pjstadig-rena 0.0.1

data/Rakefile

@@ -0,0 +1,50 @@
+require 'rubygems'
+require 'rake'
+require 'spec/rake/spectask'
+
+task :default => [:spec]
+
+desc "Install dependencies"
+task :dependencies do
+  require ''
+  gems = ['addressable/uri']
+  gems.each do |g|
+    g2 = g.split('/')[0]
+    begin
+      require g2
+    rescue
+      sh "sudo gem install " + g2
+    end
+  end
+end
+
+desc "Pushes to git"
+task :push do
+  sh "git push --all"
+  sh "growlnotify -m \"Updates pushed\" \"Git\""
+end
+
+desc "Runs specs"
+task :spec do
+  sh "spec --colour --pattern test/spec/*.spec.rb"
+end
+
+desc "Turns spec results into HTML and publish to web (Tom only!)"
+task :spec_html do
+  sh "spec --pattern test/spec/*.spec.rb --format html:rena_new_spec.html"
+  sh "scp rena_new_spec.html bbcityco@bbcity.co.uk:www/tom/files/rena_new_spec.html"
+  sh "rm rena_new_spec.html"
+end
+
+desc "Turns spec results into local HTML"
+task :spec_local do
+  sh "spec --pattern test/spec/*.spec.rb --format html:rena_new_spec.html"
+#  sh "open rena_new_spec.html"
+end
+
+desc "Run specs through RCov"
+Spec::Rake::SpecTask.new('coverage') do |t|
+  t.spec_files = FileList['test/spec/**/*.rb']
+  t.rcov = true
+  t.rcov_opts = ['--exclude', 'test,\/Library\/Ruby\/Gems\/1.8\/gems']
+end

data/lib/rena.rb

@@ -0,0 +1,5 @@
+$:.unshift File.dirname(__FILE__)
+Dir.glob(File.join(File.dirname(__FILE__), 'rena/**.rb')).each { |f| require f }
+
+module Rena
+end

data/lib/rena/bnode.rb

@@ -0,0 +1,63 @@
+class BNode
+  attr_accessor :identifier
+  def initialize(identifier = nil)
+    if identifier != nil && self.valid_id?(identifier) != false
+      @identifier = identifier
+    else
+      @identifier = "bn" + self.hash.to_s
+    end
+  end
+  
+  def eql? (eql)
+    if self.identifier == eql.identifier
+      true
+    else
+      false
+    end
+  end
+  
+  ## 
+  # Exports the BNode in N-Triples form.
+  #
+  # ==== Example
+  #   b = BNode.new; b.to_n3  # => returns a string of the BNode in n3 form
+  #
+  # ==== Returns
+  # @return [String] The BNode in n3.
+  #
+  # @author Tom Morris
+  
+  def to_n3
+    "_:" + @identifier
+  end
+  
+  
+  ## 
+  # Exports the BNode in N-Triples form.
+  #
+  # ==== Example
+  #   b = BNode.new; b.to_ntriples  # => returns a string of the BNode in N-Triples form
+  #
+  # ==== Returns
+  # @return [String] The BNode in N-Triples.
+  #
+  # @author Tom Morris
+  
+  def to_ntriples
+    self.to_n3
+  end
+  
+  def to_s
+    @identifier
+  end  
+  
+  # TODO: add valid bnode name exceptions?
+  protected
+  def valid_id? name
+    if name =~ /^[a-zA-Z_][a-zA-Z0-9]*$/
+      true
+    else
+      false
+    end
+  end
+end

data/lib/rena/exceptions/about_each_exception.rb

@@ -0,0 +1,2 @@
+class AboutEachException
+end

data/lib/rena/exceptions/uri_relative_exception.rb

@@ -0,0 +1,2 @@
+class UriRelativeException < RuntimeError
+end

data/lib/rena/graph.rb

@@ -0,0 +1,150 @@
+require 'rena/namespace'
+require 'rena/bnode'
+require 'rena/uriref'
+require 'rena/literal'
+require 'rena/triple'
+
+class Graph
+  attr_accessor :triples, :nsbinding
+  
+  def initialize
+    @triples = []
+    @nsbinding = {}
+  end
+  
+  def size
+    @triples.size
+  end
+  
+  def each
+    @triples.each { |value| yield value }
+  end
+  
+  def each_with_subject(subject)
+    @triples.each {|value|
+      if value.subject == subject
+        yield value
+      end
+    }
+  end
+  
+  ## 
+  # Adds a triple to a graph directly from the intended subject, predicate, and object.
+  #
+  # ==== Example
+  #   g = Graph.new; g.add_triple(BNode.new, URIRef.new("http://xmlns.com/foaf/0.1/knows"), BNode.new) # => results in the triple being added to g; returns an array of g's triples
+  #
+  # @param [URIRef, BNode] s the subject of the triple
+  # @param [URIRef] p the predicate of the triple
+  # @param [URIRef, BNode, Literal, TypedLiteral] o the object of the triple
+  #
+  # ==== Returns
+  # @return [Array] An array of the triples (leaky abstraction? consider returning the graph instead)
+  #
+  # @raise [Error] Checks parameter types and raises if they are incorrect.
+  # @author Tom Morris
+  
+  def add_triple(s, p, o)
+    @triples += [ Triple.new(s, p, o) ]
+  end
+  
+  ## 
+  # Adds an extant triple to a graph
+  #
+  # ==== Example
+  #   g = Graph.new; t = Triple.new(BNode.new, URIRef.new("http://xmlns.com/foaf/0.1/knows"), BNode.new); g << t) # => results in the triple being added to g; returns an array of g's triples
+  #
+  # @param [Triple] t the triple to be added to the graph
+  #
+  # ==== Returns
+  # @return [Array] An array of the triples (leaky abstraction? consider returning the graph instead)
+  #
+  # @author Tom Morris
+  
+  
+  def << (triple)
+#    self.add_triple(s, p, o)
+    @triples += [ triple ]
+  end
+  
+  ## 
+  # Exports the graph to RDF in N-Triples form.
+  #
+  # ==== Example
+  #   g = Graph.new; g.add_triple(BNode.new, URIRef.new("http://xmlns.com/foaf/0.1/knows"), BNode.new); g.to_ntriples  # => returns a string of the graph in N-Triples form
+  #
+  # ==== Returns
+  # @return [String] The graph in N-Triples.
+  #
+  # @author Tom Morris
+  
+  def to_ntriples
+    str = ""
+    @triples.each do |t|
+      str << t.to_ntriples + "\n"
+    end
+    return str
+  end
+  
+  ## 
+  # Creates a new namespace given a URI and the short name and binds it to the graph.
+  #
+  # ==== Example
+  #   g = Graph.new; g.namespace("http://xmlns.com/foaf/0.1/", "foaf") # => binds the Foaf namespace to g
+  #
+  # @param [String] uri the URI of the namespace
+  # @param [String] short the short name of the namespace
+  #
+  # ==== Returns
+  # @return [Namespace] The newly created namespace.
+  #
+  # @raise [Error] Checks validity of the desired shortname and raises if it is incorrect.
+  # @raise [Error] Checks that the newly created Namespace is of type Namespace and raises if it is incorrect.
+  # @author Tom Morris
+  
+  def namespace(uri, short)
+    self.bind Namespace.new(uri, short)
+  end
+  
+  def bind(namespace)
+    if namespace.class == Namespace
+      @nsbinding["#{namespace.short}"] = namespace
+    else
+      raise
+    end
+  end
+  
+  def has_bnode_identifier?(bnodeid)
+    temp_bnode = BNode.new(bnodeid)
+    returnval = false
+    @triples.each { |triple|
+      if triple.subject.eql?(temp_bnode)
+        returnval = true
+        break
+      end
+      if triple.object.eql?(temp_bnode)
+        returnval = true
+        break
+      end
+    }
+    return returnval
+  end
+  
+  def get_bnode_by_identifier(bnodeid)
+    temp_bnode = BNode.new(bnodeid)
+    returnval = false
+    @triples.each { |triple|
+      if triple.subject.eql?(temp_bnode)
+        returnval = triple.subject
+        break
+      end
+      if triple.object.eql?(temp_bnode)
+        returnval = triple.object
+        break
+      end
+    }
+    return returnval
+  end
+#  alias :add, :add_triple
+#  alias (=+, add_triple)
+end

data/lib/rena/literal.rb

@@ -0,0 +1,89 @@
+class Literal
+  attr_accessor :contents, :lang
+  def initialize(contents, lang = nil)
+    @contents = contents
+    if lang != nil && lang != false
+      @lang = lang.downcase
+    end
+  end
+  
+  def == (obj)
+    if obj.class == Literal && obj.contents == @contents && (obj.lang == @lang || (obj.lang == nil && @lang == nil))
+      true
+    else
+      false
+    end
+  end
+  
+  def to_n3
+    out = "\"" + @contents + "\""
+    out += "@" + @lang if @lang != nil
+    out += "^^" + @encoding if @encoding != nil
+    return out
+  end
+  
+  def to_ntriples
+    return self.to_n3
+  end
+
+  def to_trix
+    if @lang != nil && @lang != false
+      out = "<plainLiteral xml:lang=\"" + @lang + "\">"
+    else
+      out = "<plainLiteral>"
+    end
+    out += @contents
+    out += "</plainLiteral>"
+    return out
+  end
+  
+end
+
+class TypedLiteral < Literal
+  attr_accessor :contents, :encoding
+  def initialize(contents, encoding)
+    @contents = contents
+    @encoding = encoding
+    if @encoding == "http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral"
+      @xmlliteral = true
+    else
+      @xmlliteral = false
+    end
+  end
+  
+  def == (obj)
+    if obj.class == TypedLiteral && obj.contents == @contents && obj.encoding == @encoding
+      return true
+    else
+      return false
+    end
+  end
+  
+  def to_n3
+    if @encoding == "http://www.w3.org/2001/XMLSchema#int"
+      out = @contents.to_s
+    else
+      out = "\"" + @contents.to_s + "\""
+    end
+    out += "^^<" + @encoding + ">" if @encoding != nil
+    return out
+  end
+  
+  def to_trix
+    "<typedLiteral datatype=\"" + @encoding + "\">" + @contents + "</typedLiteral>"
+  end
+  
+  def xmlliteral?
+    @xmlliteral
+  end
+  
+  def infer!
+    if @contents.class == Fixnum
+      @encoding = "http://www.w3.org/2001/XMLSchema#int"
+    elsif @contents.class == Float 
+      @encoding = "http://www.w3.org/2001/XMLSchema#float"
+    else
+      @encoding = "http://www.w3.org/2001/XMLSchema#string"
+    end
+  end
+end

data/lib/rena/namespace.rb

@@ -0,0 +1,74 @@
+require 'rena/uriref'
+require 'rena/graph'
+
+class Namespace
+  attr_accessor :short, :uri, :fragment
+   
+  ## 
+  # Creates a new namespace given a URI and the short name.
+  #
+  # ==== Example
+  #   Namespace.new("http://xmlns.com/foaf/0.1/", "foaf") # => returns a new Foaf namespace
+  #
+  # @param [String] uri the URI of the namespace
+  # @param [String] short the short name of the namespace
+  # @param [Boolean] fragment are the identifiers on this resource fragment identifiers? (e.g. '#')  Defaults to false.
+  #
+  # ==== Returns
+  # @return [Namespace] The newly created namespace.
+  #
+  # @raise [Error] Checks validity of the desired shortname and raises if it is incorrect.
+  # @author Tom Morris, Pius Uzamere
+  
+  def initialize(uri, short, fragment = false)
+    @uri = uri
+    @fragment = fragment
+    if shortname_valid?(short)
+      @short = short
+    else
+      raise
+    end
+  end
+  
+  ## 
+  # Allows the construction of arbitrary URIs on the namespace.
+  #
+  # ==== Example
+  #   foaf = Namespace.new("http://xmlns.com/foaf/0.1/", "foaf"); foaf.knows # => returns a new URIRef with URI "http://xmlns.com/foaf/0.1/knows"
+  #   foaf = Namespace.new("http://xmlns.com/foaf/0.1/", "foaf", true); foaf.knows # => returns a new URIRef with URI "http://xmlns.com/foaf/0.1/#knows"
+  #
+  # @param [String] uri the URI of the namespace
+  # @param [String] short the short name of the namespace
+  # @param [Boolean] fragment are the identifiers on this resource fragment identifiers? (e.g. '#')  Defaults to false.
+  #
+  # ==== Returns
+  # @return [URIRef] The newly created URIRegerence.
+  #
+  # @raise [Error] Checks validity of the desired shortname and raises if it is incorrect.
+  # @author Tom Morris, Pius Uzamere
+  
+  def method_missing(methodname, *args)
+    unless fragment
+      URIRef.new(@uri + methodname.to_s)
+    else
+      URIRef.new(@uri + '#' + methodname.to_s)
+    end
+  end
+  
+  def bind(graph)
+    if graph.class == Graph
+      graph.bind(self)
+    else
+      raise
+    end
+  end
+  
+  private
+  def shortname_valid?(shortname)
+    if shortname =~ /[a-zA-Z_][a-zA-Z0-9_]+/
+      return true
+    else
+      return false
+    end
+  end
+end

data/lib/rena/rdfxmlparser.rb

@@ -0,0 +1,182 @@
+require 'rena/uriref'
+require 'rena/graph'
+require 'rena/literal'
+require 'rena/exceptions/uri_relative_exception'
+require 'rena/exceptions/about_each_exception'
+require 'rena/rexml_hacks'
+
+class RdfXmlParser
+  attr_accessor :xml, :graph
+  def initialize (xml_str, uri = nil)
+    @excl = ["http://www.w3.org/1999/02/22-rdf-syntax-ns#resource", "http://www.w3.org/1999/02/22-rdf-syntax-ns#nodeID", "http://www.w3.org/1999/02/22-rdf-syntax-ns#about", "http://www.w3.org/1999/02/22-rdf-syntax-ns#ID"]
+    if uri != nil
+      @uri = Addressable::URI.parse(uri)
+    end
+    @xml = REXML::Document.new(xml_str)
+#    self.iterator @xml.root.children
+    if self.is_rdf?
+      @graph = Graph.new
+      @xml.root.each_element { |e|
+        self.parse_element e
+      }
+#      puts @graph.size
+    end
+  end
+  
+  def is_rdf?
+    trigger = false
+    @xml.each_element do |e|
+      if e.namespaces.has_value? "http://www.w3.org/1999/02/22-rdf-syntax-ns#"
+        trigger = true
+      end
+    end
+    return trigger
+  end
+  
+  protected
+  def get_uri_from_atts (element, aboutmode = false)
+    if aboutmode == false
+      resourceuri = "http://www.w3.org/1999/02/22-rdf-syntax-ns#resource"
+    else
+      resourceuri = "http://www.w3.org/1999/02/22-rdf-syntax-ns#about"
+    end
+    
+    subject = nil
+    element.attributes.each_attribute { |att|
+      uri = att.namespace + att.name
+      value = att.to_s
+      if uri == "http://www.w3.org/1999/02/22-rdf-syntax-ns#aboutEach"
+        raise AboutEachException, "Failed as per RDFMS-AboutEach-Error001.rdf test from 2004 test suite"
+      end
+      if uri == "http://www.w3.org/1999/02/22-rdf-syntax-ns#aboutEachPrefix"
+        raise AboutEachException, "Failed as per RDFMS-AboutEach-Error002.rdf test from 2004 test suite"
+      end
+      if uri == "http://www.w3.org/1999/02/22-rdf-syntax-ns#bagID"
+        raise
+        if name =~ /^[a-zA-Z_][a-zA-Z0-9]*$/
+          # TODO: do something intelligent with the bagID
+        else
+          raise
+        end
+      end
+      
+      if uri == resourceuri #specified resource
+        begin
+          possible_subject = URIRef.new(value)
+        rescue UriRelativeException
+          if value[0..0].to_s != "#"
+            value = "#" + value
+          end
+          begin
+            value = URIRef.new(element.base + value)
+          rescue UriRelativeException
+            # still not a URI
+            raise
+          else
+            subject = value
+          end
+        else
+          subject = possible_subject
+          break
+        end
+      end
+      
+      if uri == "http://www.w3.org/1999/02/22-rdf-syntax-ns#nodeID" #BNode with ID
+        # we have a BNode with an identifier. First, we need to do syntax checking.
+        if value =~ /^[a-zA-Z_][a-zA-Z0-9]*$/
+          # now we check to see if the graph has the value
+          if @graph.has_bnode_identifier?(value)
+            # if so, pull it in - no need to recreate objects.
+            subject = @graph.get_bnode_by_identifier(value)
+          else
+            # if not, create a new one.
+            subject = BNode.new(value)
+          end
+        end
+      end
+      
+      if uri == "http://www.w3.org/1999/02/22-rdf-syntax-ns#ID"
+        begin
+          # check for base
+          if att.element.base?
+            subject = att.element.base.to_s + value
+          elsif @uri != nil
+            compound = @uri.to_s + "#" + value
+            subject = compound.to_s
+          else
+            raise "Needs to have an ID"
+          end
+#        rescue UriRelativeException
+        end
+      end
+
+      # add other subject detection subroutines here
+    }
+    if subject.class == NilClass
+      subject = BNode.new
+    end
+    return subject
+  end
+  
+  protected
+  def parse_element (element, subject = nil, resource = false)
+    if subject == nil
+      # figure out subject
+      subject = self.get_uri_from_atts(element, true)
+    end
+
+    # type parsing
+    if resource == true
+      type = URIRef.new(element.namespace + element.name)
+      unless type.to_s == "http://www.w3.org/1999/02/22-rdf-syntax-ns#Description"
+        @graph.add_triple(subject, URIRef.new("http://www.w3.org/1999/02/22-rdf-syntax-ns#type"), type)
+      end
+    end
+    
+    # attribute parsing
+    element.attributes.each_attribute { |att|
+      uri = att.namespace + att.name
+      value = att.to_s
+      
+      unless @excl.member? uri
+        @graph.add_triple(subject, URIRef.new(uri), Literal.new(value))
+      end
+    }
+
+    # element parsing
+    element.each_element { |e|
+      self.parse_resource_element e, subject
+    }
+  end
+  
+  def parse_resource_element e, subject
+    uri = e.namespace + e.name
+    if e.attributes.get_attribute_ns("http://www.w3.org/1999/02/22-rdf-syntax-ns#", "parseType").to_s == "Literal"
+      @graph.add_triple(subject, URIRef.new(uri), TypedLiteral.new(e.children.to_s.strip, "http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral"))
+    elsif e.has_elements?
+      # subparsing
+      e.each_element { |se| #se = 'striped element'
+        if e.attributes.get_attribute_ns("http://www.w3.org/1999/02/22-rdf-syntax-ns#", "parseType").to_s == "Resource"
+          object = BNode.new
+        else
+          object = self.get_uri_from_atts(se, true)
+        end
+        @graph.add_triple(subject, URIRef.new(uri), object)
+        self.parse_element(se, object, true)
+      }
+    elsif e.attributes.get_attribute_ns("http://www.w3.org/1999/02/22-rdf-syntax-ns#", "datatype")
+      @graph.add_triple(subject, URIRef.new(uri), TypedLiteral.new(e.text, e.attributes.get_attribute_ns("http://www.w3.org/1999/02/22-rdf-syntax-ns#", "datatype").to_s.strip))
+    elsif e.has_attributes?
+      # get object out
+      object = self.get_uri_from_atts(e)
+      @graph.add_triple(subject, URIRef.new(uri), object)
+    elsif e.has_text?
+      if e.lang?
+        @graph.add_triple(subject, URIRef.new(uri), Literal.new(e.text, e.lang))
+      else
+        @graph.add_triple(subject, URIRef.new(uri), Literal.new(e.text))                    
+      end
+    end
+  end
+  
+end