text_parser 0.1.6 → 0.1.7

data/lib/text_parser.rb

@@ -16,7 +16,8 @@ module TextParser
     options[:dictionary] = text.split(" ") unless options[:dictionary]
     return [] if options[:dictionary].count < 1
     regex = Regexp.new(options[:dictionary].join('\\b|\\b'), Regexp::IGNORECASE)
-    match_result = text.scan(regex).map{|i| i.downcase}   
+    match_result = text.scan(regex).map{|i| i.downcase}  
+    match_result.select!{|i| i.size >= options[:minimum_length]} if options[:minimum_length]
     match_result.each do |w|
       result << {:hits => match_result.count(w), :word => w} unless result.select{|r| r[:word] == w}.shift || options[:negative_dictionary].map{|i| i.downcase}.include?(w)
     end 

data/test/text_parser_test.rb

@@ -100,6 +100,18 @@ class TextParserTest < Test::Unit::TestCase
                   {:word => "pão",    :hits => 1}], "Pão açúcar".parse
     assert_equal [{:word => "ãéç",    :hits => 1}], "ãéç".parse
   end
+  
+  def test_minimum_length
+    text = "a ab   abc "
+    assert_equal [{:word => "a",    :hits => 1},
+                  {:word => "ab",   :hits => 1},
+                  {:word => "abc",  :hits => 1}], text.parse(:minimum_length => 1)
+    assert_equal [{:word => "ab",   :hits => 1},
+                  {:word => "abc",  :hits => 1}], text.parse(:minimum_length => 2)
+    assert_equal [{:word => "abc",  :hits => 1}], text.parse(:minimum_length => 3)
+    assert_equal [{:word => "abc",  :hits => 1}], text.parse(:minimum_length => 2, :negative_dictionary => ["ab"])
+    assert_equal [],                              text.parse(:minimum_length => 3, :dictionary => ["a"])
+  end
 end
 
  

data/text_parser.gemspec

@@ -1,6 +1,6 @@
 Gem::Specification.new do |s|
   s.name        = "text_parser"
-  s.version     = "0.1.6"
+  s.version     = "0.1.7"
   s.author      = "Frederico de Paula"
   s.email       = "fpaula@gmail.com"
   s.summary     = "A easy way to parse a text."

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: text_parser
 version: !ruby/object:Gem::Version
-  version: 0.1.6
+  version: 0.1.7
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2011-12-15 00:00:00.000000000Z
+date: 2011-12-19 00:00:00.000000000Z
 dependencies: []
 description: Using method parse in the String object you can parse any text.
 email: fpaula@gmail.com