From 9ee59712fab04f7345e3f44dd02321e50cd460c5 Mon Sep 17 00:00:00 2001
From: Christoph Helma <helma@in-silico.ch>
Date: Thu, 11 Jul 2013 14:59:41 +0200
Subject: duplicated and empty data_entries fixed

---
 application.rb | 21 +++++----------------
 helper.rb      | 30 ++++++++++++++++++++----------
 2 files changed, 25 insertions(+), 26 deletions(-)

diff --git a/application.rb b/application.rb
index edfbb91..61406ec 100644
--- a/application.rb
+++ b/application.rb
@@ -12,21 +12,10 @@ require_relative 'test.rb'
 # Library code
 $logger.debug "Dataset booting: #{$dataset.collect{|k,v| "#{k}: '#{v}'"} }"
 
-require 'profiler'
-
 # Entry point
 module OpenTox
   class Application < Service
 
-    before do
-      #Profiler__::start_profile
-    end
-
-    after do
-      #Profiler__::stop_profile
-      #Profiler__::print_profile($stdout)
-    end
-
     before "/#{SERVICE}/:id/:property" do
       @uri = uri("/#{SERVICE}/#{params[:id]}")
     end
@@ -73,10 +62,10 @@ module OpenTox
     get '/dataset/:id/metadata' do
       case @accept
       when "application/rdf+xml", "text/turtle", "text/plain"
-        sparql = "CONSTRUCT {?s ?p ?o.} FROM <#{@uri}> WHERE { 
-          { ?s ?p ?o.  <#{@uri}> ?p ?o. } UNION
-          { ?s ?p ?o. ?s <#{RDF.type}> <#{RDF::OT.Parameter}> . }
-          FILTER (?p != <#{RDF::OT.dataEntry}>)
+        sparql = "CONSTRUCT {?s ?p ?o.} FROM <#{@uri}> WHERE {
+          { <#{@uri}> ?p ?o.  ?s <#{RDF.type}> <#{RDF::OT.Dataset}>. }
+          UNION { ?s ?p ?o. ?s <#{RDF.type}> <#{RDF::OT.Parameter}> . }
+          MINUS { ?s <#{RDF::OT.dataEntry}> ?o. }
         } "
         FourStore.query sparql, @accept
       else
@@ -107,7 +96,7 @@ module OpenTox
       when "application/rdf+xml", "text/turtle", "text/plain"
         sparql = "CONSTRUCT {?s ?p ?o.} FROM <#{@uri}> WHERE {?s <#{RDF.type}> <#{RDF::OT.Compound}>; ?p ?o. }"
       when "text/uri-list"
-        sparql = "SELECT DISTINCT ?s FROM <#{@uri}> WHERE {?s <#{RDF.type}> <#{RDF::OT.Compound}>. ?s <#{RDF::OLO.index}> ?idx } ORDER BY ?idx"
+        sparql = "SELECT ?s FROM <#{@uri}> WHERE {?s <#{RDF.type}> <#{RDF::OT.Compound}>. ?s <#{RDF::OLO.index}> ?idx } ORDER BY ?idx"
       else
         bad_request_error "'#{@accept}' is not a supported content type."
       end
diff --git a/helper.rb b/helper.rb
index 54b4540..cb2e1c1 100644
--- a/helper.rb
+++ b/helper.rb
@@ -102,8 +102,10 @@ module OpenTox
 
         # compounds and values
         compound_uris = []
+        r = -1
         table.each_with_index do |values,j|
           compound = values.shift
+          compound_uri = nil
           begin
             case compound_format
             when /URI|URL/i
@@ -112,6 +114,7 @@ module OpenTox
               c = OpenTox::Compound.from_smiles(compound)
               if c.inchi.empty?
                 @warnings << "Cannot parse compound '#{compound}' at position #{j+2}, all entries are ignored."
+                next
               else
                 compound_uri = c.uri
               end
@@ -119,6 +122,7 @@ module OpenTox
               compound = OpenTox::Compound.from_inchi(compound)
               if c.inchi.empty?
                 @warnings << "Cannot parse compound '#{compound}' at position #{j+2}, all entries are ignored."
+                next
               else
                 compound_uri = c.uri
               end
@@ -130,26 +134,32 @@ module OpenTox
             @warnings << "Cannot parse compound '#{compound}' at position #{j+2}, all entries are ignored." # be careful with double quotes in literals! \C in smiles is an illegal Turtle string
             next
           end
+          
+          r += 1
           compound_uris << compound_uri
           unless values.size == features.size
             @warnings << "Number of values at position #{j+2} (#{values.size}) is different than header size (#{features.size}), all entries are ignored."
             next
           end
-          ntriples << "<#{compound_uri}> <#{RDF.type}> <#{RDF::OT.Compound}>."
-          ntriples << "<#{compound_uri}> <#{RDF::OLO.index}> #{j} ."
 
-          data_entry_node = "_:dataentry"+ j.to_s
+          ntriples << "<#{compound_uri}> <#{RDF.type}> <#{RDF::OT.Compound}>."
+          ntriples << "<#{compound_uri}> <#{RDF::OLO.index}> #{r} ."
+          #data_entry_node = "<#{File.join @uri,"dataentry",j.to_s}>" # too slow or not accepted by 4store
+          data_entry_node = "_:dataentry"+ r.to_s
           ntriples << "<#{@uri}> <#{RDF::OT.dataEntry}> #{data_entry_node} ."
           ntriples << "#{data_entry_node} <#{RDF.type}> <#{RDF::OT.DataEntry}> ."
-          ntriples << "#{data_entry_node} <#{RDF::OLO.index}> #{j} ."
+          ntriples << "#{data_entry_node} <#{RDF::OLO.index}> #{r} ."
           ntriples << "#{data_entry_node} <#{RDF::OT.compound}> <#{compound_uri}> ."
           values.each_with_index do |v,i|
-            @warnings << "Empty value for compound '#{compound}' (row #{j+2}) and feature '#{feature_names[i]}' (column #{i+2})." if v.blank?
-
-            value_node = data_entry_node+ "_value"+ i.to_s
-            ntriples << "#{data_entry_node} <#{RDF::OT.values}> #{value_node} ."
-            ntriples << "#{value_node} <#{RDF::OT.feature}> <#{features[i].uri}> ."
-            ntriples << "#{value_node} <#{RDF::OT.value}> \"#{v}\" ."
+            if v.blank?
+              @warnings << "Empty value for compound '#{compound}' (row #{r+2}) and feature '#{feature_names[i]}' (column #{i+2})."
+              next
+            else
+              value_node = data_entry_node+ "_value"+ i.to_s
+              ntriples << "#{data_entry_node} <#{RDF::OT.values}> #{value_node} ."
+              ntriples << "#{value_node} <#{RDF::OT.feature}> <#{features[i].uri}> ."
+              ntriples << "#{value_node} <#{RDF::OT.value}> \"#{v}\" ."
+            end
 
           end
         end
-- 
cgit v1.2.3