lib/dwc-archive/classification_normalizer.rb in dwc-archive-0.4.2 vs lib/dwc-archive/classification_normalizer.rb in dwc-archive-0.4.3

@@ -1,10 +1,10 @@
 # encoding: utf-8
-require 'biodiversity'
+require 'parsley-store'
 
 class DarwinCore 
-  
+    
   class TaxonNormalized
     attr_accessor :id, :parent_id, :classification_path, :current_name, :current_name_canonical, :synonyms, :vernacular_names, :rank, :status
 
     def initialize
       @id = @parent_id = @classification_path = @current_name = @current_name_canonical = @rank = @status = nil
@@ -16,16 +16,19 @@
 
   class SynonymNormalized < Struct.new(:name, :canonical_name, :status);end
   class VernacularNormalized < Struct.new(:name, :language);end
 
   class ClassificationNormalizer
-    def initialize(dwc_instance)
+
+    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 = ScientificNameParser.new
+      @parser = ParsleyStore.new(1,2)
+      @verbose = verbose
+      @verbose_count = 1000
     end
 
     def normalize
       injest_core
       calculate_classification_path
@@ -40,11 +43,11 @@
         a_scientific_name.force_encoding('utf-8')
       end
       begin
         parsed_name = @parser.parse(a_scientific_name)[:scientificName]
       rescue
-        @parser = ScientificNameParser.new
+        @parser = ParsleyStore.new(1,2)
         parsed_name = @parser.parse(a_scientific_name)[:scientificName]
       end
       parsed_name[:parsed] ? parsed_name[:canonical] : a_scientific_name
     end
     
@@ -61,16 +64,21 @@
     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]]), 
-        row[@core[:taxonomicstatus]])
+        @core[:taxonomicstatus] ? row[@core[:taxonomicstatus]] : nil)
     end
 
     def injest_core
       raise RuntimeError, "Darwin Core core fields must contain taxon id and scientific name" unless (@core[:id] && @core[:scientificname])
-      @dwc.core.read[0].each do |r|
+      puts "Reading core information" if @verbose
+      rows = @dwc.core.read[0]
+      puts "Injesting information from the core" if @verbose
+      rows.each_with_index do |r, i|
+        count = i + 1
+        puts "Injesting %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]])
           add_synonym_from_core(parent_id, r)
@@ -100,11 +108,11 @@
     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
+        parent_cp = @res[taxon.parent_id].classification_path
         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]
@@ -119,25 +127,32 @@
         injest_vernaculars(e) if fields.keys.include? :vernacularname
       end
     end
     
     def injest_synonyms(extension)
+      puts "Injesting synonyms extension" if @verbose
       ext, fields = *extension
-      ext.read[0].each do |r|
+      ext.read[0].each_with_index do |r, i|
+        count = i + 1
+        puts "Injesting %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]]), 
           fields[:taxonomicstatus] ? r[fields[:taxonomicstatus]] : nil)
       end
     end
 
     def injest_vernaculars(extension)
+      puts "Injesting vernacular names" if @verbose
       ext, fields = *extension
-      ext.read[0].each do |r|
+      ext.read[0].each_with_index do |r, i|
+        count = i + 1
+        puts "Injesting %s'th record" % count if @verbose && count % @verbose_count == 0
         @res[r[fields[:id]]].vernacular_names << VernacularNormalized.new(
           r[fields[:vernacularname]],
           fields[:languagecode] ? r[fields[:languagecode]] : nil)
       end
     end
 
   end
 end
+