bio-publisci 0.0.3 → 0.0.4

data/lib/bio-publisci/dataset/dataset_for.rb

@@ -3,6 +3,12 @@ module R2RDF
     extend R2RDF::Interactive
 
     def self.for(object, options={}, ask_on_ambiguous=true)
+
+      if options == false || options == true
+        ask_on_ambiguous = options
+        options = {}
+      end
+
       if object.is_a? String
         if File.exist? object
           if File.extname(object).size > 0
@@ -21,7 +27,7 @@ module R2RDF
           end
         else
           raise "Unable to find reader for File or String"
-        end 
+        end
       elsif object.is_a? Rserve::REXP
         r_object(object, options, ask_on_ambiguous)
       else
@@ -60,11 +66,11 @@ module R2RDF
               options[:measures] = selection.split(',').map(&:to_i).map{|i| meas[i]}
             end
           end
-          
+
           df.generate_n3(con.eval(var),var,options)
-        
+
         elsif r_classes.include? "cross"
-          bc = R2RDF::Reader::BigCross.new
+          bc = R2RDF::Reader::RCross.new
 
           unless options[:measures] || !ask_on_ambiguous
             pheno_names = con.eval("names(#{var}$pheno)").to_ruby
@@ -87,7 +93,7 @@ module R2RDF
         elsif r_classes.include? "matrix"
           mat = R2RDF::Reader::RMatrix.new
 
-          unless options[:measures] || !ask_on_ambiguous  
+          unless options[:measures] || !ask_on_ambiguous
             puts "Row label"
             rows = gets.chomp
             rows = "row" unless rows.size > 0
@@ -102,7 +108,7 @@ module R2RDF
 
             options[:measures] = [cols,rows,vals]
           end
-          
+
           base = var
           if ask_on_ambiguous
             puts "Output file base?"
@@ -117,7 +123,7 @@ module R2RDF
 
       elsif object.is_a? Rserve::REXP
         if object.attr.payload["class"].payload.first
-          
+
           df = R2RDF::Reader::Dataframe.new
 
           var = nil
@@ -136,7 +142,7 @@ module R2RDF
             meas = object.payload.names
             options[:measures] = interact("Which measures?",meas,meas)
           end
-          
+
           df.generate_n3(object,var,options)
         else
           raise "support for other Rserve objects coming shortly"

data/lib/bio-publisci/metadata/metadata.rb

@@ -15,8 +15,8 @@ module R2RDF
     }
     end
 
-    def basic(fields, options={} )
-      #TODO don't assume base dataset is "ns:dataset-var", 
+    def basic(fields)
+      #TODO don't assume base dataset is "ns:dataset-var",
       #make it just "var", and try to make that clear to calling classes
 
       fields[:var] = sanitize([fields[:var]]).first
@@ -31,8 +31,6 @@ module R2RDF
 
       fields[:date] = Time.now.strftime("%Y-%m-%d") unless fields[:date]
 
-      options = defaults().merge(options)
-
       #TODO some of these should probably be resources, eg dct:creator, or put under DC namespace
       str = <<-EOF.unindent
       ns:dataset-#{fields[:var]} rdfs:label "#{fields[:title]}";
@@ -64,67 +62,207 @@ module R2RDF
       str + "\n" + end_str
     end
 
-    def provenance(fields, options={})
+    def provenance(original, triplified, chain, options={})
       #TODO: should either add a prefixes method or replace some with full URIs
-      var = sanitize([fields[:var]]).first
-      creator = fields[:creator] if fields[:creator] #should be URI
-      org = fields[:organization] if fields[:organization] #should be URI
-      source_software = fields[:software] # software name, object type, optionally steps list for, eg, R
-      str = "ns:dataset-#{var} a prov:Entity.\n\n"
-      assoc_id = Time.now.nsec.to_s(32)
-      endstr = <<-EOF.unindent      
-        </ns/R2RDF> a prov:Agent .
-        ns:dataset-#{var} prov:wasGeneratredBy ns:activity-0 .
+      raise "MissingOriginal: must specify a provenance source" unless original && original[:resource]
 
-        ns:activity-0 a prov:Activity ;
-          prov:qualifiedAssociation ns:assoc-0_#{assoc_id};
-          prov:generated ns:dataset-#{var} .
+      #TODO include file type etc, or create a separate method for it
 
-        ns:assoc-0_#{assoc_id} a prov:Assocation ;
-          prov:entity </ns/R2RDF>;
-          prov:hadPlan ns:plan-0.
+      str = <<-EOF.unindent
+        <#{original[:resource]}> a prov:Entity ;
+          prov:wasGeneratredBy ns:activity-1 .
 
-        ns:plan-0 a prov:Plan ;
-          rdfs:comment "generation of dataset-#{var} by R2RDF gem".
+        ns:activity-1 a prov:Activity ;
+          prov:generated <#{original[:resource]}> .
 
       EOF
 
-      if creator
-        str << "<#{creator}> a prov:Agent, prov:Person .\n"
-        str << "</ns/R2RDF> prov:actedOnBehalfOf <#{creator}> .\n\n"
+      if original[:software]
+        original_assoc_id = Time.now.nsec.to_s(32)
+
+
+        str << <<-EOF.unindent
+          <#{original[:software]}> a prov:Entity.
+
+          ns:activity-1 prov:qualifiedAssociation ns:assoc-1_#{original_assoc_id} .
+
+          ns:assoc-1_#{original_assoc_id} a prov:Assocation ;
+            prov:entity <#{original[:software]}> .
+
+        EOF
+
+        if original[:process]
+          original[:process] = IO.read(original[:process]) if File.exist? original[:process]
+
+          steps = '"' + original[:process].split("\n").join('" "') + '"'
+          str << <<-EOF.unindent
+            ns:assoc-1_#{original_assoc_id} prov:hadPlan ns:plan-1.
+
+            ns:plan-1 a prov:Plan ;
+              rdfs:comment (#{steps});
+
+          EOF
+        end
+      end
+
+      if original[:author]
+        str << "<#{original[:author]}> a prov:Agent, prov:Person .\n"
+        str << "ns:activity-1 prov:wasAssociatedWith <#{original[:author]}> .\n"
 
-        if org
-          str << "<#{org}> a prov:Agent, prov:Organization .\n"
-          str << "<#{creator}> prov:actedOnBehalfOf <#{org}> .\n"
+        str << "<#{original[:author]}> foaf:givenName \"#{original[:author_name]}\" .\n" if original[:author_name]
+
+        if original[:organization]
+          str << "<#{original[:author]}> prov:actedOnBehalfOf <#{original[:organization]}> .\n\n"
+          str << "<#{original[:organization]}> a prov:Agent, prov:Organization.\n"
+          if original[:organization_name]
+            str << "<#{original[:organization]}> foaf:name \"#{original[:organization_name]}\" .\n\n"
+          else
+            str << "\n"
+          end
+        else
+          str << "\n"
         end
       end
 
-      if source_software
-        source_software = [source_software] unless source_software.is_a? Array
-        source_software.each_with_index.map{|soft,i|
-          str << "</ns/prov/software/#{soft[:name]}> a prov:Agent .\n"
+      if triplified
+        triples_assoc_id = Time.now.nsec.to_s(32)
+
+        str << <<-EOF.unindent
+          <#{triplified[:resource]}> a prov:Entity;
+            prov:wasGeneratredBy ns:activity-0 .
+
+          </ns/R2RDF> a prov:Agent, prov:SoftwareAgent ;
+            rdfs:label "Semantic Publishing Toolkit" .
+
+          ns:activity-0 a prov:Activity ;
+            prov:qualifiedAssociation ns:assoc-0_#{triples_assoc_id};
+            prov:generated <#{triplified[:resource]}> ;
+            prov:used <#{original[:resource]}> .
+
+          ns:assoc-0_#{triples_assoc_id} a prov:Assocation ;
+            prov:entity </ns/R2RDF>;
+            prov:hadPlan ns:plan-0.
+
+          ns:plan-0 a prov:Plan ;
+            rdfs:comment "generation of <#{triplified[:resource]}> by R2RDF gem" .
+
+        EOF
+
+        if triplified[:author]
+          str << "<#{triplified[:author]}> a prov:Agent, prov:Person .\n"
 
-          endstr << "ns:activity-0 prov:used </ns/dataset/#{var}#var> .\n"
-          endstr << "ns:dataset-#{var} prov:wasDerivedFrom </ns/dataset/#{var}#var> .\n\n"
+          str << "<#{triplified[:author]}> foaf:givenName \"#{triplified[:author_name]}\" .\n" if triplified[:author_name]
 
-          if soft[:process]
-            if File.exist? soft[:process]
-              soft[:process] = IO.read(soft[:process])
+          if triplified[:organization]
+            str << "<#{triplified[:author]}> prov:actedOnBehalfOf <#{triplified[:organization]}> .\n\n"
+            str << "<#{triplified[:organization]}> a prov:Agent, prov:Organization.\n"
+            if triplified[:organization_name]
+              str << "<#{triplified[:organization]}> foaf:name \"#{triplified[:organization_name]}\" .\n\n"
+            else
+              str << "\n"
             end
-            endstr << "</ns/dataset/#{var}#var> prov:wasGeneratredBy ns:activity-#{i+1} .\n" 
-            endstr << process(i+1, soft[:process],"/ns/prov/software/#{soft[:name]}", var)
+          else
+            str << "\n"
+          end
+        end
+      end
+
+      if chain
+        str << "ns:activity-1 prov:used <#{chain.first[:resource]}> .\n"
+        str << "<#{original[:resource]}> prov:wasDerivedFrom <#{chain.first[:resource]}> .\n\n"
+        chain.each_with_index{ |src,i|
+          if i == chain.size-1
+            str << activity(src[:resource],nil,src)
+          else
+            str << activity(src[:resource],chain[i+1][:resource],src)
           end
         }
       end
-      str + "\n" + endstr
+
+      str
     end
 
-    def process(id, step_string, software_resource, software_var, options={})      
+    def activity(entity, used, options={})
+      assoc_id = Time.now.nsec.to_s(32)
+      activity_id = Time.now.nsec.to_s(32)
+      plan_id = Time.now.nsec.to_s(32)
+
+      raise "NoEntityGiven: activity generation requires a subject entity" unless entity
+
+      entity_str = <<-EOF.unindent
+        <#{entity}> a prov:Entity ;
+          prov:wasGeneratredBy ns:activity-a_#{activity_id} ;
+      EOF
+
+      activity_str = <<-EOF.unindent
+        ns:activity-a_#{activity_id} a prov:Activity ;
+          prov:generated <#{entity}> ;
+      EOF
+
+      if used
+        entity_str << "\tprov:wasDerivedFrom <#{used}> . \n\n"
+        activity_str << "\tprov:used <#{used}> . \n\n"
+      else
+        entity_str[-2] = ".\n"
+        activity_str[-2] = ".\n"
+      end
+
+      activity_str << <<-EOF.unindent
+        ns:activity-a_#{activity_id} prov:qualifiedAssociation ns:assoc-s_#{assoc_id} .
+
+        ns:assoc-s_#{assoc_id} a prov:Assocation .
+
+      EOF
+
+      if options[:software]
+
+        activity_str << <<-EOF.unindent
+          <#{options[:software]}> a prov:Entity .
+
+          ns:assoc-s_#{assoc_id} prov:agent <#{options[:software]}> .
+        EOF
+
+        if options[:process]
+          options[:process] = IO.read(options[:process]) if File.exist? options[:process]
+
+          steps = '"' + options[:process].split("\n").join('" "') + '"'
+          activity_str << <<-EOF.unindent
+            ns:assoc-s_#{assoc_id} prov:hadPlan ns:plan-p_#{plan_id}.
+
+            ns:plan-p_#{plan_id} a prov:Plan ;
+              rdfs:comment (#{steps});
+          EOF
+        end
+      end
+
+      if options[:author]
+        entity_str << "<#{options[:author]}> a prov:Agent, prov:Person .\n"
+        entity_str << "<#{options[:author]}> foaf:givenName \"#{options[:author_name]}\" .\n" if options[:author_name]
+
+        activity_str << "ns:activity-a_#{activity_id} prov:wasAssociatedWith <#{options[:author]}> .\n"
+        activity_str << "ns:assoc-s_#{assoc_id} prov:agent <#{options[:author]}> .\n"
+
+        if options[:organization]
+          entity_str << "<#{options[:organization]}> a prov:Agent, prov:Organization .\n"
+          activity_str << "<#{options[:author]}> prov:actedOnBehalfOf <#{options[:organization]}> .\n\n"
+          if options[:organization_name]
+            entity_str << "<#{options[:organization]}> foaf:name \"#{options[:organization_name]}\" .\n\n"
+          end
+        else
+          activity_str << "\n"
+          # entity_str << "\n"
+        end
+      end
+
+      entity_str + "\n" + activity_str
+    end
+
+    def process(id, step_string, software_resource, software_var, options={})
       #TODO a better predicate for the steplist than rdfs:comment
       # and make sure it looks good.
       steps = '"' + step_string.split("\n").join('" "') + '"'
       assoc_id = Time.now.nsec.to_s(32)
-      str = <<-EOF.unindent     
+      str = <<-EOF.unindent
         ns:activity-#{id} a prov:Activity ;
           prov:qualifiedAssociation ns:assoc-#{assoc_id} ;
           prov:used </ns/dataset/#{software_var}#var>.

data/lib/bio-publisci/metadata/prov/activity.rb

@@ -0,0 +1,106 @@
+module PubliSci
+module Prov
+  class Activity
+    include Prov::Element
+    class Associations < Array
+      def [](index)
+        if self.fetch(index).is_a? Symbol
+            Prov.agents[self.fetch(index)]
+        else
+          self.fetch(index)
+        end
+      end
+    end
+
+    def generated(entity=nil)
+      if entity
+        # if entity.is_a? Symbol
+        #   entity = Prov.entities[entity.to_sym]
+        # end
+
+        if entity.is_a? Entity
+          entity.generated_by self
+        end
+
+        (@generated ||= []) << entity
+      elsif @generated.is_a? Symbol
+        @generated = Prov.entities[@generated]
+      else
+        @generated
+      end
+    end
+
+    def generated=(gen)
+      @generated = gen
+    end
+
+    def associated_with(agent=nil, &block)
+      if agent
+        (@associated ||= Associations.new) << agent
+        # Prov.register(nil,assoc)
+      elsif block_given?
+        assoc = Association.new
+        assoc.instance_eval(&block)
+        (@associated ||= Associations.new) << assoc
+        Prov.register(nil,assoc)
+      else
+        @associated
+      end
+    end
+
+    def used(entity=nil)
+      if entity
+        (@used ||= []) << entity
+      elsif @used
+        @used.map{|u| u.is_a?(Symbol) ? Prov.entities[u] : u}
+      else
+        @used
+      end
+    end
+
+    def to_n3
+      str = "<#{subject}> a prov:Activity ;\n"
+
+      if generated
+        str << "\tprov:generated "
+        generated.map{|src|
+          src = Prov.entities[src] if src.is_a?(Symbol) && Prov.entities[src]
+          str << "<#{src}>, "
+        }
+        str[-2]=" "
+        str[-1]=";\n"
+      end
+
+      if used
+        str << "\tprov:used "
+        used.map{|used|
+          str << "<#{used}>, "
+        }
+        str[-2]=";"
+        str[-1]="\n"
+      end
+
+      if associated_with
+        associated_with.map{|assoc|
+          assoc = Prov.agents[assoc] if assoc.is_a?(Symbol) && Prov.agents[assoc]
+
+          if assoc.is_a? Association
+            str << "\tprov:wasAssociatedWith <#{assoc.agent}> ;\n"
+            str << "\tprov:qualifiedAssociation <#{assoc}> ;\n"
+          else
+            str << "\tprov:wasAssociatedWith <#{assoc}> ;\n"
+          end
+        }
+      end
+
+      add_custom(str)
+
+      str << "\trdfs:label \"#{__label}\" .\n\n"
+    end
+
+    def to_s
+      subject
+    end
+  end
+end
+end

data/lib/bio-publisci/metadata/prov/agent.rb

@@ -0,0 +1,94 @@
+module PubliSci
+module Prov
+  class Agent
+    include Prov::Element
+
+    def type(t=nil)
+      if t
+        @type = t.to_sym
+      else
+        @type
+      end
+    end
+
+    def type=(t)
+      @type = t.to_sym
+    end
+
+    def name(name=nil)
+      if name
+        @name = name
+      else
+        @name
+      end
+    end
+
+    def name=(name)
+      @name = name
+    end
+
+    def organization(organization=nil)
+      if organization
+        @organization = organization
+      elsif @organization.is_a? Symbol
+        @organization = Prov.agents[@organization]
+      else
+        @organization
+      end
+    end
+
+    def organization=(organization)
+      @organization = organization
+    end
+
+    def on_behalf_of(agent=nil)
+      if agent
+        if agent.is_a? Symbol
+          raise "UnknownAgent: #{agent}" unless Prov.agents.has_key?(agent)
+          agent = Prov.agents[agent]
+        end
+        @on_behalf_of = agent
+      else
+        @on_behalf_of
+      end
+    end
+    alias_method :worked_for, :on_behalf_of
+
+    def to_n3
+      str = "<#{subject}> a prov:Agent"
+      if type
+        case type.to_sym
+        when :software
+          str << ", prov:SoftwareAgent ;\n"
+        when :person
+          str << ", prov:Person ;\n"
+        when :organization
+          str << ", prov:Organization ;\n"
+        end
+      else
+        str << " ;\n"
+      end
+
+      if name
+        if type && type.to_sym == :person
+          str << "\tfoaf:givenName \"#{name}\" ;\n"
+        else
+          str << "\tfoaf:name \"#{name}\" ;\n"
+        end
+      end
+
+      if on_behalf_of
+        str << "\tprov:actedOnBehalfOf <#{on_behalf_of}> ;\n"
+      end
+
+      add_custom(str)
+
+      str << "\trdfs:label \"#{__label}\" .\n\n"
+    end
+
+    def to_s
+      subject
+    end
+  end
+end
+end

data/lib/bio-publisci/metadata/prov/association.rb

@@ -0,0 +1,73 @@
+module PubliSci
+module Prov
+  class Association
+    def subject(sub=nil)
+      if sub
+        @subject = sub
+      else
+        @subject ||= "#{Prov.base_url}/assoc/#{Time.now.nsec.to_s(32)}"
+      end
+    end
+
+    def agent(agent=nil)
+      if agent
+        # agent = Prov.agents[agent.to_sym] if agent.is_a?(String) || agent.is_a?(Symbol)
+        # raise "UnkownAgent #{ag}" unless agent
+        # puts "Warning: overwriting agent #{@agent.subject}" if @agent
+        @agent = agent
+      elsif @agent.is_a? Symbol
+        @agent = Prov.agents[@agent]
+      else
+        @agent
+      end
+    end
+
+    def had_plan(*args, &block)
+      if block_given?
+        p = Prov::Plan.new
+        p.instance_eval(&block)
+        p.__label=args[0]
+        @plan = p
+        Prov.register(args[0], p)
+      elsif args.size == 0
+        if @plan.is_a? Symbol
+          @plan = Prov.plans[@plan]
+        end
+        @plan
+      elsif args.size == 1
+        if args[0].is_a? Symbol
+          raise "UnknownPlan: #{args[0]}" unless Prov.plans[args[0]]
+          @plan = Prov.plans[args[0]]
+        else
+          @plan = args[0]
+        end
+      else
+        name = args.shift
+        args = Hash[*args]
+        p = Prov::Plan.new
+
+        p.__label=name
+        p.subject args[:subject]
+        (args.keys - [:subject]).map{|k|
+          raise "Unkown plan setting #{k}" unless try_auto_set(p,k,args[k])
+        }
+        @plan = p
+        Prov.register(name, p)
+      end
+    end
+    alias_method :plan, :had_plan
+
+    def to_n3
+      str = "<#{subject}> a prov:Association ;\n"
+      str << "\tprov:agent <#{agent}> ;\n"
+      str << "\tprov:hadPlan <#{plan}> ;\n" if plan
+      str[-2] = ".\n"
+      str
+    end
+
+    def to_s
+      subject
+    end
+  end
+end
+end

data/lib/bio-publisci/metadata/prov/derivation.rb

@@ -0,0 +1,53 @@
+module PubliSci
+  module Prov
+    class Derivation
+
+      include PubliSci::CustomPredicate
+
+      def subject(sub=nil)
+        if sub
+          @subject = sub
+        else
+          @subject ||= "#{Prov.base_url}/derivation/#{Time.now.nsec.to_s(32)}"
+        end
+      end
+
+      def had_activity(activity=nil)
+        if activity
+          @had_activity = activity
+        elsif @had_activity.is_a? Symbol
+          @had_activity = Prov.activities[@had_activity]
+        else
+          @had_activity
+        end
+      end
+      alias_method :activity, :had_activity
+
+      def entity(entity=nil)
+        if entity
+          @entity = entity
+        elsif @entity.is_a? Symbol
+          @entity = Prov.entities[@entity]
+        else
+          @entity
+        end
+      end
+      alias_method :data, :entity
+
+      def to_n3
+        str = "<#{subject}> a prov:Derivation ;\n"
+        str << "\tprov:entity <#{entity}> ;\n" if entity
+        str << "\tprov:hadActivity <#{had_activity}> ;\n" if had_activity
+
+        add_custom(str)
+
+        str[-2] = ".\n"
+        str
+      end
+
+      def to_s
+        subject
+      end
+    end
+  end
+end