rdf 3.1.0 → 3.1.5

data/lib/rdf/cli.rb

@@ -8,7 +8,7 @@ begin
   require 'linkeddata'
 rescue LoadError
   # Silently load without linkeddata, but try some others
-  %w(reasoner rdfa rdfxml turtle vocab json/ld ld/patch).each do |ser|
+  %w(microdata n3 rdfa rdfxml reasoner tabular trig trix turtle vocab xsd json/ld ld/patch).each do |ser|
     begin
       require ser.include?('/') ? ser : "rdf/#{ser}"
     rescue LoadError
@@ -296,6 +296,12 @@ module RDF
         control: :none,
         on: ["-o", "--output FILE"],
         description: "File to write output, defaults to STDOUT") {|arg| File.open(arg, "w")},
+      RDF::CLI::Option.new(
+        symbol: :ordered,
+        control: :checkbox,
+        datatype: TrueClass,
+        on: ["--ordered"],
+        description: "Use order preserving repository"),
       RDF::CLI::Option.new(
         symbol: :format,
         control: :select,
@@ -495,14 +501,16 @@ module RDF
       options[:format] = options[:format].to_sym if options[:format]
       options[:output_format] = options[:output_format].to_sym if options[:output_format]
 
-      @repository = RDF::Repository.new
+      @repository = options[:ordered] ?
+        [].extend(RDF::Enumerable, RDF::Queryable) :
+        RDF::Repository.new
 
       # Parse input files if any command requires it
       if cmds.any? {|c| COMMANDS[c.to_sym][:parse]}
         start = Time.new
         count = 0
         self.parse(args, **options) do |reader|
-          @repository << reader
+          reader.each_statement {|st| @repository << st}
         end
         secs = Time.new - start
         options[:logger].info "Parsed #{repository.count} statements with #{@readers.join(', ')} in #{secs} seconds @ #{count/secs} statements/second."

data/lib/rdf/mixin/enumerable.rb

@@ -71,6 +71,7 @@ module RDF
     #   * `:literal_equality' preserves [term-equality](https://www.w3.org/TR/rdf11-concepts/#dfn-literal-term-equality) for literals. Literals are equal only if their lexical values and datatypes are equal, character by character. Literals may be "inlined" to value-space for efficiency only if `:literal_equality` is `false`.
     #   * `:validity` allows a concrete Enumerable implementation to indicate that it does or does not support valididty checking. By default implementations are assumed to support validity checking.
     #   * `:skolemize` supports [Skolemization](https://www.w3.org/wiki/BnodeSkolemization) of an `Enumerable`. Implementations supporting this feature must implement a `#skolemize` method, taking a base URI used for minting URIs for BNodes as stable identifiers and a `#deskolemize` method, also taking a base URI used for turning URIs having that prefix back into the same BNodes which were originally skolemized.
+    #   * `:rdfstar` supports RDF* where statements may be subjects or objects of other statements.
     #
     # @param  [Symbol, #to_sym] feature
     # @return [Boolean]

data/lib/rdf/mixin/mutable.rb

@@ -9,6 +9,7 @@ module RDF
     extend  RDF::Util::Aliasing::LateBound
     include RDF::Readable
     include RDF::Writable
+    include RDF::Util::Coercions
 
     ##
     # Returns `true` if `self` is mutable.
@@ -154,20 +155,9 @@ module RDF
     def delete(*statements)
       raise TypeError.new("#{self} is immutable") if immutable?
 
-      statements.map! do |value|
-        case
-          when value.respond_to?(:each_statement)
-            delete_statements(value)
-            nil
-          when (statement = Statement.from(value)).constant?
-            statement
-          else
-            delete_statements(query(value))
-            nil
-        end
+      coerce_statements(statements, query: true, constant: true) do |value|
+        delete_statements(value)
       end
-      statements.compact!
-      delete_statements(statements) unless statements.empty?
 
       return self
     end

data/lib/rdf/mixin/queryable.rb

@@ -128,6 +128,14 @@ module RDF
     # method in order to provide for storage-specific optimized triple
     # pattern matching.
     #
+    # ## RDFStar (RDF*)
+    #
+    # Statements may have embedded statements as either a subject or object, recursively.
+    #
+    # Patterns may also have embedded patterns as either a subject or object, recursively.
+    #
+    # When matching, match an embedded pattern against embedded statements, recursively. (see {RDF::Query::Pattern#eql?})
+    #
     # @param  [RDF::Query::Pattern] pattern
     #   the query pattern to match
     # @param  [Hash{Symbol => Object}] options ({})

data/lib/rdf/mixin/writable.rb

@@ -7,6 +7,7 @@ module RDF
   # @see RDF::Repository
   module Writable
     extend RDF::Util::Aliasing::LateBound
+    include RDF::Util::Coercions
 
     ##
     # Returns `true` if `self` is writable.
@@ -53,31 +54,20 @@ module RDF
     # @overload insert(*statements)
     #   @param  [Array<RDF::Statement>] statements
     #   @return [self]
+    #   @raise [ArgumentError] on an attempt to insert an embedded statement when it is not supported
     #
     # @overload insert(statements)
     #   @param  [Enumerable<RDF::Statement>] statements
     #   @return [self]
+    #   @raise [ArgumentError] on an attempt to insert an embedded statement when it is not supported
     def insert(*statements)
-      statements.map! do |value|
-        case
-          when value.respond_to?(:each_statement)
-            insert_statements(value)
-            nil
-          when (statement = Statement.from(value))
-            statement
-          else
-            raise ArgumentError.new("not a valid statement: #{value.inspect}")
-        end
-      end
-      statements.compact!
-      insert_statements(statements) unless statements.empty?
+      coerce_statements(statements) { |value| insert_statements value }
 
       return self
     end
     alias_method :insert!, :insert
 
   protected
-
     ##
     # Inserts statements from the given RDF reader into the underlying
     # storage or output stream.
@@ -132,10 +122,14 @@ module RDF
     #
     # @param  [RDF::Enumerable] statements
     # @return [void]
+    # @raise [ArgumentError] on an attempt to insert an embedded statement when it is not supported
     # @since  0.1.6
     def insert_statements(statements)
       each = statements.respond_to?(:each_statement) ? :each_statement : :each
       statements.__send__(each) do |statement|
+        if statement.embedded? && respond_to?(:supports?) && !supports?(:rdfstar)
+          raise ArgumentError, "Wriable does not support embedded statements"
+        end
         insert_statement(statement)
       end
     end
@@ -150,6 +144,7 @@ module RDF
     #
     # @param  [RDF::Statement] statement
     # @return [void]
+    # @raise [ArgumentError] on an attempt to insert an embedded statement when it is not supported
     # @abstract
     def insert_statement(statement)
       raise NotImplementedError.new("#{self.class}#insert_statement")

data/lib/rdf/model/dataset.rb

@@ -104,7 +104,7 @@ module RDF
     # @private
     # @see RDF::Enumerable#supports?
     def supports?(feature)
-      return true if feature == :graph_name
+      return true if [:graph_name, :rdfstar].include?(feature)
       super
     end
 

data/lib/rdf/model/graph.rb

@@ -283,6 +283,9 @@ module RDF
     # @private
     # @see RDF::Mutable#insert
     def insert_statement(statement)
+      if statement.embedded? && !@data.supports?(:rdfstar)
+        raise ArgumentError, "Graph does not support embedded statements"
+      end
       statement = statement.dup
       statement.graph_name = graph_name
       @data.insert(statement)

data/lib/rdf/model/statement.rb

@@ -26,7 +26,7 @@ module RDF
   #   RDF::Statement(s, p, "o")
   #
   class Statement
-    include RDF::Value
+    include RDF::Resource
 
     ##
     # @private
@@ -112,7 +112,7 @@ module RDF
       elsif @subject.nil?
         nil
       else
-        raise ArgumentError, "expected subject to be nil or a term, was #{@subject.inspect}"
+        raise ArgumentError, "expected subject to be nil or a resource, was #{@subject.inspect}"
       end
       @predicate = Node.intern(@predicate) if @predicate.is_a?(Symbol)
       @object    = if @object.is_a?(Value)
@@ -124,6 +124,15 @@ module RDF
       else
         Literal.new(@object)
       end
+      @graph_name = if @graph_name.is_a?(Value)
+        @graph_name.to_term
+      elsif @graph_name.is_a?(Symbol)
+        Node.intern(@graph_name)
+      elsif !@graph_name
+        @graph_name
+      else
+        raise ArgumentError, "expected graph_name to be nil or a resource, was #{@graph_name.inspect}"
+      end
     end
 
     ##
@@ -140,10 +149,18 @@ module RDF
     #
     # @return [Boolean]
     def variable?
-      !(has_subject?    && subject.resource? &&
-        has_predicate?  && predicate.resource? &&
-        has_object?     && (object.resource? || object.literal?) &&
-        (has_graph?     ? graph_name.resource? : true))
+      !(has_subject?    && subject.constant? &&
+        has_predicate?  && predicate.constant? &&
+        has_object?     && object.constant? &&
+        (has_graph?     ? graph_name.constant? : true))
+    end
+
+    ##
+    # Returns `true` if any element of the statement is, itself, a statement.
+    #
+    # @return [Boolean]
+    def embedded?
+      subject && subject.statement? || object && object.statement?
     end
 
     ##
@@ -386,7 +403,13 @@ module RDF
     # @return [String]
     def to_s
       (graph_name ? to_quad : to_triple).map do |term|
-        term.respond_to?(:to_base) ? term.to_base : term.inspect
+        if term.is_a?(Statement)
+          "<<#{term.to_s[0..-3]}>>"
+        elsif term.respond_to?(:to_base)
+          term.to_base
+        else
+          term.inspect
+        end
       end.join(" ") + " ."
     end
 

data/lib/rdf/model/uri.rb

@@ -1,5 +1,5 @@
 # coding: utf-8
-require 'uri'
+require 'cgi'
 
 module RDF
   ##
@@ -400,7 +400,7 @@ module RDF
     # @example Joining two URIs
     #     RDF::URI.new('http://example.org/foo/bar').join('/foo')
     #     #=> RDF::URI('http://example.org/foo')
-    # @see <http://github.com/ruby-rdf/rdf-spec/blob/master/lib/rdf/spec/uri.rb>
+    # @see <https://github.com/ruby-rdf/rdf-spec/blob/master/lib/rdf/spec/uri.rb>
     # @see <http://tools.ietf.org/html/rfc3986#section-5.2>
     # @see RDF::URI#/
     # @see RDF::URI#+
@@ -471,7 +471,7 @@ module RDF
     # @see RDF::URI#+
     # @see RDF::URI#join
     # @see <http://tools.ietf.org/html/rfc3986#section-5.2>
-    # @see <http://github.com/ruby-rdf/rdf-spec/blob/master/lib/rdf/spec/uri.rb>
+    # @see <https://github.com/ruby-rdf/rdf-spec/blob/master/lib/rdf/spec/uri.rb>
     # @example Building a HTTP URL
     #     RDF::URI.new('http://example.org') / 'jhacker' / 'foaf.ttl'
     #     #=> RDF::URI('http://example.org/jhacker/foaf.ttl')
@@ -846,7 +846,7 @@ module RDF
         parts[:user] = (user.dup.force_encoding(Encoding::UTF_8) if user)
         parts[:password] = (password.dup.force_encoding(Encoding::UTF_8) if password)
         parts[:host] = (host.dup.force_encoding(Encoding::UTF_8) if host)
-        parts[:port] = (URI.decode(port).to_i if port)
+        parts[:port] = (CGI.unescape(port).to_i if port)
         parts[:path] = (path.to_s.dup.force_encoding(Encoding::UTF_8) unless path.empty?)
         parts[:query] = (query[1..-1].dup.force_encoding(Encoding::UTF_8) if query)
         parts[:fragment] = (fragment[1..-1].dup.force_encoding(Encoding::UTF_8) if fragment)
@@ -902,7 +902,7 @@ module RDF
     # Normalized version of user
     # @return [String]
     def normalized_user
-      URI.encode(URI.decode(user), /[^#{IUNRESERVED}|#{SUB_DELIMS}]/) if user
+      URI.encode(CGI.unescape(user), /[^#{IUNRESERVED}|#{SUB_DELIMS}]/).force_encoding(Encoding::UTF_8) if user
     end
 
     ##
@@ -928,7 +928,7 @@ module RDF
     # Normalized version of password
     # @return [String]
     def normalized_password
-      URI.encode(URI.decode(password), /[^#{IUNRESERVED}|#{SUB_DELIMS}]/) if password
+      URI.encode(CGI.unescape(password), /[^#{IUNRESERVED}|#{SUB_DELIMS}]/).force_encoding(Encoding::UTF_8) if password
     end
 
     HOST_FROM_AUTHORITY_RE = /(?:[^@]+@)?([^:]+)(?::.*)?$/.freeze
@@ -1180,8 +1180,8 @@ module RDF
         inject(return_type == Hash ? {} : []) do |memo,kv|
           k,v = kv.to_s.split('=', 2)
           next if k.to_s.empty?
-          k = URI.decode(k)
-          v = URI.decode(v) if v
+          k = CGI.unescape(k)
+          v = CGI.unescape(v) if v
           if return_type == Hash
             case memo[k]
             when nil then memo[k] = v
@@ -1293,9 +1293,9 @@ module RDF
     def normalize_segment(value, expr, downcase = false)
       if value
         value = value.dup.force_encoding(Encoding::UTF_8)
-        decoded = URI.decode(value)
+        decoded = CGI.unescape(value)
         decoded.downcase! if downcase
-        URI.encode(decoded, /[^(?:#{expr})]/)
+        URI.encode(decoded, /[^(?:#{expr})]/).force_encoding(Encoding::UTF_8)
       end
     end
 

data/lib/rdf/nquads.rb

@@ -69,9 +69,9 @@ module RDF
 
           begin
             unless blank? || read_comment
-              subject   = read_uriref || read_node || fail_subject
+              subject   = read_uriref || read_node || read_rdfstar || fail_subject
               predicate = read_uriref(intern: true) || fail_predicate
-              object    = read_uriref || read_node || read_literal || fail_object
+              object    = read_uriref || read_node || read_literal || read_rdfstar || fail_object
               graph_name    = read_uriref || read_node
               if validate? && !read_eos
                 log_error("Expected end of statement (found: #{current_line.inspect})", lineno: lineno, exception: RDF::ReaderError)

data/lib/rdf/ntriples.rb

@@ -15,15 +15,17 @@ module RDF
   #
   #     <https://rubygems.org/gems/rdf> <http://purl.org/dc/terms/title> "rdf" .
   #
-  # Installation
-  # ------------
+  # ## RDFStar (RDF*)
+  #
+  # Supports statements as resources using `<<s p o>>`.
+  #
+  # ## Installation
   #
   # This is the only RDF serialization format that is directly supported by
   # RDF.rb. Support for other formats is available in the form of add-on
   # gems, e.g. 'rdf-xml' or 'rdf-json'.
   #
-  # Documentation
-  # -------------
+  # ## Documentation
   #
   # * {RDF::NTriples::Format}
   # * {RDF::NTriples::Reader}

data/lib/rdf/ntriples/reader.rb

@@ -1,4 +1,7 @@
 # -*- encoding: utf-8 -*-
+
+require 'strscan'
+
 module RDF::NTriples
   ##
   # N-Triples parser.
@@ -25,6 +28,10 @@ module RDF::NTriples
   #     end
   #   end
   #
+  # ** RDFStar (RDF*)
+  #
+  # Supports statements as resources using `<<s p o>>`.
+  #
   # @see http://www.w3.org/TR/rdf-testcases/#ntriples
   # @see http://www.w3.org/TR/n-triples/
   class Reader < RDF::Reader
@@ -70,6 +77,10 @@ module RDF::NTriples
     # 22
     STRING_LITERAL_QUOTE = /"((?:[^\"\\\n\r]|#{ECHAR}|#{UCHAR})*)"/.freeze
 
+    # RDF*
+    ST_START              = /^<</.freeze
+    ST_END                = /^\s*>>/.freeze
+
     # @see http://www.w3.org/TR/rdf-testcases/#ntrip_grammar
     COMMENT               = /^#\s*(.*)$/.freeze
     NODEID                = /^#{BLANK_NODE_LABEL}/.freeze
@@ -202,7 +213,7 @@ module RDF::NTriples
       begin
         read_statement
       rescue RDF::ReaderError
-        value = read_uriref || read_node || read_literal
+        value = read_uriref || read_node || read_literal || read_rdfstar
         log_recover
         value
       end
@@ -218,9 +229,9 @@ module RDF::NTriples
 
         begin
           unless blank? || read_comment
-            subject   = read_uriref || read_node || fail_subject
+            subject   = read_uriref || read_node || read_rdfstar || fail_subject
             predicate = read_uriref(intern: true) || fail_predicate
-            object    = read_uriref || read_node || read_literal || fail_object
+            object    = read_uriref || read_node || read_literal || read_rdfstar || fail_object
 
             if validate? && !read_eos
               log_error("Expected end of statement (found: #{current_line.inspect})", lineno: lineno, exception: RDF::ReaderError)
@@ -234,6 +245,20 @@ module RDF::NTriples
       end
     end
 
+    ##
+    # @return [RDF::Statement]
+    def read_rdfstar
+      if @options[:rdfstar] && match(ST_START)
+        subject   = read_uriref || read_node || read_rdfstar || fail_subject
+        predicate = read_uriref(intern: true) || fail_predicate
+        object    = read_uriref || read_node || read_literal || read_rdfstar || fail_object
+        if !match(ST_END)
+          log_error("Expected end of statement (found: #{current_line.inspect})", lineno: lineno, exception: RDF::ReaderError)
+        end
+        RDF::Statement.new(subject, predicate, object)
+      end
+    end
+
     ##
     # @return [Boolean]
     # @see    http://www.w3.org/TR/rdf-testcases/#ntrip_grammar (comment)

data/lib/rdf/ntriples/writer.rb

@@ -221,6 +221,15 @@ module RDF::NTriples
       format_triple(*statement.to_triple, **options)
     end
 
+    ##
+    # Returns the N-Triples representation of an RDF* reified statement.
+    #
+    # @param  [RDF::Statement] statement
+    # @param  [Hash{Symbol => Object}] options ({})
+    # @return [String]
+    def format_rdfstar(statement, **options)
+      "<<%s %s %s>>" % statement.to_a.map { |value| format_term(value, **options) }
+    end
     ##
     # Returns the N-Triples representation of a triple.
     #

data/lib/rdf/query.rb

@@ -289,6 +289,8 @@ module RDF
     #   any additional keyword options
     # @option options [Hash{Symbol => RDF::Term}] bindings
     #   optional variable bindings to use
+    # @option options [Boolean] :optimize
+    #   Optimize query before execution.
     # @option options [RDF::Query::Solutions] solutions
     #   optional initial solutions for chained queries
     # @yield  [solution]
@@ -311,6 +313,7 @@ module RDF
         return @solutions
       end
 
+      self.optimize! if options[:optimize]
       patterns = @patterns
       graph_name = name if graph_name.nil?
       @graph_name = graph_name unless graph_name.nil?
@@ -505,7 +508,7 @@ module RDF
     # @return [RDF::Query]
     def dup
       patterns = @patterns.map {|p| p.dup}
-      Query.new(patterns, solutions: @solutions.dup, **options)
+      Query.new(patterns, graph_name: graph_name, solutions: @solutions.dup, **options)
     end
 
     ##