oddb2xml 2.0.6 → 2.0.7

data/lib/oddb2xml/parslet_compositions.rb

@@ -51,7 +51,7 @@ module ParseUtil
         result = result.gsub(entry.pattern,  entry.replacement)
         unless result.eql?(intermediate)
             entry.nr_occurrences += 1
-            puts "Fixed #{result}" if VERBOSE_MESSAGES
+            puts "#{File.basename(__FILE__)}:#{__LINE__}: fixed \nbefore: #{intermediate}\nafter:  #{result}"
         end
       }
       @nrLines += 1
@@ -93,6 +93,122 @@ class QtyLit   < Struct.new(:qty)
 end
 
 class CompositionTransformer < Parslet::Transform
+  def CompositionTransformer.get_ratio(parse_info)
+    if parse_info[:ratio]
+      if parse_info[:ratio].to_s.length > 0 and parse_info[:ratio].to_s != ', '
+        parse_info[:ratio].to_s.sub(/^,\s+/, '').sub(/,\s+$/,'')
+      else
+        nil
+      end
+    else
+      nil
+    end
+  end
+
+  rule(:corresp => simple(:corresp),
+       ) {
+    |dictionary|
+      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
+      @@corresp = dictionary[:corresp].to_s
+  }
+  rule( :substance_name => simple(:substance_name),
+        :dose => simple(:dose),
+      ) {
+    |dictionary|
+        puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
+        dose = dictionary[:dose].is_a?(ParseDose) ? dictionary[:dose] : nil
+        substance = ParseSubstance.new(dictionary[:substance_name], dose)
+        @@substances <<  substance
+        substance
+  }
+
+  rule( :more_info => simple(:more_info),
+      ) {
+    |dictionary|
+        puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
+        @@corresp = dictionary[:more_info].to_s.strip.sub(/:$/, '')
+  }
+  rule( :more_info => simple(:more_info),
+        :substance_name => simple(:substance_name),
+        :dose => simple(:dose),
+      ) {
+    |dictionary|
+        puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
+        dose = dictionary[:dose].is_a?(ParseDose) ? dictionary[:dose] : nil
+        substance = ParseSubstance.new(dictionary[:substance_name].to_s, dose)
+        substance.more_info = dictionary[:more_info].to_s.strip.sub(/:$/, '') if dictionary[:more_info] and dictionary[:more_info].to_s.length > 0
+        @@substances <<  substance
+        substance
+  }
+
+  rule(:lebensmittel_zusatz => simple(:lebensmittel_zusatz),
+       :more_info => simple(:more_info),
+       :digits => simple(:digits)) {
+    |dictionary|
+      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
+      substance =  ParseSubstance.new("#{dictionary[:lebensmittel_zusatz]} #{dictionary[:digits]}")
+      substance.more_info =  dictionary[:more_info].to_s.sub(/:\s+$/, '').strip if dictionary[:more_info]
+      @@substances <<  substance
+      substance
+  }
+  rule(:excipiens => subtree(:excipiens),
+       ) {
+    |dictionary|
+      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
+      info = dictionary[:excipiens].is_a?(Hash) ? dictionary[:excipiens] : dictionary[:excipiens].first
+      @@excipiens           =  ParseSubstance.new(info[:excipiens_description] ? info[:excipiens_description] : 'Excipiens')
+      @@excipiens.dose      = info[:dose] if info[:dose]
+      @@excipiens.more_info = CompositionTransformer.get_ratio(dictionary)
+      @@excipiens.cdose     = info[:dose_corresp] if info[:dose_corresp]
+      @@excipiens.more_info = info[:more_info] if info[:more_info]
+      binding.pry if dictionary[:dose_2]
+      nil
+  }
+  rule(:composition => subtree(:composition),
+       ) {
+    |dictionary|
+      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
+      info = dictionary[:composition].is_a?(Hash) ? dictionary[:composition] : dictionary[:composition].first
+      if info.is_a?(Hash)
+        @@excipiens           =  ParseSubstance.new(info[:excipiens_description] ? info[:excipiens_description] : 'Excipiens')
+        @@excipiens.dose      = info[:dose] if info[:dose]
+        @@excipiens.more_info = CompositionTransformer.get_ratio(dictionary)
+        @@excipiens.cdose     = info[:dose_corresp] if info[:dose_corresp]
+        @@excipiens.more_info = info[:more_info] if info[:more_info]
+        binding.pry if dictionary[:dose_2]
+        @@excipiens
+       else
+        info
+       end
+       }
+  rule(:substance => simple(:substance),
+       :chemical_substance => simple(:chemical_substance),
+       :substance_ut => sequence(:substance_ut),
+       :ratio => simple(:ratio),
+       ) {
+    |dictionary|
+       puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
+       ratio = CompositionTransformer.get_ratio(dictionary)
+       if ratio and ratio.length > 0
+        if dictionary[:substance].more_info
+          dictionary[:substance].more_info += ' ' + ratio.strip
+        else
+          dictionary[:substance].more_info = ratio.strip
+        end
+       end
+       if  dictionary[:chemical_substance]
+        dictionary[:substance].chemical_substance = dictionary[:chemical_substance]
+        @@substances -= [dictionary[:chemical_substance]]
+       end
+       if  dictionary[:substance_ut].size > 0
+        dictionary[:substance].salts += dictionary[:substance_ut].last.salts
+        dictionary[:substance_ut].last.salts = []
+        dictionary[:substance].salts << dictionary[:substance_ut].last
+        @@substances -= dictionary[:substance_ut]
+       end
+       dictionary[:substance]
+  }
+
   rule(:int => simple(:int))        { IntLit.new(int) }
   rule(:number => simple(:nb)) {
     nb.match(/[eE\.]/) ? Float(nb) : Integer(nb)
@@ -115,8 +231,17 @@ class CompositionTransformer < Parslet::Transform
     :unit    => simple(:unit))  { ParseDose.new(nil, unit) }
   rule(
     :qty    => simple(:qty))  { ParseDose.new(qty, nil) }
+  rule(
+    :qty    => simple(:qty),
+    :unit    => simple(:unit),
+    :dose_right    => simple(:dose_right),
+  )  {
+    dose = ParseDose.new(qty, unit)
+    dose.unit = dose.unit.to_s + ' et ' + ParseDose.new(dose_right).to_s
+    dose
+  }
 
-  @@substances ||= []
+@@substances ||= []
   @@excipiens  = nil
   def CompositionTransformer.clear_substances
     @@substances = []
@@ -132,311 +257,6 @@ class CompositionTransformer < Parslet::Transform
   def CompositionTransformer.corresp
     @@corresp ? @@corresp.clone : nil
   end
-
-  rule(:ratio => simple(:ratio) ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-      @@substances.last.more_info = dictionary[:ratio].to_s if @@substances.last
-  }
-  rule(:substance => sequence(:substance),
-       :ratio => simple(:ratio)) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-      @@substances.last.more_info = dictionary[:ratio].to_s if @@substances.last
-  }
-
-  rule(:solvens => simple(:solvens) ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-      substance =  ParseSubstance.new(dictionary[:solvens].to_s)
-      substance.more_info =  'Solvens'
-      @@substances <<  substance
-  }
-  rule(:lebensmittel_zusatz => simple(:lebensmittel_zusatz),
-       :more_info => simple(:more_info),
-       :digits => simple(:digits)) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-      substance =  ParseSubstance.new("#{dictionary[:lebensmittel_zusatz]} #{dictionary[:digits]}")
-      substance.more_info =  dictionary[:more_info].to_s.sub(/:$/, '')
-      @@substances <<  substance
-  }
-  rule(:lebensmittel_zusatz => simple(:lebensmittel_zusatz),
-       :digits => simple(:digits)) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}"  if VERBOSE_MESSAGES
-      @@substances << ParseSubstance.new("#{dictionary[:lebensmittel_zusatz]} #{dictionary[:digits]}")
-      dictionary[:substance]
-  }
-  rule(:substance => simple(:substance)) {
-    |dictionary|
-    puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-  }
-  rule(:substance_name => simple(:substance_name),
-       :dose => simple(:dose),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-      @@substances << ParseSubstance.new(dictionary[:substance_name].to_s, dictionary[:dose])
-  }
-  rule(:substance_ut => sequence(:substance_ut),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-      nil
-  }
-  rule(:for_ut => sequence(:for_ut),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-      if dictionary[:for_ut].size > 1
-        @@substances[-2].salts << dictionary[:for_ut].last.clone
-        @@substances.delete(dictionary[:for_ut].last)
-      end
-      nil
-  }
-
-  rule(:substance_name => simple(:substance_name),
-       :dose => simple(:dose),
-       :substance_corresp => sequence(:substance_corresp),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-      substance = ParseSubstance.new(dictionary[:substance_name].to_s, dictionary[:dose])
-      substance.chemical_substance = @@substances.last
-      @@substances.delete_at(-1)
-      @@substances <<  substance
-  }
-
-  rule(:mineralia => simple(:mineralia),
-       :more_info => simple(:more_info),
-       :substance_name => simple(:substance_name),
-       :dose => simple(:dose),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-      substance = ParseSubstance.new(dictionary[:substance_name].to_s, dictionary[:dose])
-      substance.more_info = dictionary[:mineralia].to_s + ' ' + dictionary[:more_info].to_s
-       # TODO: fix alia
-      @@substances <<  substance
-  }
-  rule(:substance_name => simple(:substance_name),
-       :conserv => simple(:conserv),
-       :dose => simple(:dose),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}"
-      substance = ParseSubstance.new(dictionary[:substance_name], ParseDose.new(dictionary[:dose].to_s))
-      @@substances <<  substance
-      substance.more_info =  dictionary[:conserv].to_s.sub(/:$/, '')
-  }
-
-  rule(:substance_name => simple(:substance_name),
-       :mineralia => simple(:mineralia),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}"
-      substance = ParseSubstance.new(dictionary[:substance_name])
-      substance.more_info =  dictionary[:mineralia].to_s.sub(/:$/, '')
-      @@substances <<  substance
-  }
-  rule(:substance_name => simple(:substance_name),
-       :more_info => simple(:more_info),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-      substance = ParseSubstance.new(dictionary[:substance_name])
-      @@substances <<  substance
-      substance.more_info =  dictionary[:more_info].to_s.sub(/:$/, '')
-  }
-  rule(:substance_name => simple(:substance_name),
-       :residui => simple(:residui),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-       binding.pry
-      substance = ParseSubstance.new(dictionary[:substance_name])
-      @@substances <<  substance
-      substance.more_info =  dictionary[:residui].to_s.sub(/:$/, '')
-  }
-  rule(:qty => simple(:qty),
-       :unit => simple(:unit),
-       :dose_right => simple(:dose_right),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-      ParseDose.new(dictionary[:qty].to_s, dictionary[:unit].to_s + ' et ' +  dictionary[:dose_right].to_s )
-  }
-
-  rule(:substance_name => simple(:substance_name),
-       :qty => simple(:qty),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}"
-      @@substances << ParseSubstance.new(dictionary[:substance_name].to_s.strip, ParseDose.new(dictionary[:qty].to_s))
-  }
-
-  rule(:substance_name => simple(:substance_name),
-       :dose_corresp => simple(:dose_corresp),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}"
-      @@substances << ParseSubstance.new(dictionary[:substance_name].to_s, dictionary[:dose_corresp])
-  }
-  rule(:description => simple(:description),
-       :substance_name => simple(:substance_name),
-       :qty => simple(:qty),
-       :more_info => simple(:more_info),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-      substance = ParseSubstance.new(dictionary[:substance_name], ParseDose.new(dictionary[:qty].to_s))
-      @@substances <<  substance
-      substance.more_info =  dictionary[:more_info].to_s
-      substance.description =  dictionary[:description].to_s
-      substance
-  }
-  rule(:der => simple(:der),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-      @@substances << ParseSubstance.new(dictionary[:der].to_s)
-  }
-  rule(:der => simple(:der),
-       :substance_corresp => sequence(:substance_corresp),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-      substance = ParseSubstance.new(dictionary[:der].to_s)
-      substance.chemical_substance = @@substances.last
-      @@substances.delete_at(-1)
-      @@substances <<  substance
-  }
-  rule(:histamin => simple(:histamin),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: histamin dictionary #{dictionary}"
-      @@substances << ParseSubstance.new(dictionary[:histamin].to_s)
-  }
-  rule(:substance_name => simple(:substance_name),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}"  if VERBOSE_MESSAGES
-      @@substances << ParseSubstance.new(dictionary[:substance_name].to_s)
-  }
-  rule(:one_substance => sequence(:one_substance)) {
-    |dictionary|
-       puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}"
-      @@substances << ParseSubstance.new(dictionary[:one_substance])
-  }
-  rule(:one_substance => sequence(:one_substance)) {
-    |dictionary|
-       puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}"
-      @@substances << ParseSubstance.new(dictionary[:one_substance])
-  }
-
-  rule(:substance_name => simple(:substance_name),
-       :substance_ut => sequence(:substance_ut),
-       :dose => simple(:dose),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}"
-      @@substances.last.salts << ParseSubstance.new(dictionary[:substance_name].to_s, dictionary[:dose])
-      nil
-  }
-
-  rule(:mineralia => simple(:mineralia),
-       :dose => simple(:dose),
-       :substance_name => simple(:substance_name),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}"  if VERBOSE_MESSAGES
-      dose = dictionary[:dose].is_a?(ParseDose) ? dictionary[:dose] : ParseDose.new(dictionary[:dose].to_s)
-      substance = ParseSubstance.new(dictionary[:substance_name], dose)
-      substance.more_info = dictionary[:mineralia].to_s
-      @@substances <<  substance
-      # @@substances << ParseSubstance.new(dictionary[:substance_name].to_s, dictionary[:dose])
-  }
-
-  rule(:mineralia => simple(:mineralia),
-       :dose => simple(:dose),
-       :substance_ut => simple(:substance_ut),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}"
-      dose = dictionary[:dose].is_a?(ParseDose) ? dictionary[:dose] : ParseDose.new(dictionary[:dose].to_s)
-      substance = ParseSubstance.new(dictionary[:substance_ut], dose)
-      substance.more_info = dictionary[:mineralia].to_s
-       binding.pry
-      @@substances <<  substance
-      nil
-  }
-
-
-  rule(:mineralia => simple(:mineralia),
-       :substance_ut => simple(:substance_ut),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}"
-       binding.pry
-      @@substances.last.salts << ParseSubstance.new(dictionary[:substance_name].to_s, dictionary[:dose])
-      nil
-  }
-  rule( :more_info => simple(:more_info),
-        :substance_name => simple(:substance_name),
-        :dose => simple(:dose),
-      ) {
-    |dictionary|
-        puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-        dose = dictionary[:dose].is_a?(ParseDose) ? dictionary[:dose] : ParseDose.new(dictionary[:dose].to_s)
-        substance = ParseSubstance.new(dictionary[:substance_name], dose)
-        substance.more_info = dictionary[:more_info].to_s
-        @@substances <<  substance
-  }
-
-  rule(:excipiens => simple(:excipiens),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-      @@excipiens = dictionary[:excipiens].is_a?(ParseDose) ? ParseSubstance.new('excipiens', dictionary[:excipiens]) : nil
-  }
-
-  rule(:substance_name => simple(:substance_name),
-       :dose_pro => simple(:dose_pro),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-      dose = dictionary[:dose_pro].is_a?(ParseDose) ? dictionary[:dose_pro] : ParseDose.new(dictionary[:dose_pro].to_s)
-      substance = ParseSubstance.new(dictionary[:substance_name], dose)
-      @@excipiens = dose
-      @@substances <<  substance
-  }
-  rule(:substance_name => simple(:substance_name),
-       :dose => simple(:dose),
-       :dose_pro => simple(:dose_pro),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-      dose = dictionary[:dose_pro].is_a?(ParseDose) ? dictionary[:dose_pro] : ParseDose.new(dictionary[:dose_pro].to_s)
-      dose_pro = dictionary[:dose_pro].is_a?(ParseDose) ? dictionary[:dose_pro] : ParseDose.new(dictionary[:dose_pro].to_s)
-      substance = ParseSubstance.new(dictionary[:substance_name], dose)
-      @@excipiens = dose_pro
-      @@substances <<  substance
-  }
-
-  rule(:dose_pro => simple(:dose_pro),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-      dictionary[:dose_pro]
-  }
-
-  rule(:corresp => simple(:corresp),
-       ) {
-    |dictionary|
-      puts "#{File.basename(__FILE__)}:#{__LINE__}: dictionary #{dictionary}" if VERBOSE_MESSAGES
-      @@corresp = dictionary[:corresp].to_s
-  }
 end
 
 class ParseDose
@@ -509,15 +329,21 @@ end
 
 class ParseComposition
   attr_accessor   :source, :label, :label_description, :substances, :galenic_form, :route_of_administration,
-                  :corresp
+                  :corresp, :excipiens
 
-  ErrorsToFix = { /(sulfuris D6\s[^\s]+\smg)\s([^,]+)/ => '\1, \2',
-                  /(\d+)\s+\-\s*(\d+)/ => '\1-\2',
+  ErrorsToFix = { /(\d+)\s+\-\s*(\d+)/ => '\1-\2',
                   'o.1' => '0.1',
-                  'g DER:' => 'g, DER:',
-                  /(excipiens ad solutionem pro \d+ ml), corresp\./ => '\1 corresp.',
-                  /^(pollinis allergeni extractum[^\:]+\:)/ => 'A): \1',
-                  /^(acari allergeni extractum 5000 U\.\:)/ => 'A): \1',
+                  /\s+(mg|g) DER:/ => ' \1, DER:',
+                  ' mind. ' => ' min. ',
+                  ' streptococci pyogen. ' => ' streptococci pyogen ',
+                  ' ut excipiens' => ', excipiens',
+                  ' Corresp. ' => ' corresp. ',
+                  ',,' => ',',
+                  'avena elatior,dactylis glomerata' => 'avena elatior, dactylis glomerata',
+                  ' color.: corresp. ' => ' corresp.',
+                  / U\.: (excipiens) / =>  ' U. \1 ',
+                  / U\.: (alnus|betula|betula|betulae) / =>  ' U., \1 ',
+                  /^(acari allergeni extractum (\(acarus siro\)|).+\s+U\.\:)/ => 'A): \1',
                 }
   @@errorHandler = ParseUtil::HandleSwissmedicErrors.new( ErrorsToFix )
 
@@ -536,7 +362,6 @@ class ParseComposition
     return nil if string == nil or  string.eql?('.') or string.eql?('')
     stripped = string.gsub(/^"|["\n]+$/, '')
     return nil unless stripped
-    @@errorHandler.nrParsingErrors += 1
     if /(U\.I\.|U\.)$/.match(stripped)
       cleaned = stripped
     else
@@ -547,7 +372,6 @@ class ParseComposition
 
     cleaned = @@errorHandler.apply_fixes(cleaned)
     puts "ParseComposition.new cleaned #{cleaned}" if VERBOSE_MESSAGES and not cleaned.eql?(stripped)
-
     CompositionTransformer.clear_substances
     result = ParseComposition.new(cleaned)
     parser3 = CompositionParser.new
@@ -555,28 +379,30 @@ class ParseComposition
     begin
       if defined?(RSpec)
         ast = transf3.apply(parser3.parse_with_debug(cleaned))
-        puts "#{File.basename(__FILE__)}:#{__LINE__}: ==>  #{ast}" if VERBOSE_MESSAGES
+        puts "#{File.basename(__FILE__)}:#{__LINE__}: ==> " if VERBOSE_MESSAGES
+        pp ast if VERBOSE_MESSAGES
       else
         ast = transf3.apply(parser3.parse(cleaned))
       end
     rescue Parslet::ParseFailed => error
+      @@errorHandler.nrParsingErrors += 1
       puts "#{File.basename(__FILE__)}:#{__LINE__}: failed parsing ==>  #{cleaned}"
       return nil
     end
     result.source = string
     return result unless ast
     return result if ast.is_a?(Parslet::Slice)
-    # pp ast; binding.pry
 
     result.substances = CompositionTransformer.substances
-    excipiens = CompositionTransformer.excipiens
+    result.excipiens = CompositionTransformer.excipiens
     result.corresp = CompositionTransformer.corresp if CompositionTransformer.corresp
-    if excipiens and excipiens.unit
-      pro_qty = "/#{excipiens.qty} #{excipiens.unit}".sub(/\/1\s+/, '/')
+    if result.excipiens and result.excipiens.unit
+      pro_qty = "/#{result.excipiens.qty} #{result.excipiens.unit}".sub(/\/1\s+/, '/')
       result.substances.each {
         |substance|
+            next unless substance.is_a?(ParseSubstance)
           substance.chemical_substance.unit = "#{substance.chemical_substance.unit}#{pro_qty}"    if substance.chemical_substance
-          substance.dose.unit               = "#{substance.dose.unit}#{pro_qty}"                  if substance.unit and not substance.unit.eql?(excipiens.unit)
+          substance.dose.unit               = "#{substance.dose.unit}#{pro_qty}"                  if substance.unit and not substance.unit.eql?(result.excipiens.unit)
       }
     end
     if ast.is_a?(Array) and  ast.first.is_a?(Hash)
@@ -592,7 +418,6 @@ class ParseComposition
       end
       result.label_description = label_description
     end
-    @@errorHandler.nrParsingErrors -=1 if result.substances.size > 0 or result.corresp
     return result
   end
 end

data/lib/oddb2xml/version.rb

@@ -1,3 +1,3 @@
 module Oddb2xml
-  VERSION = "2.0.6"
+  VERSION = "2.0.7"
 end

data/spec/calc_spec.rb

@@ -293,7 +293,7 @@ if RunAllTests
       m = />.*  /.match(xml)
       m.should eq nil
       doc = REXML::Document.new xml
-#      puts xml; binding.pry
+      # puts xml; binding.pry
       gtin = '7680540151009'
       ean12 = '7680' + sprintf('%05d',tst_naropin.iksnr_A) + sprintf('%03d',tst_naropin.pack_K)
       ean13 = (ean12 + Oddb2xml.calc_checksum(ean12))
@@ -328,14 +328,20 @@ if RunAllTests
 
       XPath.match( doc, "//ARTICLE[GTIN='7680434541015']/COMPOSITIONS/COMPOSITION/SUBSTANCES/SUBSTANCE/SUBSTANCE_NAME").
         find{|x| x.text.eql?(matri_name)}.text.should eq matri_name
-      XPath.match( doc, "//ARTICLE[GTIN='7680434541015']/COMPOSITIONS/COMPOSITION/SUBSTANCES/SUBSTANCE/CHEMICAL_SUBSTANCE").last.text.should eq 'Levomenolum'
+      XPath.match( doc, "//ARTICLE[GTIN='7680434541015']/NAME").last.text.should eq 'Kamillin Medipharm, Bad'
       XPath.match( doc, "//ARTICLE[GTIN='7680434541015']/COMPOSITIONS/COMPOSITION/SUBSTANCES/SUBSTANCE/QTY").first.text.should eq '98.9'
-      XPath.match( doc, "//ARTICLE[GTIN='7680434541015']/COMPOSITIONS/COMPOSITION/SUBSTANCES/SUBSTANCE/MORE_INFO").first.text.should eq 'ratio: 1:2-2.8'
       XPath.match( doc, "//ARTICLE[GTIN='7680300150105']/COMPOSITIONS/COMPOSITION/SUBSTANCES/SUBSTANCE/SUBSTANCE_NAME").first.text.should eq 'Lidocaini Hydrochloridum'
       XPath.match( doc, "//ARTICLE[GTIN='7680300150105']/COMPOSITIONS/COMPOSITION/SUBSTANCES/SUBSTANCE/UNIT").first.text.should eq 'mg/ml'
+      XPath.match( doc, "//ARTICLE[GTIN='7680434541015']/COMPOSITIONS/COMPOSITION/SUBSTANCES/SUBSTANCE/CHEMICAL_SUBSTANCE/SUBSTANCE_NAME").first.text.should eq 'Levomenolum'
+      XPath.match( doc, "//ARTICLE[GTIN='7680434541015']/COMPOSITIONS/COMPOSITION/SUBSTANCES/SUBSTANCE/CHEMICAL_SUBSTANCE/MORE_INFO").first.text.should eq 'ratio: 1:2-2.8'
+      XPath.match( doc, "//ARTICLE[GTIN='7680434541015']/COMPOSITIONS/COMPOSITION/SUBSTANCES/SUBSTANCE/CHEMICAL_SUBSTANCE/DOSE_TEXT").first.text.should eq '10-50 mg'
+      XPath.match( doc, "//ARTICLE[GTIN='7680434541015']/COMPOSITIONS/COMPOSITION/SUBSTANCES/SUBSTANCE/CHEMICAL_SUBSTANCE/QTY").first.should eq nil
+      XPath.match( doc, "//ARTICLE[GTIN='7680434541015']/COMPOSITIONS/COMPOSITION/SUBSTANCES/SUBSTANCE/CHEMICAL_SUBSTANCE/UNIT").first.should eq nil
 
-      XPath.match( doc, "//ARTICLE[GTIN='7680434541015']/COMPOSITIONS/COMPOSITION/SUBSTANCES/SUBSTANCE/CHEMICAL_QTY").first.text.should eq '10-50'
-      XPath.match( doc, "//ARTICLE[GTIN='7680434541015']/COMPOSITIONS/COMPOSITION/SUBSTANCES/SUBSTANCE/CHEMICAL_UNIT").first.text.should eq 'mg/100 g'
+      XPath.match( doc, "//ARTICLE[GTIN='7680446250592']/COMPOSITIONS/COMPOSITION/SUBSTANCES/SUBSTANCE/SALTS/SALT/SUBSTANCE_NAME").first.text.should eq 'Ceftriaxonum Natricum'
+
+      XPath.match( doc, "//ARTICLE[GTIN='7680611860045']/NAME").first.text.should eq 'Nutriflex Omega special, Infusionsemulsion 2500 ml'
+      XPath.match( doc, "//ARTICLE[GTIN='7680611860045']/SELLING_UNITS").first.text.should eq '5'
 
     end
   end
@@ -464,14 +470,14 @@ if RunAllTests
     specify { expect(fluticasoni.dose.to_s).to eq  "100 µg/25 mg" }
     lactosum =  info.first.substances.find{ |x| x.name.match(/Lactosum/i) }
     specify { expect(lactosum.name).to eq "Lactosum Monohydricum" }
-    specify { expect(lactosum.dose.to_s).to eq  "25 mg" }
+    specify { expect(lactosum.dose).to eq  nil }
   end
 
   context 'find correct result compositions for stuff with percents' do
     txt = 'calcium carbonicum hahnemanni C7 5 %, chamomilla recutita D5 22.5 %, magnesii hydrogenophosphas trihydricus C5 50 %, passiflora incarnata D5 22.5 %, xylitolum, excipiens ad globulos.'
     info = ParseUtil.parse_compositions(txt)
     specify { expect(info.size).to eq  1 }
-    specify { expect(info.first.substances.size).to eq ExcipiensIs_a_Substance ? 6 : 5 }
+    specify { expect(info.first.substances.size).to eq 5 }
     recutita =  info.first.substances.find{ |x| x.name.match(/recutita/i) }
     specify { expect(recutita.name).to eq  'Chamomilla Recutita D5' }
     specify { expect(recutita.qty.to_f).to eq  22.5 }
@@ -498,11 +504,11 @@ if RunAllTests
                       text
                       )
     specify { expect(info.compositions.size).to eq  2 }
-    specify { expect(info.compositions.first.substances.size).to eq ExcipiensIs_a_Substance ? 7 : 6 }
+    specify { expect(info.compositions.first.substances.size).to eq 6 }
     poloxamerum =  info.compositions.first.substances.find{ |x| x.name.match(/poloxamerum/i) }
     skip { expect(poloxamerum.name).to eq  'Poloxamerum 238' }
     skip { expect(poloxamerum.qty.to_f).to eq  "" }
-    specify { expect(poloxamerum.unit).to eq  "" }
+    specify { expect(poloxamerum.unit).to eq  nil }
   end
 
   context 'find correct result for 61676 Phostal 3-Bäume A): ' do
@@ -566,7 +572,7 @@ Solvens: aqua ad iniectabilia q.s. ad suspensionem pro 1 ml."
                       text
                       )
     specify { expect(info.compositions.size).to eq  2 }
-    specify { expect(info.compositions.first.label).to eq  nil }
+    specify { expect(info.compositions.first.label).to eq  "Praeparatio cryodesiccata:" }
     substance1 =  info.compositions.first.substances.find{ |x| x.name.match(/virus rabiei inactivatu/i) }
     specify { expect(substance1).should_not be  nil }
     if substance1
@@ -612,14 +618,14 @@ Die HILFSSTOFFE sind Aqua ad iniectabilia und Natrii chloridum.
                     text)
     specify { expect(info.pkg_size).to eq '2 x 7' }
     specify { expect(info.selling_units).to eq 14 }
-    specify { expect(info.compositions.first.substances.size).to eq ExcipiensIs_a_Substance ? 3 : 2 }
+    specify { expect(info.compositions.first.substances.size).to eq 2 }
     viscum =  info.compositions.first.substances.find{ |x| x.name.match(/viscum/i) }
     specify { expect(viscum).not_to eq nil}
     natrii =  info.compositions.first.substances.find{ |x| x.name.match(/natrii chloridum/i) }
     specify { expect(natrii).not_to eq nil}
     if viscum
       specify { expect(viscum.name).to eq  'Extractum Aquosum Liquidum Fermentatum 0.05 Mg Ex Viscum Album (mali) Recens' }
-      specify { expect(viscum.is_active_agent).to eq  true }
+      skip { expect(viscum.is_active_agent).to eq  true } # TODO: Grenzfall. Active-Agent viscum album (mali) recens ist sub-string
       specify { expect(viscum.dose.to_s).to eq  '0.01 mg/ml' }
       specify { expect(viscum.qty.to_f).to eq  0.01}
       specify { expect(viscum.unit).to eq  'mg/ml'}
@@ -633,7 +639,7 @@ Die HILFSSTOFFE sind Aqua ad iniectabilia und Natrii chloridum.
                     text)
     specify { expect(info.pkg_size).to eq '2 x 7' }
     specify { expect(info.selling_units).to eq 14 }
-    specify { expect(info.compositions.first.substances.size).to eq ExcipiensIs_a_Substance ? 4 : 3 }
+    specify { expect(info.compositions.first.substances.size).to eq 3 }
     viscum =  info.compositions.first.substances.find{ |x| x.name.match(/viscum/i) }
     specify { expect(viscum).not_to eq nil}
     if viscum