Class: RDFObject::NTriplesParser

Inherits:

Parser

• Object
• Parser
• RDFObject::NTriplesParser

Defined in:

lib/rdf_objects/parsers.rb

Instance Attribute Summary

Attributes inherited from Parser

#base_uri, #collection

Instance Method Summary

• #data=(ntriples) ⇒ Object
• #parse ⇒ Object
• #parse_ntriple(ntriple) ⇒ Object

Methods inherited from Parser

init_parser, #initialize, parse, #sanitize_uri

Constructor Details

This class inherits a constructor from RDFObject::Parser

Instance Method Details

#data=(ntriples) ⇒ Object

# File 'lib/rdf_objects/parsers.rb', line 222

def data=(ntriples)
  if ntriples.is_a?(String)
    @ntriples = ntriples.split("\n")
  elsif ntriples.is_a?(Array)
    @ntriples = ntriples
  elsif ntriples.respond_to?(:read)
    @ntriples = ntriples.readlines
  end      
end

#parse ⇒ Object

# File 'lib/rdf_objects/parsers.rb', line 232

def parse
  @ntriples.each do | assertion |
    next if assertion[0, 1] == "#" # Ignore comments
    triple = parse_ntriple(assertion)
    resource = @collection.find_or_create(triple[:subject])
    object = case triple[:type]
    when "literal" then triple[:object]
    when "uri" then @collection.find_or_create(triple[:object])
    when "bnode" then @collection.find_or_create(triple[:object])
    end
    resource.assert(triple[:predicate],object)
  end
  @collection
end

#parse_ntriple(ntriple) ⇒ Object

# File 'lib/rdf_objects/parsers.rb', line 167

def parse_ntriple(ntriple)
  if ntriple.respond_to?(:force_encoding)
    ntriple.force_encoding("ASCII-8BIT")
  end      
  scanner = StringScanner.new(ntriple)
  if ntriple[0,1] == "<"
    subject = scanner.scan_until(/> /)
    subject.sub!(/^</,'')
    subject.sub!(/> $/,'')
  else
    subject = scanner.scan_until(/\w /)
    subject.strip!
  end
  predicate = scanner.scan_until(/> /)
  predicate.sub!(/^</,'')
  predicate.sub!(/> $/,'')
  if scanner.match?(/</)
    tmp_object = scanner.scan_until(/>\s?\.\s*\n?$/)
    tmp_object.sub!(/^</,'')
    tmp_object.sub!(/>\s?\.\s*\n?$/,'')
    object = self.sanitize_uri(tmp_object)
    type = "uri"
  elsif scanner.match?(/_:/)
    object = scanner.scan_until(/\w\s?\.\s*\n?$/)
    object.sub!(/\s?\.\s*\n?$/,'')
    type = "bnode"
  else
    language = nil
    data_type = nil
    scanner.getch
    tmp_object = scanner.scan_until(/("\s?\.\s*\n?$)|("@[A-z])|("\^\^)/)
    scanner.pos=(scanner.pos-2)
    tmp_object.sub!(/"..?$/,'')
    if tmp_object.respond_to?(:force_encoding)
      tmp_object.force_encoding('utf-8').chomp!
    else
      uscan = UTF8Parser.new(tmp_object)
      tmp_object = uscan.parse_string.chomp
    end
    if scanner.match?(/@/)
      scanner.getch
      language = scanner.scan_until(/\s?\.\n?$/)
      language.sub!(/\s?\.\n?$/,'')
      language = language.to_sym
    elsif scanner.match?(/\^\^/)
      scanner.skip_until(/</)
      data_type = scanner.scan_until(/>/)
      data_type.sub!(/>$/,'')
    end
    object = RDF::Literal.new(tmp_object,{:datatype=>data_type,:language=>language})
    type = "literal"      
  end
  {:subject=>subject, :predicate=>predicate, :object=>object, :type=>type}
end