wrote a shakespeare analyzer

README.md

@@ -1,33 +1,11 @@
-***An exercise for Prime subscribers. Visit http://learn.thoughtbot.com/prime to learn more.***
+# Shakespeare Play Analyzer
+(like wc for theater)
 
-### Difficulty level: intermediate.
 
-## Your Task
+#### How Can I Use This Wonderful Library?
 
-As a Shakespeare buff, statistics junkie, and unix lover, Ben finds himself wanting a command-line tool for analyzing Macbeth.
-
-Write a command-line program that prints the number of lines spoken by each character in the play.
-
-Sample usage/output (using made-up numbers):
-
-    $ ruby macbeth_analyzer.rb
-      543 Macbeth
-      345 Banquo
-      220 Duncan
-      (etc.)
-
-You can find an XML-encoded version of Macbeth here: http://www.ibiblio.org/xml/examples/shakespeare/macbeth.xml. Your program should download and parse this file at runtime.
-
-Your solution must be tested, preferably via TDD.
-
-## Working/Submitting
-
-1. To work on this exercise, fork the repo and begin implementing your solution.
-2. When you are done, copy the output of your program into a file in this repository. 
-3. Create a pull request so your code can be reviewed.
-4. Perform a code review on at least one other person's solution. Your comments should follow our code review guidelines: https://github.com/thoughtbot/guides/tree/master/code-review. Most important: be friendly. Make suggestions, not demands. 
-5. Improve your solution based on the comments you've received and approaches you've learned from reviewing others' attempts.
-
-## Bounty
-
-While knowledge and skill improvement are their own rewards, the author with the best solution (as judged by thoughtbot) will receive a cool thoughtbot t-shirt.
+```
+git clone https://github.com/shreve/shakespeare_analyzer.git
+cd shakespeare_analyzer
+ruby app.rb
+```

app

@@ -0,0 +1,16 @@
+#!/usr/bin/env ruby
+
+require_relative 'lib/shakespeare_analyzer.rb'
+
+MACBETH_URL = "http://www.ibiblio.org/xml/examples/shakespeare/macbeth.xml"
+HAMLET_URL = "http://www.ibiblio.org/xml/examples/shakespeare/hamlet.xml"
+
+macbeth_analyzer = ShakespeareAnalyzer.new MACBETH_URL
+hamlet_analyzer = ShakespeareAnalyzer.new HAMLET_URL
+
+puts "Lines	Speaker"
+puts "Macbeth"
+print macbeth_analyzer.lines_per_speaker
+
+puts "Hamlet"
+print hamlet_analyzer.lines_per_speaker

lib/play.rb

@@ -0,0 +1,40 @@
+require 'nokogiri'
+require 'open-uri'
+require_relative 'speech'
+
+class Play
+	attr_accessor :document
+
+	def initialize(play_location)
+		@play_location = play_location
+	end
+
+	def speeches
+		document.css('SPEECH').map do |speech|
+			Speech.new speech
+		end
+	end
+
+	def document
+		@document ||= Nokogiri::XML(play_data)
+	end
+
+	private
+		def play_data
+			if location_is_a_file?
+				@play_location
+			elsif location_is_a_url?
+				open(@play_location)
+			else
+				File.open(@play_location, 'r')
+			end
+		end
+
+		def location_is_a_file?
+			@play_location.is_a? File
+		end
+
+		def location_is_a_url?
+			@play_location.match(/^http/)
+		end
+end

lib/shakespeare_analyzer.rb

@@ -0,0 +1,34 @@
+require_relative 'play'
+
+class ShakespeareAnalyzer
+	attr_accessor :play, :speeches, :speakers
+
+	def initialize(play_location)
+		@play = Play.new(play_location)
+		@speakers = Hash.new(0)
+	end
+
+	def lines_per_speaker
+		most_verbose_speakers.map { |name, line_count| "#{line_count}\t#{name}\n" }.join
+	end
+
+	private
+		def most_verbose_speakers
+			analyze_speeches
+			@speakers.sort_by { |name, line_count| -line_count }
+		end
+
+		def get_speeches
+			@speeches ||= @play.speeches
+		end
+
+		def analyze_speeches
+			get_speeches.each do |speech|
+				# nested iterators :/
+				speech.speakers.each do |speaker|
+					# new value is previous value or zero plus the number of lines in the speech
+					@speakers[speaker] += speech.lines.count
+				end
+			end
+		end
+end

lib/speech.rb

@@ -0,0 +1,11 @@
+class Speech
+    attr_accessor :speakers, :lines
+
+    def initialize(speech_xml)
+        @speakers = speech_xml.css('SPEAKER').map do |speaker|
+            # I'd do map(&:content), but I hate CAPSLOCKED SPEAKER NAMES
+            speaker.content.downcase.split.map(&:capitalize).join(' ')
+        end
+        @lines = speech_xml.css('LINE').map(&:content)
+    end
+end

shakespeare_analyzer_out.txt

@@ -0,0 +1,41 @@
+719	Macbeth
+265	Lady Macbeth
+212	Malcolm
+180	Macduff
+135	Ross
+113	Banquo
+74	Lennox
+70	Duncan
+62	First Witch
+46	Porter
+45	Doctor
+41	Lady Macduff
+39	Hecate
+35	Sergeant
+30	Siward
+30	First Murderer
+27	Third Witch
+27	Second Witch
+24	All
+23	Gentlewoman
+23	Messenger
+21	Angus
+21	Lord
+20	Son
+15	Second Murderer
+12	Menteith
+11	Old Man
+11	Caithness
+10	Donalbain
+8	Third Murderer
+7	Young Siward
+5	Third Apparition
+5	Seyton
+5	Servant
+4	Second Apparition
+3	Lords
+2	First Apparition
+2	Both Murderers
+2	Fleance
+1	Attendant
+1	Soldiers

test/example_play.xml

@@ -0,0 +1,22 @@
+<PLAY>
+    <TITLE>A Tale of Two Rubyists: The Chunkiest Bacon</TITLE>
+    <ACT>
+        <TITLE>ACT I</TITLE>
+        <SCENE>
+            <TITLE>SCENE I. An Introduction</TITLE>
+            <SPEECH>
+                <SPEAKER>FOX 1</SPEAKER>
+                <LINE>Lorem ipsum dolor sit amet...</LINE>
+            </SPEECH>
+            <SPEECH>
+                <SPEAKER>FOX 2</SPEAKER>
+                <LINE>This is boring, I'm out.</LINE>
+            </SPEECH>
+            <SPEECH>
+                <SPEAKER>FOX 1</SPEAKER>
+                <SPEAKER>FOX 2</SPEAKER>
+                <LINE>CHUNKY BACON!</LINE>
+            </SPEECH>
+        </SCENE>
+    </ACT>
+</PLAY>

test/play_test.rb

@@ -0,0 +1,27 @@
+class TestPlay < MiniTest::Unit::TestCase
+    def setup
+        @play = Play.new(example_play)
+    end
+
+    def test_play_initializes_with_file_path
+        play = Play.new(example_play_path)
+        assert play.document, "play was unable to initialize when given a file path as a string"
+    end
+
+    def test_play_initializes_with_file_instance
+        assert @play.document, "play was unable to initialize when given a File object"
+    end
+
+#  This passed, but my internet sucks, so I don't
+#  want to fetch the play every time I run my tests
+    def test_play_initializes_with_url
+        play = Play.new("http://www.ibiblio.org/xml/examples/shakespeare/macbeth.xml")
+        assert play.document, "play was unable to initialize when given a url"
+    end
+
+    def test_speeches_initializes_an_array_of_speech_objects
+        speeches = @play.speeches
+        assert speeches.is_a? Array
+        assert speeches.first.is_a? Speech
+    end
+end

test/shakespeare_analyzer_test.rb

@@ -0,0 +1,16 @@
+require_relative '../lib/shakespeare_analyzer.rb'
+require 'minitest/autorun'
+require_relative 'test_helper'
+require_relative 'play_test'
+require_relative 'speech_test'
+
+class TestShakespeareAnalyzer < MiniTest::Unit::TestCase
+    def setup
+        @analyzer = ShakespeareAnalyzer.new example_play
+    end
+
+    def test_lines_per_speaker_outputs_data_correctly
+        assert_equal "2\tFox 1\n2\tFox 2\n", @analyzer.lines_per_speaker,
+            "lines_per_speaker output didn't match expected"
+    end
+end

test/speech_test.rb

@@ -0,0 +1,23 @@
+require 'nokogiri'
+
+class TestSpeech < MiniTest::Unit::TestCase
+    def setup
+        @speech = Speech.new(speech_xml.first)
+    end
+
+    def test_speech_initializes_with_attributes
+        assert @speech.speakers && @speech.lines
+    end
+
+    def test_speakers_are_properly_capitalized
+        assert_equal "Fox 1", @speech.speakers.first
+    end
+
+    def test_lines_are_properly_added
+        assert_equal 1, @speech.lines.count
+    end
+
+    def speech_xml
+        nk_example_play.css('SPEECH')
+    end
+end

test/test_helper.rb

@@ -0,0 +1,11 @@
+def example_play_path
+    File.join(File.dirname(__FILE__), 'example_play.xml')
+end
+
+def example_play
+    File.open(example_play_path, 'r')
+end
+
+def nk_example_play
+    Nokogiri::XML(example_play)
+end