wrapper adjustments for validation

8 files changed, 195 insertions, 99 deletions

diff --git a/lib/dataset.rb b/lib/dataset.rb
index f416921..c472d84 100644
--- a/lib/dataset.rb
+++ b/lib/dataset.rb
@@ -1,7 +1,7 @@
 LOGGER.progname = File.expand_path(__FILE__)
 
 module OpenTox
-
+  
 	class Dataset 
 
 		attr_accessor :uri, :title, :source, :identifier, :data, :features, :compounds
@@ -13,26 +13,96 @@ module OpenTox
 		end
 
 		def self.find(uri)
-			if uri.match(/webservices.in-silico.ch|localhost/) # try to get YAML first
-				YAML.load RestClient.get(uri, :accept => 'application/x-yaml').to_s 
+    
+			if uri.match(/webservices.in-silico.ch|localhost|ot.dataset.de/) # try to get YAML first
+				d = YAML.load RestClient.get(uri, :accept => 'application/x-yaml').to_s 
 			else # get default rdf+xml
 				owl = OpenTox::Owl.from_uri(uri)
-				@title = owl.title
-				@source = owl.source
-				@identifier = owl.identifier.sub(/^\[/,'').sub(/\]$/,'')
-				@uri = @identifier
-				@data = owl.data
-				halt 404, "Dataset #{uri} empty!" if @data.empty?
-				@data.each do |compound,features|
-					@compounds << compound
+        
+        d = Dataset.new
+				d.title = owl.title
+				d.source = owl.source
+				d.identifier = owl.identifier.sub(/^\[/,'').sub(/\]$/,'')
+				d.uri = d.identifier
+				d.data = owl.data
+				halt 404, "Dataset #{uri} empty!" if d.data.empty?
+				d.data.each do |compound,features|
+					d.compounds << compound
 					features.each do |f,v|
-						@features << f
+						d.features << f
 					end
 				end
-				@compounds.uniq!
-				@features.uniq!
-			end
+				d.compounds.uniq!
+				d.features.uniq!
+		  end
+      return d
 		end
+    
+    # creates a new dataset, using only those compounsd specified in new_compounds
+    # returns uri of new dataset
+    def create_new_dataset( new_compounds, new_title, new_source )
+      
+      dataset = OpenTox::Dataset.new
+      dataset.title = new_title
+      dataset.source = new_source
+      dataset.features = @features
+      dataset.compounds = new_compounds
+      new_compounds.each do |c|
+        dataset.data[c] = @data[c] 
+      end
+      return dataset.save
+    end
+    
+    # returns classification value
+    def get_predicted_class(compound, feature)
+      v = get_value(compound, feature)
+      if v.is_a?(Hash)
+        if v.has_key?(:classification)
+          return v[:classification]
+        else
+          return "no classification key"
+        end
+      else
+        raise "invalid value type"
+      end
+      
+    end
+    
+    # returns prediction confidence if available
+    def get_prediction_confidence(compound, feature)
+      v = get_value(compound, feature)
+      if v.is_a?(Hash)
+        if v.has_key?(:confidence)
+          return v[:confidence].abs
+        else
+          # PENDING: return nil isntead of raising an exception
+          raise "no confidence key"
+        end
+      else
+        raise "invalid value type"
+      end
+    end
+    
+    # return compound-feature value
+    def get_value(compound, feature)
+      v = @data[compound]
+      raise "no values for compound "+compound.to_s if v==nil
+      if v.is_a?(Array)
+        # PENDING: why using an array here?
+        v.each do |e|
+          if e.is_a?(Hash)
+            if e.has_key?(feature)
+              return e[feature]
+            end
+          else
+            raise "invalid internal value type"
+          end
+        end
+        raise "feature value no found: "+feature.to_s
+      else
+        raise "invalid value type"
+      end
+    end
 
 
 		def save
@@ -120,21 +190,7 @@ module OpenTox
       resource = RestClient::Resource.new(@@config[:services]["opentox-dataset"], :user => @@users[:users].keys[0], :password => @@users[:users].values[0])		  
 		  uri = resource.post data, :content_type => content_type
 			dataset = Dataset.new
-<<<<<<< HEAD
-			dataset.read uri.to_s
-			dataset
-		end
-
-		def self.find(uri)
-			dataset = Dataset.new
-      LOGGER.debug "Getting data from #{uri}"
-      data = `curl "#{uri}" 2> /dev/null`
-			#LOGGER.debug data
-			#data = RestClient.get(uri, :accept => 'application/rdf+xml') # unclear why this does not work for complex uris, Dataset.find works from irb
-			dataset.rdf = data
-=======
 			dataset.read uri.chomp.to_s
->>>>>>> helma/development
 			dataset
 		end
 
@@ -146,61 +202,6 @@ module OpenTox
 			features
 		end
 
-<<<<<<< HEAD
-		def data
-			data = {}
-			@model.subjects(RDF['type'], OT['DataEntry']).each do |data_entry|
-				compound_node  = @model.object(data_entry, OT['compound'])
-				compound_uri = @model.object(compound_node, DC['identifier']).to_s
-				@model.find(data_entry, OT['values'], nil) do |s,p,values|
-					feature_node = @model.object values, OT['feature']
-					feature_uri = @model.object(feature_node, DC['identifier']).to_s.sub(/\^\^.*$/,'') # remove XML datatype
-					type = @model.object(values, RDF['type'])
-					if type == OT['FeatureValue']
-						value = @model.object(values, OT['value']).to_s
-						case value.to_s
-						when TRUE_REGEXP # defined in environment.rb
-							value = true
-						when FALSE_REGEXP # defined in environment.rb
-							value = false
-				  	when /.*\^\^<.*XMLSchema#.*>/
-              case value.to_s
-              when /XMLSchema#string/
-                value = value.to_s[0..(value.to_s.index("^^")-1)]
-              when /XMLSchema#double/
-                value = value.to_s[0..(value.to_s.index("^^")-1)].to_f
-              else
-                LOGGER.warn " ILLEGAL TYPE "+compound_uri + " has value '" + value.to_s + "' for feature " + feature_uri
-                value = nil
-              end
-						else
-							LOGGER.warn compound_uri + " has value '" + value.to_s + "' for feature " + feature_uri
-							value = nil
-						end
-						data[compound_uri] = {} unless data[compound_uri]
-						data[compound_uri][feature_uri] = [] unless data[compound_uri][feature_uri]
-						data[compound_uri][feature_uri] << value unless value.nil?
-					elsif type == OT['Tuple']
-						entry = {}
-						data[compound_uri] = {} unless data[compound_uri]
-						data[compound_uri][feature_uri] = [] unless data[compound_uri][feature_uri]
-						@model.find(values, OT['complexValue'],nil) do |s,p,complex_value|
-							name_node = @model.object complex_value, OT['feature']
-							name = @model.object(name_node, DC['title']).to_s
-							value = @model.object(complex_value, OT['value']).to_s
-							v = value.sub(/\^\^.*$/,'') # remove XML datatype
-							v = v.to_f if v.match(/^[\.|\d]+$/) # guess numeric datatype
-							entry[name] = v
-						end
-						data[compound_uri][feature_uri] << entry
-					end
-				end
-			end
-			data
-		end
-=======
->>>>>>> helma/development
-
 		def compounds
 			compounds = []
 			@model.subjects(RDF['type'], OT["Compound"]).each do |compound_node|
diff --git a/lib/environment.rb b/lib/environment.rb
index 4c501b4..2838c29 100644
--- a/lib/environment.rb
+++ b/lib/environment.rb
@@ -95,12 +95,14 @@ class MyLogger < Logger
 
 end
 
-logfile = "#{LOG_DIR}/#{ENV["RACK_ENV"]}.log"
 
+logfile = "#{LOG_DIR}/#{ENV["RACK_ENV"]}.log"
 LOGGER = MyLogger.new(logfile,'daily') # daily rotation
+
 #LOGGER = MyLogger.new(STDOUT)
+#LOGGER.datetime_format = "%Y-%m-%d %H:%M:%S "
 
-LOGGER.level = Logger::DEBUG
+#LOGGER.level = Logger::DEBUG
 
 if File.exist?(user_file)
   @@users = YAML.load_file(user_file)
diff --git a/lib/model.rb b/lib/model.rb
index 8877b84..e8eee09 100644
--- a/lib/model.rb
+++ b/lib/model.rb
@@ -3,11 +3,16 @@ module OpenTox
 
 		class Generic
 
-			attr_accessor :predicted_variables, :independent_variables, :dependent_variables, :activity_dataset_uri, :feature_dataset_uri, :effects, :activities, :p_values, :fingerprints, :features, :algorithm
+			attr_accessor :uri, :title, :source, :identifier, :predicted_variables, :independent_variables, :dependent_variables, :activity_dataset_uri, :feature_dataset_uri, :effects, :activities, :p_values, :fingerprints, :features, :algorithm
 
 			def self.find(uri)
 				owl = OpenTox::Owl.from_uri(uri)
-				@title = owl.title
+        return self.new(owl)
+      end
+      
+      protected
+      def initialize(owl)
+        @title = owl.title
 				@source = owl.source
 				@identifier = owl.identifier.sub(/^\[/,'').sub(/\]$/,'')
 				@uri = @identifier
@@ -16,8 +21,46 @@ module OpenTox
 				@independent_variables = owl.independentVariables
 				@predicted_variables = owl.predictedVariables
 			end
-			
-		end
+	 end
+  
+  
+   class PredictionModel < Generic
+     
+     def self.build( algorithm_uri, algorithm_params )
+        
+       LOGGER.debug "Build model, algorithm_uri:"+algorithm_uri.to_s+", algorithm_parms: "+algorithm_params.to_s
+       uri = OpenTox::RestClientWrapper.post(algorithm_uri,algorithm_params).to_s
+       uri = OpenTox::Task.find(uri).wait_for_resource.to_s if Utils.task_uri?(uri)
+       return PredictionModel.find(uri)
+     end
+    
+     def predict_dataset( dataset_uri )
+
+       LOGGER.debug "Predict dataset: "+dataset_uri.to_s+" with model "+@uri.to_s
+       
+       #HACK using curl
+       uri = ""
+       IO.popen("curl -X POST -d dataset_uri='"+dataset_uri+"' "+@uri.to_s+" 2> /dev/null") do |f| 
+         while line = f.gets
+           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
+     end
+    
+     def classification?
+       #HACK replace with request to ontology server
+       case @title
+       when /lazar classification/
+         return true
+       else
+         raise "unknown model: "+@title.to_s
+       end
+     end
+   end
+  
    
 		class Lazar < Generic
 
diff --git a/lib/opentox-ruby-api-wrapper.rb b/lib/opentox-ruby-api-wrapper.rb
index 849043c..c795ca8 100644
--- a/lib/opentox-ruby-api-wrapper.rb
+++ b/lib/opentox-ruby-api-wrapper.rb
@@ -8,6 +8,6 @@ rescue LoadError
 	puts "Please install Openbabel with 'rake openbabel:install' in the compound component"
 end
 
-['owl', 'compound','dataset','algorithm','model','task','validation','utils','authorization'].each do |lib|
+['owl', 'compound','dataset','algorithm','model','task','validation','utils','authorization','features'].each do |lib|
 	require lib
 end
diff --git a/lib/owl.rb b/lib/owl.rb
index 1d47d2c..2cd89c4 100644
--- a/lib/owl.rb
+++ b/lib/owl.rb
@@ -15,15 +15,18 @@ 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
-				parser.parse_into_model(owl.model,uri)
-			rescue => e
-				raise "Error parsing #{uri}: #{e.message + e.backtrace}"
+        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
diff --git a/lib/spork.rb b/lib/spork.rb
index e8cf37f..c77b5b5 100644
--- a/lib/spork.rb
+++ b/lib/spork.rb
@@ -53,7 +53,7 @@ module Spork
         yield
 
       rescue => ex
-        raise ex
+        #raise ex
         logger.error "spork> Exception in child[#{Process.pid}] - #{ex.class}: #{ex.message}" if logger
       ensure
         logger.info "spork> child[#{Process.pid}] took #{Time.now - start} sec" if logger
diff --git a/lib/task.rb b/lib/task.rb
index 5591a34..75cc2d2 100644
--- a/lib/task.rb
+++ b/lib/task.rb
@@ -90,7 +90,37 @@ module OpenTox
 			until self.completed? or self.failed?
 				sleep dur
 			end
-		end
+	  end
+  
+    def self.as_task
+      task = OpenTox::Task.create
+      LOGGER.debug "Starting task"
+      pid = Spork.spork(:logger => LOGGER) do
+        task.started
+        LOGGER.debug "Task #{task.uri} started #{Time.now}"
+        begin
+          result = yield
+          task.completed(result)
+        rescue => ex
+          raise ex
+          LOGGER.error ex.message
+          task.failed
+        end
+        raise "Invalid task state" unless task.completed? || task.failed?
+      end  
+      LOGGER.debug "task PID: " + pid.to_s
+      task.pid = pid
+      task.uri
+    end  
+  
+    def wait_for_resource
+      wait_for_completion
+      if failed?
+        LOGGER.error "task failed: "+uri.to_s
+        return nil
+      end
+      return resource
+    end
 
 	end
 
diff --git a/lib/utils.rb b/lib/utils.rb
index cccb5ea..bbb750d 100644
--- a/lib/utils.rb
+++ b/lib/utils.rb
@@ -4,6 +4,23 @@ module OpenTox
 		def self.gauss(sim, sigma = 0.3) 
 			x = 1.0 - sim
 			Math.exp(-(x*x)/(2*sigma*sigma))
-		end
+	  end
+    
+    def self.task_uri?(uri)
+      is_uri?(uri) && uri.to_s =~ /task/
+    end
+    
+    def self.dataset_uri?(uri)
+     is_uri?(uri) && uri.to_s =~ /dataset/
+    end
+  
+    def self.is_uri?(uri)
+      begin
+        URI::parse(uri)
+      rescue URI::InvalidURIError
+        false
+      end
+    end
+  
 	end
 end