#!/usr/bin/ruby -w
## Frequency Analysis library, planned

##Frequency_Analysis runs basic frequency analysis routines on a provided
##String corpus. In the standard invocation the constructor is
##called with a String as its' only argument, analyses are performed,
##and the results are then availble through the accessors. A useful
##reset_corpus method allow for finer grained operation as well as 
##testing and debugging. There is a driver method planned that will
##allow command line operation. Unit tests are in Test_Frequency_Analysis.
class Frequency_Analysis 
  attr_reader :words_list, :characters_list, :word_frequency_table, 
    :character_frequency_table, :corpus
    
  ## TODO etna constant?
  
  @corpus = '''Lorum ipsum dolor sic amet.'''
  
  ## Print friendly message and exit on unknown or bad input.
  def usage
    p 'FreqAnal: a frequency analysis library (planned)... '
    exit
  end

  ## Create a new Frequency Analyzer and run all analyses on the 
  ## provided String corpus. Checks for String duck type before
  ## doing any analysis. Downcases all input for consistency.
  def initialize(victim_text)
    usage if not victim_text.class == 'String'.class
    @corpus = victim_text.downcase
    build_words_list
    build_characters_list
    build_word_frequency_table
    build_character_frequency_table
  end
  
  ## Sets a new corpus after initialization, intended 
  ## primarily for testing and debugging of individual methods.
  ## Downcases all input for consistency.
  def reset_corpus(new_victim)
    usage if not new_victim.class == 'String'.class    
    @corpus = new_victim.downcase
  end
  
  ## Builds list of words in corpus sorted by their frequency of
  ## appearance descending (most common words first).
  def build_words_list
    @wl = []
    @words_list = @wl
  end
  
  ## Builds list of characters in corpus sorted by their frequency
  ## of appearance descending (most common characters first).
  def build_characters_list
    @cl = []
    @characters_list = @cl
  end
  
  ## Builds hashtable of words and their frequency in corpus. Frequencies
  ## are keyed to words. Eg:
  ##   { 'the' => 40, 'a' => 121, 'i' => 87 ... }
  def build_word_frequency_table
    @wft = []
    @word_frequency_table = @wft
  end

  ## Builds hashtable of characters and their frequency in corpus. 
  ## Frequencies are keyed to characters. Eg: 
  ##   { 'e' => 40, 't' => 121, 'n' => 87 ... }
  def build_character_frequency_table
    @cft = []
    @character_frequency_table = @cft
  end
      
end

## Driver will eventually will check for args and return reports based on
##   input arguments and flags, even...
if __FILE__ == $0
  ## TODO check and parse args
  fa = Frequency_Analysis.new('Lorum ipsum nil carbonadrum est.')
  ## TODO print some reports to stdout
  fa.usage
end