eaternet 0.4.3 → 0.4.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 9f09792a0e3e7fe94b6c3379288c8e49093792f2
-  data.tar.gz: 4aa2df5d05299c963e82f774540cad1f54d5a0c3
+  metadata.gz: d229e39d8e34a1c791ce2d212de62da1e1001ca3
+  data.tar.gz: c8f41c3455bcdbacca49900fd6691acae380d689
 SHA512:
-  metadata.gz: 0c7e7427fc67460dcca75658313f734534574dd77fb40fdadd94e73bf6d0be4425b44daf82233004d217b1cac11e123b2c798c56e1c8b3a1c58f07e3ac60b2e4
-  data.tar.gz: 87a118b0c3a3ce819f81b92c94b645ed413a09d89982468248d308e61b6f842d679e961cbf71c2fcdfcf364f3717a81f3fdf76f050b6091d398d5c112ef36090
+  metadata.gz: fe5fd0d5c67daeb5c173315e7f6e9bf01e903b6455995e0bdfddd61524e3630b9f5ac5b0e383d570825e5b1a2e62f241242eff877e200ef62ef6c4c629273392
+  data.tar.gz: 8a491c8d04d2022c1d75e34c2e196e070f6289c28bbb52514570cfc88eac9f5fdf4ef647ccd066b394870f72b45a5d5f26d9333e7edc4c826428843d3f9cb311

data/lib/eaternet/util.rb

@@ -67,7 +67,7 @@ module Eaternet
     # @return [String] the cleaned up string
     def self.cleanup_title(a_string)
       return nil if a_string.nil?
-      cleanup(a_string).titleize.sub("Mc Donald's", "McDonald's")
+      titleize(cleanup(a_string))
     end
 
     # Remove extraneous whitespace from the string
@@ -83,7 +83,6 @@ module Eaternet
       (Time.now - File.mtime(path)).to_i / 86_400.0
     end
 
-
     private
 
     def self.expired?(cache_path)
@@ -101,5 +100,50 @@ module Eaternet
       `mkdir -p #{cache_dir}`
       cache_dir
     end
+
+    #
+    # A titleize that creates a usable title according to grammar rules.
+    #
+    ARTICLES = Set.new %w(a an and by di et for in is not of on or over the to under with)
+
+    def self.titleize(a_string)
+      # Only fix if all uppercase
+      return a_string unless all_uppercase?(a_string)
+
+      result = []
+
+      for word in a_string.downcase.split(/[[:space:]]/) # handle unicode
+        if irish?(word)
+          result << inflect_irish(word)
+        else
+          word = word.capitalize unless ARTICLES.include? word
+          result << word
+        end
+      end
+
+      just_capitalize(result.join(' '))
+    end
+
+    def self.all_uppercase?(a_string)
+      a_string.upcase == a_string
+    end
+
+    def self.irish?(a_string)
+      a_string =~ /^(ma?c)|(o').+$/ && !italian?(a_string)
+    end
+
+    def self.italian?(a_string)
+      %w(macaron macaroni macchiato machiavelli).include? a_string
+    end
+
+    def self.inflect_irish(a_string)
+      a_string =~ /^((ma?c)|(o'))(.+)$/
+      Regexp.last_match(1).capitalize + Regexp.last_match(4).capitalize
+    end
+
+    def self.just_capitalize(a_string)
+      head, tail = a_string.split(//, 2)
+      head.upcase + tail
+    end
   end
 end

data/lib/eaternet/version.rb

@@ -1,3 +1,3 @@
 module Eaternet
-  VERSION = '0.4.3'
+  VERSION = '0.4.4'
 end

data/test/eaternet/agencies/snhd_lives_test.rb

@@ -18,7 +18,7 @@ class SnhdLivesTest < Minitest::Test
     VCR.use_cassette(SNHD_CASSETTE) do
       b = @@snhd.businesses.first
       assert_equal 'PR0000002', b.business_id
-      assert_equal "McDonald's #3549 D Hotel", b.name
+      assert_equal "McDonald's #3549 D HOTEL", b.name
       assert_equal '301 Fremont St', b.address
       assert_equal 'Las Vegas', b.city
       assert_equal '89101-5600', b.postal_code

data/test/eaternet/util_test.rb

@@ -5,7 +5,6 @@ require 'timecop'
 require 'eaternet/util'
 
 class UtilTest < Minitest::Test
-
   def setup
     @file_contents = 'This is the file'
     @url = "http://downloadtest.com/file-#{rand(1_000_000)}.txt"
@@ -45,9 +44,67 @@ class UtilTest < Minitest::Test
     assert_requested :get, @url, times: 2
   end
 
-  def test_properly_cleans_up_mcdonalds
-    input = "McDonald's #3549 D HOTEL"
-    desired_output = "McDonald's #3549 D Hotel"
-    assert_equal desired_output, Eaternet::Util.cleanup_title(input)
+  #
+  # #cleanup_title
+  #
+
+  def assert_cleans_up_to(output, input)
+    assert_equal output, Eaternet::Util.cleanup_title(input)
+  end
+
+  def test_doesnt_mess_up_numeric_streets
+    assert_cleans_up_to '1654 86th St', '1654 86TH ST'
+  end
+
+  # Irish
+
+  def test_mcdonald_avenue
+    assert_cleans_up_to '1158 McDonald Avenue', '1158 MCDONALD AVENUE'
+  end
+
+  def test_mckennas_pub
+    assert_cleans_up_to "McKenna's Pub", "MCKENNA'S PUB"
+  end
+
+  def test_macdougal_street
+    assert_cleans_up_to '122 MacDougal St', '122 MACDOUGAL ST'
+  end
+
+  def test_o_sullivans_pub
+    assert_cleans_up_to "O'Sullivan's Pub", "O'SULLIVAN'S PUB"
+  end
+
+  # Italian
+
+  def test_macchiato
+    assert_cleans_up_to 'Macchiato Espresso Bar', 'MACCHIATO ESPRESSO BAR'
+  end
+
+  def test_di_vittorio
+    assert_cleans_up_to 'La Lanterna di Vittorio', 'LA LANTERNA DI VITTORIO'
+  end
+
+  def test_machiavelli
+    assert_cleans_up_to 'Machiavelli', 'MACHIAVELLI'
+  end
+
+  def test_macaroni
+    assert_cleans_up_to 'Macaroni', 'MACARONI'
+  end
+
+  # French
+
+  def test_rouge_et_blanc
+    assert_cleans_up_to 'Rouge et Blanc', 'ROUGE ET BLANC'
+  end
+
+  def test_macaron
+    assert_cleans_up_to 'Macaron', 'MACARON'
+  end
+
+  # Weird
+
+  def test_weird_stuff
+    assert_cleans_up_to 'aBcDeFg', 'aBcDeFg'
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: eaternet
 version: !ruby/object:Gem::Version
-  version: 0.4.3
+  version: 0.4.4
 platform: ruby
 authors:
 - Robb Shecter