biointerchange 1.0.1 → 1.0.2

data/lib/biointerchange/textmining/text_mining_rdf_ntriples.rb

@@ -19,8 +19,7 @@ class RDFWriter < BioInterchange::Writer
   #
   # +ostream+:: instance of an IO class or derivative that is used for RDF serialization
   def initialize(ostream)
-    raise BioInterchange::Exceptions::ImplementationWriterError, 'The output stream is not an instance of IO or its subclasses.' unless ostream.kind_of?(IO) || ostream.kind_of?(StringIO)
-    @ostream = ostream
+    super(ostream)
   end
 
   # Serializes a model as RDF.
@@ -103,7 +102,9 @@ private
     graph = RDF::Graph.new
     document_uri = RDF::URI.new(uri_prefix) if uri_prefix
     document_uri = RDF::URI.new(model.uri) unless document_uri
-    graph.insert(RDF::Statement.new(document_uri, RDF.type, BioInterchange::SIO.document))
+    set_base(document_uri)
+    add_prefix('http://semanticscience.org/resource/', 'sio')
+    create_triple(document_uri, RDF.type, BioInterchange::SIO.document)
     model.contents.each { |content|
       if content.kind_of?(BioInterchange::TextMining::Content)
         serialize_content(graph, document_uri, content)
@@ -113,7 +114,7 @@ private
         raise BioInterchange::Exceptions::ImplementationWriterError, "Can only serialize Content and ContentConnection from a Document."
       end
     }
-    RDF::NTriples::Writer.dump(graph, @ostream)
+    close
   end
 
 
@@ -124,7 +125,7 @@ private
   # +content+:: an instance that describes the content
   def serialize_content(graph, document_uri, content)
     content_uri = RDF::URI.new(content.uri)
-    graph.insert(RDF::Statement.new(document_uri, BioInterchange::SIO.has_attribute, content_uri))
+    create_triple(document_uri, BioInterchange::SIO.has_attribute, content_uri)
     serialize_process(graph, document_uri, content_uri, content.process) if content.process
     
     sio_url = BioInterchange::SIO.language_entity
@@ -155,11 +156,10 @@ private
       sio_url = BioInterchange::SIO.character
     end
     
-    graph.insert(RDF::Statement.new(content_uri, RDF.type, sio_url))
+    create_triple(content_uri, RDF.type, sio_url)
     
-    graph.insert(RDF::Statement.new(content_uri, BioInterchange::SIO.has_attribute, serialize_content_start(graph, document_uri, content_uri, content)))
-    graph.insert(RDF::Statement.new(content_uri, BioInterchange::SIO.has_attribute, serialize_content_stop(graph, document_uri, content_uri, content)))
-  
+    create_triple(content_uri, BioInterchange::SIO.has_attribute, serialize_content_start(graph, document_uri, content_uri, content))
+    create_triple(content_uri, BioInterchange::SIO.has_attribute, serialize_content_stop(graph, document_uri, content_uri, content))
   end
 
   # Serializes a ContentConnection object for a given document URI.
@@ -169,31 +169,29 @@ private
   # +content+:: an instance that describes the content
   def serialize_contentconnection(graph, document_uri, contentcon)
     contentcon_uri = RDF::URI.new(contentcon.uri)
-    graph.insert(RDF::Statement.new(document_uri, BioInterchange::SIO.has_attribute, contentcon_uri))
+    create_triple(document_uri, BioInterchange::SIO.has_attribute, contentcon_uri)
     serialize_process(graph, document_uri, contentcon_uri, contentcon.process) if contentcon.process
     
-
     #TODO these sio tags need confirming - there are here as a initial proof of concept
     #next issue, some of these are relations and some are labels, need to separate out which
     #I seem to recall that the only relationship types that should be used are "has_attribute" and "RDF::type", in which case these need adjusting for that.
     #I presume this'd mean making a "has_attribute" link between the content1 and the contentconnection relationship in some way.
     case contentcon.type
     when ContentConnection::UNSPECIFIED
-      graph.insert(RDF::Statement.new(contentcon.content1.uri, BioInterchange::SIO.has_attribute, BioInterchange::SIO.language_entity))
+      create_triple(contentcon.content1.uri, BioInterchange::SIO.has_attribute, BioInterchange::SIO.language_entity)
     when ContentConnection::EQUIVALENCE
-      graph.insert(RDF::Statement.new(contentcon.content1.uri, BioInterchange::SIO.is_equal_to, contentcon.content2.uri))
+      create_triple(contentcon.content1.uri, BioInterchange::SIO.is_equal_to, contentcon.content2.uri)
     when ContentConnection::SUBCLASS
       #TODO this class needs more information, the relationship is between a contentcon.content, and 'something'... I've yet to work out what
-      graph.insert(RDF::Statement.new(contentcon.content2.uri, BioInterchange::SIO.has_attribute, BioInterchange::SIO.in_relation_to))
+      create_triple(contentcon.content2.uri, BioInterchange::SIO.has_attribute, BioInterchange::SIO.in_relation_to)
     when ContentConnection::THEME
       #TODO there are other more specific options for this that need investigating as options. 
-      graph.insert(RDF::Statement.new(contentcon.content1.uri, BioInterchange::SIO.has_target, contentcon.content2.uri))
+      create_triple(contentcon.content1.uri, BioInterchange::SIO.has_target, contentcon.content2.uri)
     when ContentConnection::SPECULATION
-      graph.insert(RDF::Statement.new(contentcon.content1.uri, BioInterchange::SIO.has_attribute, BioInterchange::SIO.speculation))
+      create_triple(contentcon.content1.uri, BioInterchange::SIO.has_attribute, BioInterchange::SIO.speculation)
     when ContentConnection::NEGATION
-      graph.insert(RDF::Statement.new(contentcon.content1.uri, BioInterchange::SIO.denotes, BioInterchange::SIO.negative_regulation))
+      create_triple(contentcon.content1.uri, BioInterchange::SIO.denotes, BioInterchange::SIO.negative_regulation)
     end
-  
   end
 
   # Serializes a process object for a specific document uri
@@ -201,17 +199,17 @@ private
   #
   def serialize_process(graph, document_uri, content_uri, process)
     process_uri = process_uri(process, :process)
-    graph.insert(RDF::Statement.new(content_uri, BioInterchange::SIO.is_direct_part_of, process_uri))
+    create_triple(content_uri, BioInterchange::SIO.is_direct_part_of, process_uri)
     # If this is an email address, then create a FOAF representation, otherwise, do something else:
     if process.type == BioInterchange::TextMining::Process::MANUAL then
-      graph.insert(RDF::Statement.new(process_uri, RDF.type, RDF::FOAF.Person))
-      graph.insert(RDF::Statement.new(process_uri, RDF::FOAF.name, RDF::Literal.new(process.name)))
-      graph.insert(RDF::Statement.new(process_uri, RDF::FOAF.name, RDF::URI.new(process.uri)))
+      create_triple(process_uri, RDF.type, RDF::FOAF.Person)
+      create_triple(process_uri, RDF::FOAF.name, RDF::Literal.new(process.name))
+      create_triple(process_uri, RDF::FOAF.name, RDF::URI.new(process.uri))
     else
-      graph.insert(RDF::Statement.new(process_uri, RDF.type, BioInterchange::SIO.process))
-      graph.insert(RDF::Statement.new(process_uri, BioInterchange::SIO.has_attribute, serialize_process_name(graph, document_uri, content_uri, process_uri, process)))
-      graph.insert(RDF::Statement.new(process_uri, BioInterchange::SIO.has_attribute, serialize_process_uri(graph, document_uri, content_uri, process_uri, process)))
-      graph.insert(RDF::Statement.new(process_uri, BioInterchange::SIO.has_attribute, serialize_process_date(graph, document_uri, content_uri, process_uri, process))) if process.date
+      create_triple(process_uri, RDF.type, BioInterchange::SIO.process)
+      create_triple(process_uri, BioInterchange::SIO.has_attribute, serialize_process_name(graph, document_uri, content_uri, process_uri, process))
+      create_triple(process_uri, BioInterchange::SIO.has_attribute, serialize_process_uri(graph, document_uri, content_uri, process_uri, process))
+      create_triple(process_uri, BioInterchange::SIO.has_attribute, serialize_process_date(graph, document_uri, content_uri, process_uri, process)) if process.date
     end
   end
 
@@ -220,8 +218,8 @@ private
   #
   def serialize_process_name(graph, document_uri, content_uri, process_uri, process)
     kind_uri = process_uri(process, :name)
-    graph.insert(RDF::Statement.new(kind_uri, RDF.type, BioInterchange::SIO.name))
-    graph.insert(RDF::Statement.new(kind_uri, BioInterchange::SIO.has_attribute, RDF::Literal.new("#{process.name}")))
+    create_triple(kind_uri, RDF.type, BioInterchange::SIO.name)
+    create_triple(kind_uri, BioInterchange::SIO.has_attribute, RDF::Literal.new("#{process.name}"))
   end
 
   #
@@ -229,8 +227,8 @@ private
   #
   def serialize_process_uri(graph, document_uri, content_uri, process_uri, process)
     kind_uri = process_uri(process, :uri)
-    graph.insert(RDF::Statement.new(kind_uri, RDF.type, BioInterchange::SIO.software_entity))
-    graph.insert(RDF::Statement.new(kind_uri, BioInterchange::SIO.has_attribute, RDF::URI.new(process.uri)))
+    create_triple(kind_uri, RDF.type, BioInterchange::SIO.software_entity)
+    create_triple(kind_uri, BioInterchange::SIO.has_attribute, RDF::URI.new(process.uri))
   end
 
   #
@@ -238,7 +236,7 @@ private
   #
   def serialize_process_date(graph, document_uri, content_uri, process_uri, process)
     kind_uri = process_uri(process, :date)
-    graph.insert(RDF::Statement.new(kind_uri, RDF::DC.date, RDF::Literal.new(Date.parse(process.date))))
+    create_triple(kind_uri, RDF::DC.date, RDF::Literal.new(Date.parse(process.date)))
   end
 
   #
@@ -246,8 +244,8 @@ private
   #
   def serialize_content_start(graph, document_uri, content_uri, content)
     kind_uri = content_uri(content, :start)
-    graph.insert(RDF::Statement.new(kind_uri, RDF::type, BioInterchange::SIO.start_position))
-    graph.insert(RDF::Statement.new(kind_uri, BioInterchange::SIO.has_attribute, RDF::Literal.new(content.offset)))
+    create_triple(kind_uri, RDF::type, BioInterchange::SIO.start_position)
+    create_triple(kind_uri, BioInterchange::SIO.has_attribute, RDF::Literal.new(content.offset))
   end
   
   #
@@ -255,8 +253,8 @@ private
   #
   def serialize_content_stop(graph, document_uri, content_uri, content)
     kind_uri = content_uri(content, :stop)
-    graph.insert(RDF::Statement.new(kind_uri, RDF::type, BioInterchange::SIO.stop_position))
-    graph.insert(RDF::Statement.new(kind_uri, BioInterchange::SIO.has_attribute, RDF::Literal.new((content.offset+content.length).to_s)))
+    create_triple(kind_uri, RDF::type, BioInterchange::SIO.stop_position)
+    create_triple(kind_uri, BioInterchange::SIO.has_attribute, RDF::Literal.new((content.offset+content.length).to_s))
   end
 
 end

data/lib/biointerchange/writer.rb

@@ -12,6 +12,13 @@ class Writer
   def initialize(ostream)
     raise BioInterchange::Exceptions::ImplementationWriterError, 'The output stream is not an instance of IO or its subclasses.' unless ostream.kind_of?(IO)
     @ostream = ostream
+    @prefixes = {
+      '<http://purl.org/dc/terms/'                   => 'dc:',
+      '<http://www.w3.org/1999/02/22-rdf-syntax-ns#' => 'rdf:',
+      '<http://www.w3.org/2000/01/rdf-schema#'       => 'rdfs:',
+      '<http://www.w3.org/2002/07/owl#'              => 'owl:',
+      '<http://www.w3.org/2001/XMLSchema#'           => 'xsd:'
+    }
   end
 
   # Serializes an object model instance.
@@ -43,7 +50,6 @@ class Writer
       object_representation = "<#{object_uri}>".sub(/\s/, '%20')
     else
       if datatype then
-        # TODO Append type.
         object_representation = "\"#{object.to_s}\"^^<#{datatype.to_s}>"
       else
         object_representation = "\"#{object.to_s}\""
@@ -54,7 +60,7 @@ class Writer
       if BioInterchange::format == :turtle then
         serialize_turtle(subject_uri, predicate_uri, object_representation)
       else
-        @ostream.puts("#{subject_uri} #{predicate_uri} #{object_representation}")
+        @ostream.puts("#{subject_uri} #{predicate_uri} #{object_representation} .")
       end
     rescue Errno::EPIPE
       # Whenever an output pipe disappears, then the user may be happy with what he/she
@@ -62,6 +68,8 @@ class Writer
       # such as "head".
       exit 0
     end
+
+    subject
   end
 
   # Finishes serializing triples.
@@ -73,13 +81,11 @@ class Writer
 
   # Sets the base URI prefix that is output/used when serializing triples in Turtle.
   def set_base(uri_prefix)
-    uri_prefix(nil)
     @prefixes["<#{uri_prefix}"] = '<'
   end
 
   # Adds a URI prefix that should be abbreviated when serializing triples in Turtle.
   def add_prefix(uri_prefix, abbreviation_prefix)
-    uri_prefix(nil)
     @prefixes["<#{uri_prefix}"] = "#{abbreviation_prefix}:"
   end
 
@@ -89,21 +95,10 @@ private
   #
   # +uri+:: URI that should be shortened
   def uri_prefix(uri)
-    @prefixes = {
-      '<http://biohackathon.org/resource/faldo#'     => 'faldo:',
-      '<http://purl.obolibrary.org/obo/'             => 'obo:',
-      '<http://purl.org/dc/terms/'                   => 'dc:',
-      '<http://www.biointerchange.org/gfvo#'         => 'gfvo:',
-      '<http://www.w3.org/1999/02/22-rdf-syntax-ns#' => 'rdf:',
-      '<http://www.w3.org/2000/01/rdf-schema#'       => 'rdfs:',
-      '<http://www.w3.org/2001/XMLSchema#'           => 'xsd:',
-      '<http://www.w3.org/2002/07/owl#'              => 'owl:'
-    } unless @prefixes
-
     return nil unless uri
 
     @prefixes.keys.each { |prefix|
-      return prefix if uri.start_with?(prefix)
+      return prefix if uri.start_with?(prefix) and not "#{prefix}>" == uri
     }
     nil
   end

data/make.sh

@@ -0,0 +1,4 @@
+#!/usr/bin/env bash
+
+bundle exec rake gemspec ; bundle exec gem build biointerchange.gemspec ; gem install --no-rdoc --no-ri biointerchange
+

data/spec/exceptions_spec.rb

@@ -2,13 +2,7 @@
 require 'rubygems'
 require 'rspec'
 
-# Turn off verbose reporting here, since class definitions may be loaded multiple
-# times here. That reports that constants have been already been initialized, which
-# is true, but they are only "re-initialized" with the very same values.
-v, $VERBOSE = $VERBOSE, nil
-load 'lib/biointerchange/core.rb'
-load 'lib/biointerchange/exceptions.rb'
-$VERBOSE = v
+require 'biointerchange'
 
 describe BioInterchange::Exceptions do
   describe 'error and exception creation tests' do

data/spec/gff3_rdfwriter_spec.rb

@@ -2,21 +2,7 @@
 require 'rubygems'
 require 'rspec'
 
-# Turn off verbose reporting here, since class definitions may be loaded multiple
-# times here. That reports that constants have been already been initialized, which
-# is true, but they are only "re-initialized" with the very same values.
-v, $VERBOSE = $VERBOSE, nil
-load 'lib/biointerchange/core.rb'
-load 'lib/biointerchange/gfvo.rb'
-load 'lib/biointerchange/sofa.rb'
-load 'lib/biointerchange/reader.rb'
-load 'lib/biointerchange/model.rb'
-load 'lib/biointerchange/writer.rb'
-load 'lib/biointerchange/genomics/gff3_rdf_ntriples.rb'
-load 'lib/biointerchange/genomics/gff3_feature_set.rb'
-load 'lib/biointerchange/genomics/gff3_feature.rb'
-load 'lib/biointerchange/genomics/gff3_feature_sequence.rb'
-$VERBOSE = v
+require 'biointerchange'
 
 describe BioInterchange::Genomics::RDFWriter do
   describe 'serialization of GFF3 models' do
@@ -73,7 +59,7 @@ describe BioInterchange::Genomics::RDFWriter do
       lines.each { |line|
         feature_no += 1 if line.match(/\sa\s+gfvo:Feature\s+[.,;]$/)
       }
-      lines.count.should be == 64
+      lines.count.should be == 91
       feature_no.should be == 3
     end
   end

data/spec/gvf_rdfwriter_spec.rb

@@ -2,24 +2,7 @@
 require 'rubygems'
 require 'rspec'
 
-# Turn off verbose reporting here, since class definitions may be loaded multiple
-# times here. That reports that constants have been already been initialized, which
-# is true, but they are only "re-initialized" with the very same values.
-v, $VERBOSE = $VERBOSE, nil
-load 'lib/biointerchange/core.rb'
-load 'lib/biointerchange/gfvo.rb'
-load 'lib/biointerchange/sofa.rb'
-load 'lib/biointerchange/reader.rb'
-load 'lib/biointerchange/model.rb'
-load 'lib/biointerchange/writer.rb'
-load 'lib/biointerchange/genomics/gvf_feature_set.rb'
-load 'lib/biointerchange/genomics/gvf_feature.rb'
-# The GVF implementation extends the GFF3 implementation, so load those classes too:
-load 'lib/biointerchange/genomics/gff3_rdf_ntriples.rb'
-load 'lib/biointerchange/genomics/gff3_feature_set.rb'
-load 'lib/biointerchange/genomics/gff3_feature.rb'
-load 'lib/biointerchange/genomics/gff3_feature_sequence.rb'
-$VERBOSE = v
+require 'biointerchange'
 
 describe BioInterchange::Genomics::RDFWriter do
   describe 'serialization of GVF models' do
@@ -73,7 +56,7 @@ describe BioInterchange::Genomics::RDFWriter do
       lines.each { |line|
         feature_no += 1 if line.match(/\sa\s+gfvo:Feature\s+[.,;]$/)
       }
-      lines.count.should be == 64
+      lines.count.should be == 91
       feature_no.should be == 3
     end
   end

data/spec/phylogenetics_spec.rb

@@ -3,19 +3,7 @@ require 'rubygems'
 require 'rspec'
 require 'bio'
 
-# Turn off verbose reporting here, since class definitions may be loaded multiple
-# times here. That reports that constants have been already been initialized, which
-# is true, but they are only "re-initialized" with the very same values.
-v, $VERBOSE = $VERBOSE, nil
-load 'lib/biointerchange/core.rb'
-load 'lib/biointerchange/cdao.rb'
-load 'lib/biointerchange/reader.rb'
-load 'lib/biointerchange/model.rb'
-load 'lib/biointerchange/writer.rb'
-load 'lib/biointerchange/phylogenetics/newick_reader.rb'
-load 'lib/biointerchange/phylogenetics/tree_set.rb'
-load 'lib/biointerchange/phylogenetics/cdao_rdf_ntriples.rb'
-$VERBOSE = v
+require 'biointerchange'
 
 describe BioInterchange::Phylogenetics::NewickReader do
   describe 'deserialization of Newick trees' do

data/spec/text_mining_pdfx_xml_reader_spec.rb

@@ -2,19 +2,7 @@
 require 'rubygems'
 require 'rspec'
 
-# Turn off verbose reporting here, since class definitions may be loaded multiple
-# times here. That reports that constants have been already been initialized, which
-# is true, but they are only "re-initialized" with the very same values.
-v, $VERBOSE = $VERBOSE, nil
-load 'lib/biointerchange/core.rb'
-load 'lib/biointerchange/reader.rb'
-load 'lib/biointerchange/model.rb'
-load 'lib/biointerchange/textmining/text_mining_reader.rb'
-load 'lib/biointerchange/textmining/pdfx_xml_reader.rb'
-load 'lib/biointerchange/textmining/document.rb'
-load 'lib/biointerchange/textmining/content.rb'
-load 'lib/biointerchange/textmining/process.rb'
-$VERBOSE = v
+require 'biointerchange'
 
 describe BioInterchange::TextMining::PDFxXMLReader do
   describe 'deserialization of pdfx text-mining documents' do

data/spec/text_mining_pubannos_json_reader_spec.rb

@@ -2,20 +2,7 @@
 require 'rubygems'
 require 'rspec'
 
-# Turn off verbose reporting here, since class definitions may be loaded multiple
-# times here. That reports that constants have been already been initialized, which
-# is true, but they are only "re-initialized" with the very same values.
-v, $VERBOSE = $VERBOSE, nil
-load 'lib/biointerchange/core.rb'
-load 'lib/biointerchange/reader.rb'
-load 'lib/biointerchange/model.rb'
-load 'lib/biointerchange/textmining/text_mining_reader.rb'
-load 'lib/biointerchange/textmining/pubannos_json_reader.rb'
-load 'lib/biointerchange/textmining/document.rb'
-load 'lib/biointerchange/textmining/content.rb'
-load 'lib/biointerchange/textmining/content_connection.rb'
-load 'lib/biointerchange/textmining/process.rb'
-$VERBOSE = v
+require 'biointerchange'
 
 describe BioInterchange::TextMining::PubAnnosJSONReader do
   describe 'deserialization of pubannos json text-mining documents' do

data/spec/text_mining_rdfwriter_spec.rb

@@ -1,21 +1,9 @@
 
 require 'rubygems'
 require 'rspec'
+require 'tempfile'
 
-# Turn off verbose reporting here, since class definitions may be loaded multiple
-# times here. That reports that constants have been already been initialized, which
-# is true, but they are only "re-initialized" with the very same values.
-v, $VERBOSE = $VERBOSE, nil
-load 'lib/biointerchange/core.rb'
-load 'lib/biointerchange/sio.rb'
-load 'lib/biointerchange/reader.rb'
-load 'lib/biointerchange/model.rb'
-load 'lib/biointerchange/writer.rb'
-load 'lib/biointerchange/textmining/text_mining_rdf_ntriples.rb'
-load 'lib/biointerchange/textmining/document.rb'
-load 'lib/biointerchange/textmining/content.rb'
-load 'lib/biointerchange/textmining/process.rb'
-$VERBOSE = v
+require 'biointerchange'
 
 describe BioInterchange::TextMining::RDFWriter do
   describe 'serialization of text-mining documents' do
@@ -23,7 +11,7 @@ describe BioInterchange::TextMining::RDFWriter do
       istream, ostream = IO.pipe
       BioInterchange::TextMining::RDFWriter.new(ostream).serialize(BioInterchange::TextMining::Document.new('http://example.org'))
       ostream.close
-      istream.read.lines.count.should eq(1)
+      istream.read.lines.count.should eq(8)
     end
 
     it 'document with two entities' do
@@ -61,14 +49,15 @@ describe BioInterchange::TextMining::RDFWriter do
     end
 
     it 'full advanced json document' do
-      ostream = StringIO.new
+      ofile = Tempfile.new('full_advanced_json_document.ttl')
       reader = BioInterchange::TextMining::PubAnnosJSONReader.new("Test", "http://test.com", "2012-12-09", BioInterchange::TextMining::Process::UNSPECIFIED, "0.0")
         
       model = reader.deserialize(File.new('examples/pubannotation.2626671.json'))
       
-      BioInterchange::TextMining::RDFWriter.new(ostream).serialize(model)
-      ostream.close_write
-      ostream.string.lines.count.should > 100
+      BioInterchange::TextMining::RDFWriter.new(File.new(ofile, 'w')).serialize(model)
+      ofile.close
+      IO.readlines(ofile.path).count.should be > 100
+      ofile.unlink
     end
   end
 end