dwc-archive 0.8.0 → 0.8.1

data/CHANGELOG

@@ -1 +1,3 @@
+0.8.1 Linnean classification path is now only for species and infraspecies with canonical forms. It ends with a canonical form of the taxon
+
 0.8.0 Added linnean classification path to normalized data from DwCA. It consists of data associated with clades like 'kingdom', 'order' etc.

data/VERSION

@@ -1 +1 @@
-0.8.0
+0.8.1

data/lib/dwc-archive/classification_normalizer.rb

@@ -166,8 +166,8 @@ class DarwinCore
             taxon.status = r[@core_fields[:taxonomicstatus]] if @core_fields[:taxonomicstatus]
             taxon.source = r[@core_fields[:source]] if @core_fields[:source]
             taxon.local_id = r[@core_fields[:localid]] if @core_fields[:localid]
-            taxon.linnean_classification_path = get_linnean_classification_path(r)
             taxon.global_id = r[@core_fields[:globalid]] if @core_fields[:globalid]
+            taxon.linnean_classification_path = get_linnean_classification_path(r, taxon)
             add_name_string(taxon.current_name)
             add_name_string(taxon.current_name_canonical) if taxon.current_name_canonical && !taxon.current_name_canonical.empty?
           end
@@ -313,12 +313,20 @@ class DarwinCore
         end
       end
     end
-
-    def get_linnean_classification_path(row)
+    
+    #Collect linnean classification path only on species level
+    def get_linnean_classification_path(row, taxon)
       res = []
       [:kingdom, :phylum, :class, :order, :family, :genus, :subgenus].each do |clade|
         res << [row[@core_fields[clade]], clade] if @core_fields[clade] 
       end
+      if !res.empty?
+        if taxon.current_name_canonical && taxon.current_name_canonical.split(" ").size > 1 && taxon.current_name != res[-1][0]
+          res << [taxon.current_name_canonical, taxon.rank]
+        else
+          res = []
+        end
+      end
       res
     end
 

data/spec/lib/dwc-archive_spec.rb

@@ -157,7 +157,7 @@ describe DarwinCore do
       dwc = DarwinCore.new(file)
       cn = DarwinCore::ClassificationNormalizer.new(dwc)
       norm = cn.normalize
-      cn.normalized_data.first.last.linnean_classification_path.should == [["Animalia", :kingdom], ["Arthropoda", :phylum], ["Insecta", :class], ["Diptera", :order], ["Cecidomyiidae", :family], ["Resseliella", :genus]]
+      cn.normalized_data.first.last.linnean_classification_path.should == [["Animalia", :kingdom], ["Arthropoda", :phylum], ["Insecta", :class], ["Diptera", :order], ["Cecidomyiidae", :family], ["Resseliella", :genus], ["Resseliella theobaldi", "species"]]
     end
 
     it "should keep linnean classification empty if terms are not there" do

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: dwc-archive
 version: !ruby/object:Gem::Version
-  version: 0.8.0
+  version: 0.8.1
   prerelease: 
 platform: ruby
 authors: