name-spotter 0.1.9 → 0.1.10

data/VERSION

@@ -1 +1 @@
-0.1.9
+0.1.10

data/lib/name-spotter/taxon_finder_client.rb

@@ -14,7 +14,7 @@ class NameSpotter
       @current_string = ''
       @current_string_state = ''
       @word_list_matches = 0
-      @cursor = 5.times.inject([]) { |res| res << ['',-1] }
+      @cursor = 8.times.inject([]) { |res| res << ['',-1] }
       @current_index = nil
       words = str.split(/\s/)
       words.each do |word|
@@ -44,15 +44,13 @@ class NameSpotter
       if output = socket.gets
         response = parse_socket_response(output)
         return if not response
-
-        unless response.return_string.blank?
-          verbatim_string, scientific_string, start_position = process_response(response.return_string)
-          add_name NameSpotter::ScientificName.new(verbatim_string, :start_position => start_position, :scientific_name => scientific_string)
-        end
-        unless response.return_string_2.blank?
-          verbatim_string, scientific_string, start_position = process_response(response.return_string_2)
+        
+        [response.return_string, response.return_string_2].each do |str|
+          next if !str || str.split(" ").size > 6
+          verbatim_string, scientific_string, start_position = process_response(str)
           add_name NameSpotter::ScientificName.new(verbatim_string, :start_position => start_position, :scientific_name => scientific_string)
         end
+        @current_index = nil
       end
     end
 
@@ -85,8 +83,7 @@ class NameSpotter
       if @current_index
         start_position = @current_index
         words, indices = @cursor.transpose
-        verbatim_string = words[indices.index(start_position)...-1].join(" ")
-        @current_index = nil
+        verbatim_string = words[indices.index(start_position)...-1].join(" ") rescue (require 'ruby-debug'; debugger)
       else
         verbatim_string, start_position = @cursor[-1]
       end

data/name-spotter.gemspec

@@ -5,7 +5,7 @@
 
 Gem::Specification.new do |s|
   s.name = "name-spotter"
-  s.version = "0.1.9"
+  s.version = "0.1.10"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Anthony Goddard", "Chuck Ha", "Dmitry Mozzherin"]

data/spec/name-spotter_spec.rb

@@ -100,4 +100,10 @@ describe "NameSpotter" do
     res.should == {:names=>[{:verbatim=>"Ophloihrix nidis", :scientificName=>"Ophloihrix nidis", :offsetStart=>26, :offsetEnd=>41}]}
   end
 
+  it "should not parse ridiculously long infraspecies names by taxon finder" do
+    text = "If we encounter Plantago major it is ok, but if it is Plantago quercus quercus quercus quercus quercus quercus quercus quercus quercus quercus quercus quercus quercus quercus, something is probably not right. However we take Plantago quercus quercus quercus quercus quercus by some strange reason. Well, the reason is this kind of thing -- Pardosa moesta var. moesta f. moesta or something like that"
+    res = @tf.find(text)
+    res.should == {:names=>[{:verbatim=>"Plantago major", :scientificName=>"Plantago major", :offsetStart=>16, :offsetEnd=>29}, {:verbatim=>"Plantago quercus quercus quercus quercus quercus", :scientificName=>"Plantago quercus quercus quercus quercus quercus", :offsetStart=>225, :offsetEnd=>272}, {:verbatim=>"Pardosa moesta var. moesta f. moesta", :scientificName=>"Pardosa moesta var. moesta f. moesta", :offsetStart=>340, :offsetEnd=>375}]}
+  end
+
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: name-spotter
 version: !ruby/object:Gem::Version
-  version: 0.1.9
+  version: 0.1.10
   prerelease: 
 platform: ruby
 authors:
@@ -269,7 +269,7 @@ required_ruby_version: !ruby/object:Gem::Requirement
       version: '0'
       segments:
       - 0
-      hash: 3944184144538713044
+      hash: 1111528646489484932
 required_rubygems_version: !ruby/object:Gem::Requirement
   none: false
   requirements: