classification_normalizer.rb in dwc-archive-0.7.3

- old
+ new

@@ -58,11 +58,11 @@
   private
 
     def get_canonical_name(a_scientific_name)
       a_scientific_name.force_encoding('utf-8')
       canonical_name = @parser.parse(a_scientific_name, :canonical_only => true)
-      canonical_name.to_s.empty? ? a_scientific_name : canonical_name.force_encoding('utf-8')
+      canonical_name.to_s.empty? ? a_scientific_name : canonical_name
     end
     
     def get_fields(element)
       data = element.fields.inject({}) { |res, f| res[f[:term].split('/')[-1].downcase.to_sym] = f[:index].to_i; res }
       data[:id] = element.id[:index] 
@@ -87,14 +87,14 @@
     end
 
     def set_scientific_name(row, fields)
       row[fields[:scientificname]] = 'N/A' unless row[fields[:scientificname]]
       canonical_name = ''
-      scientific_name = row[fields[:scientificname]].strip
+      scientific_name = row[fields[:scientificname]].strip.force_encoding('utf-8')
       if separate_canonical_and_authorship?(row, fields)
-        canonical_name = row[fields[:scientificname]].strip
-        scientific_name += " #{row[fields[:scientificnameauthorship]].strip}"
+        canonical_name = row[fields[:scientificname]].strip.force_encoding('utf-8')
+        scientific_name += " #{row[fields[:scientificnameauthorship]].strip.force_encoding('utf-8')}"
       else
         canonical_name = get_canonical_name(row[fields[:scientificname]])
       end
       fields[:canonicalname] = row.size
       row << canonical_name
@@ -122,12 +122,12 @@
           elsif !@core_fields[:acceptednameusageid] && @core_fields[:taxonomicstatus] && status_synonym?(r[@core_fields[:taxonomicstatus]])
             add_synonym_from_core(parent_id, r)
           else
             taxon = @normalized_data[r[@core_fields[:id]]] ? @normalized_data[r[@core_fields[:id]]] : @normalized_data[r[@core_fields[:id]]] = DarwinCore::TaxonNormalized.new
             taxon.id = r[@core_fields[:id]]
-            taxon.current_name = r[@core_fields[:scientificname]].force_encoding("utf-8")
-            taxon.current_name_canonical = r[@core_fields[:canonicalname]].force_encoding("utf-8")
+            taxon.current_name = r[@core_fields[:scientificname]]
+            taxon.current_name_canonical = r[@core_fields[:canonicalname]]
             taxon.parent_id = r[parent_id] 
             taxon.rank = r[@core_fields[:taxonrank]] if @core_fields[:taxonrank]
             taxon.status = r[@core_fields[:taxonomicstatus]] if @core_fields[:taxonomicstatus]
             add_name_string(taxon.current_name)
             add_name_string(taxon.current_name_canonical)
@@ -246,7 +246,5 @@
       end
     end
 
   end
 end
-
-