From 18454895a9e426e3a6499ff268044687a996993e Mon Sep 17 00:00:00 2001
From: Christoph Helma <helma@in-silico.ch>
Date: Fri, 12 Jul 2013 18:12:18 +0200
Subject: value type identification/conversion fixed

---
 lib/dataset.rb    | 35 ++++++++++++++---------------------
 lib/opentox.rb    |  3 ---
 lib/validation.rb |  4 ++--
 3 files changed, 16 insertions(+), 26 deletions(-)

diff --git a/lib/dataset.rb b/lib/dataset.rb
index c2ff833..cb64406 100644
--- a/lib/dataset.rb
+++ b/lib/dataset.rb
@@ -64,24 +64,13 @@ module OpenTox
           RestClientWrapper.get(service_uri,{:query => sparql},{:accept => "text/uri-list", :subjectid => @subjectid}).split("\n").each do |row|
             r,c,v = row.split("\t")
             @data_entries[r.to_i] ||= []
-            #v = v.to_f if v.numeric?
-            #v = nil if v.is_a? String and v.empty?
-            @data_entries[r.to_i][c.to_i] = v
-          end
-        # TODO: fallbacks for external and unordered datasets
-        features.each_with_index do |feature,i|
-          if feature[RDF.type].include? RDF::OT.NumericFeature
-            if feature[RDF.type].include? RDF::OT.NominalFeature
-              if feature[RDF.type].include? RDF::OT.StringFeature
-                @data_entries.each { |row| row[i] = row[i].to_s if row[i] }
-              else
-                @data_entries.each { |row| row[i] = row[i] if row[i] }
-              end
-            else
-              @data_entries.each { |row| row[i] = row[i].to_f if row[i] }
+            # adjust value class depending on feature type, StringFeature takes precedence over NumericFeature
+            if features[c.to_i][RDF.type].include? RDF::OT.NumericFeature and ! features[c.to_i][RDF.type].include? RDF::OT.StringFeature
+              v = v.to_f if v
             end
+            @data_entries[r.to_i][c.to_i] = v if v 
           end
-        end
+        # TODO: fallbacks for external and unordered datasets
       end
       @data_entries
     end
@@ -131,7 +120,8 @@ module OpenTox
       if prediction_feature and confidence_feature
         compounds.each do |compound|
           value = values(compound,prediction_feature).first
-          confidence = values(compound,confidence_feature).first
+          value = value.to_f if prediction_feature[RDF.type].include? RDF::OT.NumericFeature and ! prediction_feature[RDF.type].include? RDF::OT.StringFeature
+          confidence = values(compound,confidence_feature).first.to_f
           predictions << {:compound => compound, :value => value, :confidence => confidence} if value and confidence
         end
       end
@@ -147,6 +137,8 @@ module OpenTox
     def upload filename, wait=true
       uri = RestClientWrapper.put(@uri, {:file => File.new(filename)}, {:subjectid => @subjectid})
       wait_for_task uri if URI.task?(uri) and wait
+      compounds true
+      features true
       metadata true
       @uri
     end
@@ -160,13 +152,14 @@ module OpenTox
       row = @compounds.collect{|c| c.uri}.index(compound.uri)
       @features << feature unless @features.collect{|f| f.uri}.include?(feature.uri)
       col = @features.collect{|f| f.uri}.index(feature.uri)
-      @data_entries[row] ||= []
-      if @data_entries[row][col] # duplicated values
-        #row = @compounds.size
+      if @data_entries[row] and @data_entries[row][col] # duplicated values
         @compounds << compound
         row = @compounds.collect{|c| c.uri}.rindex(compound.uri)
       end
-      @data_entries[row][col] = value
+      if value
+        @data_entries[row] ||= []
+        @data_entries[row][col] = value
+      end
     end
 
     # TODO: remove? might be dangerous if feature ordering is incorrect
diff --git a/lib/opentox.rb b/lib/opentox.rb
index 93e2c52..73a33d3 100644
--- a/lib/opentox.rb
+++ b/lib/opentox.rb
@@ -156,10 +156,7 @@ module OpenTox
 
     # rdf serialization methods for all formats e.g. to_rdfxml
     send :define_method, "to_#{format}".to_sym do
-      puts format
-      puts self.inspect
       create_rdf
-      puts @rdf.to_s
       RDF::Writer.for(format).buffer(:encoding => Encoding::ASCII) do |writer|
         writer << @rdf
       end
diff --git a/lib/validation.rb b/lib/validation.rb
index 3b1cf13..565a366 100644
--- a/lib/validation.rb
+++ b/lib/validation.rb
@@ -53,7 +53,7 @@ module OpenTox
     # @param [OpenTox::Task,optional] waiting_task (can be a OpenTox::Subtask as well), progress is updated accordingly
     # @return [OpenTox::Validation]
     def self.create_training_test_split( params, subjectid=nil, waiting_task=nil )
-      params[:subjectid] = subjectid if subjectid
+      #params[:subjectid] = subjectid if subjectid
       uri = OpenTox::RestClientWrapper.post( File.join($validation[:uri],"training_test_split"),
         params,{:content_type => "text/uri-list", :subjectid => subjectid},waiting_task )
       Validation.new(wait_for_task(uri))
@@ -185,7 +185,7 @@ module OpenTox
     # @param [OpenTox::Task,optional] waiting_task (can be a OpenTox::Subtask as well), progress is updated accordingly
     # @return [OpenTox::Crossvalidation]
     def self.create( params, subjectid=nil, waiting_task=nil )
-      params[:subjectid] = subjectid if subjectid
+      #params[:subjectid] = subjectid if subjectid
       uri = OpenTox::RestClientWrapper.post( File.join($validation[:uri],"crossvalidation"),
         params,{:content_type => "text/uri-list", :subjectid => subjectid},waiting_task )
       uri = wait_for_task(uri)
-- 
cgit v1.2.3