frequency_analyser 1.1.0 → 1.2.0

data/README.md

@@ -36,15 +36,34 @@ FrequencyAnalyser.analyse('foo', File.new('bar'), StringIO.new('baz'), ['q', 'u'
 #=> { 'a'=>2, 'b'=>2, 'f'=>1, 'o'=>2, 'q'=>1, 'r'=>1, 'u'=>1, 'x'=>1, 'z'=>1 }
 ```
 
+## Probabilities and Percentages
+
+If you'd like to calculate the frequency probabilities instead, you can pass in an optional symbol:
+
+```ruby
+FrequencyAnalyser.analyse('Hello, world!', :probability)
+=> { 'd'=>0.1, 'e'=>0.1, 'h'=>0.1, 'l'=>0.3, 'o'=>0.2, 'r'=>0.1, 'w'=>0.1 }
+```
+
+The same goes for percentages:
+
+```ruby
+FrequencyAnalyser.analyse('Hello, world!', :probability)
+=> { 'd'=>10, 'e'=>10, 'h'=>10, 'l'=>30, 'o'=>20, 'r'=>10, 'w'=>10 }
+```
+
 ## Counting other things
 
-By defeault, Frequency Analyser counts alphabetic characters. You can
-change this by instantiating your own counter:
+By default, Frequency Analyser counts alphabetic characters. You can
+change this by instantiating your own support classes:
 
 ```ruby
-counter = FrequencyAnalyser::Counter.new(%w(1 3 5 !))
-analyser = FrequencyAnalyser.new(counter)
+counter    = FrequencyAnalyser::Counter.new(%w(1 3 5 !))
+aggregator = FrequencyAnalyser::Aggregator.new(counter)
+analyser   = FrequencyAnalyser.new(aggregator)
 
 analyser.analyse('!12321!')
 #=> { '!'=>2, '1'=>2, '3'=>1 }
 ```
+
+Most of the gem is architected in this way, so it should be straightforward to add new modes, for example.

data/lib/frequency_analyser.rb

@@ -1,3 +1,5 @@
 require 'frequency_analyser/base'
 require 'frequency_analyser/counter'
 require 'frequency_analyser/aggregation'
+require 'frequency_analyser/aggregator'
+require 'frequency_analyser/modifier'

data/lib/frequency_analyser/aggregator.rb

@@ -0,0 +1,26 @@
+class FrequencyAnalyser::Aggregator < Struct.new(:counter, :aggregation)
+
+  def initialize(counter = fa::Counter, aggregation = fa::Aggregation.new)
+    super
+  end
+
+  def self.aggregate(*files)
+    new.aggregate(*files)
+  end
+
+  def aggregate(*files)
+    files = [files].flatten
+    files.each do |file|
+      file.each_line do |line|
+        aggregation << counter.count(line)
+      end
+    end
+    aggregation
+  end
+
+  private
+  def fa
+    FrequencyAnalyser
+  end
+
+end

data/lib/frequency_analyser/base.rb

@@ -1,23 +1,25 @@
-class FrequencyAnalyser < Struct.new(:counter, :aggregation)
+class FrequencyAnalyser < Struct.new(:aggregator, :modifier)
 
-  def initialize(counter = Counter, aggregation = Aggregation.new)
+  def initialize(aggregator = Aggregator, modifier = Modifier)
     super
   end
 
-  def self.analyse(*files)
-    new.analyse(*files)
+  def self.analyse(*args)
+    new.analyse(*args)
   end
 
-  def analyse(*files)
-    files = *files.flatten
+  def analyse(*args)
+    files, mode = coerce(args)
+    aggregation = aggregator.aggregate(files)
+    modifier.modify(aggregation, mode)
+  end
 
-    files.each do |file|
-      file.each_line do |line|
-        aggregation << counter.count(line)
-      end
-    end
+  private
+  def coerce(args)
+    files = [args].flatten
+    mode = files.pop if files.last.is_a? Symbol
 
-    aggregation
+    [files, mode]
   end
 
 end

data/lib/frequency_analyser/modifier.rb

@@ -0,0 +1,37 @@
+class FrequencyAnalyser::Modifier
+
+  def self.modify(aggregation, mode)
+    new.modify(aggregation, mode)
+  end
+
+  def modify(aggregation, mode)
+    case mode
+    when :probability
+      probability(aggregation)
+    when :percentage
+      percentage(aggregation)
+    else
+      aggregation
+    end
+  end
+
+  private
+  def probability(aggregation)
+    sum = sum(aggregation)
+    aggregation.inject(aggregation) do |hash, (k, v)|
+      hash.merge!(k => v.to_f / sum)
+    end
+  end
+
+  def percentage(aggregation)
+    sum = sum(aggregation)
+    aggregation.inject(aggregation) do |hash, (k, v)|
+      hash.merge!(k => v.to_f / sum * 100)
+    end
+  end
+
+  def sum(aggregation)
+    aggregation.values.inject(:+)
+  end
+
+end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: frequency_analyser
 version: !ruby/object:Gem::Version 
-  hash: 19
+  hash: 31
   prerelease: 
   segments: 
   - 1
-  - 1
+  - 2
   - 0
-  version: 1.1.0
+  version: 1.2.0
 platform: ruby
 authors: 
 - Christopher Patuzzo
@@ -15,7 +15,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2012-09-15 00:00:00 Z
+date: 2012-09-16 00:00:00 Z
 dependencies: 
 - !ruby/object:Gem::Dependency 
   name: rspec
@@ -42,8 +42,10 @@ extra_rdoc_files: []
 files: 
 - README.md
 - lib/frequency_analyser/aggregation.rb
+- lib/frequency_analyser/aggregator.rb
 - lib/frequency_analyser/base.rb
 - lib/frequency_analyser/counter.rb
+- lib/frequency_analyser/modifier.rb
 - lib/frequency_analyser.rb
 homepage: https://github.com/cpatuzzo/frequency_analyser
 licenses: []