lib/dwc-archive/classification_normalizer.rb in dwc-archive-0.4.13 vs lib/dwc-archive/classification_normalizer.rb in dwc-archive-0.4.14

@@ -25,11 +25,11 @@
     attr_accessor :verbose
     attr_reader :error_names, :tree
 
     def initialize(dwc_instance, verbose = false)
       @dwc = dwc_instance
-      @core = get_fields(@dwc.core)
+      @core_fields = get_fields(@dwc.core)
       @extensions = @dwc.extensions.map { |e| [e, get_fields(e)] }
       @res = {}
       @parser = ParsleyStore.new(1,2)
       @verbose = verbose
       @verbose_count = 10000
@@ -54,11 +54,11 @@
       @res
     end
 
   private
 
-    def canonical_name(a_scientific_name)
+    def get_canonical_name(a_scientific_name)
       if R19
         a_scientific_name.force_encoding('utf-8')
       end
       begin
         parsed_name = @parser.parse(a_scientific_name)[:scientificName]
@@ -82,42 +82,51 @@
     end
     
     def add_synonym_from_core(taxon_id, row)
       taxon = @res[row[taxon_id]] ? @res[row[taxon_id]] : @res[row[taxon_id]] = DarwinCore::TaxonNormalized.new
       taxon.synonyms << SynonymNormalized.new(
-        row[@core[:scientificname]], 
-        canonical_name(row[@core[:scientificname]]), 
-        @core[:taxonomicstatus] ? row[@core[:taxonomicstatus]] : nil)
+        row[@core_fields[:scientificname]], 
+        row[@core_fields[:canonicalname]], 
+        @core_fields[:taxonomicstatus] ? row[@core_fields[:taxonomicstatus]] : nil)
     end
 
+    def set_scientific_name(row, fields)
+      canonical_name = fields[:scientificnameauthorship] ? row[fields[:scientificname]] : get_canonical_name(row[fields[:scientificname]])
+      fields[:canonicalname] = row.size
+      row << canonical_name
+      scientific_name = (fields[:scientificnameauthorship] && row[fields[:scientificnameauthorship]].to_s.strip != '') ? row[fields[:scientificname]].strip + ' ' + row[fields[:scientificnameauthorship]].strip : row[fields[:scientificname]].strip
+      row[fields[:scientificname]] = scientific_name
+    end
+
     def ingest_core
-      raise RuntimeError, "Darwin Core core fields must contain taxon id and scientific name" unless (@core[:id] && @core[:scientificname])
+      raise RuntimeError, "Darwin Core core fields must contain taxon id and scientific name" unless (@core_fields[:id] && @core_fields[:scientificname])
       puts "Reading core information" if @verbose
       rows = @dwc.core.read[0]
       puts "Ingesting information from the core" if @verbose
       rows.each_with_index do |r, i|
         count = i + 1
+        set_scientific_name(r, @core_fields)
         puts "Ingesting %s'th record" % count if @verbose and count % @verbose_count == 0
         #core has AcceptedNameUsageId
-        if @core[:acceptednameusageid] && r[@core[:acceptednameusageid]] && r[@core[:acceptednameusageid]] != r[@core[:id]]
-          add_synonym_from_core(@core[:acceptednameusageid], r)
-        elsif !@core[:acceptednameusageid] && status_synonym?(r[@core[:taxonomicstatus]])
+        if @core_fields[:acceptednameusageid] && r[@core_fields[:acceptednameusageid]] && r[@core_fields[:acceptednameusageid]] != r[@core_fields[:id]]
+          add_synonym_from_core(@core_fields[:acceptednameusageid], r)
+        elsif !@core_fields[:acceptednameusageid] && status_synonym?(r[@core_fields[:taxonomicstatus]])
           add_synonym_from_core(parent_id, r)
         else
-          taxon = @res[r[@core[:id]]] ? @res[r[@core[:id]]] : @res[r[@core[:id]]] = DarwinCore::TaxonNormalized.new
-          taxon.id = r[@core[:id]]
-          taxon.current_name = r[@core[:scientificname]]
-          taxon.current_name_canonical = canonical_name(r[@core[:scientificname]])
+          taxon = @res[r[@core_fields[:id]]] ? @res[r[@core_fields[:id]]] : @res[r[@core_fields[:id]]] = DarwinCore::TaxonNormalized.new
+          taxon.id = r[@core_fields[:id]]
+          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[:taxonrank]] if @core[:taxonrank]
-          taxon.status = r[@core[:taxonomicstatus]] if @core[:taxonomicstatus]
+          taxon.rank = r[@core_fields[:taxonrank]] if @core_fields[:taxonrank]
+          taxon.status = r[@core_fields[:taxonomicstatus]] if @core_fields[:taxonomicstatus]
         end
       end
     end
     
     def parent_id
-      parent_id_field = @core[:highertaxonid] || @core[:parentnameusageid]
+      parent_id_field = @core_fields[:highertaxonid] || @core_fields[:parentnameusageid]
     end
 
     def calculate_classification_path
       @res.each do |taxon_id, taxon|
         next if !taxon.classification_path.empty?
@@ -170,13 +179,14 @@
     def ingest_synonyms(extension)
       puts "Ingesting synonyms extension" if @verbose
       ext, fields = *extension
       ext.read[0].each_with_index do |r, i|
         count = i + 1
+        set_scientific_name(r, fields)
         puts "Ingesting %s'th record" % count if @verbose && count % @verbose_count == 0
         @res[r[fields[:id]]].synonyms << SynonymNormalized.new(
           r[fields[:scientificname]], 
-          canonical_name(r[fields[:scientificname]]), 
+          r[fields[:canonicalname]], 
           fields[:taxonomicstatus] ? r[fields[:taxonomicstatus]] : nil)
       end
     end
 
     def ingest_vernaculars(extension)