dwc-archive 0.4.2 → 0.4.3

data/VERSION

@@ -1 +1 @@
-0.4.2
+0.4.3

data/lib/dwc-archive.rb

@@ -25,6 +25,9 @@ require 'dwc-archive/generator_eml_xml'
 require 'dwc-archive/classification_normalizer'
 
 class DarwinCore
+  
+  VERSION = open(File.join(File.dirname(__FILE__), '..', 'VERSION')).readline.strip
+
   attr_reader :archive, :core, :metadata, :extensions
   alias :eml :metadata
   
@@ -52,9 +55,9 @@ class DarwinCore
   end
 
   # generates a hash from a classification data with path to each node, list of synonyms and vernacular names.
-  def normalize_classification
+  def normalize_classification(verbose = false)
     return nil unless has_parent_id?
-    DarwinCore::ClassificationNormalizer.new(self).normalize
+    DarwinCore::ClassificationNormalizer.new(self, verbose).normalize
   end
 
   def has_parent_id?

data/lib/dwc-archive/classification_normalizer.rb

@@ -1,8 +1,8 @@
 # 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
 
@@ -18,12 +18,15 @@ class DarwinCore
   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
@@ -42,7 +45,7 @@ class DarwinCore
       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
@@ -63,12 +66,17 @@ class DarwinCore
       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)
@@ -102,7 +110,7 @@ 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
+        parent_cp = @res[taxon.parent_id].classification_path
         if parent_cp
           taxon.classification_path = parent_cp + [taxon.current_name_canonical]
         else
@@ -121,8 +129,11 @@ class DarwinCore
     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]]), 
@@ -131,8 +142,11 @@ class DarwinCore
     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)
@@ -141,3 +155,4 @@ class DarwinCore
 
   end
 end
+

data/spec/lib/dwc-archive_spec.rb

@@ -5,6 +5,12 @@ describe DarwinCore do
     @file_dir = File.join(File.dirname(__FILE__), '..', 'files')
   end
 
+  describe "VERSION" do
+    it "should return VERSION number" do
+      DarwinCore::VERSION.split('.').join('').to_i.should > 41
+    end
+  end
+
   describe "::nil_field?" do
     it "should return true for entries which normally mean nil" do
       [nil, '/N', ''].each do |i|

metadata

@@ -1,12 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: dwc-archive
 version: !ruby/object:Gem::Version 
+  hash: 9
   prerelease: false
   segments: 
   - 0
   - 4
-  - 2
-  version: 0.4.2
+  - 3
+  version: 0.4.3
 platform: ruby
 authors: 
 - Dmitry Mozzherin
@@ -14,7 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-09-10 00:00:00 -04:00
+date: 2010-09-13 00:00:00 -04:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -25,6 +26,7 @@ dependencies:
     requirements: 
     - - ">="
       - !ruby/object:Gem::Version 
+        hash: 13
         segments: 
         - 1
         - 2
@@ -40,6 +42,7 @@ dependencies:
     requirements: 
     - - ">="
       - !ruby/object:Gem::Version 
+        hash: 3
         segments: 
         - 0
         version: "0"
@@ -112,6 +115,7 @@ required_ruby_version: !ruby/object:Gem::Requirement
   requirements: 
   - - ">="
     - !ruby/object:Gem::Version 
+      hash: 3
       segments: 
       - 0
       version: "0"
@@ -120,6 +124,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
   requirements: 
   - - ">="
     - !ruby/object:Gem::Version 
+      hash: 3
       segments: 
       - 0
       version: "0"