tao_rdfizer 0.9.13 → 0.11.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: '00759ca80638a1d6f0d730af7c919ccf51c1828a614158987ffe4f5d99ff9b9a'
-  data.tar.gz: bfb0c62f1e6e39ec615323d67edc36b16fa41ce0d7cb440e25a7f3d10013573b
+  metadata.gz: 95aaa9006671c09cfcf08d12876932c5783f3d784ad3855b4a3819ad79febf87
+  data.tar.gz: 6f3970db2a93208bd760c68651bc33a212d7e384396314deaebe1fdc744b20b0
 SHA512:
-  metadata.gz: fe5fe5d6b21b68fe9a6e565641f8a9f1ed0c3ae10f714995a4a1f5282103246556b02570f5ac18d53d38d41f6dcbb04cb3c690382b2427f9b9e1d78f5055aacf
-  data.tar.gz: 729279be8a26a908095195c90fb76900adc7f32b2cbbbd01ff7ef6e5115ec10c05c53fb8cd809dc7d3900d989c342301ddbdfea70eee6468fd8970f2aa69deca
+  metadata.gz: 6ffe933f1e7ecfc91b7e51d010fc84d42c5afabfc0cac59df717f264a45d337a4e67d5a2e56082cc00a67b9b20594156fd60fe51e700a4a78cef8c3d2c50fe20
+  data.tar.gz: efc71f7e1765dff987554031bbf4edcea8881cabf232ac5504b67fb33d4ac5b514b96f02410b2293b3a37b443508a4167786bd67cf696543802dcba150146299

data/bin/tao_rdfizer

@@ -3,6 +3,7 @@ require 'tao_rdfizer'
 require 'json'
 
 mode = nil
+options = {}
 
 ## command line option processing
 require 'optparse'
@@ -17,6 +18,14 @@ optparse = OptionParser.new do |opts|
     mode = :spans
   end
 
+  opts.on('-x', '--x-prefixes', 'without prefixes.') do
+    options[:with_prefixes] = false
+  end
+
+  opts.on('-o', '--only-prefixes', 'only prefixes.') do
+    options[:only_prefixes] = true
+  end
+
   opts.on('-h', '--help', 'displays this screen.') do
     puts opts
     exit
@@ -34,7 +43,7 @@ begin
 	annotations = JSON.parse File.read(ARGV[0]), :symbolize_names => true
 	annotations = [annotations] unless annotations.class == Array
 	rdfizer = TAO::RDFizer.new(mode)
-	puts rdfizer.rdfize(annotations)
+	puts rdfizer.rdfize(annotations, options)
 rescue ArgumentError, IOError => e
 	puts e.message
 end

data/lib/tao_rdfizer/tao_rdfizer.rb

@@ -14,18 +14,45 @@ class TAO::RDFizer
 		else
 			ERB_ANNOTATIONS_TTL
 		end
-
 		@tao_ttl_erb = ERB.new(template, nil, '-')
 		@prefix_ttl_erb = ERB.new(ERB_PREFIXES_TTL, nil, '-')
 	end
 
-	def rdfize(annotations_col)
+	def rdfize(annotations_col, options = nil)
+		options ||= {}
+		only_prefixes = options.has_key?(:only_prefixes) ? options[:only_prefixes] == true : false
+		with_prefixes = options.has_key?(:with_prefixes) ? options[:with_prefixes] == true : true
+
+		# check the format
+		annotations_col.each do |annotations|
+			raise "'target' is missing" unless annotations.has_key? :target
+		end
+
 		# namespaces
 		namespaces = {}
 
 		anns = annotations_col.first
 		anns[:namespaces].each {|n| namespaces[n[:prefix]] = n[:uri]} unless anns[:namespaces].nil?
 
+		prefixes_ttl = @prefix_ttl_erb.result_with_hash(namespaces:namespaces) if only_prefixes || with_prefixes
+
+		if only_prefixes
+			prefixes_ttl
+		else
+			annotations_ttl = get_annotations_ttl(annotations_col, namespaces)
+			if with_prefixes
+				prefixes_ttl + annotations_ttl
+			else
+				annotations_ttl
+			end
+		end
+	end
+
+	private
+
+	def get_annotations_ttl(annotations_col, namespaces)
+		anns = annotations_col.first
+
 		unless @mode ==:spans
 			raise ArgumentError, "A project name has to be specified." unless anns.has_key?(:project)
 			prefix_for_this = anns[:project].downcase.gsub(/ /, '_')
@@ -35,6 +62,7 @@ class TAO::RDFizer
 		end
 
 		denotations = []
+		attributes = []
 		relations = []
 		spans = []
 
@@ -47,25 +75,36 @@ class TAO::RDFizer
 			end
 
 			# denotations and relations
-			_denotations = annotations[:denotations]
-			_relations = annotations[:relations]
-			_denotations = [] if _denotations.nil?
-			_relations = [] if _relations.nil?
+			_denotations = annotations[:denotations] || []
+			_attributes = annotations[:attributes] || []
+			_relations = annotations[:relations] || []
 			if @mode == :spans && annotations.has_key?(:tracks)
 				annotations[:tracks].each do |track|
-					_denotations += track[:denotations]
-					_relations += track[:relations]
+					_denotations += track[:denotations] if track.has_key? :denotations
+					_attributes += track[:attributes] if track.has_key? :attributes
+					_relations += track[:relations] if track.has_key? :relations
 				end
 			end
 
 			begin
-				if @mode == :annotations
+				unless @mode == :span
+					# index attributes
+					attributesh = _attributes.inject({}) do |h, a|
+						if a[:pred].end_with?('_id')
+							subj = a[:subj]
+							h[subj] = [] unless h.has_key? subj
+							h[subj] << a[:obj]
+						end
+						h
+					end
+
 					# denotations preprocessing
 					_denotations.each do |d|
 						span_uri = "<#{text_uri}/spans/#{d[:span][:begin]}-#{d[:span][:end]}>"
 						d[:span_uri] = span_uri
 						d[:obj_uri] = "#{prefix_for_this}:#{text_id}-#{d[:id]}"
-						d[:cls_uri] = find_uri(d[:obj], namespaces, prefix_for_this)
+						class_uris = (attributesh[d[:id]] || []).push(d[:obj])
+						d[:class_uris] = class_uris.map{|uri| find_uri(uri, namespaces, prefix_for_this)}
 					rescue ArgumentError => e
 						raise ArgumentError, "[#{sourcedb}-#{sourceid}-#{d[:id]}] " + e.message
 					end
@@ -101,7 +140,7 @@ class TAO::RDFizer
 					s[:text] = text[s[:begin] ... s[:end]]
 				end
 
-				# index
+				# index spans
 				spanh = _spans.inject({}){|r, s| r[s[:span_uri]] = s; r}
 
 				# add denotation information
@@ -155,11 +194,9 @@ class TAO::RDFizer
 			spans += _spans unless @mode == :annotations
 		end
 
-		ttl = @prefix_ttl_erb.result(binding) + @tao_ttl_erb.result(binding)
+		@tao_ttl_erb.result(binding)
 	end
 
-	private
-
 	def include_parent?(spans, span)
 		# spans.each{|s| return true if (s[:begin] <= span[:begin] && s[:end] > span[:end]) || (s[:begin] < span[:begin] && s[:end] >= span[:end])}
 		spans.each{|s| return true if s[:begin] <= span[:begin] && s[:end] >= span[:end]}
@@ -175,7 +212,10 @@ class TAO::RDFizer
 	end
 
 	def find_uri (label, namespaces, prefix_for_this)
-		raise ArgumentError, "A label including a whitespace character found: #{label}." if label.match(/\s/)
+		if label.match(/\s/)
+			# raise ArgumentError, "A label including a whitespace character found: #{label}."
+			label.gsub(/\s/, '_')
+		end
 		delimiter_position = label.index(':')
 		if !delimiter_position.nil? && namespaces.keys.include?(label[0...delimiter_position])
 			label.gsub('(', '\(').gsub(')', '\)') # brackets have to be escaped
@@ -191,10 +231,11 @@ class TAO::RDFizer
 		end
 	end
 
+	# variable: denotations, relations
 	ERB_ANNOTATIONS_TTL = <<~HEREDOC
 		<% denotations.each do |d| -%>
 		<%= d[:obj_uri] %> tao:denoted_by <%= d[:span_uri] %> ;
-			rdf:type <%= d[:cls_uri] %> .
+		<%= d[:class_uris].map{|c| "\trdf:type " + c}.join(" ;\n") + " ." %>
 		<% end -%>
 		<%# relations -%>
 		<% relations.each do |r| -%>
@@ -202,6 +243,7 @@ class TAO::RDFizer
 		<% end -%>
 	HEREDOC
 
+	# variable: spans
 	ERB_SPANS_TTL = <<~HEREDOC
 		<% spans.each do |s| -%>
 		<%= s[:span_uri] %> rdf:type tao:Text_span ;
@@ -218,6 +260,7 @@ class TAO::RDFizer
 		<% end -%>
 	HEREDOC
 
+	# variable: namespaces
 	ERB_PREFIXES_TTL = <<~HEREDOC
 		@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
 		@prefix tao: <http://pubannotation.org/ontology/tao.owl#> .

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: tao_rdfizer
 version: !ruby/object:Gem::Version
-  version: 0.9.13
+  version: 0.11.1
 platform: ruby
 authors:
 - Jin-Dong Kim
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2020-04-06 00:00:00.000000000 Z
+date: 2021-05-23 00:00:00.000000000 Z
 dependencies: []
 description: It uses TAO (text annotation ontology) for representation of annotations
   to text.
@@ -40,8 +40,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubyforge_project: 
-rubygems_version: 2.7.9
+rubygems_version: 3.0.9
 signing_key: 
 specification_version: 4
 summary: A RDF statement generator for annotations in the PubAnnotation JSON format.