adding ambit/ntua hacks

3 files changed, 79 insertions, 25 deletions

diff --git a/lib/dataset.rb b/lib/dataset.rb
index c472d84..0baec2f 100644
--- a/lib/dataset.rb
+++ b/lib/dataset.rb
@@ -29,11 +29,20 @@ module OpenTox
 				d.data.each do |compound,features|
 					d.compounds << compound
 					features.each do |f,v|
-						d.features << f
+						d.features << f.keys[0]
 					end
 				end
 				d.compounds.uniq!
 				d.features.uniq!
+        
+        #PENDING: remove debug checks
+        d.data.each do |c,f|
+          f.each do |ff,v|
+            raise "illegal data: feature is no string "+ff.inspect unless ff.is_a?(Hash)
+          end
+        end
+        raise "illedal dataset data\n"+d.data.inspect+"\n" unless d.data.is_a?(Hash) and d.data.values.is_a?(Array)
+        raise "illegal dataset features:\n"+d.features.inspect+"\n" unless d.features.size>0 and d.features[0].is_a?(String)
 		  end
       return d
 		end
diff --git a/lib/model.rb b/lib/model.rb
index 9f03c4b..d8d751b 100644
--- a/lib/model.rb
+++ b/lib/model.rb
@@ -15,13 +15,13 @@ module OpenTox
         @title = owl.title
 				@source = owl.source
 				@identifier = owl.identifier.sub(/^\[/,'').sub(/\]$/,'')
-				@uri = @identifier
+				@uri = owl.uri.to_s #@identifier
 				@algorithm = owl.algorithm
 				@dependent_variables = owl.dependentVariables
 				@independent_variables = owl.independentVariables
 				@predicted_variables = owl.predictedVariables
         
-        raise "invalid model: "+self.to_yaml unless @dependent_variables.to_s.size>0 &&  @independent_variables.to_s.size>0 && @predicted_variables.to_s.size>0
+        raise "invalid model:\n"+self.to_yaml+"\n" unless Utils.is_uri?(@uri) && @dependent_variables.to_s.size>0 &&  @independent_variables.to_s.size>0 && @predicted_variables.to_s.size>0
 			end
 	 end
   
@@ -47,18 +47,37 @@ module OpenTox
            uri += line
          end
        end
-       uri = OpenTox::Task.find(uri).wait_for_resource.to_s if Utils.task_uri?(uri)
-       return uri if Utils.dataset_uri?(uri)
-       raise "not sure about prediction result: "+uri.to_s
+         
+       if uri.to_s =~ /ambit.*task/
+         #HACK handle redirect
+         LOGGER.debug "AMBIT TASK "+uri.to_s
+         redirect = ""
+         while (redirect.size == 0)
+           IO.popen("bin/redirect.sh "+uri.to_s) do |f| 
+             while line = f.gets
+               redirect += line.chomp
+             end
+           end
+           sleep 0.3
+         end
+         LOGGER.debug "REDIRECT to: "+redirect.to_s
+         raise "invalid redirect result" unless redirect =~ /ambit.*dataset/
+         return uri
+       else
+         uri = OpenTox::Task.find(uri).wait_for_resource.to_s if Utils.task_uri?(uri)
+         return uri if Utils.dataset_uri?(uri)
+         raise "not sure about prediction result: "+uri.to_s
+       end
      end
     
      def classification?
        #HACK replace with request to ontology server
-       case @title
-       when /lazar classification/
+       if @title =~ /lazar classification/
          return true
+       elsif @uri =~/ntua/ and @title =~ /mlr/
+         return false
        else
-         raise "unknown model: "+@title.to_s
+         raise "unknown model, uri:"+@uri.to_s+" title:"+@title.to_s
        end
      end
    end
diff --git a/lib/owl.rb b/lib/owl.rb
index 2cd89c4..11da4d0 100644
--- a/lib/owl.rb
+++ b/lib/owl.rb
@@ -15,29 +15,34 @@ module OpenTox
 			owl.model.add owl.uri, RDF['type'], OT[owl.ot_class]
 			owl.model.add owl.uri, DC['identifier'], owl.uri
 			owl
-	 end
+	  end
   
-		def self.from_uri(uri)
-			owl = OpenTox::Owl.new
-			parser = Redland::Parser.new
-			begin
-        data = RestClient.get(uri,:accept => "application/rdf+xml").to_s
-        parser.parse_string_into_model(owl.model, data, uri)
-				#parser.parse_into_model(owl.model,uri)
-		  rescue => e
-        raise "Error parsing #{uri}: "+e.message
-				#raise "Error parsing #{uri}: #{e.message.to_s + e.backtrace.to_s}"
-			end
-			owl.uri =  Redland::Uri.new(uri.chomp)
-			owl
+    def self.from_data(data,uri)
+      owl = OpenTox::Owl.new
+      parser = Redland::Parser.new
+      begin
+         parser.parse_string_into_model(owl.model, data, uri)
+      rescue => e
+         raise "Error parsing #{uri}: "+e.message
+      end
+      owl.uri =  Redland::Uri.new(uri.chomp)
+      owl
+    end
+  
+	  def self.from_uri(uri)
+     return from_data(RestClient.get(uri,:accept => "application/rdf+xml").to_s, uri) 
 		end
 
 		def rdf
 			@model.to_string
 		end
 
+    #def predictedVariables
+       #
+    #end
+
 		def method_missing(name, *args)
-			methods = ['title', 'source', 'identifier', 'algorithm', 'independentVariables', 'dependentVariables', 'predictedVariables', 'date','trainingDataset' ]
+			methods = ['title', 'source', 'identifier', 'algorithm', 'independentVariables', 'dependentVariables', 'predictedVariables', 'date','trainingDataset', 'hasStatus', "percentageCompleted" ]
 			if methods.include? name.to_s.sub(/=/,'')
 				if /=/ =~ name.to_s # setter
 					name = name.to_s.sub(/=/,'')
@@ -48,6 +53,16 @@ module OpenTox
 					end
 					@model.add @uri, DC[name], args.first
 				else # getter
+          #HACK for reading Panteli's models
+          if @uri.to_s =~ /ntua.*model/ and !["title", "source", "identifier"].include?(name.to_s)
+            me = @model.subject(RDF['type'],OT['Model'])
+            #puts "going for "+name.to_s
+            return @model.object(me, OT[name.to_s]).uri.to_s
+          elsif @uri.to_s =~ /ambit.*task/ and ["hasStatus", "percentageCompleted"].include?(name.to_s)
+            me = @model.subject(RDF['type'],OT['Task'])
+            return @model.object(me, OT[name.to_s]).literal.value.to_s
+          end
+          #raise "stop there "+name.to_s
 					@model.object(@uri, DC[name.to_s]).to_s
 				end
 			else
@@ -145,8 +160,19 @@ module OpenTox
 							value = true
 						when FALSE_REGEXP # defined in environment.rb
 							value = false
+					 when /.*\^\^<.*XMLSchema#.*>/
+              #HACK for reading ambit datasets
+              case value.to_s
+              when /XMLSchema#string/
+                value = value.to_s[0..(value.to_s.index("^^")-1)]
+              when /XMLSchema#double/
+                value = value.to_s[0..(value.to_s.index("^^")-1)].to_f
+              else
+                LOGGER.warn " ILLEGAL TYPE "+compound_uri + " has value '" + value.to_s + "' for feature " + feature_uri
+                value = nil
+              end
 						else
-							LOGGER.warn compound_uri + " has value '" + value.to_s + "' for feature " + feature_uri
+	 						LOGGER.warn compound_uri + " has value '" + value.to_s + "' for feature " + feature_uri
 							value = nil
 						end
 						data[compound_uri] = [] unless data[compound_uri]