classification_normalizer.rb in dwc-archive-0.4.4

- old
+ new

@@ -16,19 +16,22 @@
 
   class SynonymNormalized < Struct.new(:name, :canonical_name, :status);end
   class VernacularNormalized < Struct.new(:name, :language);end
 
   class ClassificationNormalizer
+    attr_accessor :verbose
+    attr_reader :error_names
 
     def initialize(dwc_instance, verbose = false)
       @dwc = dwc_instance
       @core = get_fields(@dwc.core)
       @extensions = @dwc.extensions.map { |e| [e, get_fields(e)] }
       @res = {}
       @parser = ParsleyStore.new(1,2)
       @verbose = verbose
-      @verbose_count = 1000
+      @verbose_count = 10000
+      @error_names = []
     end
 
     def normalize
       injest_core
       calculate_classification_path
@@ -99,20 +102,30 @@
     end
 
     def calculate_classification_path
       @res.each do |taxon_id, taxon|
         next if taxon.classification_path 
-        get_classification_path(taxon)
+        begin
+          get_classification_path(taxon)
+        rescue DarwinCore::ParentNotCurrentError
+          next
+        end
       end
     end
 
     def get_classification_path(taxon)
       return if taxon.classification_path
       if DarwinCore.nil_field?(taxon.parent_id)
         taxon.classification_path = [taxon.current_name_canonical]
       else
-        parent_cp = @res[taxon.parent_id].classification_path
+        begin
+          parent_cp = @res[taxon.parent_id].classification_path
+        rescue NoMethodError #name has a parent which is not a current name
+          error = "The parent of the taxon \'#{taxon.current_name}\' is deprecated"
+          @error_names << {:name => taxon, :error => error}
+          raise DarwinCore::ParentNotCurrentError, error
+        end
         if parent_cp
           taxon.classification_path = parent_cp + [taxon.current_name_canonical]
         else
           get_classification_path(@res[taxon.parent_id]) 
           taxon.classification_path = @res[taxon.parent_id].classification_path + [taxon.current_name_canonical]
@@ -153,6 +166,7 @@
       end
     end
 
   end
 end
+
 