tao_rdfizer 0.9.9 → 0.10

checksums.yaml

@@ -1,7 +1,7 @@
 ---
-SHA1:
-  metadata.gz: 0254770e5fbff5ef2071765b19a324af5ad1a3cd
-  data.tar.gz: 639d52929cfe8123766b540cc4f2a92ccf616f36
+SHA256:
+  metadata.gz: a2f897b03c9e6856aaf752fd645808c47e6d52422ea1b6934b44b694d8d0d9a1
+  data.tar.gz: 8d9fb426258bf3fceabb6471a81fcaeb99cd18dcace251575a0f2f715fa4d787
 SHA512:
-  metadata.gz: c9feeec71d327a4781036d16d5d2773f1fd77678ef23ae2f01b627bdb41f2ab04bf401c57204fdd96c27d237c65c8a9ec9da3060fd2206c4d43d7ca1cad4da50
-  data.tar.gz: 67cd10929f38deb58b0e988547f7bde80fb8dd88e3c771fc034274ef374233bfbe13a5d72c8fd26127a1e8cf5c18c34d158493e11c16d718310b5ca40f37651b
+  metadata.gz: f0d230f25945bf45aed05f8f0cb3d46113d5eb4213b17ce4812cbfbcfa5636d540343b8ee8fe94cf1bbbc1a1eb4db35ac076bf1d7cd3cb7abdd92db446c94f41
+  data.tar.gz: 6f00c1aeb735b8e76da2e9185ad141c0c29c1d45d94faad1b3d89ec903a1c931e22957f4f356dce24ff6beadd53a77ac7fcd5aad2c17cb281a5d0128cd062b89

data/lib/tao_rdfizer/tao_rdfizer.rb

@@ -4,11 +4,11 @@ require 'erb'
 module TAO; end unless defined? TAO
 
 class TAO::RDFizer
-  # if mode == :spans then produces span descriptions
-  # if mode == :annotations then produces annotation descriptions
-  # if mode == nil then produces both
-  def initialize(mode = nil)
-  	@mode = mode
+	# if mode == :spans then produces span descriptions
+	# if mode == :annotations then produces annotation descriptions
+	# if mode == nil then produces both
+	def initialize(mode = nil)
+		@mode = mode
 		template = if !mode.nil? && mode == :spans
 			ERB_SPANS_TTL
 		else
@@ -17,7 +17,7 @@ class TAO::RDFizer
 
 		@tao_ttl_erb = ERB.new(template, nil, '-')
 		@prefix_ttl_erb = ERB.new(ERB_PREFIXES_TTL, nil, '-')
-  end
+	end
 
 	def rdfize(annotations_col)
 		# namespaces
@@ -27,6 +27,7 @@ class TAO::RDFizer
 		anns[:namespaces].each {|n| namespaces[n[:prefix]] = n[:uri]} unless anns[:namespaces].nil?
 
 		unless @mode ==:spans
+			raise ArgumentError, "A project name has to be specified." unless anns.has_key?(:project)
 			prefix_for_this = anns[:project].downcase.gsub(/ /, '_')
 			raise ArgumentError, "'#{prefix_for_this}' is a reserved prefix for this project." if namespaces.has_key?(prefix_for_this)
 			project_uri = 'http://pubannotation.org/projects/' + anns[:project]
@@ -34,6 +35,7 @@ class TAO::RDFizer
 		end
 
 		denotations = []
+		attributes = []
 		relations = []
 		spans = []
 
@@ -46,30 +48,57 @@ 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]
+					_attributes += track[:attributes]
 					_relations += track[:relations]
 				end
 			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)
-			end
+			begin
+				if @mode == :annotations
+					# 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
 
-			# relations preprocessing
-			_relations.each do |r|
-				r[:subj_uri] = "#{prefix_for_this}:#{text_id}-#{r[:subj]}"
-				r[:obj_uri] = "#{prefix_for_this}:#{text_id}-#{r[:obj]}"
-				r[:pred_uri] = find_uri(r[:pred], namespaces, prefix_for_this)
+					# 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]}"
+						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
+
+					# relations preprocessing
+					_relations.each do |r|
+						r[:subj_uri] = "#{prefix_for_this}:#{text_id}-#{r[:subj]}"
+						r[:obj_uri] = "#{prefix_for_this}:#{text_id}-#{r[:obj]}"
+						r[:pred_uri] = find_uri(r[:pred], namespaces, prefix_for_this)
+					rescue ArgumentError => e
+						raise ArgumentError, "[#{sourcedb}-#{sourceid}-#{r[:id]}] " + e.message
+					end
+				else
+					# denotations preprocessing
+					_denotations.each do |d|
+						span_uri = "<#{text_uri}/spans/#{d[:span][:begin]}-#{d[:span][:end]}>"
+						d[:span_uri] = span_uri
+					rescue ArgumentError => e
+						raise ArgumentError, "[#{sourcedb}-#{sourceid}-#{d[:id]}] " + e.message
+					end
+				end
 			end
 
 			unless @mode == :annotations
@@ -84,7 +113,7 @@ 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
@@ -158,25 +187,26 @@ 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/)
 		delimiter_position = label.index(':')
 		if !delimiter_position.nil? && namespaces.keys.include?(label[0...delimiter_position])
-			label
+			label.gsub('(', '\(').gsub(')', '\)') # brackets have to be escaped
 		elsif label =~ %r[^https?://]
 			"<#{label}>"
 		else
 			clabel = if label.match(/^\W+$/)
 				'SYM'
 			else
-				label.sub(/^\W+/, '').sub(/\W+$/, '').gsub(/ +/, '_')
+				label.sub(/^\W+/, '').sub(/[^a-zA-Z0-9_)]+$/, '').gsub(/ +/, '_')
 			end
-			namespaces.has_key?('_base') ? "<#{clabel}>" : "#{prefix_for_this}:#{clabel}"
+			namespaces.has_key?('_base') ? "<#{clabel}>" : "#{prefix_for_this}:#{clabel.gsub('(', '\(').gsub(')', '\)')}"
 		end
 	end
 
 	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| -%>
@@ -193,7 +223,7 @@ class TAO::RDFizer
 		<% s[:children].each do |s| -%>
 			tao:contains <%= s[:span_uri] %> ;
 		<% end -%>
-			tao:has_text <%= s[:text].dump %> ;
+			tao:has_text <%= s[:text].inspect %> ;
 			tao:belongs_to <<%= s[:source_uri] %>> ;
 			tao:begins_at <%= s[:begin] %> ;
 			tao:ends_at <%= s[:end] %> .

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: tao_rdfizer
 version: !ruby/object:Gem::Version
-  version: 0.9.9
+  version: '0.10'
 platform: ruby
 authors:
 - Jin-Dong Kim
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2018-01-24 00:00:00.000000000 Z
+date: 2020-12-20 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: 2.6.11
+rubygems_version: 3.0.8
 signing_key: 
 specification_version: 4
 summary: A RDF statement generator for annotations in the PubAnnotation JSON format.