relaton-w3c 1.7.1 → 1.9.1

data/lib/relaton_w3c/data_parser.rb

@@ -0,0 +1,205 @@
+module RelatonW3c
+  class DataParser
+    #
+    # Document parser initalization
+    #
+    # @param [RDF::Query::Solution] sol entry from the SPARQL query
+    # @param [RelatonW3c::DataFetcher] fetcher data fetcher
+    #
+    def initialize(sol, fetcher)
+      @sol = sol
+      @fetcher = fetcher
+    end
+
+    #
+    # Initialize document parser and run it
+    #
+    # @param [RDF::Query::Solution] sol entry from the SPARQL query
+    # @param [RelatonW3c::DataFetcher] fetcher data fetcher
+    #
+    # @return [RelatonW3c:W3cBibliographicItem, nil] bibliographic item
+    #
+    def self.parse(sol, fetcher)
+      new(sol, fetcher).parse
+    end
+
+    #
+    # Parse document
+    #
+    # @return [RelatonW3c:W3cBibliographicItem, nil] bibliographic item
+    #
+    def parse # rubocop:disable Metrics/MethodLength, Metrics/AbcSize
+      return unless @fetcher.class::USED_TYPES.include? type
+
+      RelatonW3c::W3cBibliographicItem.new(
+        type: "standard",
+        doctype: parse_doctype,
+        fetched: Date.today.to_s,
+        language: ["en"],
+        script: ["Latn"],
+        title: parse_title,
+        link: parse_link,
+        docid: parse_docid,
+        docnumber: identifier(@sol.link.to_s),
+        series: parse_series,
+        date: parse_date,
+        relation: parse_relation,
+        contributor: parse_contrib,
+        editorialgroup: parse_editorialgroup,
+      )
+    end
+
+    #
+    # Parse title
+    #
+    # @return [RelatonBib::TypedTitleStringCollection] title
+    #
+    def parse_title
+      t = RelatonBib::TypedTitleString.new title: @sol.title.to_s
+      RelatonBib::TypedTitleStringCollection.new [t]
+    end
+
+    #
+    # Parse link
+    #
+    # @return [Array<RelatonBib::TypedUri>] link
+    #
+    def parse_link
+      [RelatonBib::TypedUri.new(type: "src", content: @sol.link.to_s)]
+    end
+
+    #
+    # Parse docidentifier
+    #
+    # @return [Arra<RelatonBib::DocumentIdentifier>] docidentifier
+    #
+    def parse_docid
+      id = pub_id(@sol.link.to_s)
+      [RelatonBib::DocumentIdentifier.new(type: "W3C", id: id)]
+    end
+
+    #
+    # Generate PubID
+    #
+    # @param [String] url url
+    #
+    # @return [String] PubID
+    #
+    def pub_id(url)
+      "W3C #{identifier(url)}"
+    end
+
+    def identifier(url)
+      /.+\/(\w+(?:-[\w.]+)+(?:\/\w+)?)/.match(url)[1].to_s
+    end
+
+    #
+    # Parse series
+    #
+    # @return [Array<RelatonBib::Series>] series
+    #
+    def parse_series
+      title = RelatonBib::TypedTitleString.new content: "W3C #{type}"
+      [RelatonBib::Series.new(title: title, number: identifier(@sol.link.to_s))]
+    end
+
+    def type # rubocop:disable Metrics/MethodLength
+      @type ||= begin
+        sse = SPARQL.parse(%(
+          PREFIX : <http://www.w3.org/2001/02pd/rec54#>
+          PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
+          SELECT ?type
+          WHERE {
+            { <#{@sol.link}> rdf:type ?type }
+          }
+        ))
+        tps = @fetcher.data.query(sse).map { |s| s.type.to_s.split("#").last }
+        tps.detect { |t| Scrapper::DOCTYPES.key?(t) }
+      end
+    end
+
+    #
+    # Parse doctype
+    #
+    # @return [Strinf] doctype
+    #
+    def parse_doctype
+      Scrapper::DOCTYPES[type]
+    end
+
+    def parse_date
+      [RelatonBib::BibliographicDate.new(type: "published", on: @sol.date.to_s)]
+    end
+
+    #
+    # Parse relation
+    #
+    # @return [Array<RelatonBib::DocumentRelation>] relation
+    #
+    def parse_relation # rubocop:disable Metrics/MethodLength, Metrics/AbcSize
+      sse = SPARQL.parse(%(
+        PREFIX doc: <http://www.w3.org/2000/10/swap/pim/doc#>
+        SELECT ?obsoletes
+        WHERE {
+          VALUES ?p { doc:obsoletes }
+          { <#{@sol.link}> ?p ?obsoletes }
+        }
+      ))
+      @fetcher.data.query(sse).order_by(:obsoletes).map do |r|
+        tp, url = r.to_h.first
+        fr = RelatonBib::LocalizedString.new pub_id(url.to_s)
+        bib = W3cBibliographicItem.new formattedref: fr
+        RelatonBib::DocumentRelation.new(type: tp.to_s, bibitem: bib)
+      end
+    end
+
+    #
+    # Parse contributor
+    #
+    # @return [Array<RelatonBib::ContributionInfo>] contributor
+    #
+    def parse_contrib # rubocop:disable Metrics/MethodLength
+      sse = SPARQL.parse(%(
+        PREFIX : <http://www.w3.org/2001/02pd/rec54#>
+        PREFIX contact: <http://www.w3.org/2000/10/swap/pim/contact#>
+        SELECT ?full_name
+        WHERE {
+          <#{@sol.link}> :editor/contact:fullName ?full_name
+        }
+      ))
+      @fetcher.data.query(sse).order_by(:full_name).map do |ed|
+        cn = RelatonBib::LocalizedString.new(ed.full_name.to_s, "en", "Latn")
+        n = RelatonBib::FullName.new completename: cn
+        p = RelatonBib::Person.new name: n
+        RelatonBib::ContributionInfo.new entity: p, role: [type: "editor"]
+      end
+    end
+
+    #
+    # Parse editorialgroup
+    #
+    # @return [RelatonBib::EditorialGroup] editorialgroup
+    #
+    def parse_editorialgroup # rubocop:disable Metrics/MethodLength, Metrics/AbcSize
+      sse = SPARQL.parse(%(
+        PREFIX org: <http://www.w3.org/2001/04/roadmap/org#>
+        PREFIX contact: <http://www.w3.org/2000/10/swap/pim/contact#>
+        SELECT ?home_page
+        WHERE {
+          <#{@sol.link}> org:deliveredBy/contact:homePage ?home_page
+        }
+      ))
+      res = @fetcher.data.query(sse).order_by(:home_page)
+      tc = res.each_with_object([]) do |edg, obj|
+        wg = @fetcher.group_names[edg.home_page.to_s.sub(/\/$/, "")]
+        if wg
+          rwg = RelatonBib::WorkGroup.new name: wg["name"]
+          obj << RelatonBib::TechnicalCommittee.new(rwg)
+        else
+          warn "Working group name not found for #{edg.home_page}"
+        end
+      end
+      RelatonBib::EditorialGroup.new tc
+    end
+  end
+end

data/lib/relaton_w3c/hash_converter.rb

@@ -4,7 +4,7 @@ module RelatonW3c
       # @param item_hash [Hash]
       # @return [RelatonW3c::W3cBibliographicItem]
       def bib_item(item_hash)
-        W3cBibliographicItem.new **item_hash
+        W3cBibliographicItem.new(**item_hash)
       end
     end
   end

data/lib/relaton_w3c/hit_collection.rb

@@ -22,8 +22,8 @@ module RelatonW3c
     # @param ref [String] reference to search
     def initialize(ref)
       %r{
-        ^(W3C\s)?
-        (?<type>(CR|NOTE|PER|PR|REC|RET|WD|Candidate\sRecommendation|
+        ^(?:W3C\s)?
+        (?<type>(?:CR|NOTE|PER|PR|REC|RET|WD|Candidate\sRecommendation|
           Group\sNote|Proposed\sEdited\sRecommendation|Proposed\sRecommendation|
           Recommendation|Retired|Working\sDraft))? # type
         \s?
@@ -41,7 +41,7 @@ module RelatonW3c
     # @param title_date [String]
     # @param type [String]
     # @return [Array<Hash>]
-    def from_yaml(title_date, type)
+    def from_yaml(title_date, type) # rubocop:disable Metrics/AbcSize,Metrics/CyclomaticComplexity,Metrics/MethodLength,Metrics/PerceivedComplexity
       /(?<title>.+)\s(?<date>\d{4}-\d{2}-\d{2})$/ =~ title_date
       title ||= title_date
       result = data.select do |hit|
@@ -50,7 +50,7 @@ module RelatonW3c
           type_date_filter(hit, type, date)
       end
       if result.empty?
-        result = data.select { |h| h["link"].split("/").last.match? /#{title}/ }
+        result = data.select { |h| h["link"].split("/").last.match?(/#{title}/) }
       end
       result.map { |h| Hit.new(h, self) }
     end
@@ -109,7 +109,7 @@ module RelatonW3c
     # @param type [String]
     # @return [String]
     def short_type(type)
-      tp = TYPES.select { |k,v| v == type }.keys
+      tp = TYPES.select { |_, v| v == type }.keys
       tp.first || type
     end
 
@@ -137,7 +137,7 @@ module RelatonW3c
     # fetch data form server and save it to file.
     #
     def fetch_data
-      resp = Net::HTTP.get_response URI.parse(DOMAIN + "/TR/")
+      resp = Net::HTTP.get_response URI.parse("#{DOMAIN}/TR/")
       # return if there aren't any changes since last fetching
       return unless resp.code == "200"
 
@@ -153,7 +153,7 @@ module RelatonW3c
     # @param h_el [Nokogiri::XML::Element]
     # @param link [Nokogiri::XML::Element]
     # @param pubdetails [Nokogiri::XML::Element]
-    def fetch_hit(h_el, link, pubdetails)
+    def fetch_hit(h_el, link, pubdetails) # rubocop:disable Metrics/AbcSize,Metrics/MethodLength
       datepub = pubdetails.at("text()").text.match(/\d{4}-\d{2}-\d{2}/).to_s
       editor = h_el.xpath("ul[@class='editorlist']/li").map { |e| e.text.strip }
       keyword = h_el.xpath("ul[@class='taglist']/li").map { |e| e.text.strip }

data/lib/relaton_w3c/processor.rb

@@ -4,11 +4,12 @@ module RelatonW3c
   class Processor < Relaton::Processor
     attr_reader :idtype
 
-    def initialize
+    def initialize # rubocop:disable Lint/MissingSuper
       @short = :relaton_w3c
       @prefix = "W3C"
       @defaultprefix = %r{^W3C\s}
       @idtype = "W3C"
+      @datasets = %w[w3c-rdf]
     end
 
     # @param code [String]
@@ -19,6 +20,18 @@ module RelatonW3c
       ::RelatonW3c::W3cBibliography.get(code, date, opts)
     end
 
+    #
+    # Fetch all the documents from http://www.w3.org/2002/01/tr-automation/tr.rdf
+    #
+    # @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)
+      DataFetcher.fetch(**opts)
+    end
+
     # @param xml [String]
     # @return [RelatonCalconnect::CcBibliographicItem]
     def from_xml(xml)
@@ -29,7 +42,7 @@ module RelatonW3c
     # @return [RelatonIsoBib::CcBibliographicItem]
     def hash_to_bib(hash)
       item_hash = ::RelatonW3c::HashConverter.hash_to_bib(hash)
-      ::RelatonW3c::W3cBibliographicItem.new item_hash
+      ::RelatonW3c::W3cBibliographicItem.new(**item_hash)
     end
 
     # Returns hash of XML grammar

data/lib/relaton_w3c/scrapper.rb

@@ -1,16 +1,16 @@
 module RelatonW3c
   class Scrapper
-    class << self
-      DOCTYPES = {
-        "CR" => "candidateRecommendation",
-        "NOTE" => "groupNote",
-        "PER" => "proposedEditedRecommendation",
-        "PR" => "proposedRecommendation",
-        "REC" => "recommendation",
-        "RET" => "retired",
-        "WD" => "workingDraft",
-      }.freeze
+    DOCTYPES = {
+      "CR" => "candidateRecommendation",
+      "NOTE" => "groupNote",
+      "PER" => "proposedEditedRecommendation",
+      "PR" => "proposedRecommendation",
+      "REC" => "recommendation",
+      "RET" => "retired",
+      "WD" => "workingDraft",
+    }.freeze
 
+    class << self
       # @param hit [Hash]
       # @return [RelatonW3c::W3cBibliographicItem]
       def parse_page(hit) # rubocop:disable Metrics/AbcSize, Metrics/MethodLength
@@ -29,7 +29,7 @@ module RelatonW3c
           doctype: fetch_doctype(hit, doc),
           contributor: fetch_contributor(hit, doc),
           relation: fetch_relation(doc),
-          keyword: hit["keyword"]
+          keyword: hit["keyword"],
         )
       end
 
@@ -53,7 +53,7 @@ module RelatonW3c
             titles << { content: title.gsub(/\n/, " "), type: "main" }
           end
           subtitle = doc.at(
-            "//h2[@id='subtitle']|//p[contains(@class, 'subline')]"
+            "//h2[@id='subtitle']|//p[contains(@class, 'subline')]",
           )&.text
           titles << { content: subtitle, tipe: "subtitle" } if subtitle
         end
@@ -62,7 +62,7 @@ module RelatonW3c
         end
         titles.map do |t|
           title = RelatonBib::FormattedString.new(
-            content: t[:content], language: "en", script: "Latn"
+            content: t[:content], language: "en", script: "Latn",
           )
           RelatonBib::TypedTitleString.new(type: t[:type], title: title)
         end
@@ -88,7 +88,7 @@ module RelatonW3c
       # @param hit [Hash]
       # @param doc [Nokogiri::HTML::Document, NilClass]
       # @return [Array<RelatonBib::BibliographicDate>]
-      def fetch_date(hit, doc)
+      def fetch_date(hit, doc) # rubocop:disable Metrics/CyclomaticComplexity
         on = hit["datepub"] || doc&.at("//h2/time[@datetime]")&.attr(:datetime)
         on ||= fetch_date1(doc) || fetch_date2(doc)
         [RelatonBib::BibliographicDate.new(type: "published", on: on)] if on
@@ -143,7 +143,7 @@ module RelatonW3c
             end
             mem
           end
-          contribs.map { |c| contrib_info **c }
+          contribs.map { |c| contrib_info(**c) }
         else
           hit["editor"].map do |ed|
             contrib_info name: ed, role: [{ type: "editor" }]
@@ -162,7 +162,7 @@ module RelatonW3c
       # @param element [Nokogiri::XML::Element]
       # @param type [String]
       # @return [Hash]
-      def parse_contrib(element, type)
+      def parse_contrib(element, type) # rubocop:disable Metrics/MethodLength
         p = element.at("a")
         return unless p
 
@@ -187,7 +187,7 @@ module RelatonW3c
         name = RelatonBib::FullName.new completename: completename
         af = []
         if args[:org]
-          org = RelatonBib::Organization.new **args[:org]
+          org = RelatonBib::Organization.new(**args[:org])
           af << RelatonBib::Affiliation.new(organization: org)
         end
         en = RelatonBib::Person.new name: name, url: args[:url], affiliation: af

data/lib/relaton_w3c/version.rb

@@ -1,3 +1,3 @@
 module RelatonW3c
-  VERSION = "1.7.1".freeze
+  VERSION = "1.9.1".freeze
 end

data/lib/relaton_w3c/w3c_bibliographic_item.rb

@@ -10,7 +10,7 @@ module RelatonW3c
       if args[:doctype] && !TYPES.include?(args[:doctype])
         warn "[relaton-w3c] invalid document type: #{args[:doctype]}"
       end
-      super **args
+      super
     end
   end
 end