dimus-taxamatch_rb 0.5.3 → 0.5.5

data/lib/taxamatch_rb.rb

@@ -21,10 +21,11 @@ module Taxamatch
    
    
     #takes two scientific names and returns true if names match and false if they don't
-    def taxamatch(str1, str2) 
+    def taxamatch(str1, str2, return_boolean = true) 
       preparsed_1 = @parser.parse(str1)
       preparsed_2 = @parser.parse(str2)
-      taxamatch_preparsed(preparsed_1, preparsed_2)['match'] rescue false
+      match = taxamatch_preparsed(preparsed_1, preparsed_2) rescue nil
+      return_boolean && match ? match['match'] : match
     end
   
     #takes two hashes of parsed scientific names, analyses them and returns back 
@@ -34,7 +35,7 @@ module Taxamatch
       result =  match_uninomial(preparsed_1, preparsed_2) if preparsed_1[:uninomial] && preparsed_2[:uninomial] 
       result =  match_multinomial(preparsed_1, preparsed_2) if preparsed_1[:genus] && preparsed_2[:genus]
       if result && result['match']
-        result['match'] = false if match_authors(preparsed_1, preparsed_2) == 0 
+        result['match'] = match_authors(preparsed_1, preparsed_2) == 0 ? false : true 
       end
       return result
     end
@@ -46,17 +47,16 @@ module Taxamatch
     def match_multinomial(preparsed_1, preparsed_2)
       gen_match = match_genera(preparsed_1[:genus], preparsed_2[:genus])
       sp_match = match_species(preparsed_1[:species], preparsed_2[:species])
-      au_match = match_authors(preparsed_1, preparsed_2)
       total_length = preparsed_1[:genus][:epitheton].size + preparsed_2[:genus][:epitheton].size + preparsed_1[:species][:epitheton].size + preparsed_2[:species][:epitheton].size
       match = match_matches(gen_match, sp_match)
-      match.merge({'score' => (1- match['edit_distance']/(total_length/2))})
+      match.merge({'score' => (1 - match['edit_distance']/(total_length/2))})
     end
   
     def match_genera(genus1, genus2)
       genus1_length = genus1[:normalized].size
       genus2_length = genus2[:normalized].size
       match = false
-      ed = @dlm.distance(genus1[:normalized], genus2[:normalized],2,3)
+      ed = @dlm.distance(genus1[:normalized], genus2[:normalized],1,3) #TODO put block = 2
       return {'edit_distance' => ed, 'phonetic_match' => true, 'match' => true} if genus1[:phonetized] == genus2[:phonetized] 
     
       match = true if ed <= 3 && ([genus1_length, genus2_length].min > ed * 2) && (ed < 2 || genus1[0] == genus2[0])
@@ -69,7 +69,8 @@ module Taxamatch
       sp1[:phonetized] = Taxamatch::Phonetizer.normalize_ending sp1[:phonetized]
       sp2[:phonetized] = Taxamatch::Phonetizer.normalize_ending sp2[:phonetized]
       match = false
-      ed = @dlm.distance(sp1[:normalized], sp2[:normalized], 4, 4)
+      ed = @dlm.distance(sp1[:normalized], sp2[:normalized], 1, 4) #TODO put block 4
+      #puts 's: %s, %s, %s' % [sp1[:normalized], sp2[:normalized], ed]
       return {'edit_distance' => ed, 'phonetic_match' => true, 'match' => true} if sp1[:phonetized] == sp2[:phonetized]
     
       match = true if ed <= 4 && ([sp1_length, sp2_length].min >= ed * 2) && (ed < 2 || sp1[:normalized][0] == sp2[:normalized][0]) && (ed < 4 || sp1[:normalized][0...3] == sp2[:normalized][0...3])

data/spec/damerau_levenshtein_mod_test.txt

@@ -58,3 +58,6 @@ trimerophyton|mertriophyton|10|3|3
 #it should stop trying if distance exceeds maximum allowed distance
 Pxxxxomus|Pomatomus|10|1|4
 Pxxxxomus|Pomatomus|2|1|3
+
+#
+PUNCTATA|PUNCTATA|10|1|0

data/spec/taxamatch_rb_spec.rb

@@ -55,12 +55,13 @@ describe 'Taxamatch::Base' do
   
   it 'should get txt tests' do
     dl = Taxamatch::DamerauLevenshteinMod.new
-    read_test_file(File.expand_path(File.dirname(__FILE__)) + '/taxamatch_test.txt', 3) do |y|
+    read_test_file(File.expand_path(File.dirname(__FILE__)) + '/taxamatch_test.txt', 4) do |y|
       if y
         y[2] = y[2] == 'true' ? true : false
-        res = @tm.taxamatch(y[0], y[1])
-        puts "%s, %s, %s" % [y[0], y[1], y[2]] if res != y[2]
-        res.should == y[2]
+        res = @tm.taxamatch(y[0], y[1], false)
+        #puts "%s, %s, %s, %s" % [y[0], y[1], y[2], y[3]] if res != y[2]
+        res['match'].should == y[2]
+        res['edit_distance'].should == y[3].to_i
       end
     end
   end

data/spec/taxamatch_test.txt

@@ -1,28 +1,34 @@
 ###
 #
 # Tests for string comparison by taxamatch algorithm
+# name1|name2|match|edit_distance
 #
 ##
 
-# additional authorship should match
-Puma concolor|Puma concolor L.|true
-
-# one-letter misspeling in species epithet should match
-Puma concolor|Puma cancolor|true
-
-Pomatomus saltatrix|Pomatomus saltratix|true
-Pomatomus saltator|Pomatomus saltatrix|true
-
-Loligo pealeii|Loligo plei|false
-
-# different authors should not match
-Puma concolor Linnaeus|Puma concolor Kurtz|false
+## additional authorship should match
+Puma concolor|Puma concolor L.|true|0
+#
+## one-letter misspeling in species epithet should match
+Puma concolor|Puma cancolor|true|1
+#
+Pomatomus saltatrix|Pomatomus saltratix|true|2
+Pomatomus saltator|Pomatomus saltatrix|true|3
+#
+Loligo pealeii|Loligo plei|false|3
+#
+## different authors should not match
+Puma concolor Linnaeus|Puma concolor Kurtz|false|0
+#
+##real life examples
+Biatora borealis|Bactra borealis Diakonoff 1964|false|3
+#
+Homo sapien|Homo sapiens|true|1
+Homo sapiens Linnaeus|Homo sapens (Linn. 1758) |true|1
+Homo sapiens Mozzherin|Homo sapiens Linneaus|false|0
+#
+Quinqueloculina punctata|Quinqueloculina punctata d'Orbigny 1905|true|0
+Pomatomus saltator (Linnaeus, 1766)|Pomatomus saltatrix (Linnaeus, 1766)|true|0|3
 
-#real life examples
-Biatora borealis|Bactra borealis Diakonoff 1964|false
 
-Homo sapien|Homo sapiens|true
-Homo sapiens Linnaeus|Homo sapens (Linn. 1758) |true
-Homo sapiens Mozzherin|Homo sapiens Linneaus|false
 
 

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: dimus-taxamatch_rb
 version: !ruby/object:Gem::Version 
-  version: 0.5.3
+  version: 0.5.5
 platform: ruby
 authors: 
 - Dmitry Mozzherin
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2009-08-09 00:00:00 -07:00
+date: 2009-08-16 00:00:00 -07:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -32,7 +32,7 @@ dependencies:
       - !ruby/object:Gem::Version 
         version: "0"
     version: 
-description: 
+description: This gem implements algorithsm for fuzzy matching scientific names developed by Tony Rees
 email: dmozzherin@eol.org
 executables: []
 
@@ -55,7 +55,7 @@ files:
 - spec/taxamatch_rb_spec.rb
 - spec/taxamatch_test.txt
 - LICENSE
-has_rdoc: true
+has_rdoc: false
 homepage: http://github.com/dimus/taxamatch_rb
 licenses: 
 post_install_message: 
@@ -80,8 +80,8 @@ requirements: []
 rubyforge_project: 
 rubygems_version: 1.3.5
 signing_key: 
-specification_version: 2
-summary: TODO
+specification_version: 3
+summary: Implementation of Tony Rees Taxamatch algorithms
 test_files: 
 - spec/spec_helper.rb
 - spec/taxamatch_rb_spec.rb