Add version option and accessor to RDF::Reader.

gkellogg · gkellogg · commit b3f9a12ca9de · 2025-07-01T17:08:58.000-07:00
Add VERSION support in N-Triples parser, with warning when unexpected combination is used.
diff --git a/lib/rdf/ntriples/reader.rb b/lib/rdf/ntriples/reader.rb
@@ -92,6 +92,7 @@ class Reader < RDF::Reader
 
     # LANGTAG is deprecated
     LANGTAG               = LANG_DIR
+    RDF_VERSION          = /VERSION/.freeze
 
     ##
     # Reconstructs an RDF value from its serialized N-Triples
@@ -154,7 +155,8 @@ def self.parse_uri(input, intern: false, **options)
     def self.parse_literal(input, **options)
       case input
         when LITERAL_WITH_LANGUAGE
-          RDF::Literal.new(unescape($1), language: $4)
+          language, direction = $4.split('--')
+          RDF::Literal.new(unescape($1), language: language, direction: direction)
         when LITERAL_WITH_DATATYPE
           RDF::Literal.new(unescape($1), datatype: $4)
         when LITERAL_PLAIN
@@ -216,7 +218,11 @@ def read_triple
         line = @line    # for backtracking input in case of parse error
 
         begin
-          unless blank? || read_comment
+          if blank? || read_comment
+            # No-op
+          elsif version = read_version
+            @options[:version] = version
+          else
             subject   = read_uriref || read_node || read_quotedTriple || fail_subject
             predicate = read_uriref(intern: true) || fail_predicate
             object    = read_uriref || read_node || read_literal || read_tripleTerm || read_quotedTriple || fail_object
@@ -237,6 +243,9 @@ def read_triple
     # @return [RDF::Statement]
     def read_tripleTerm
       if @options[:rdfstar] && match(TT_START)
+        if version && version != "1.2"
+          log_warn("Triple term used with version #{version}")
+        end
         subject   = read_uriref || read_node || fail_subject
         predicate = read_uriref(intern: true) || fail_predicate
         object    = read_uriref || read_node || read_literal || read_tripleTerm || fail_object
@@ -307,6 +316,7 @@ def read_literal
           when lang_dir = match(LANG_DIR)
             language, direction = lang_dir.split('--')
             raise ArgumentError if direction && !@options[:rdfstar]
+            log_warn("Literal base direction used with version #{version}") if version && version == "1.1"
             RDF::Literal.new(literal_str, language: language, direction: direction)
           when datatype = match(/^(\^\^)/) # FIXME
             RDF::Literal.new(literal_str, datatype: read_uriref || fail_object)
@@ -323,6 +333,18 @@ def read_literal
       log_error("Invalid Literal (found: \"#{v}\")", lineno: lineno, token: "#v", exception: RDF::ReaderError)
     end
 
+    ##
+    # @return [String]
+    def read_version
+      if match(RDF_VERSION)
+        ver_tok = match(LITERAL_PLAIN)
+        unless RDF::Format::VERSIONS.include?(ver_tok)
+          log_warn("Expected version to be one of #{RDF::Format::VERSIONS.join(', ')}, was #{ver_tok}")
+        end
+        ver_tok
+      end
+    end
+
     ##
     # @return [Boolean]
     # @see http://www.w3.org/TR/rdf-testcases/#ntrip_grammar (triple)
diff --git a/lib/rdf/reader.rb b/lib/rdf/reader.rb
@@ -176,7 +176,13 @@ def self.options
           default: true,
           control: :checkbox,
           on: ["--[no-]verifySSL"],
-          description: "Verify SSL results on HTTP GET")
+          description: "Verify SSL results on HTTP GET"),
+        RDF::CLI::Option.new(
+          symbol: :version,
+          control: :select,
+          datatype: RDF::Format::VERSIONS, # 1.1, 1.2, or 1.2-basic
+          on: ["--version"],
+          description: "RDF Version."),
       ]
     end
 
@@ -284,6 +290,8 @@ def to_sym
     #   any additional options
     # @param [Boolean]  validate     (false)
     #   whether to validate the parsed statements and values
+    # @option options [String] :version ("1.2")
+    #   Parse a specific version of RDF ("1.1', "1.2", or "1.2-basic"")
     # @yield  [reader] `self`
     # @yieldparam  [RDF::Reader] reader
     # @yieldreturn [void] ignored
@@ -309,6 +317,9 @@ def initialize(input = $stdin,
         validate:       validate
       })
 
+      # The rdfstar option implies version 1.2, but can be overridden
+      @options[:version] ||= "1.2" if @options[:rdfstar]
+
       @input = case input
         when String then StringIO.new(input)
         else input
@@ -391,6 +402,18 @@ def prefix(name, uri = nil)
     end
     alias_method :prefix!, :prefix
 
+    ##
+    # Returns the RDF version determined by this reader.
+    #
+    # @example
+    #   reader.version  #=> RDF::URI('http://purl.org/dc/terms/')
+    #
+    # @return [String]
+    # @since  3.3.2
+    def version
+      @options[:version]
+    end
+
     ##
     # Iterates the given block for each RDF statement.
     #
diff --git a/spec/ntriples_spec.rb b/spec/ntriples_spec.rb
@@ -152,6 +152,10 @@
     it "should accept strings" do
       expect { reader.new('') }.not_to raise_error
     end
+
+    it "sets version from reader option" do
+      expect(reader.new('', version: '1.2').version).to eql '1.2'
+    end
   end
 
   describe ".open" do
@@ -566,6 +570,16 @@
             end
           end
         end
+
+        it "warns if version is not 1.2 or 1.2-basic" do
+          expect {parse('<http://subj> <http://pred>  "Hello"@en--ltr .', rdfstar: true, version: '1.1', logger: logger)}.not_to raise_error
+          expect(logger.to_s).to include("WARN")
+        end
+
+        it "warns if inline version is not 1.2 or 1.2-basic" do
+          expect {parse(%{VERSION "1.1"\n<http://subj> <http://pred>  "Hello"@en--ltr .}, rdfstar: true, logger: logger)}.not_to raise_error
+          expect(logger.to_s).to include("WARN")
+        end
       end
 
       context 'should parse a value that was written without passing through the writer encoding' do
@@ -664,6 +678,43 @@
       end
     end
 
+    context "version" do
+      {
+        'VERSION "1.2"': %(
+          VERSION "1.2"
+          <http://example/s> <http://example/p> <http://example/o>
+        ),
+        'VERSION "1.2-basic"': %(
+          VERSION "1.2-basic"
+          <http://example/s> <http://example/p> <http://example/o>
+        ),
+        'VERSION "1.1"': %(
+          VERSION "1.1"
+          <http://example/s> <http://example/p> <http://example/o>
+        ),
+      }.each do |name, input|
+        it name do
+          expect do
+            parse(input, logger: logger)
+            expect(logger.to_s).not_to include("WARN")
+          end.not_to write.to(:error)
+        end
+      end
+
+      it 'version "1.2" is an error' do
+        expect do
+          parse('version "1.2"', logger: logger)
+        end.to raise_error(RDF::ReaderError)
+      end
+
+      it 'VERSION "1.0" is a warning' do
+        expect do
+          parse('VERSION "1.0"', logger: logger)
+        end.not_to write.to(:error)
+        expect(logger.to_s).to include('WARN')
+      end
+    end
+
     context "triple terms" do
       ill_statements = {
         "subject-iii": '<<(<http://example/s1> <http://example/p1> <http://example/o1>)>> <http://example/p> <http://example/o> .',
@@ -709,20 +760,35 @@
 
         statements.each do |name, st|
           context name do
-            let(:graph) {parse(st, rdfstar: true)}
+            let(:graph) {parse(st, rdfstar: true, logger: logger)}
 
             it "creates an unquoted statement" do
               expect(graph.count).to eql(1)
               graph.statements.each do |stmt|
                 expect(stmt).not_to be_quoted
               end
+              expect(logger.to_s).not_to include("WARN")
             end
 
             it "statements which are object of another statement are triple terms" do
               referencing = graph.statements.first
               expect(referencing).to be_a_statement
               expect(referencing.object).to be_a_statement
               expect(referencing.object).to be_tripleTerm
+              expect(logger.to_s).not_to include("WARN")
+            end
+
+            it "warns if version is not 1.2" do
+              expect {parse(st, rdfstar: true, version: '1.1', logger: logger)}.not_to raise_error
+              expect(logger.to_s).to include("WARN")
+            end
+
+            it "warns if inline version is not 1.2" do
+              expect {parse(%{VERSION "1.1"\n#{st}}, rdfstar: true, logger: logger)}.not_to raise_error
+              expect(logger.to_s).to include("WARN")
+
+              expect {parse(%{VERSION "1.2-basic"\n#{st}}, rdfstar: true, logger: logger)}.not_to raise_error
+              expect(logger.to_s).to include("WARN")
             end
           end
         end
