialize(number:, **args) # rubocop:disable Metrics/MethodLength, Metrics/AbcSize
+          @publisher = args[:publisher]
+          @std = args[:std]
+          @stage = args[:stage]
+          @number = number
+          @part = args[:part]
+          @status = args[:status]
+          @approval = args[:approval]
+          @edition = args[:edition]
+          @draft = args[:draft]
+          @rev = args[:rev]
+          @corr = args[:corr]
+          @amd = args[:amd]
+          @year = args[:year]
+          @month = args[:month]
+          @redline = args[:redline]
+        end
+
+        #
+        # PubId string representation
+        #
+        # @param [Boolean] trademark if true, add trademark symbol
+        #
+        # @return [String]
+        #
+        def to_s(trademark: false) # rubocop:disable Metrics/AbcSize, Metrics/MethodLength, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
+          out = number
+          out = "Std #{out}" if std
+          out = "#{stage} #{out}" if stage
+          out = "#{approval} #{out}" if approval
+          out = "#{status} #{out}" if status
+          out = "#{publisher} #{out}" if publisher
+          out += ".#{part}" if part
+          if trademark
+            out += out.match?(/^IEEE\s(Std\s)?(802|2030)/) ? "\u00AE" : "\u2122"
+          end
+          out += edition_to_s + draft_to_s + rev_to_s + corr_to_s + amd_to_s
+          out + year_to_s + month_to_s + redline_to_s
+        end
+
+        def edition_to_s
+          edition ? "/E-#{edition}" : ""
+        end
+
+        def draft_to_s
+          draft ? "/D-#{draft}" : ""
+        end
+
+        def rev_to_s
+          rev ? "/R-#{rev}" : ""
+        end
+
+        def corr_to_s
+          corr ? "/Cor#{corr}" : ""
+        end
+
+        def amd_to_s
+          amd ? "/Amd#{amd}" : ""
+        end
+
+        def year_to_s
+          year ? "-#{year}" : ""
+        end
+
+        def month_to_s
+          month ? "-#{month}" : ""
+        end
+
+        def redline_to_s
+          redline ? " Redline" : ""
+        end
+      end
+
+      # @return [Array<RelatonIeee::PubId::Id>]
+      attr_reader :pubid
+
+      #
+      # IEEE publication id
+      #
+      # @param [Array<Hash>, Hash] pubid
+      #
+      def initialize(pubid)
+        @pubid = array(pubid).map { |id| Id.new(**id) }
+      end
+
+      #
+      # Convert to array
+      #
+      # @param [Array<Hash>, Hash] pid
+      #
+      # @return [Array<Hash>]
+      #
+      def array(pid)
+        pid.is_a?(Array) ? pid : [pid]
+      end
+
+      #
+      # PubId string representation
+      #
+      # @param [Boolean] trademark if true, add trademark symbol
+      #
+      # @return [String]
+      #
+      def to_s(trademark: false)
+        pubid.map { |id| id.to_s(trademark: trademark) }.join("/")
+      end
+
+      #
+      # Generate ID without publisher and second number
+      #
+      # @return [String]
+      #
+      def to_id # rubocop:disable Metrics/MethodLength,Metrics/AbcSize,Metrics/CyclomaticComplexity,Metrics/PerceivedComplexity
+        out = pubid[0].to_s
+        if pubid.size > 1
+          out += pubid[1].edition_to_s if pubid[0].edition.nil?
+          out += pubid[1].draft_to_s if pubid[0].draft.nil?
+          out += pubid[1].rev_to_s if pubid[0].rev.nil?
+          out += pubid[1].corr_to_s if pubid[0].corr.nil?
+          out += pubid[1].amd_to_s if pubid[0].amd.nil?
+          out += pubid[1].year_to_s if pubid[0].year.nil?
+          out += pubid[1].month_to_s if pubid[0].month.nil?
+          out += pubid[1].redline_to_s unless pubid[0].redline
+        end
+        out
+      end
+    end
+  end
+end