dwc-archive 0.4.10 → 0.4.11

data/VERSION

@@ -1 +1 @@
-0.4.10
+0.4.11

data/features/dwca-reader.feature

@@ -61,3 +61,5 @@ Feature: Reading of a Darwing Core Archive
     Then I am able to use DarwinCore#normalize_classification method
     And get normalized classification in expected format
     And there are paths, synonyms and vernacular names in normalized classification
+    And names used in classification can be accessed by "name_strings" method
+    And nodes_ids organized in trees can be accessed by "tree" method

data/features/step_definitions/dwc-reader_steps.rb

@@ -162,7 +162,8 @@ Then /^I can read extensions content using block$/ do
 end
 
 Then /^I am able to use DarwinCore\#normalize_classification method$/ do
-  @normalized_classification = @dwc.normalize_classification
+  @cn = DarwinCore::ClassificationNormalizer.new(@dwc)
+  @normalized_classification = @cn.normalize
 end
 
 Then /^get normalized classification in expected format$/ do
@@ -191,3 +192,27 @@ Then /^there are paths, synonyms and vernacular names in normalized classificati
   @vernaculars_are_generated.should be_true
   @synonyms_are_generated.should be_true
 end
+
+Then /^names used in classification can be accessed by "([^"]*)" method$/ do |name_strings|
+  names = @cn.send(name_strings.to_sym) 
+  names.size.should > @normalized_classification.size
+end
+
+
+Then /^nodes_ids organized in trees can be accessed by "([^"]*)" method$/ do |tree|
+  def flatten_tree(data, keys)
+    data.each do |k, v|
+      keys << k
+      if v != {}
+        debugger if v.class != Hash
+        flatten_tree(v, keys)
+      end
+    end
+  end
+  tree = @cn.send(tree.to_sym)
+  tree.class.should == Hash
+  keys = []
+  flatten_tree(tree, keys) 
+  @normalized_classification.size.should == keys.size
+end
+

data/lib/dwc-archive/classification_normalizer.rb

@@ -4,13 +4,14 @@ require 'parsley-store'
 class DarwinCore 
     
   class TaxonNormalized
-    attr_accessor :id, :parent_id, :classification_path, :current_name, :current_name_canonical, :synonyms, :vernacular_names, :rank, :status
+    attr_accessor :id, :parent_id, :classification_path_id, :classification_path, :current_name, :current_name_canonical, :synonyms, :vernacular_names, :rank, :status
 
     def initialize
       @id = @parent_id = @rank = @status = nil
       @current_name = ''
       @current_name_canonical = ''
       @classification_path = []
+      @classification_path_id = []
       @synonyms = []
       @vernacular_names = []
     end
@@ -22,7 +23,7 @@ class DarwinCore
 
   class ClassificationNormalizer
     attr_accessor :verbose
-    attr_reader :error_names
+    attr_reader :error_names, :tree
 
     def initialize(dwc_instance, verbose = false)
       @dwc = dwc_instance
@@ -32,7 +33,17 @@ class DarwinCore
       @parser = ParsleyStore.new(1,2)
       @verbose = verbose
       @verbose_count = 10000
+      @name_strings = {}
       @error_names = []
+      @tree = {}
+    end
+
+    def add_name_string(name_string)
+      @name_strings[name_string] = 1 unless @name_strings[name_string]
+    end
+
+    def name_strings
+      @name_strings.keys
     end
 
     def normalize
@@ -55,6 +66,8 @@ class DarwinCore
         @parser = ParsleyStore.new(1,2)
         parsed_name = @parser.parse(a_scientific_name)[:scientificName]
       end
+      add_name_string(a_scientific_name)
+      add_name_string(parsed_name[:canonical]) if parsed_name[:parsed]
       parsed_name[:parsed] ? parsed_name[:canonical] : a_scientific_name
     end
     
@@ -118,8 +131,11 @@ class DarwinCore
 
     def get_classification_path(taxon)
       return if !taxon.classification_path.empty?
+      current_node = {taxon.id => {}}
       if DarwinCore.nil_field?(taxon.parent_id)
         taxon.classification_path << taxon.current_name_canonical
+        taxon.classification_path_id << taxon.id
+        @tree.merge!(current_node)
       else
         begin
           parent_cp = @res[taxon.parent_id].classification_path
@@ -131,8 +147,14 @@ class DarwinCore
         if parent_cp.empty?
           get_classification_path(@res[taxon.parent_id]) 
           taxon.classification_path += @res[taxon.parent_id].classification_path + [taxon.current_name_canonical]
+          taxon.classification_path_id += @res[taxon.parent_id].classification_path_id + [taxon.id]
+          parent_node = @res[taxon.parent_id].classification_path_id.inject(@tree) {|node, id| node[id]}
+          parent_node.merge!(current_node)
         else
           taxon.classification_path += parent_cp + [taxon.current_name_canonical]
+          taxon.classification_path_id += @res[taxon.parent_id].classification_path_id + [taxon.id]
+          parent_node = @res[taxon.parent_id].classification_path_id.inject(@tree) {|node, id| node[id]}
+          parent_node.merge!(current_node)
         end
       end
     end
@@ -167,6 +189,7 @@ class DarwinCore
         @res[r[fields[:id]]].vernacular_names << VernacularNormalized.new(
           r[fields[:vernacularname]],
           fields[:languagecode] ? r[fields[:languagecode]] : nil)
+        add_name_string(r[fields[:vernacularname]])
       end
     end
 

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: dwc-archive
 version: !ruby/object:Gem::Version 
-  hash: 27
+  hash: 25
   prerelease: false
   segments: 
   - 0
   - 4
-  - 10
-  version: 0.4.10
+  - 11
+  version: 0.4.11
 platform: ruby
 authors: 
 - Dmitry Mozzherin
@@ -15,7 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-09-18 00:00:00 -04:00
+date: 2010-09-21 00:00:00 -04:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency