lingua 0.6.1 → 0.6.2

data/CHANGELOG.markdown

@@ -0,0 +1,3 @@
+0.6.2
+-----
+* Fix for sentences that end in abbreviations (e.g. dr, mrs, ms) [chad]

data/VERSION

@@ -1 +1 @@
-0.6.1
+0.6.2

data/lib/lingua/en/sentence.rb

@@ -78,7 +78,7 @@ module Lingua
       end
 
       def self.set_abbr_regex!
-        @abbr_regex = /(#{abbreviations.join("|")})\.#{EOS}/i
+        @abbr_regex = / (#{abbreviations.join("|")})\.#{EOS}/i
       end
 
       initialize_abbreviations!

data/lingua.gemspec

@@ -5,11 +5,11 @@
 
 Gem::Specification.new do |s|
   s.name = %q{lingua}
-  s.version = "0.6.1"
+  s.version = "0.6.2"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["David Balatero"]
-  s.date = %q{2010-05-17}
+  s.date = %q{2010-07-25}
   s.description = %q{Provides sentence splitting, syllable, and text-quality algorithms.}
   s.email = %q{dbalatero@gmail.com}
   s.extra_rdoc_files = [
@@ -19,6 +19,7 @@ Gem::Specification.new do |s|
   s.files = [
     ".document",
      ".gitignore",
+     "CHANGELOG.markdown",
      "LICENSE",
      "README.rdoc",
      "Rakefile",

data/spec/lingua/en/sentence_spec.rb

@@ -2,7 +2,7 @@ require File.dirname(__FILE__) + "/../../spec_helper"
 
 describe Lingua::EN::Sentence do
   klass = Lingua::EN::Sentence
-
+  
   describe "#sentences" do
     describe "multi-paragraph text" do
       before(:each) do
@@ -10,11 +10,11 @@ describe Lingua::EN::Sentence do
         text << "Visit http://www.google.com and check out my site. Thanks very much!"
         @sentences = klass.sentences(text)
       end
-
+  
       it "should get the correct number of sentences" do
         @sentences.should have(5).things
       end
-
+  
       it "should get the correct sentences" do
         @sentences[0].should == "As Milton Bradley once said, \"board games are the shit.\""
         @sentences[1].should == "And I'm inclined to agree."
@@ -23,109 +23,126 @@ describe Lingua::EN::Sentence do
         @sentences[4].should == "Thanks very much!"
       end
     end
-
+  
     describe "quoted sentences" do
       before(:each) do
         text = "As Milton Bradley once said, \"board games are the shit.\" And I'm inclined to agree. \"Why can't we be friends?\""
         @sentences = klass.sentences(text)
       end
-
+  
       it "should get the correct number of sentences" do
         @sentences.should have(3).things
       end
-
+  
       it "should get the correct sentences" do
         @sentences[0].should == "As Milton Bradley once said, \"board games are the shit.\""
         @sentences[1].should == "And I'm inclined to agree."
         @sentences[2].should == "\"Why can't we be friends?\""
       end
     end
-
+  
     describe "ellipses correction" do
       before(:each) do
         text = "Well... why would you do that? Let's not fight."
         @sentences = klass.sentences(text)
       end
-
+  
       it "should get the correct number of sentences" do
         @sentences.should have(2).things
       end
-
+  
       it "should get the right sentences" do
         @sentences[0].should == "Well... why would you do that?"
         @sentences[1].should == "Let's not fight."
       end
     end
-
+  
     describe "simple URL matching" do
       before(:each) do
         text = "Hello, visit http://www.google.com/index.php?ok=ok for more info. Ok?"
         @sentences = klass.sentences(text)
       end
-
+  
       it "should get the correct number of sentences" do
         @sentences.should have(2).things
       end
-
+  
       it "should get the right sentences" do
         @sentences[0].should == "Hello, visit http://www.google.com/index.php?ok=ok for more info."
         @sentences[1].should == "Ok?"
       end
     end
-
+  
     describe "ending a sentence with an abbreviation" do
       before(:each) do
         text = "I was born in the U.S.S.R. My parents were from the U.S. This is not weird."
         @sentences = klass.sentences(text)
       end
-
+  
       it "should get the correct number of sentences" do
         @sentences.should have(3).things
       end
-
+  
       it "should get the correct sentences" do
         @sentences[0].should == "I was born in the U.S.S.R."
         @sentences[1].should == "My parents were from the U.S."
         @sentences[2].should == "This is not weird."
       end
+      
+      describe "which is hard-coded (like st, dr, mrs...)" do
+        before(:each) do
+          text = "This is a test. The word 'test' ends with the abbreviation for 'street'. This should still be three sentences."
+          @sentences = klass.sentences(text)
+        end
+        
+        it "should have the correct number of sentences" do
+          @sentences.should have(3).things
+        end
+        
+        it "should get the correct sentences" do
+          @sentences[0].should == "This is a test."
+          @sentences[1].should == "The word 'test' ends with the abbreviation for 'street'."
+          @sentences[2].should == "This should still be three sentences."
+        end
+      end
     end
-
+    
     describe "basic sentences" do
       before(:each) do
         text = "Hello, my name is David. What is your name?"
         @sentences = klass.sentences(text)
       end
-
+  
       it "should get the correct number of sentences" do
         @sentences.should have(2).things
       end
     end
-
+  
     describe "short sentences w/ line breaks" do
       before(:each) do
         @doc = <<-EOF
         So how does the 401(k) plan work?  Let's see -
-
+  
         The 401(k) consists of - first, asking your employer to set aside a portion (upto 15% of your total income) in keeping with the plan.
         EOF
         @sentences = klass.sentences(@doc)
       end
-
+  
       it "should find 3 sentences" do
         @sentences.should have(3).things
       end
-
+  
       it "should stop at line breaks" do
         @sentences[1].should == "Let's see -"
       end
     end
-
+  
     describe "sentences with URLs and abbreviation" do
       before(:each) do
         text = "Many of these leading names now have their own website, e.g.  http://www.kaptest.com/. Hello, e.g. you don't know what you mean. I'm so angry about what you said about the U.S.A. or the u.S. or the U.S.S.R. ok."
         @sentences = klass.sentences(text)
       end
-
+  
       it "should get the correct number of sentences" do
         @sentences[0].should == "Many of these leading names now have their own website, e.g.  http://www.kaptest.com/."
         @sentences[1].should == "Hello, e.g. you don't know what you mean."
@@ -134,22 +151,23 @@ describe Lingua::EN::Sentence do
       end
     end
   end
-
+  
   describe "#abbreviation" do
     it "should change the abbreviations list" do
       klass.abbreviation('monkey', 'pig')
       klass.abbreviations.should include('monkey')
       klass.abbreviations.should include('pig')
     end
-
+  
     it "should change the regex for abbreviations" do
       lambda {
         klass.abbreviation('monkey')
       }.should change(klass, :abbr_regex)
     end
-
+  
     after(:each) do
       klass.initialize_abbreviations!
     end
   end
+
 end

metadata

@@ -5,8 +5,8 @@ version: !ruby/object:Gem::Version
   segments: 
   - 0
   - 6
-  - 1
-  version: 0.6.1
+  - 2
+  version: 0.6.2
 platform: ruby
 authors: 
 - David Balatero
@@ -14,7 +14,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-05-17 00:00:00 -07:00
+date: 2010-07-25 00:00:00 -07:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -43,6 +43,7 @@ extra_rdoc_files:
 files: 
 - .document
 - .gitignore
+- CHANGELOG.markdown
 - LICENSE
 - README.rdoc
 - Rakefile