lib/dwc-archive/classification_normalizer.rb in dwc-archive-0.7.4 vs lib/dwc-archive/classification_normalizer.rb in dwc-archive-0.7.5

@@ -30,18 +30,23 @@
       @extensions = @dwc.extensions.map { |e| [e, get_fields(e)] }
       @normalized_data = {}
       @synonyms = {}
       @parser = ParsleyStore.new(1,2)
       @name_strings = {}
+      @vernacular_name_strings = {}
       @error_names = []
       @tree = {}
     end
 
     def add_name_string(name_string)
       @name_strings[name_string] = 1 unless @name_strings[name_string]
     end
 
+    def add_vernacular_name_string(name_string)
+      @vernacular_name_strings[name_string] = 1 unless @vernacular_name_strings[name_string]
+    end
+
     def name_strings
       @name_strings.keys
     end
 
     def normalize(opts = {:with_canoical_names => true})
@@ -111,12 +116,10 @@
         authorship = row[fields[:scientificnameauthorship]].to_s.strip
       end
       !(authorship.empty? || row[fields[:scientificname]].index(authorship))
     end
 
-
-
     def ingest_core
       @normalized_data = {}
       raise RuntimeError, "Darwin Core core fields must contain taxon id and scientific name" unless (@core_fields[:id] && @core_fields[:scientificname])
       @dwc.core.read do |rows|
         rows[1].each do |error|
@@ -256,10 +259,10 @@
           vernacular = VernacularNormalized.new(
             r[fields[:vernacularname]],
             language,
             locality)
           @normalized_data[r[fields[:id]]].vernacular_names << vernacular
-          add_name_string(vernacular.name)
+          add_vernacular_name_string(vernacular.name)
         end
       end
     end
 
   end