Fix handling several substances

lib/oddb2xml/parslet_compositions.rb

@@ -162,21 +162,20 @@ class SubstanceParser < DoseParser
   rule(:substance_residui) { str('residui:')    >> space >> substance }
   rule(:substance_conserv) { str('conserv.:')   >> space >> substance }
   rule(:substance_corresp) { substance.as(:substance) >>str('corresp.') >> space >>  (str('suspensio reconstituta') >> space).maybe >>
-                             (substance_et | substance).as(:substance_corresp)  }
+                             (substance).as(:substance_corresp)  }
   rule(:substance_ut) { substance.as(:substance_ut) >> space >> str('ut') >> space >> substance }
-  rule(:substance_et) { (substance.as(:substance) >> str('et') >> space).repeat(1) >> (substance_corresp | substance).as(:substance) }
   rule(:praeparatio) { ((identifier >> space?).repeat(1).as(:description) >> str(':') >> space).maybe>>
                        (identifier >> space?).repeat(1).as(:substance_name) >>
                         number.as(:qty) >> space >> str('U.:') >> space >>
                         (identifier >> space?).repeat(1).as(:more_info) >>
                         space?
                        }
 
-  #rule(:one_substance) { (str(',') >> space).maybe >> (excipiens) } # Sometimes it is handy for debugging to be able to debug just one the different variants
-  rule(:one_substance) { (str(',') >> space).maybe >> (der | excipiens | praeparatio | histamin | named_substance | substance_residui | substance_conserv | substance_et | substance_ut | substance_corresp | substance ) }
-#  rule(:one_substance) { (str(',') >> space).maybe >> ( substance | excipiens) }
-  rule(:one_substance) { ( substance_et | praeparatio | substance | excipiens) >> (str(',') >> space).maybe  }
-  rule(:all_substances) { one_substance.repeat(1) }
+  rule(:substance_separator) { (comma | str('et')) >> space? }
+# rule(:one_substance) { (der | excipiens | praeparatio | histamin | named_substance | substance_residui | substance_conserv | substance_ut | substance_corresp | substance ) >> substance_separator }
+# rule(:one_substance) { excipiens >> substance_separator.maybe} # Sometimes it is handy for debugging to be able to debug just one the different variants
+  rule(:one_substance) { praeparatio | substance_corresp | substance }
+  rule(:all_substances) { (one_substance >> substance_separator.maybe).repeat(1) }
   root :all_substances
 end
 

spec/parslet_spec.rb

@@ -9,7 +9,7 @@
 require "#{Dir.pwd}/lib/oddb2xml/parslet_compositions"
 require 'parslet/rig/rspec'
 
-RunAllCompositionsTests = false # takes over a minute!
+RunAllCompositionsTests = false # takes over two minutes!
 RunFailingSpec = false
 RunExcipiensTest = true
 RunDoseTests = true
@@ -307,9 +307,9 @@ def run_substance_tests(hash_string_to_name)
     }
   run_substance_tests(failing_tests)   if RunFailingSpec
   run_substance_tests(excipiens_tests) if RunExcipiensTest
-  # run_substance_tests(tests)           if RunAllTests
+  run_substance_tests(tests)           if RunFailingSpec
   run_composition_tests( ["acari allergeni extractum 50'000 U.",
-                          "pollinis allergeni extractum 50'000 U.: fraxinus excelsior, conserv.: phenolum, excipiens ad solutionem pro 1 ml."]) if false
+                          "pollinis allergeni extractum 50'000 U.: fraxinus excelsior, conserv.: phenolum, excipiens ad solutionem pro 1 ml."])
   if RunMostImportantParserTests
     context "should return correct substance for given with et (IKSNR 11879)" do
       string = "calcii lactas pentahydricus 25 mg et calcii hydrogenophosphas anhydricus 300 mg"
@@ -329,6 +329,7 @@ def run_substance_tests(hash_string_to_name)
       string = "calcii lactas pentahydricus 25 mg et calcii hydrogenophosphas anhydricus 300 mg corresp. calcium 100 mg"
 
       composition = ParseComposition.from_string(string)
+      # pp composition.substances; binding.pry
       specify { expect(composition.substances.size).to eq 3 }
       calcium = composition.substances.find{ |x| /calcium/i.match(x.name) }
       pentahydricus = composition.substances.find{ |x| /pentahydricus/i.match(x.name) }
@@ -499,8 +500,6 @@ def run_substance_tests(hash_string_to_name)
     source =
 "I) DTPa-IPV-Komponente (Suspension): toxoidum diphtheriae 30 U.I., toxoidum pertussis 25 µg et haemagglutininum filamentosum 25 µg"
     composition = ParseComposition.from_string(source)
-    pp composition
-    # binding.pry
     specify { expect(composition.source).to eq source }
 
     specify { expect(composition.label).to eq 'I' }