ridic 0.6.2 → 0.6.3

data/README.md

@@ -43,6 +43,8 @@ category_number refers to the “column”, or category hierarchy. In the case o
 
     "ALE" => ["ORALITY","NEED","PRIMARY"]
 
+The full dictionary can be found in its original format at the URL provided at the top of the page, or in hash table format  
+within lib/ridic/dictionary.rb
 ## Contributing
 
 1. Fork it

data/lib/ridic/version.rb

@@ -1,3 +1,3 @@
 module RiDic
-  VERSION = "0.6.2"
+  VERSION = "0.6.3"
 end

data/lib/ridic.rb

@@ -1,21 +1,33 @@
 require "ridic/version"
 require "ridic/dictionary"
+require 'benchmark'
 
 module RiDic
   def self.word_match(text_word)
-    dictionary_1 = RiDic::Dictionary.words[text_word.upcase]
-    dictionary_1 == nil ? RiDic::Dictionary.word_stems[text_word.upcase] : dictionary_1
+    text_word.upcase!
+    dictionary_1 = RiDic::Dictionary.words[text_word]
+    dictionary_1 == nil ? RiDic::Dictionary.word_stems[text_word] : dictionary_1
   end
 
   def self.stem_match(text_word)
     text_word.upcase!
-    RiDic::Dictionary.word_stems.each {|key, value| (return value) if text_word.match("^#{key}")}
+    until text_word.length < 3
+      text_word = text_word[0...-1]
+      dictionary_2 = RiDic::Dictionary.word_stems[text_word]
+      (return dictionary_2) if dictionary_2 != nil
+    end
     nil
   end  
 
   def self.all_categories_in_document(document_text, result = [])
-    sanitize(document_text).split(' ').each do |elem| 
-      word_match(elem) == nil ? result << stem_match(elem) : result << word_match(elem)
+    sanitize(document_text).split(' ').each do |elem|
+      word_res = word_match(elem) 
+      if word_res == nil
+        stem_res = stem_match(elem)
+        result.push(stem_res) if stem_res != nil
+      else
+        result << word_res
+      end
     end
     result
   end
@@ -26,13 +38,15 @@ module RiDic
   end
 
   def self.category_distribution(document_text, category_number = 1, result = Hash.new(0))
-    first_categories = category_in_document(sanitize(document_text), category_number).delete_if {|i| i == nil}
+    first_categories = category_in_document(sanitize(document_text.upcase), category_number)
     first_categories.each {|elem| result[elem.first] += 1}
     sort_distribution(result)
   end
 
   private
 
+  # general private methods
+
   def self.sort_distribution(distribution_set, result = Hash.new)
     distribution_set.sort_by {|key, value| value}.reverse.each {|i| result[i[0]] = i[1]}
     result
@@ -42,3 +56,4 @@ module RiDic
     document_text.split(' ').each {|word| word.gsub!(/\W/, '')}.join(' ')
   end
 end
+

data/spec/ridic_spec.rb

@@ -25,8 +25,8 @@ describe 'RiDic' do
     let(:document_text) {"splendid and sublime ale critiqued and bit the harlot of an apple"}
     let(:categories_result){RiDic.all_categories_in_document(document_text)}
     
-    it 'returns an element for every word in the document' do
-      categories_result.length.should eql(document_text.split.length)
+    it 'returns an element for every word match in the document' do
+      categories_result.length.should eql(5)
     end
 
     it 'correctly evaluates the first dictionary item' do

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: ridic
 version: !ruby/object:Gem::Version
-  version: 0.6.2
+  version: 0.6.3
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-12-28 00:00:00.000000000 Z
+date: 2013-01-02 00:00:00.000000000 Z
 dependencies: []
 description: Ruby gem wrapper for the Regressive Imagery Dictionary
 email:
@@ -18,11 +18,13 @@ executables: []
 extensions: []
 extra_rdoc_files: []
 files:
+- .DS_Store
 - .gitignore
 - Gemfile
 - LICENSE.txt
 - README.md
 - Rakefile
+- lib/.DS_Store
 - lib/ridic.rb
 - lib/ridic/dictionary.rb
 - lib/ridic/version.rb