relaton-w3c 1.11.3 → 1.11.6

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 4c96856baa51c84d8397b62be6dee1b8225e7ab854cd70176187e68fc269a23c
-  data.tar.gz: 54d5cc018e0de6ecca6d402ccbe41947047190d08e922c8f46b7cb20164cea22
+  metadata.gz: faf78f2abf348e9aa6dd74f2f0fe94899d9d6fc34af3368849a818b6dd856e26
+  data.tar.gz: c4cabbd9559b9f122f34fbb204f7780606bfc87c9db999492fbdd2b6b02639b3
 SHA512:
-  metadata.gz: 070eb14907a49f99b7c0f45841d83250c1c051cd9a16b9cddfc83f6d874274f401ac95c9554f6994c6702e557ef8a200e576ee5fe1d0b7c7f7583afb53074794
-  data.tar.gz: cab9e48e248b889c5d15e3449a93dc5cb100b7680a87f025ebfd090a1afba36e77730d9e0bc85a10bc9208b9ea2d813e444e0a28b2ac77d412650395479468f3
+  metadata.gz: cc6e6c80914b8973091795b1f234ea57a49499bb0466ae04a14358c30e526c38a6f607031e9dd0d4963c2caccca26dd9418939c871b4f47016aa094156d783a3
+  data.tar.gz: 041625ce66ae2949112bc20856de89b05fd4772c1dd1f3413a3a5b04059ec7cb84f8b92517a3634294080031760fca8795b7ae113bbfae03986f775c5f52acb0

data/.github/workflows/rake.yml

@@ -10,27 +10,4 @@ on:
 
 jobs:
   rake:
-    name: Test on Ruby ${{ matrix.ruby }} ${{ matrix.os }}
-    runs-on: ${{ matrix.os }}
-    continue-on-error: ${{ matrix.experimental }}
-    strategy:
-      fail-fast: false
-      matrix:
-        ruby: [ '3.0', '2.7', '2.6', '2.5' ]
-        os: [ ubuntu-latest, windows-latest, macos-latest ]
-        experimental: [ false ]
-    steps:
-      - uses: actions/checkout@v2
-        with:
-          submodules: true
-
-      # https://github.com/ruby-debug/debase/issues/89#issuecomment-686827382
-      - if: matrix.os == 'macos-latest' && matrix.ruby == '2.5'
-        run: echo BUNDLE_BUILD__DEBASE="--with-cflags=\"-Wno-error=implicit-function-declaration\"" >> $GITHUB_ENV
-
-      - uses: ruby/setup-ruby@v1
-        with:
-          ruby-version: ${{ matrix.ruby }}
-          bundler-cache: true
-
-      - run: bundle exec rake
+    uses: relaton/support/.github/workflows/rake.yml@master

data/.rubocop.yml

@@ -7,6 +7,6 @@ require: rubocop-rails
 inherit_from:
   - https://raw.githubusercontent.com/riboseinc/oss-guides/master/ci/rubocop.yml
 AllCops:
-  TargetRubyVersion: 2.5
+  TargetRubyVersion: 2.6
 Rails:
   Enabled: false

data/data/reference.W3C.DSig-label.xml

@@ -28,5 +28,5 @@
     </author>
     <date day="27" month="May" year="1998"/>
   </front>
-  <seriesInfo name="W3C Recommendation" value="REC-DSig-label"/>
+  <seriesInfo name="W3C" value="REC-DSig-label"/>
 </reference>

data/data/reference.W3C.P3P-rdfschema.xml

@@ -22,5 +22,5 @@
     </author>
     <date day="25" month="January" year="2002"/>
   </front>
-  <seriesInfo name="W3C Note" value="P3P-rdfschema"/>
+  <seriesInfo name="W3C" value="NOTE-P3P-rdfschema"/>
 </reference>

data/data/reference.W3C.P3P.xml

@@ -34,5 +34,5 @@
     </author>
     <date day="16" month="April" year="2002"/>
   </front>
-  <seriesInfo name="W3C Recommendation" value="REC-PICS-services"/>
+  <seriesInfo name="W3C" value="REC-PICS-services"/>
 </reference>

data/data/reference.W3C.PICS-labels.xml

@@ -39,5 +39,5 @@
     </author>
     <date day="31" month="October" year="1996"/>
   </front>
-  <seriesInfo name="W3C Recommendation" value="REC-PICS-labels"/>
+  <seriesInfo name="W3C" value="REC-PICS-labels"/>
 </reference>

data/data/reference.W3C.PICS-rules.xml

@@ -34,5 +34,5 @@
     </author>
     <date day="29" month="December" year="1997"/>
   </front>
-  <seriesInfo name="W3C Recommendation" value="REC-PICSRules"/>
+  <seriesInfo name="W3C" value="REC-PICSRules"/>
 </reference>

data/data/reference.W3C.PICS-services.xml

@@ -33,5 +33,5 @@
     </author>
     <date day="31" month="October" year="1996"/>
   </front>
-  <seriesInfo name="W3C Recommendation" value="REC-PICS-services"/>
+  <seriesInfo name="W3C" value="REC-PICS-services"/>
 </reference>

data/data/reference.W3C.daml-oil-reference.xml

@@ -35,5 +35,5 @@
     </author>
     <date day="18" month="December" year="2001"/>
   </front>
-  <seriesInfo value="daml+oil-reference" name="W3C Note"/>
+  <seriesInfo value="NOTE-daml+oil-reference" name="W3C"/>
 </reference>

data/data/reference.W3C.soap11.xml

@@ -52,5 +52,5 @@
     </author>
     <date day="8" month="May" year="2000"/>
   </front>
-  <seriesInfo value="soap11" name="W3C Note"/>
+  <seriesInfo value="NOTE-soap11" name="W3C"/>
 </reference>

data/data/reference.W3C.soap12-part1.xml

@@ -34,5 +34,5 @@
     </author>
     <date day="26" month="June" year="2002"/>
   </front>
-  <seriesInfo value="soap12-part1" name="W3C Working Draft"/>
+  <seriesInfo value="WD-soap12-part1" name="W3C"/>
 </reference>

data/data/reference.W3C.soap12-part2.xml

@@ -34,5 +34,5 @@
     </author>
     <date day="26" month="June" year="2002"/>
   </front>
-  <seriesInfo value="soap12-part2" name="W3C Working Draft"/>
+  <seriesInfo value="WD-soap12-part2" name="W3C"/>
 </reference>

data/data/reference.W3C.xkms.xml

@@ -46,5 +46,5 @@
     </author>
     <date day="30" month="March" year="2001"/>
   </front>
-  <seriesInfo value="xkms" name="W3C Note"/>
+  <seriesInfo value="NOTE-xkms" name="W3C"/>
 </reference>

data/data/reference.W3C.xml-c14n.xml

@@ -11,5 +11,5 @@
     </author>
     <date day="15" month="March" year="2001"/>
   </front>
-  <seriesInfo name="W3C Recommendation" value="xpath"/>
+  <seriesInfo name="W3C" value="REC-xpath"/>
 </reference>

data/data/reference.W3C.xmldsig-core.xml

@@ -22,5 +22,5 @@
     </author>
     <date day="6" month="October" year="2000"/>
   </front>
-  <seriesInfo value="xmldsig-core" name="W3C Recommendation"/>
+  <seriesInfo value="REC-xmldsig-core" name="W3C"/>
 </reference>

data/data/reference.W3C.xmlenc-core.xml

@@ -16,5 +16,5 @@
     </author>
     <date day="2" month="August" year="2002"/>
   </front>
-  <seriesInfo value="xmlenc-core" name="W3C Candidate Recommendation"/>
+  <seriesInfo value="CR-xmlenc-core" name="W3C"/>
 </reference>

data/data/reference.W3C.xpath.xml

@@ -18,5 +18,5 @@
     </author>
     <date day="16" month="November" year="1999"/>
   </front>
-  <seriesInfo name="W3C Recommendation" value="xpath"/>
+  <seriesInfo name="W3C" value="REC-xpath"/>
 </reference>

data/lib/relaton_w3c/data_fetcher.rb

@@ -68,60 +68,7 @@ module RelatonW3c
     end
 
     #
-    # Create index file
-    #
-    # def create_index
-    #   index_file = "index-w3c.yaml"
-    #   index_yaml = @index.sort do |a, b|
-    #     compare_index_items a, b
-    #   end.to_yaml
-    #   File.write index_file, index_yaml, encoding: "UTF-8"
-    # end
-
-    #
-    # Compare index items
-    #
-    # @param [Hash] aid first item
-    # @param [Hash] bid second item
-    #
-    # @return [Integer] comparison result
-    #
-    # def compare_index_items(aid, bid) # rubocop:disable Metrics/AbcSize
-    #   ret = aid[:code] <=> bid[:code]
-    #   ret = stage_weight(bid[:stage]) <=> stage_weight(aid[:stage]) if ret.zero?
-    #   ret = date_weight(bid[:date]) <=> date_weight(aid[:date]) if ret.zero?
-    #   # ret = aid[:type] <=> bid[:type] if ret.zero?
-    #   ret
-    # end
-
-    #
-    # Weight of stage
-    #
-    # @param [String, nil] stage stage
-    #
-    # @return [Integer] weight
-    #
-    # def stage_weight(stage)
-    #   return DataParser::STAGES.size if stage.nil?
-
-    #   DataParser::STAGES.keys.index(stage)
-    # end
-
-    #
-    # Weight of date
-    #
-    # @param [String] date date
-    #
-    # @return [String] weight
-    #
-    # def date_weight(date)
-    #   return "99999999" if date.nil?
-
-    #   date
-    # end
-
-    #
-    # Query RDF source for documents
+    # Query RDF source for versioned documents
     #
     # @return [RDF::Query::Solutions] query results
     #
@@ -140,6 +87,11 @@ module RelatonW3c
       data.query sse
     end
 
+    #
+    # Query RDF source for unversioned documents
+    #
+    # @return [Array<RDF::Query::Solution>] query results
+    #
     def query_unversioned_docs
       sse = SPARQL.parse(%(
         PREFIX doc: <http://www.w3.org/2000/10/swap/pim/doc#>

data/lib/relaton_w3c/data_index.rb

@@ -13,25 +13,6 @@ module RelatonW3c
       @index = index
     end
 
-    #
-    # Create index from a GitHub repository
-    #
-    # @return [RelatonW3c::DataIndex] data index
-    #
-    def self.create_from_repo # rubocop:disable Metrics/MethodLength, Metrics/AbcSize
-      resp = Zip::InputStream.new URI("#{W3cBibliography::SOURCE}index-w3c.zip").open
-      zip = resp.get_next_entry
-
-      # Newer versions of Psych uses the `permitted_classes:` parameter
-      index = if YAML.method(:safe_load).parameters.collect(&:last).index(:permitted_classes)
-                YAML.safe_load(zip.get_input_stream.read, permitted_classes: [Symbol])
-              else
-                YAML.safe_load(zip.get_input_stream.read, [Symbol])
-              end
-
-      DataIndex.new index: index
-    end
-
     #
     # Add document to index
     #
@@ -39,7 +20,9 @@ module RelatonW3c
     # @param [String] file path to document file
     #
     def add(docnumber, file)
-      @index << docnumber_to_parts(docnumber, file)
+      dnparts = self.class.docnumber_to_parts docnumber
+      dnparts[:file] = file
+      @index << dnparts
     end
 
     #
@@ -67,11 +50,14 @@ module RelatonW3c
     # @return [String] document's filename
     #
     def search(ref) # rubocop:disable Metrics/AbcSize, Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
-      dparts = docnumber_to_parts(ref)
+      dparts = self.class.docnumber_to_parts(ref)
+      return if dparts[:code].nil?
+
       @index.detect do |parts|
         parts[:code].match?(/^#{Regexp.escape dparts[:code]}/i) &&
           (dparts[:stage].nil? || dparts[:stage].casecmp?(parts[:stage])) &&
-          (dparts[:type].nil? || dparts[:type].casecmp?(parts[:type])) &&
+          (dparts[:type].nil? || dparts[:type].casecmp?(parts[:type]) ||
+            (parts[:type].nil? && dparts[:type] == "TR")) &&
           (dparts[:date].nil? || dparts[:date] == parts[:date]) &&
           (dparts[:suff].nil? || dparts[:suff].casecmp?(parts[:suff]))
       end&.fetch(:file)
@@ -86,7 +72,7 @@ module RelatonW3c
     # @return [Integer] comparison result
     #
     def compare_index_items(aid, bid) # rubocop:disable Metrics/AbcSize
-      ret = aid[:code] <=> bid[:code]
+      ret = aid[:code].downcase <=> bid[:code].downcase
       ret = stage_weight(bid[:stage]) <=> stage_weight(aid[:stage]) if ret.zero?
       ret = date_weight(bid[:date]) <=> date_weight(aid[:date]) if ret.zero?
       # ret = aid[:type] <=> bid[:type] if ret.zero?
@@ -119,28 +105,47 @@ module RelatonW3c
       date
     end
 
-    #
-    # Parse document number to parts
-    #
-    # @param [String] docnumber document number
-    # @param [String, nil] file path to document file
-    #
-    # @return [Hash{Symbol=>String}] document parts
-    #
-    def docnumber_to_parts(docnumber, file = nil) # rubocop:disable Metrics/MethodLength
-      %r{
-        ^(?:(?:(?<stage>WD|CRD|CR|PR|PER|REC|SPSD|OBSL|RET)|(?<type>D?NOTE))-)?
-        (?<code>\w+(?:[+-][\w.]+)*?)
-        (?:-(?<date>\d{8}|\d{6}))?
-        (?:/(?<suff>\w+))?$
-      }xi =~ docnumber
-      entry = { code: code }
-      entry[:file] = file if file
-      entry[:stage] = stage if stage
-      entry[:type] = type if type
-      entry[:date] = date if date
-      entry[:suff] = suff if suff
-      entry
+    class << self
+      #
+      # Create index from a GitHub repository
+      #
+      # @return [RelatonW3c::DataIndex] data index
+      #
+      def create_from_repo # rubocop:disable Metrics/MethodLength, Metrics/AbcSize
+        resp = Zip::InputStream.new URI("#{W3cBibliography::SOURCE}index-w3c.zip").open
+        zip = resp.get_next_entry
+
+        # Newer versions of Psych uses the `permitted_classes:` parameter
+        index = if YAML.method(:safe_load).parameters.collect(&:last).index(:permitted_classes)
+                  YAML.safe_load(zip.get_input_stream.read, permitted_classes: [Symbol])
+                else
+                  YAML.safe_load(zip.get_input_stream.read, [Symbol])
+                end
+
+        DataIndex.new index: index
+      end
+
+      #
+      # Parse document number to parts
+      #
+      # @param [String] docnumber document number
+      #
+      # @return [Hash{Symbol=>String}] document parts
+      #
+      def docnumber_to_parts(docnumber) # rubocop:disable Metrics/MethodLength
+        %r{
+          ^(?:(?:(?<stage>WD|CRD|CR|PR|PER|REC|SPSD|OBSL|RET)|(?<type>D?NOTE|TR))-)?
+          (?<code>\w+(?:[+-][\w.]+)*?)
+          (?:-(?<date>\d{8}|\d{6}|\d{4}))?
+          (?:/(?<suff>\w+))?$
+        }xi =~ docnumber
+        entry = { code: code }
+        entry[:stage] = stage if stage
+        entry[:type] = type if type
+        entry[:date] = date if date
+        entry[:suff] = suff if suff
+        entry
+      end
     end
   end
 end

data/lib/relaton_w3c/data_parser.rb

@@ -86,9 +86,10 @@ module RelatonW3c
     # @return [RelatonBib::TypedTitleStringCollection] title
     #
     def parse_title
-      return [] unless @sol.respond_to?(:title)
-
-      t = RelatonBib::TypedTitleString.new content: @sol.title.to_s
+      content = if @sol.respond_to?(:title) then @sol.title.to_s
+                else document_versions.max_by { |dv| dv.date.to_s }.title.to_s
+                end
+      t = RelatonBib::TypedTitleString.new content: content
       RelatonBib::TypedTitleStringCollection.new [t]
     end
 
@@ -99,7 +100,6 @@ module RelatonW3c
     #
     def parse_link
       link = @sol.respond_to?(:link) ? @sol.link : @sol.version_of
-
       [RelatonBib::TypedUri.new(type: "src", content: link.to_s)]
     end
 
@@ -109,9 +109,7 @@ module RelatonW3c
     # @return [Arra<RelatonBib::DocumentIdentifier>] docidentifier
     #
     def parse_docid
-      return [] unless @sol.respond_to?(:link)
-
-      id = pub_id(@sol.link)
+      id = @sol.respond_to?(:link) ? pub_id(@sol.link) : pub_id(@sol.version_of)
       [RelatonBib::DocumentIdentifier.new(type: "W3C", id: id, primary: true)]
     end
 
@@ -169,7 +167,7 @@ module RelatonW3c
     #
     def type
       # thre are many types, we need to find the right one
-      @type ||= types_stages&.detect { |t| USED_TYPES.include?(t) }
+      @type ||= types_stages&.detect { |t| USED_TYPES.include?(t) } || "technicalReport"
     end
 
     #
@@ -221,10 +219,16 @@ module RelatonW3c
     def parse_relation
       if @sol.respond_to?(:link)
         relations + editor_drafts
-      else document_versions
+      else
+        document_versions.map { |r| create_relation(r.link.to_s, "hasEdition") }
       end
     end
 
+    #
+    # Create relations
+    #
+    # @return [Array<RelatonBib::DocumentRelation>] relations
+    #
     def relations # rubocop:disable Metrics/MethodLength, Metrics/AbcSize
       {
         "doc:obsoletes" => { type: "obsoletes" },
@@ -234,14 +238,16 @@ module RelatonW3c
         ":previousEdition" => { type: "editionOf" },
       }.reduce([]) do |acc, (predicate, tp)|
         acc + relation_query(predicate).map do |r|
-          fr = RelatonBib::LocalizedString.new pub_id(r.rel.to_s)
-          bib = W3cBibliographicItem.new formattedref: fr
-          tp[:description] = RelatonBib::FormattedString.new content: tp[:description] if tp[:description]
-          RelatonBib::DocumentRelation.new(**tp, bibitem: bib)
+          create_relation(r.rel.to_s, tp[:type], tp[:description])
         end
       end
     end
 
+    #
+    # Parse editor drafts relation
+    #
+    # @return [Array<RelatonBib::DocumentRelation>] relation
+    #
     def editor_drafts # rubocop:disable Metrics/MethodLength
       sse = SPARQL.parse(%(
         PREFIX : <http://www.w3.org/2001/02pd/rec54#>
@@ -250,15 +256,17 @@ module RelatonW3c
         WHERE { <#{@sol.link}> :ED ?rel . }
       ))
       @fetcher.data.query(sse).map do |s|
-        fr = RelatonBib::LocalizedString.new pub_id(s.rel.to_s)
-        bib = W3cBibliographicItem.new formattedref: fr
-        desc = RelatonBib::FormattedString.new content: "Editor's draft"
-        RelatonBib::DocumentRelation.new(
-          type: "hasDraft", description: desc, bibitem: bib,
-        )
+        create_relation(s.rel.to_s, "hasDraft", "Editor's draft")
       end
     end
 
+    #
+    # Query for relations
+    #
+    # @param [String] predicate relation type
+    #
+    # @return [RDF::Query::Solutions] query result
+    #
     def relation_query(predicate)
       sse = SPARQL.parse(%(
         PREFIX : <http://www.w3.org/2001/02pd/rec54#>
@@ -270,19 +278,43 @@ module RelatonW3c
       @fetcher.data.query(sse).order_by(:rel)
     end
 
-    def document_versions
-      sse = SPARQL.parse(%(
-        PREFIX doc: <http://www.w3.org/2000/10/swap/pim/doc#>
-        SELECT ?link
-        WHERE { ?link doc:versionOf <#{@sol.version_of}> }
-      ))
-      @fetcher.data.query(sse).map do |r|
-        fref = RelatonBib::FormattedRef.new content: pub_id(r.link)
-        bib = W3cBibliographicItem.new formattedref: fref
-        RelatonBib::DocumentRelation.new(type: "hasEdition", bibitem: bib)
+    #
+    # Query document versions relations
+    #
+    # @return [RDF::Query::Solutions] query results
+    #
+    def document_versions # rubocop:disable Metrics/MethodLength
+      @document_versions ||= begin
+        sse = SPARQL.parse(%(
+          PREFIX : <http://www.w3.org/2001/02pd/rec54#>
+          PREFIX dc: <http://purl.org/dc/elements/1.1/>
+          PREFIX doc: <http://www.w3.org/2000/10/swap/pim/doc#>
+          PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
+          SELECT ?link ?title ?date
+          WHERE { ?link doc:versionOf <#{@sol.version_of}> ; dc:title ?title ; dc:date ?date }
+        ))
+        @fetcher.data.query(sse)
       end
     end
 
+    #
+    # Create relation
+    #
+    # @param [String] url relation URL
+    # @param [String] type relation type
+    # @param [String, nil] desc relation description
+    #
+    # @return [RelatonBib::DocumentRelation] <description>
+    #
+    def create_relation(url, type, desc = nil)
+      id = pub_id(url)
+      fref = RelatonBib::FormattedRef.new content: id
+      docid = RelatonBib::DocumentIdentifier.new(type: "W3C", id: id, primary: true)
+      bib = W3cBibliographicItem.new formattedref: fref, docid: [docid]
+      dsc = RelatonBib::FormattedString.new content: desc if desc
+      RelatonBib::DocumentRelation.new(type: type, bibitem: bib, description: dsc)
+    end
+
     #
     # Parse formattedref
     #

data/lib/relaton_w3c/version.rb

@@ -1,3 +1,3 @@
 module RelatonW3c
-  VERSION = "1.11.3".freeze
+  VERSION = "1.11.6".freeze
 end

data/lib/relaton_w3c/w3c_bibliography.rb

@@ -9,10 +9,10 @@ module RelatonW3c
 
     class << self
       # @param text [String]
-      # @return [RelatonW3c::HitCollection]
-      def search(text) # rubocop:disable Metrics/MethodLength
+      # @return [RelatonW3c::W3cBibliographicItem]
+      def search(text) # rubocop:disable Metrics/MethodLength, Metrics/AbcSize
         ref = DataParser.parse_identifier text.sub(/^W3C\s/, "")
-        file = DataIndex.create_from_repo.search(ref)
+        file = DataIndex.create_from_repo.search ref.gsub(" ", "-").squeeze("-")
         return unless file
 
         url = "#{SOURCE}#{file}"

data/lib/relaton_w3c/workgroups.yaml

@@ -103,6 +103,8 @@
   name: Web Services Architecture Working Group
 'https://www.w3.org/2011/webappsec':
   name: Web Application Security Working Group
+'https://www.w3.org/groups/wg/webappsec':
+  name: Web Application Security Working Group
 'https://www.w3.org/Social/WG':
   name: Social Web Working Group
 'https://www.w3.org/XML/Core':

data/lib/relaton_w3c.rb

@@ -2,9 +2,6 @@ require "relaton_bib"
 require "relaton_w3c/version"
 require "relaton_w3c/w3c_bibliography"
 require "relaton_w3c/w3c_bibliographic_item"
-# require "relaton_w3c/hit_collection"
-# require "relaton_w3c/hit"
-# require "relaton_w3c/scrapper"
 require "relaton_w3c/xml_parser"
 require "relaton_w3c/bibxml_parser"
 require "relaton_w3c/hash_converter"

data/relaton_w3c.gemspec

@@ -14,7 +14,7 @@ Gem::Specification.new do |spec|
                        "using the IsoBibliographicItem model"
   spec.homepage      = "https://github.com/relaton/relaton-wc3"
   spec.license       = "BSD-2-Clause"
-  spec.required_ruby_version = Gem::Requirement.new(">= 2.5.0")
+  spec.required_ruby_version = Gem::Requirement.new(">= 2.6.0")
 
   # spec.metadata["allowed_push_host"] = "TODO: Set to 'http://mygemserver.com'"
 
@@ -39,8 +39,8 @@ Gem::Specification.new do |spec|
 
   spec.add_dependency "linkeddata", "~> 3.1.0"
   spec.add_dependency "mechanize", "~> 2.8.0"
-  spec.add_dependency "rdf", "~> 3.1.0"
-  spec.add_dependency "rdf-normalize", "~> 0.4.0"
+  spec.add_dependency "rdf", "~> 3.2.0"
+  spec.add_dependency "rdf-normalize", "~> 0.5.0"
   spec.add_dependency "relaton-bib", "~> 1.11.0"
   spec.add_dependency "rubyzip", "~> 2.3.0"
   spec.add_dependency "shex", "~> 0.6.0"

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: relaton-w3c
 version: !ruby/object:Gem::Version
-  version: 1.11.3
+  version: 1.11.6
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2022-04-20 00:00:00.000000000 Z
+date: 2022-06-01 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: equivalent-xml
@@ -114,28 +114,28 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 3.1.0
+        version: 3.2.0
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 3.1.0
+        version: 3.2.0
 - !ruby/object:Gem::Dependency
   name: rdf-normalize
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 0.4.0
+        version: 0.5.0
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 0.4.0
+        version: 0.5.0
 - !ruby/object:Gem::Dependency
   name: relaton-bib
   requirement: !ruby/object:Gem::Requirement
@@ -247,10 +247,7 @@ files:
 - lib/relaton_w3c/data_index.rb
 - lib/relaton_w3c/data_parser.rb
 - lib/relaton_w3c/hash_converter.rb
-- lib/relaton_w3c/hit.rb
-- lib/relaton_w3c/hit_collection.rb
 - lib/relaton_w3c/processor.rb
-- lib/relaton_w3c/scrapper.rb
 - lib/relaton_w3c/version.rb
 - lib/relaton_w3c/w3c_bibliographic_item.rb
 - lib/relaton_w3c/w3c_bibliography.rb
@@ -270,7 +267,7 @@ required_ruby_version: !ruby/object:Gem::Requirement
   requirements:
   - - ">="
     - !ruby/object:Gem::Version
-      version: 2.5.0
+      version: 2.6.0
 required_rubygems_version: !ruby/object:Gem::Requirement
   requirements:
   - - ">="

data/lib/relaton_w3c/hit.rb

@@ -1,15 +0,0 @@
-# frozen_string_literal: true
-
-module RelatonW3c
-  # Hit.
-  class Hit < RelatonBib::Hit
-    #
-    # Parse page.
-    #
-    # @param lang [String, NilClass]
-    # @return [RelatonW3c::W3cBibliographicItem]
-    def fetch(_lang = nil)
-      @fetch ||= Scrapper.parse_page hit
-    end
-  end
-end

data/lib/relaton_w3c/hit_collection.rb

@@ -1,172 +0,0 @@
-# frozen_string_literal: true
-
-require "fileutils"
-require "yaml"
-
-module RelatonW3c
-  # Page of hit collection.
-  class HitCollection < RelatonBib::HitCollection
-    TYPES = {
-      "CR" => "Candidate Recommendation",
-      "NOTE" => "Group Note",
-      "PER" => "Proposed Edited Recommendation",
-      "PR" => "Proposed Recommendation",
-      "REC" => "Recommendation",
-      "RET" => "Retired",
-      "WD" => "Working Draft",
-    }.freeze
-    DOMAIN = "https://www.w3.org"
-    DATADIR = File.expand_path(".relaton/w3c", Dir.home).freeze
-    DATAFILE = File.expand_path("bibliography.yml", DATADIR).freeze
-
-    # @param ref [String] reference to search
-    def initialize(ref)
-      %r{
-        ^(?: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?
-        (?<title_date>.+) # title_date
-      }x =~ ref
-      super
-      @array = from_yaml title_date, type
-    end
-
-    private
-
-    #
-    # Fetch data form yaml
-    #
-    # @param title_date [String]
-    # @param type [String]
-    # @return [Array<Hash>]
-    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|
-        (hit["title"].casecmp?(title) ||
-          hit["link"].split("/").last.match?(/-#{title}-/)) &&
-          type_date_filter(hit, type, date)
-      end
-      if result.empty?
-        result = data.select { |h| h["link"].split("/").last.match?(/#{title}/) }
-      end
-      result.map { |h| Hit.new(h, self) }
-    end
-
-    # @param hit [Hash]
-    # @param type [String]
-    # @param date [String]
-    # @return [TrueClass, FalseClass]
-    def type_date_filter(hit, type, date) # rubocop:disable Metrics/AbcSize
-      if (type && hit["type"] != short_type(type)) || (date && hit["date"] != date)
-        history = get_history hit, type, date
-        return false unless history.any?
-
-        hit["type"] = short_type type
-        hit["datepub"] = history.first.at("td").text
-        hit["link"] = history.first.at("a")[:href]
-      end
-      true
-    end
-
-    # @param hit [Hash]
-    # @param type [String]
-    # @param date [String]
-    # @return [Array<Nokogiri::XML::Element>, Nokogiri::HTML::NodeSet]
-    def get_history(hit, type, date)
-      resp = Net::HTTP.get URI.parse(HitCollection::DOMAIN + hit["history"])
-      history_doc = Nokogiri::HTML resp
-      history = history_doc.xpath(
-        "//table//a[contains(.,'#{long_type(type)}')]/../..",
-      )
-      return filter_history_by_date(history, history_doc, type, date) if date
-
-      history
-    end
-
-    # @param history [Nokogiri::XML::NodeSet]
-    # @param history_doc [Nokogiri::HTML::NodeSet]
-    # @param type [String]
-    # @param date [String]
-    # @return [Array<Nokogiri::XML::Element>, Nokogiri::HTML::NodeSet]
-    def filter_history_by_date(history, history_doc, type, date)
-      if type
-        history.select do |h|
-          h.at("td[@class='table_datecol']").text == date
-        end
-      else
-        history_doc.xpath(
-          "//table//td[@class='table_datecol'][.='#{date}']/..",
-        )
-      end
-    end
-
-    #
-    # Convetr long type name to short
-    #
-    # @param type [String]
-    # @return [String]
-    def short_type(type)
-      tp = TYPES.select { |_, v| v == type }.keys
-      tp.first || type
-    end
-
-    #
-    # Convert shot type name to long
-    #
-    # @param [String]
-    # @return [String]
-    def long_type(type)
-      TYPES[type] || type
-    end
-
-    #
-    # Fetches YAML data
-    #
-    # @return [Hash]
-    def data
-      FileUtils.mkdir_p DATADIR
-      ctime = File.ctime DATAFILE if File.exist? DATAFILE
-      fetch_data if !ctime || ctime.to_date < Date.today
-      @data ||= YAML.safe_load File.read(DATAFILE, encoding: "UTF-8")
-    end
-
-    #
-    # fetch data form server and save it to file.
-    #
-    def fetch_data
-      resp = Net::HTTP.get_response URI.parse("#{DOMAIN}/TR/")
-      # return if there aren't any changes since last fetching
-      return unless resp.code == "200"
-
-      doc = Nokogiri::HTML resp.body
-      @data = doc.xpath("//ul[@id='container']/li").map do |h_el|
-        link = h_el.at("h2/a")
-        pubdetails = h_el.at("p[@class='pubdetails']")
-        fetch_hit h_el, link, pubdetails
-      end
-      File.write DATAFILE, @data.to_yaml, encoding: "UTF-8"
-    end
-
-    # @param h_el [Nokogiri::XML::Element]
-    # @param link [Nokogiri::XML::Element]
-    # @param pubdetails [Nokogiri::XML::Element]
-    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 }
-      {
-        "title" => link.text.gsub("\u00a0", " "),
-        "link" => link[:href],
-        "type" => h_el.at("div").text.upcase,
-        "workgroup" => h_el.xpath("p[@class='deliverer']").map(&:text),
-        "datepub" => datepub,
-        "history" => pubdetails.at("a[text()='History']")[:href],
-        "editor" => editor,
-        "keyword" => keyword,
-      }
-    end
-  end
-end

data/lib/relaton_w3c/scrapper.rb

@@ -1,218 +0,0 @@
-module RelatonW3c
-  class Scrapper
-    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
-        resp = Net::HTTP.get_response URI.parse(hit["link"])
-        doc = resp.code == "200" ? Nokogiri::HTML(resp.body) : nil
-        W3cBibliographicItem.new(
-          type: "standard",
-          docid: fetch_docid(hit),
-          fetched: Date.today.to_s,
-          language: ["en"],
-          script: ["Latn"],
-          title: fetch_title(hit, doc),
-          abstract: fetch_abstract(doc),
-          link: fetch_link(hit),
-          date: fetch_date(hit, doc),
-          doctype: fetch_doctype(hit, doc),
-          contributor: fetch_contributor(hit, doc),
-          relation: fetch_relation(doc),
-          keyword: hit["keyword"],
-        )
-      end
-
-      private
-
-      # @param hit [Hash]
-      # @return [Array<RelatonBib::DocumentIdentifier>]
-      def fetch_docid(hit)
-        id = hit["link"].split("/").last
-        [RelatonBib::DocumentIdentifier.new(id: id, type: "W3C", primary: true)]
-      end
-
-      # @param hit [Hash]
-      # @param doc [Nokogiri::HTML::Document]
-      # @return [Array<RelatonBib::TypedTitleString>]
-      def fetch_title(hit, doc) # rubocop:disable Metrics/AbcSize, Metrics/CyclomaticComplexity, Metrics/MethodLength, Metrics/PerceivedComplexity
-        titles = []
-        if doc
-          title = doc.at("//*[contains(@id, 'title')]")&.text
-          if title && !title.empty?
-            titles << { content: title.gsub(/\n/, " "), type: "main" }
-          end
-          subtitle = doc.at(
-            "//h2[@id='subtitle']|//p[contains(@class, 'subline')]",
-          )&.text
-          titles << { content: subtitle, tipe: "subtitle" } if subtitle
-        end
-        if titles.empty? && hit["title"]
-          titles << { content: hit["title"], type: "main" }
-        end
-        titles.map do |t|
-          title = RelatonBib::FormattedString.new(
-            content: t[:content], language: "en", script: "Latn",
-          )
-          RelatonBib::TypedTitleString.new(type: t[:type], title: title)
-        end
-      end
-
-      # @param doc [Nokogiri::HTML::Document, NilClass]
-      # @return [Array<RelatonBib::FormattedString>]
-      def fetch_abstract(doc)
-        return [] unless doc
-
-        content = doc.at("//h2[.='Abstract']/following-sibling::p",
-                         "//div[@class='abstract']/p").text
-        [RelatonBib::FormattedString.new(content: content, language: "en",
-                                         script: "Latn")]
-      end
-
-      # @param hit [Hash]
-      # @return [Array<RelatonBib::TypedUri>]
-      def fetch_link(hit)
-        [RelatonBib::TypedUri.new(type: "src", content: hit["link"])]
-      end
-
-      # @param hit [Hash]
-      # @param doc [Nokogiri::HTML::Document, NilClass]
-      # @return [Array<RelatonBib::BibliographicDate>]
-      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
-      end
-
-      # @param doc [Nokogiri::HTML::Document, NilClass]
-      # @return [String]
-      def fetch_date1(doc)
-        d = doc&.at("//h2[@property='dc:issued']")&.attr(:content)
-        d&.match(/\d{4}-\d{2}-\d{2}/)&.to_s
-      end
-
-      # @param doc [Nokogiri::HTML::Document, NilClass]
-      # @return [String]
-      def fetch_date2(doc)
-        d = doc&.at("//h2[contains(@id, 'w3c-recommendation')]")
-        return unless d
-
-        Date.parse(d.attr(:id.match(/\d{2}-\w+-\d{4}/).to_s)).to_s
-      end
-
-      # @param hit [Hash]
-      # @param doc [Nokogiri::HTML::Document, NilClass]
-      # @return [String]
-      def fetch_doctype(hit, doc)
-        if hit["type"]
-          DOCTYPES[hit["type"]]
-        elsif doc
-          type = HitCollection::TYPES.detect do |_k, v|
-            doc.at("//h2[contains(., '#{v}')]/time[@datetime]")
-          end
-          DOCTYPES[type&.first]
-        end
-      end
-
-      # @param hit [Hash]
-      # @param doc [Nokogiri::HTML::Document, NilClass]
-      # @return [Array<RelatonBib::ContributionInfo>]
-      def fetch_contributor(hit, doc) # rubocop:disable Metrics/AbcSize, Metrics/CyclomaticComplexity, Metrics/MethodLength, Metrics/PerceivedComplexity
-        if doc
-          editors = find_contribs(doc, "Editors").reduce([]) do |mem, ed|
-            c = parse_contrib ed, "editor"
-            mem << c if c
-            mem
-          end
-          contribs = find_contribs(doc, "Authors").reduce(editors) do |mem, ath|
-            ed = mem.detect { |e| e[:id] && e[:id] == ath["data-editor-id"] }
-            if ed
-              ed[:role] << { type: "author" }
-            else
-              mem << parse_contrib(ath, "author")
-            end
-            mem
-          end
-          contribs.map { |c| contrib_info(**c) }
-        else
-          hit["editor"].map do |ed|
-            contrib_info name: ed, role: [{ type: "editor" }]
-          end
-        end
-      end
-
-      # @param doc [Nokogiri::NTML::Document]
-      # @param type [String]
-      # @return [Array<Nokogiri::XML::Element]
-      def find_contribs(doc, type)
-        doc.xpath("//dt[contains(.,'#{type}')]/following-sibling::dd"\
-                  "[preceding-sibling::dt[1][contains(.,'#{type}')]]")
-      end
-
-      # @param element [Nokogiri::XML::Element]
-      # @param type [String]
-      # @return [Hash]
-      def parse_contrib(element, type) # rubocop:disable Metrics/MethodLength
-        p = element.at("a")
-        return unless p
-
-        contrib = {
-          name: p.text,
-          url: p[:href],
-          role: [{ type: type }],
-          id: element["data-editor-id"],
-        }
-        org = element.at("a[2]")
-        contrib[:org] = { name: org.text, url: org[:href] } if org
-        contrib
-      end
-
-      # @param name [String]
-      # @param url [String, NilClass]
-      # @param role [Array<Hash>]
-      # @parma org [Hash]
-      # @return [RelatonBib::ContributionInfo]
-      def contrib_info(**args)
-        completename = RelatonBib::LocalizedString.new(args[:name])
-        name = RelatonBib::FullName.new completename: completename
-        af = []
-        if 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
-        RelatonBib::ContributionInfo.new entity: en, role: args[:role]
-      end
-
-      # @param doc [Nokogiri::HTML::Document]
-      # @return [Array<RelatonBib::DocumentRelation>]
-      def fetch_relation(doc)
-        return [] unless doc && (link = recommendation_link(doc))
-
-        hit = { "link" => link }
-        item = parse_page hit
-        [RelatonBib::DocumentRelation.new(type: "obsoletedBy", bibitem: item)]
-      end
-
-      # @param doc [Nokogiri::HTML::Document]
-      # @return [String, NilClass]
-      def recommendation_link(doc)
-        recom = doc.at("//dt[.='Latest Recommendation:']",
-                       "//dt[.='Previous Recommendation:']")
-        return unless recom
-
-        recom.at("./following-sibling::dd/a")[:href]
-      end
-    end
-  end
-end