family-reunion 0.1.2 → 0.1.3

data/VERSION

@@ -1 +1 @@
-0.1.2
+0.1.3

data/family-reunion.gemspec

@@ -5,11 +5,11 @@
 
 Gem::Specification.new do |s|
   s.name = %q{family-reunion}
-  s.version = "0.1.2"
+  s.version = "0.1.3"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Dmitry Mozzherin", "David Shorthouse"]
-  s.date = %q{2011-06-07}
+  s.date = %q{2011-06-14}
   s.description = %q{An algorithm to merge related nodes of two taxonomic hierarchies with synonym information}
   s.email = %q{dmozzherin@gmail.com}
   s.extra_rdoc_files = [

data/lib/family-reunion.rb

@@ -1,3 +1,4 @@
+require 'logger'
 require 'json'
 require 'taxamatch_rb'
 require 'family-reunion/cache'
@@ -15,6 +16,24 @@ class FamilyReunion
   attr :primary_valid_names_set, :secondary_valid_names_set
   attr :primary_synonyms_set, :secondary_synonyms_set
 
+  VERSION = open(File.join(File.dirname(__FILE__), '..', 'VERSION')).readline.strip
+
+  def self.logger
+    @@logger ||= Logger.new(nil)
+  end
+
+  def self.logger=(logger)
+    @@logger = logger
+  end
+
+  def self.logger_reset
+    self.logger = Logger.new(nil)
+  end
+
+  def self.logger_write(obj_id, message, method = :info)
+    self.logger.send(method, "|%s|%s|" % [obj_id, message])
+  end
+
   def initialize(primary_node, secondary_node)
     @primary_node = FamilyReunion::TopNode.new(primary_node)
     @secondary_node = FamilyReunion::TopNode.new(secondary_node)
@@ -29,20 +48,24 @@ class FamilyReunion
     merge_exact_matches
     merge_fuzzy_matches if with_fuzzy_matching
     merge_no_matches
+    FamilyReunion.logger_write(self.object_id, "Merging is complete")
     @merges
   end
 
   private
 
   def merge_exact_matches
+    FamilyReunion.logger_write(self.object_id, "Started merging of exact matches")
     ExactMatcher.new(self).merge
   end
 
   def merge_fuzzy_matches
+    FamilyReunion.logger_write(self.object_id, "Started merging of fuzzy matches")
     FuzzyMatcher.new(self).merge
   end
-  
+
   def merge_no_matches
+    FamilyReunion.logger_write(self.object_id, "Started gap filling, adding new species and uninomials")
     NomatchOrganizer.new(self).merge
   end
 

data/lib/family-reunion/exact_matcher.rb

@@ -7,9 +7,13 @@ class FamilyReunion
     end
 
     def merge
+      FamilyReunion.logger_write(@fr.object_id, "Merging exact matches of accepted names")
       add_valid_matches(get_valid_matches)
+      FamilyReunion.logger_write(@fr.object_id, "Merging exact matches of accepted names to synonyms")
       add_synonym_matches(get_valid_to_synonym_matches, :valid_to_synonym)
+      FamilyReunion.logger_write(@fr.object_id, "Merging exact matches of synonyms to accepted names")
       add_synonym_matches(get_synonym_to_valid_matches, :synonym_to_valid)
+      FamilyReunion.logger_write(@fr.object_id, "Merging exact matches of synonyms")
       add_synonym_matches(get_synonym_to_synonym_matches, :synonym_to_synonym)
     end
 
@@ -23,9 +27,9 @@ class FamilyReunion
       # Homonyms are treated separately, and are not matched by the algorithm,
       # they are excluded from valid_matches
       valid_matches.each do |name|
-        primary_id = @fr.primary_node.valid_names_hash[name][:id]
-        secondary_id = @fr.secondary_node.valid_names_hash[name][:id]
-        @fr.merges[primary_id] = {:matches => {secondary_id.to_s => {:match_type => :valid_to_valid}}, :nonmatches => []}
+        primary_id = @fr.primary_node.valid_names_hash[name][:id].to_s.to_sym
+        secondary_id = @fr.secondary_node.valid_names_hash[name][:id].to_s.to_sym
+        @fr.merges[primary_id] = {:matches => {secondary_id => {:match_type => :valid_to_valid}}, :nonmatches => []}
       end
     end
 
@@ -36,7 +40,7 @@ class FamilyReunion
     def get_synonym_to_valid_matches
       @fr.primary_synonyms_set & @fr.secondary_valid_names_set
     end
-    
+
     def get_synonym_to_synonym_matches
       @fr.primary_synonyms_set & @fr.secondary_synonyms_set
     end
@@ -61,9 +65,9 @@ class FamilyReunion
       valid_names = node.valid_names_hash
       synonyms = node.synonyms_hash
       if valid_names.has_key?(name)
-        return [valid_names[name][:id]]
+        return [valid_names[name][:id].to_s.to_sym]
       else
-        return synonyms[name].map {|n| n[:id]}
+        return synonyms[name].map {|n| n[:id].to_s.to_sym}
       end
     end
 

data/lib/family-reunion/fuzzy_matcher.rb

@@ -8,9 +8,13 @@ class FamilyReunion
     end
 
     def merge
+      FamilyReunion.logger_write(@fr.object_id, "Merging fuzzy matches of accepted names")
       add_matches(get_valid_matches, :fuzzy_valid_to_valid)
+      FamilyReunion.logger_write(@fr.object_id, "Merging fuzzy matches of accepted names to synonyms")
       add_matches(get_valid_to_synonym_matches, :fuzzy_valid_to_synonym)
+      FamilyReunion.logger_write(@fr.object_id, "Merging fuzzy matches of synonyms to accepted names")
       add_matches(get_synonym_to_valid_matches, :fuzzy_synonym_to_valid)
+      FamilyReunion.logger_write(@fr.object_id, "Merging fuzzy matches of synonyms")
       add_matches(get_synonym_to_synonym_matches, :fuzzy_synonym_to_synonym)
     end
 
@@ -42,8 +46,8 @@ class FamilyReunion
 
     def add_matches(matched_nodes, match_type)
       matched_nodes.each do |primary_node, secondary_nodes|
-        primary_id = primary_node[:id]
-        secondary_ids = secondary_nodes.map { |n| n[:id] }
+        primary_id = primary_node[:id].to_s.to_sym
+        secondary_ids = secondary_nodes.map { |n| n[:id].to_s.to_sym }
         secondary_id_matches = format_secondary_id_for_merge(secondary_ids, match_type)
         add_record_to_merges(primary_id, secondary_id_matches)
       end

data/lib/family-reunion/matcher_helper.rb

@@ -3,13 +3,14 @@ class FamilyReunion
     private
     def format_secondary_id_for_merge(secondary_ids, match_type)
       secondary_ids.inject({}) do |res, i|
-        i = i.to_s
+        raise "Secondary id is not a symbol" unless i.is_a?(Symbol)
         res[i] = {:match_type => match_type} unless res.has_key?(i)
         res
       end
     end
 
     def add_record_to_merges(primary_id, secondary_id_matches)
+      raise "Primary id is not a symbol" unless primary_id.is_a?(Symbol)
       if @fr.merges.has_key?(primary_id)
         secondary_id_matches.each do |key, val|
           @fr.merges[primary_id][:matches][key] = val unless @fr.merges[primary_id][:matches].has_key?(key)

data/lib/family-reunion/nomatch_organizer.rb

@@ -7,6 +7,7 @@ class FamilyReunion
     end
 
     def merge
+      FamilyReunion.logger_write(@fr.object_id, "Filling gaps with new taxa")
       organize_nonmatches(get_nomach_secondary_ids)
     end
 

data/lib/family-reunion/top_node.rb

@@ -76,7 +76,7 @@ class FamilyReunion
 
     def update_paths_hash(node)
       path = node[:path].map { |n| n.to_sym }
-      path_ids = node[:path_ids].map { |i| i.to_sym }
+      path_ids = node[:path_ids].map { |i| i.to_s.to_sym }
       until path.empty?
         populate_paths_hash(path, path_ids)
         path.pop

data/spec/family-reunion_spec.rb

@@ -6,6 +6,7 @@ describe FamilyReunion do
   end
 
   it "should generate instances of nodes" do
+    FamilyReunion.logger = Logger.new($stdout)
     @fr.primary_node.is_a?(FamilyReunion::TopNode).should be_true
     @fr.secondary_node.is_a?(FamilyReunion::TopNode).should be_true
   end

metadata

@@ -5,8 +5,8 @@ version: !ruby/object:Gem::Version
   segments: 
   - 0
   - 1
-  - 2
-  version: 0.1.2
+  - 3
+  version: 0.1.3
 platform: ruby
 authors: 
 - Dmitry Mozzherin
@@ -15,7 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-06-07 00:00:00 -04:00
+date: 2011-06-14 00:00:00 -04:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -233,7 +233,7 @@ required_ruby_version: !ruby/object:Gem::Requirement
   requirements: 
   - - ">="
     - !ruby/object:Gem::Version 
-      hash: 1186928035501677277
+      hash: 4425929328463574267
       segments: 
       - 0
       version: "0"