open_nlp 0.0.6-java → 0.0.7-java

data/.gitignore

@@ -15,3 +15,4 @@ spec/reports
 test/tmp
 test/version_tmp
 tmp
+.idea

data/lib/open_nlp.rb

@@ -17,6 +17,9 @@ require 'open_nlp/model/sentence_detector'
 require 'open_nlp/model/tokenizer'
 require 'open_nlp/model/parser'
 
+require 'open_nlp/util'
+require 'open_nlp/util/span'
+
 require 'open_nlp/tool'
 require 'open_nlp/categorizer'
 require 'open_nlp/named_entity_detector'

data/lib/open_nlp/sentence_detector.rb

@@ -6,5 +6,10 @@ module OpenNlp
       raise ArgumentError, "string must be a String" unless string.is_a?(String)
       @j_instance.sentDetect(string).to_ary
     end
+
+    def pos_detect(string)
+      raise ArgumentError, "string must be a String" unless string.is_a?(String)
+      @j_instance.sentPosDetect(string).map{|span| OpenNlp::Util::Span.new(span.getStart, span.getEnd)}
+    end
   end
 end

data/lib/open_nlp/util.rb

@@ -0,0 +1,5 @@
+module OpenNlp
+  module Util
+
+  end
+end

data/lib/open_nlp/util/span.rb

@@ -0,0 +1,38 @@
+class OpenNlp::Util::Span
+  include OpenNlp::JavaClass
+
+  self.java_class = Java::opennlp.tools.util.Span
+
+  attr_reader :j_instance
+
+  def initialize(s, e)
+    raise ArgumentError, "s should be an integer" unless s.is_a?(Fixnum)
+    raise ArgumentError, "e should be an integer" unless e.is_a?(Fixnum)
+
+    @j_instance = self.class.java_class.new(s, e)
+  end
+
+  def start
+    j_instance.getStart
+  end
+
+  def end
+    j_instance.getEnd
+  end
+
+  def type
+    j_instance.getType
+  end
+
+  def length
+    j_instance.length
+  end
+
+  def ==(obj)
+    return false unless obj.is_a?(self.class)
+
+    [:start, :end, :type].each_with_object(true) do |m,res|
+      res = res && self.public_send(m) == obj.public_send(m)
+    end
+  end
+end

data/lib/open_nlp/version.rb

@@ -1,3 +1,3 @@
 module OpenNlp
-  VERSION = '0.0.6'
+  VERSION = '0.0.7'
 end

data/spec/sentence_detector_spec.rb

@@ -17,7 +17,7 @@ describe OpenNlp::SentenceDetector do
     end
   end
 
-  describe "sentence detection" do
+  describe "#detect" do
     let(:sent_detector) { subject.new(model) }
 
     it "should detect no sentences in an empty string" do
@@ -34,4 +34,17 @@ describe OpenNlp::SentenceDetector do
       lambda { sent_detector.detect(nil) }.should raise_error(ArgumentError)
     end
   end
-end
+
+  describe "#pos_detect" do
+    let(:sent_detector) { subject.new(model) }
+
+    it "should detect sentences in a string" do
+      sentences = sent_detector.pos_detect("The sky is blue. The Grass is green.")
+      sentences.should == [OpenNlp::Util::Span.new(0, 16), OpenNlp::Util::Span.new(17, 36)]
+    end
+
+    it "should raise an ArgumentError for a non-string" do
+      expect { sent_detector.detect(nil) }.to raise_error(ArgumentError)
+    end
+  end
+end

data/spec/tokenizer_spec.rb

@@ -33,4 +33,4 @@ describe OpenNlp::Tokenizer do
       lambda { tokenizer.tokenize(nil) }.should raise_error(ArgumentError)
     end
   end
-end
+end

metadata

@@ -2,14 +2,14 @@
 name: open_nlp
 version: !ruby/object:Gem::Version
   prerelease:
-  version: 0.0.6
+  version: 0.0.7
 platform: java
 authors:
 - Hck
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2012-10-31 00:00:00.000000000 Z
+date: 2013-02-22 00:00:00.000000000 Z
 dependencies: []
 description: JRuby tools wrapper for Apache OpenNLP
 email:
@@ -45,6 +45,8 @@ files:
 - lib/open_nlp/sentence_detector.rb
 - lib/open_nlp/tokenizer.rb
 - lib/open_nlp/tool.rb
+- lib/open_nlp/util.rb
+- lib/open_nlp/util/span.rb
 - lib/open_nlp/version.rb
 - open_nlp.gemspec
 - spec/categorizer_spec.rb