dwc-archive 0.4.3 → 0.4.4

data/VERSION

@@ -1 +1 @@
-0.4.3
+0.4.4

data/lib/dwc-archive.rb

@@ -28,7 +28,7 @@ class DarwinCore
   
   VERSION = open(File.join(File.dirname(__FILE__), '..', 'VERSION')).readline.strip
 
-  attr_reader :archive, :core, :metadata, :extensions
+  attr_reader :archive, :core, :metadata, :extensions, :classification_normalizer
   alias :eml :metadata
   
   DEFAULT_TMP_DIR = "/tmp"
@@ -57,7 +57,8 @@ class DarwinCore
   # generates a hash from a classification data with path to each node, list of synonyms and vernacular names.
   def normalize_classification(verbose = false)
     return nil unless has_parent_id?
-    DarwinCore::ClassificationNormalizer.new(self, verbose).normalize
+    @classification_normalizer ||= DarwinCore::ClassificationNormalizer.new(self, verbose)
+    @classification_normalizer.normalize
   end
 
   def has_parent_id?

data/lib/dwc-archive/classification_normalizer.rb

@@ -18,6 +18,8 @@ class DarwinCore
   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
@@ -26,7 +28,8 @@ class DarwinCore
       @res = {}
       @parser = ParsleyStore.new(1,2)
       @verbose = verbose
-      @verbose_count = 1000
+      @verbose_count = 10000
+      @error_names = []
     end
 
     def normalize
@@ -101,7 +104,11 @@ class DarwinCore
     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
 
@@ -110,7 +117,13 @@ class DarwinCore
       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
@@ -156,3 +169,4 @@ class DarwinCore
   end
 end
 
+

data/lib/dwc-archive/errors.rb

@@ -6,4 +6,5 @@ class DarwinCore
   class CoreFileError < Error; end
   class ExtensionFileError < Error; end
   class GeneratorError < Error; end
+  class ParentNotCurrentError < Error; end
 end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: dwc-archive
 version: !ruby/object:Gem::Version 
-  hash: 9
+  hash: 7
   prerelease: false
   segments: 
   - 0
   - 4
-  - 3
-  version: 0.4.3
+  - 4
+  version: 0.4.4
 platform: ruby
 authors: 
 - Dmitry Mozzherin
@@ -15,7 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-09-13 00:00:00 -04:00
+date: 2010-09-14 00:00:00 -04:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency