word_count_analyzer 0.0.1

data/spec/word_count_analyzer/date_spec.rb

@@ -0,0 +1,257 @@
+require 'spec_helper'
+
+RSpec.describe WordCountAnalyzer::Date do
+  context '#includes_date?' do
+    it 'returns true if the string includes a date #001' do
+      string = 'Today is Monday, April 4th, 2011, aka 04/04/2011.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #002' do
+      string = 'Today is Monday April 4th 2011.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #003' do
+      string = 'Today is April 4th, 2011.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #004' do
+      string = 'Today is Mon., Apr. 4, 2011.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #005' do
+      string = 'Today is 04/04/2011.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #006' do
+      string = 'Today is 04.04.2011.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #007' do
+      string = 'Today is 2011.04.04.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #008' do
+      string = 'Today is 2011/04/04.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #009' do
+      string = 'Today is 2011-04-04.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #010' do
+      string = 'Today is 04-04-2011.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #011' do
+      string = 'Today is 2003 November 9.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #012' do
+      string = 'Today is 2003Nov9.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #013' do
+      string = 'Today is 2003Nov09.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #014' do
+      string = 'Today is 2003-Nov-9.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #015' do
+      string = 'Today is 2003-Nov-09.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #016' do
+      string = 'Today is 2003-Nov-9, Sunday.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #017' do
+      string = 'Today is 2003. november 9.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #018' do
+      string = 'Today is 2003.11.9.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #019' do
+      string = 'Today is Monday, Apr. 4, 2011.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #020' do
+      string = 'Today is 2003/11/09.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #021' do
+      string = 'Today is 20030109.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #022' do
+      string = 'Today is 01092003.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #023' do
+      string = 'Today is Sunday, November 9, 2014.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #024' do
+      string = 'Today is November 9, 2014.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #025' do
+      string = 'Today is Nov. 9, 2014.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #026' do
+      string = 'Today is july 1st.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #027' do
+      string = 'Today is jul. 1st.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #028' do
+      string = 'Today is 8 November 2014.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #029' do
+      string = 'Today is 8. November 2014.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #030' do
+      string = 'Today is 08-Nov-2014.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #031' do
+      string = 'Today is 08Nov14.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #032' do
+      string = 'Today is 8th November 2014.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #033' do
+      string = 'Today is the 8th of November 2014.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #034' do
+      string = 'Today is 08/Nov/2014.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #035' do
+      string = 'Today is Sunday, 8 November 2014.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns true if the string includes a date #036' do
+      string = 'Today is 8 November 2014.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(true)
+    end
+
+    it 'returns false if the string does not include a date #037' do
+      string = 'Hello world. There is no date here - $50,000. The sun is hot.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.includes_date?).to eq(false)
+    end
+  end
+
+  context '#occurences' do
+    it 'counts the date occurences in a string #001' do
+      string = 'Today is Sunday, 8 November 2014.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.occurences).to eq(1)
+    end
+
+    it 'counts the date occurences in a string #002' do
+      string = 'Today is Sunday, 8 November 2014. Yesterday was 07/Nov/2014.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.occurences).to eq(2)
+    end
+  end
+
+  context '#replace' do
+    it 'replaces the date occurences in a string #001' do
+      string = 'Today is Tues. March 3rd, 2011.'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.replace).to eq('Today is  wsdateword ')
+    end
+  end
+
+  context '#replace_number_only_date' do
+    it 'replaces only the number date occurences in a string' do
+      string = 'Today is Tues. March 3rd, 2011. 4/28/2013'
+      ws = WordCountAnalyzer::Date.new(string: string)
+      expect(ws.replace_number_only_date).to eq("Today is Tues. March 3rd, 2011.  wsdateword ")
+    end
+  end
+end

data/spec/word_count_analyzer/ellipsis_spec.rb

@@ -0,0 +1,69 @@
+require 'spec_helper'
+
+RSpec.describe WordCountAnalyzer::Ellipsis do
+  context '#includes_ellipsis?' do
+    it 'returns true if the string includes an ellipsis #001' do
+      string = 'Using an ellipsis … causes different counts.'
+      ws = WordCountAnalyzer::Ellipsis.new(string: string)
+      expect(ws.includes_ellipsis?).to eq(true)
+    end
+
+    it 'returns true if the string includes an ellipsis #002' do
+      string = 'Using an ellipsis causes different counts…depending on the style that you use.'
+      ws = WordCountAnalyzer::Ellipsis.new(string: string)
+      expect(ws.includes_ellipsis?).to eq(true)
+    end
+
+    it 'returns true if the string includes an ellipsis #003' do
+      string = 'Using an ellipsis causes different counts depending on the style . . . that you use.'
+      ws = WordCountAnalyzer::Ellipsis.new(string: string)
+      expect(ws.includes_ellipsis?).to eq(true)
+    end
+
+    it 'returns true if the string includes an ellipsis #004' do
+      string = 'Using an ellipsis causes different counts depending on the style . . . . that you use.'
+      ws = WordCountAnalyzer::Ellipsis.new(string: string)
+      expect(ws.includes_ellipsis?).to eq(true)
+    end
+
+    it 'returns true if the string includes an ellipsis #005' do
+      string = 'Using an ellipsis causes different counts depending on the style.... that you use.'
+      ws = WordCountAnalyzer::Ellipsis.new(string: string)
+      expect(ws.includes_ellipsis?).to eq(true)
+    end
+
+    it "returns false if the string doesn't include an ellipsis #006" do
+      string = 'Hello world.'
+      ws = WordCountAnalyzer::Ellipsis.new(string: string)
+      expect(ws.includes_ellipsis?).to eq(false)
+    end
+
+    it "returns false if the string includes a dotted_line #007" do
+      string = '.....'
+      ws = WordCountAnalyzer::Ellipsis.new(string: string)
+      expect(ws.includes_ellipsis?).to eq(false)
+    end
+
+    it "returns false if the string includes a dotted_line #007" do
+      string = "Here is one …………………………………………………………………… and another ......"
+      ws = WordCountAnalyzer::Ellipsis.new(string: string)
+      expect(ws.includes_ellipsis?).to eq(false)
+    end
+  end
+
+  context '#replace' do
+    it 'returns a string with the ellipsis replaced #001' do
+      string = 'Using an ellipsis … causes different counts…depending on the style . . . that you use. I never meant that.... She left the store. The practice was not abandoned. . . .'
+      ws = WordCountAnalyzer::Ellipsis.new(string: string)
+      expect(ws.replace).to eq("Using an ellipsis  wseword  causes different counts wseword depending on the style wseword that you use. I never meant that. wseword  She left the store. The practice was not abandoned wseword ")
+    end
+  end
+
+  context '#occurences' do
+    it 'returns a string with the ellipsis replaced #001' do
+      string = 'Using an ellipsis … causes different counts…depending on the style . . . that you use. I never meant that.... She left the store. The practice was not abandoned. . . .'
+      ws = WordCountAnalyzer::Ellipsis.new(string: string)
+      expect(ws.occurences).to eq(5)
+    end
+  end
+end

data/spec/word_count_analyzer/hyperlink_spec.rb

@@ -0,0 +1,77 @@
+require 'spec_helper'
+
+RSpec.describe WordCountAnalyzer::Hyperlink do
+  context '#hyperlink?' do
+    it 'returns true if the string is a hyperlink #001' do
+      string = "http://www.example.com/this-IS-a_test/hello.html"
+      ws = WordCountAnalyzer::Hyperlink.new(string: string)
+      expect(ws.hyperlink?).to eq(true)
+    end
+
+    it 'returns true if the string is a hyperlink #002' do
+      string = "http://www.google.co.uk"
+      ws = WordCountAnalyzer::Hyperlink.new(string: string)
+      expect(ws.hyperlink?).to eq(true)
+    end
+
+    it 'returns true if the string is a hyperlink #003' do
+      string = "https://google.co.uk"
+      ws = WordCountAnalyzer::Hyperlink.new(string: string)
+      expect(ws.hyperlink?).to eq(true)
+    end
+
+    it 'returns false if the string is not a hyperlink #004' do
+      string = "hello"
+      ws = WordCountAnalyzer::Hyperlink.new(string: string)
+      expect(ws.hyperlink?).to eq(false)
+    end
+
+    it 'returns false if the string is not a hyperlink #005' do
+      string = "john@gmail.com"
+      ws = WordCountAnalyzer::Hyperlink.new(string: string)
+      expect(ws.hyperlink?).to eq(false)
+    end
+
+    it 'returns false if the string is not a hyperlink #006' do
+      string = "date:"
+      ws = WordCountAnalyzer::Hyperlink.new(string: string)
+      expect(ws.hyperlink?).to eq(false)
+    end
+
+    it 'returns false if the string is not a hyperlink #007' do
+      string = 'The file location is c:\Users\johndoe.'
+      ws = WordCountAnalyzer::Hyperlink.new(string: string)
+      expect(ws.hyperlink?).to eq(false)
+    end
+  end
+
+  context '#occurences' do
+    it 'returns the occurences of hyperlink tokens in a string #001' do
+      string = "Today the date is: Jan 1. Visit https://www.example.com/hello or http://www.google.co.uk"
+      ws = WordCountAnalyzer::Hyperlink.new(string: string)
+      expect(ws.occurences).to eq(2)
+    end
+  end
+
+  context '#replace' do
+    it 'replaces the hyperlinks in a string with regular tokens #001' do
+      string = "Today the date is: Jan 1. Visit https://www.example.com/hello or http://www.google.co.uk"
+      ws = WordCountAnalyzer::Hyperlink.new(string: string)
+      expect(ws.replace).to eq("Today the date is: Jan 1. Visit  wslinkword  or  wslinkword ")
+    end
+
+    it 'replaces the hyperlinks in a string with regular tokens #002' do
+      string = 'The file location is c:\Users\johndoe or d:\Users\john\www'
+      ws = WordCountAnalyzer::Hyperlink.new(string: string)
+      expect(ws.replace).to eq('The file location is c:\Users\johndoe or d:\Users\john\www')
+    end
+  end
+
+  context '#replace_split_at_period' do
+    it 'replaces the hyperlinks in a string with regular tokens, split at periods #001' do
+      string = "http://www.google.co.uk"
+      ws = WordCountAnalyzer::Hyperlink.new(string: string)
+      expect(ws.replace_split_at_period).to eq("http://www google co uk")
+    end
+  end
+end

data/spec/word_count_analyzer/hyphenated_word_spec.rb

@@ -0,0 +1,81 @@
+require 'spec_helper'
+
+RSpec.describe WordCountAnalyzer::HyphenatedWord do
+  context '#hyphenated_word?' do
+    it 'returns true if the token is a hyphenated word #001' do
+      token = 'devil-may-care'
+      ws = WordCountAnalyzer::HyphenatedWord.new(token: token)
+      expect(ws.hyphenated_word?).to eq(true)
+    end
+
+    it 'returns true if the token is a hyphenated word #002' do
+      token = '(2R)-2-methylsulfanyl-3-hydroxybutanedioate'
+      ws = WordCountAnalyzer::HyphenatedWord.new(token: token)
+      expect(ws.hyphenated_word?).to eq(true)
+    end
+
+    it 'returns false if the token is not a hyphenated word' do
+      token = 'hello'
+      ws = WordCountAnalyzer::HyphenatedWord.new(token: token)
+      expect(ws.hyphenated_word?).to eq(false)
+    end
+
+    it 'returns false if the token is a hyperlink' do
+      token = 'https://www.example-one.com'
+      ws = WordCountAnalyzer::HyphenatedWord.new(token: token)
+      expect(ws.hyphenated_word?).to eq(false)
+    end
+
+    it 'returns false if the token is long string of dashes' do
+      token = '------------'
+      ws = WordCountAnalyzer::HyphenatedWord.new(token: token)
+      expect(ws.hyphenated_word?).to eq(false)
+    end
+
+    it 'returns true if the token is a hyphenated word (small em dashes)' do
+      token = 'devil﹘may﹘care'
+      ws = WordCountAnalyzer::HyphenatedWord.new(token: token)
+      expect(ws.hyphenated_word?).to eq(true)
+    end
+  end
+
+  context '#count_as_multiple' do
+    it 'returns the count of the individual words that are separated by the hyphen' do
+      token = 'devil-may-care'
+      ws = WordCountAnalyzer::HyphenatedWord.new(token: token)
+      expect(ws.count_as_multiple).to eq(3)
+    end
+
+    it 'handles small em dashes' do
+      token = 'devil﹘may﹘care'
+      ws = WordCountAnalyzer::HyphenatedWord.new(token: token)
+      expect(ws.count_as_multiple).to eq(3)
+    end
+
+    it 'returns the count of the individual words that are separated by the hyphen #002' do
+      token = '(2R)-2-methylsulfanyl-3-hydroxybutanedioate'
+      ws = WordCountAnalyzer::HyphenatedWord.new(token: token)
+      expect(ws.count_as_multiple).to eq(5)
+    end
+  end
+
+  context '#replace' do
+    it 'splits hyphenated words #001' do
+      token = 'devil-may-care'
+      ws = WordCountAnalyzer::HyphenatedWord.new(token: token)
+      expect(ws.replace).to eq('devil may care')
+    end
+
+    it 'splits hyphenated words #002' do
+      token = 'devil﹘may﹘care'
+      ws = WordCountAnalyzer::HyphenatedWord.new(token: token)
+      expect(ws.replace).to eq('devil may care')
+    end
+
+    it 'splits hyphenated words #003' do
+      token = '(2R)-2-methylsulfanyl-3-hydroxybutanedioate'
+      ws = WordCountAnalyzer::HyphenatedWord.new(token: token)
+      expect(ws.replace).to eq('(2R) 2 methylsulfanyl 3 hydroxybutanedioate')
+    end
+  end
+end